Introduction

Ground-level ozone (O3) is a pervasive global air pollutant that poses significant risks to human health, damages plants, and disrupts ecosystems1,2,3. Ground-level O3 mainly forms through photochemical reactions between nitrogen oxides (NOx) and volatile organic compounds (VOCs). Despite its high reactivity as a potent oxidizing agent, O3 can persist in the atmosphere at concentrations in the parts per billion (ppb) level4. Even when O3 precursors such as VOCs and NOx are well controlled, O3 levels frequently exceed 100 ppb, surpassing the World Health Organization’s guideline value of ~51 ppb for the 8-h daily maximum concentration5,6. Therefore, directly eliminating O3 from the atmosphere is highly desirable. Among the various methods for O3 removal, catalytic decomposition is recognized as an attractive remediation technology, as it efficiently converts O3 into oxygen under mild reaction conditions7,8.

Various O3 decomposition catalysts, particularly manganese oxides, have been extensively studied owing to their high efficiency, environmental friendliness, safety, and cost-effectiveness9,10,11. The introduction of anion vacancies–oxygen vacancies (Vo) in manganese oxide catalysts in a well-known strategy for enhancing ozone decomposition capability7,12,13. However, the development of catalytic O3 decomposition still faces substantial challenges, particularly its long-term effectiveness and humidity resistance. On the one hand, the deactivation of Vo-containing manganese oxide catalysts is inevitable, primarily due to the accumulation of oxygen-containing intermediates (O22− or O2−) (Eq. 5) and H2O molecules on the Vo sites, which are difficult to release even at 100 °C14,15. On the other hand, catalytic O3 decomposition is a dynamic redox process (Eqs. 1–5), requiring both electron donor sites and acceptor sites with the catalyst. Relying solely on the activity of electron -rich centers like Vo makes it difficult to achieve the rapid conversion of O₃ to O₂. Moreover, a redox catalyst with moisture resistance is highly desirable. Thus, developing reactive catalysts with innovative active centers that facilitate rapid electron shuttling while preventing intermediates and water accumulation is crucial for achieving sustained high activity, remarkable stability, and resistance to humidity.

$${{{\rm{Step}}}}\,1:\,{{{{\rm{O}}}}}_{3}+{{{{\rm{e}}}}}^{-}\to {{{{\rm{O}}}}}_{2}+\ast {{{{\rm{O}}}}}^{-}$$
(1)
$${{{{\rm{O}}}}}_{3}+2{{{{\rm{e}}}}}^{-}\to {{{{\rm{O}}}}}_{2}+\ast {{{{\rm{O}}}}}^{2-}$$
(2)
$${{{\rm{Step}}}}\,2:\,{{{{\rm{O}}}}}_{3}+{{{{\rm{e}}}}}^{-}+\ast {{{{\rm{O}}}}}^{-}\to {{{{\rm{O}}}}}_{2}+\ast {{{{\rm{O}}}}}_{{2}^{2-}}$$
(3)
$${{{{\rm{O}}}}}_{3}+\ast {{{{\rm{O}}}}}^{2-}\to {{{{\rm{O}}}}}_{2}+\ast {{{{\rm{O}}}}}_{{2}^{2-}}$$
(4)
$${{{\rm{Step}}}}\,3:\,\ast {{{{\rm{O}}}}}_{{2}^{2-}}\to {{{{\rm{O}}}}}_{2}+2{{{{\rm{e}}}}}^{-}$$
(5)

Constructing Lewis acid-base pairs (LPs) to tailor the electronic configurations of catalysts is recognized as an effective strategy to enhance catalytic activity and accelerate reaction rates16,17. Solid surface LPs comprise Lewis acidic and basic sites that provide electron donor and acceptor sites, enabling the synergetic activation of reactant molecules18,19. Ozone molecules, which contain both electron-deficient and electron-rich oxygen atoms20, can interact with these sites - acting as a Lewis base with acidic sites or as a Lewis acid with basic sites. Therefore, the formation of LPs on manganese oxide is expected to improve its affinity for ozone. Among various solid LP catalysts, those with newly engineered defect-based LP sites on metal oxides have recently shown great promise in enhancing catalytic reactivity. For instance, a defective MnO2 catalyst enriched with LPs has demonstrated high catalytic activity for polyol oxidation, where unsaturated Mnδ+ species associated with oxygen vacancies act as Lewis acids, and the nearby electron-reconstructed oxygen atom serves as Lewis bases21. Additionally, introducing cation defects such as Ni vacancies as Lewis acid sites paired with oxygen vacancies as Lewis base sites, has also been reported to enhance catalytic reactions22. Mn vacancies (VMn) in manganese oxides have been identified as capable of tuning electronic environments, oxygen atom mobility, and oxophilicity23,24,. Addressing these challenges in catalytic ozone decomposition could be achieved by the rational design of VMn-based LPs in manganese oxides, which would help overcome the current limitations in charge-transfer and intermediate oxygen dissociation during ozone decomposition. However, constructing Lewis acid-base pairs to tailor the electronic configurations of catalysts for enhancing the ozone decomposition capability has not been reported, and the intrinsic effects of VMn-based LPs on ozone decomposition remain unclear and warrant further investigation.

In this work, a hydrophobic carbon-containing mesocrystalline manganese oxide (Meso-MnO@C) enriched with VMn-based LP sites has been developed, demonstrating outstanding O3 decomposition performance across a wide range of relative humidity (RH). Different from the widely reported ozone decomposition catalytic system with oxygen vacancy (Vo) as active site, we proposed the active sites of VMn induced LPs in manganese oxide to overcome the bottleneck of traditional Vo-containing ozone decomposition catalysts. Both theoretical calculations and experiment results confirm that the presence of VMn induces charge redistribution within the Meso-MnO, leading to the formation of effective LPs composed of VMn-Mn acidic sites and electron-reconstructed oxygen atoms nearby VMn as basic sites. The synergistic effect of these LPs enhances electron transfer, lowers the energy barriers for O3 capture and facilitates the conversion of intermediate oxygen species to O2, and simultaneously improves the bulk oxygen mobility, thereby accelerating the kinetics of O3 catalytic decomposition. Additionally, the hydrophobic carbon layer prevents water vapor condensation on the Meso-MnO@C surface, thereby avoiding the competitive adsorption of water molecules on active sites. As a result, the Meso-MnO@C achieved complete elimination of 50 ppm O3 for 100 h under high humidity conditions (RH of 65 %), with a reaction rate that surpasses the performance of most previously reported ozone decomposition catalysts.

Results

Preparation, morphology, and composition of Meso-MnO@C

The Meso-MnO mesocrystal were synthesized hydrothermally using colloids- Polyvinyl alcohol (PVA) as a structure-directing agent, and the carbon in Meso-MnO@C is derived from PVA via calcine under an inert atmosphere. Meso-MnO@C with different concentrations of VMn were synthesized by adjusting the amount of PVA additions. MnO2-0G, Meso-MnO@C-0.5 G, Meso-MnO@C-1G, Meso-MnO@C-1.5G, and Meso-MnO@C-2G were synthesized with the different PVA dosages of 0, 0.5, 1.0, 1.5 g and 2.0 g, respectively. Meso-MnO were synthesized under identical experimental conditions of Meso-MnO@C-2G except that was calcined in air atmosphere to release the carbonaceous components. Com-MnO is the commercial MnO. With the PVA additions increased from 0.5 to 2 g (The amount at which the solution is saturated), the morphology of Meso-MnO@C gradually evolves from disorder nanocrystals to regular shaped cubes (Supplementary Fig. 1). The Meso-MnO@C-2G with PVA additions of 2 g exhibit cubic shapes with the size of ca. 2 μm that are assembled by ultra-small crystals (Fig. 1). The HRTEM image and (STEM-EDX) elemental mapping images show the existence of thin carbon layer in Meso-MnO@C-2G and the exposure of (111) plane of MnO (lattice fringes of 0.252 nm) (Fig. 1d, e). The corresponding selected area electron diffraction (SAED) pattern (Fig. 1f) of Meso-MnO@C-2G shows a single crystalline nature, revealing its mesocrystal nature due to the constituting crystallites being arranged along a shared crystallographic register. Mesocrystal consist of oriented aggregated crystalline nanoparticles is known that contain extensive vacancy defects on the surface25.

Fig. 1: Microstructure and morphology studies of Meso-MnO@C-2G.
figure 1

a, b SEM images, c HAADF-STEM, d EDS Mapping images, e HRTEM image, f TEM image and SAED pattern (inset) of Meso-MnO@C-2G.

Full size image

The XRD patterns of prepared materials are shown in Supplementary Fig. 2. Besides the sample synthesized under the absence of PVA demonstrates the characteristic peaks of α-MnO2 (JCPDS NO. 44-0141), all other samples exhibit the XRD peaks well index to MnO (space group Fm-3m (225), JCPDS No. 07-0230). A typical IV isotherm shown in the nitrogen adsorption and desorption isotherms (Supplementary Fig. 3a) indicates the mesoporous characteristic of Meso-MnO@C-2G. The BET surface area of Meso-MnO@C-2G is 58.8 m2/g, which is higher than that of MnO2 (30.2 m2/g) and Com-MnO (1.5 m2/g) (Supplementary Table 1). The Raman spectrum for Meso-MnO@C-2G shows two weak peaks centered at 1350 and 1610 cm−1 (Supplementary Fig. 3b) assign to the D-bond and G-bond due to the carbon species in Meso-MnO@C-2G, while no obvious Raman signal peaks can be observed in the range of 1200–1800 cm−1 in Meso-MnO, Com-MnO, and MnO2-0G. Moreover, the results of TG-DTA and ICP reveal the carbon content in Meso-MnO@C-2G was ca. 22% (Supplementary Fig. 4).

X-ray absorption fine structure (XAFS) was carried out to gain insight into the defective structure at atomic scale. As illustrated in Supplementary Fig. 5a and Fig. 2a, the k3 -weighted Fourier-transforms of Mn K-edge EXAFS spectra of Meso-MnO@C-2G, Com-MnO and MnO standard show two distinct Fourier transforms (FT) peaks that are typical features of MnO assigned to the closest Mn-O and Mn-Mn, respectively. The coordination number (N) of Mn-Mn (N = 2.0) in Meso-MnO@C-2G significantly reduced compared with that of Com-MnO and MnO standard (N = 12.0 and 12.9) (Supplementary Table 2 and Supplementary Fig. 5b). The survived Mn vacancies (VMn) can reduce the Mn-Mn coordination number. Moreover, the coordination number of Mn-O in Meso-MnO@C-2G slightly increased compared with that of Com-MnO and MnO standard, suggesting the existence of point defects-interstitial oxygen in Meso-MnO@C-2G, the presence of interstitial oxygen should render the excited bulk oxygen mobility26.

Fig. 2: Fine-structure characterizations and the electron distribution calculation of samples.
figure 2

a k3 -weighted Fourier-transforms of Mn K-edge EXAFS spectra of Meso-MnO@C-2G, Com-MnO and MnO stander. b The XPS spectra of Mn 2p of Meso-MnO@C-2G and Com-MnO. c The CO2-TPD and d NH3-TPD profiles of samples. e Electron localization function (ELF) map of MnO with VMn. f The schematic illustration of Lewis pair in Meso-MnO@C-2G.

Full size image

Moreover, defects usually lead to charge imbalance in the MnO crystal structure. The formation of VMn could raise the average valence of Mn in MnO27,28. We identified the valence states of the elements in samples by XPS. As shown in Fig. 2b, the Mn 2p1/2 and 2p3/2 peaks of XPS are located at 652.8 and 641.2 eV, indicating the major proportion of Mn(II) in Meso-MnO@C and Com-MnO. The Mn 2p3/2 peak can be deconvoluted into three peaks with the binding energy at 640.7, 641.7, and 642.7 eV, corresponding to Mn2+, Mn3+ and Mn4+ respectively9. It can be observed that the sum ratio of Mn4++Mn3+ in Meso-MnO@C-2G and Com-MnO catalysts are calculated to be 33.38%, and 18.22% (Supplementary Table 1), respectively. The ratio of Mn4++Mn3+/Mn2+ increased with the increased additions of PVA in Meso-MnO@C (Supplementary Fig. 6a), revealing the VMn concentration in Meso-MnO@C increased with the increased PVA additions during synthesis. The average oxidation state (AOS) value of surface Mn atoms in the Meso-MnO@C-2G and Com-MnO are calculated as 3.02, and 2.31 (Supplementary Fig. 7) (estimated by AOS = 8.956–1.126 ΔEs)29, respectively. The higher AOS of Mn atoms indicates higher-valence of Mn atoms on the surface of Meso-MnO@C-2G30. The AOS of Meso-MnO@C also increased with the amount of PVA additions (Supplementary Fig. 6b), indicating the VMn concentration in Meso-MnO@C can be adjusted by the PVA additions during synthesis. As shown in the XPS spectrum of O 1 s, the ratio of lattice oxygen (OLatt) in Meso-MnO@C-2G is 63.97%, which is lower than that of Com-MnO (73.05%), being attributed to the greater adsorption of surface-adsorbed oxygen (Oads) by the unsaturated Mn-O geometry in Meso-MnO@C-2G (Supplementary Fig. 9). These results indicate more VMn in Meso-MnO@C-2G than those in Com-MnO, and the VMn concentration in Meso-MnO@C can be adjusted by the amount of PVA additions.

To probe the Lewis pairs on Meso-MnO@C-2G, the Lewis acidity and basicity was investigated by the temperature-programmed NH3 desorption (NH3-TPD) and CO2 desorption (CO2-TPD) (Figs. 2c, d). The Com-MnO and MnO2-0G have negligible acidic sites and basic sites for NH3 and CO2 adsorption, while Meso-MnO@C-2G shows strong acidity and basicity with two distinct peaks at 70–130 °C and above 400 °C in both NH3-TPD and CO2-TPD, revealing the existence of the weak and strong Lewis acid sites, and the weak and strong Lewis base sites in Meso-MnO@C-2G. Notably, Meso-MnO@C-2G displays higher NH3 and CO2 desorption temperature and stronger signal intensity than that of Meso-MnO@C-1G, in which the content of VMn in Meso-MnO@C-2G is higher than that of Meso-MnO@C-1G. This indicates the presence of VMn enhances the strength of Lewis acidity and Lewis basicity. To verify the existence of LPs in Meso-MnO@C-2G, the electron localization function (ELF) analysis was manipulated for VMn-MnO to get the electron distribution of microregions around VMn. Figure 2e shows the Electron Localization Function distributions for Mn53O54(200) slab. Notably, the presence of VMn in MnO (Mn53O54) leads to the charge redistribution with the electron delocalization of Mn atoms and obviously localized electrons around O atoms, which the electron deficiency sites-VMn -Mn is supposed to function as LA and the sites nearby O- VMn plays the role of LB. Bader charge calculations reveal the constructed VMn resulted in the redistribution of electrons with a decrease in the number of Bader charge and an increase in Mn valence (Supplementary Fig. 10), where the results are in accordance with the XPS spectra. Meanwhile, the Meso-MnO@C-2G exhibit reduced electrical resistance than that of Meso-MnO, MnO2-0G and Com-MnO, and even lower than that of Meso-MnO@C-1G (Supplementary Fig. 11a). Moreover, the electrical resistance of Meso-MnO is also lower than MnO2-0G and Com-MnO. Therefore, the lower electrical resistance of Meso-MnO@C-2G can be attributed to the presence of VMn and existence of carbon layer. The ozone decomposition process is essentially based on dynamic electron shuttle (Eqs. 1–5)31,32. The electronic structures of LPs sites in Meso-MnO@C-2G could act as both electron donors and acceptors that facilitate the rapid electron shuttle (Fig. 2f).

Moreover, the redox ability of samples was investigated by H2-TPR (Supplementary Fig. 11b). The peak intensity of Meso-MnO@C-2G and Meso-MnO@C-1G are stronger than that of Com-MnO, and the reduction temperature of Meso-MnO@C-2G is significantly lower than that of Meso-MnO@C-1G by reduction of Mn4+ and Mn3+ to Mn2+, implying the higher redox ability of Meso-MnO@C-2G. The oxygen mobility of samples was analyzed by O2-TPD. As shown in Supplementary Fig. 11c, the Meso-MnO@C-2G demonstrates a large desorption peak area at 380–500 °C compared to Meso-MnO@C-1G and Com-MnO, revealing the high migration of lattice oxygen, which is beneficial to the redox reactions26.

Ozone decomposition performance

The O3 decomposition experiment was carried out in a continuous fixed membrane reactor9. As expected, the Meso-MnO@C-2G catalyst exhibits the high catalytic activity with an O3 conversion of 100% after a 6 h reaction under RH = 30% at a weight hourly space velocity (WHSV) of 1500 L⋅g−1 h−1, which the removal efficiency is higher than that of the Meso-MnO (93.4%), Meso-MnO+C (91.2%), Com-MnO (22.4%), Com-MnO/C (23.3%), and MnO2-0G (67.8%) after 2 h reaction (Fig. 3a and Supplementary Fig. 12a). Moreover, even the reaction time prolonged to 50 h, the O3 conversion over Meso-MnO@C-2G still maintained at 99.5% (Supplementary Fig. 12b). Notability, Meso-MnO@C samples exhibited an enhanced ozone decomposition performance with the increased PVA additions (Supplementary Fig. 13). As the VMn concentration in Meso-MnO@C-2G directly related with the PVA additions, suggesting that the VMn should played an important role for ozone decomposition. As shown in Supplementary Fig. 14, the intensity of the XRD diffraction peaks of used Meso-MnO@C is similar with that of the fresh material and the valence state of Mn in the XPS spectra of Meso-MnO@C-2G slightly changed (Supplementary Fig. 15 and Supplementary Table 1), indicating the high stability of Meso-MnO@C. However, the ratio of high valanced Mn3++Mn4+ in the Com-MnO and MnO2-0G significantly increased (Supplementary Fig. 16), the relative content of OLatt/Oabs dramatically decreased. The electrons are easier to be transferred from Mn2+/(Mn3++Mn4+) to reduce the O3 to O2−/O22− (Eqs. 1–4), but hard to be compensated from oxidizing O2−/O22− to O2 over Com-MnO and MnO2-0G due to the insufficient electron acceptor sites and unsatisfactory electron transport capacity (Eq. 5). The LPs in Meso-MnO@C-2G could maintain the rapid dynamic electron shuttle for charge compensation and the high migration of lattice oxygen also beneficial the charge compensation that the stability can be realized.

Fig. 3: Catalytic performance of various catalysts.
figure 3

a O3 conversion efficiency on Meso-MnO@C-2G, MnO2, Meso-MnO, Com-MnO +C and Com-MnO at 30% RH. b Influences of molecular Lewis-base or Lewis-acid on the catalytic activity of ozone decomposition. c The comparison of reaction rate of our work with other reported works. d O3 conversion efficiency on Meso-MnO@C-2G, MnO2-0G, Meso-MnO, Meso-MnO+C and Com-MnO at 65% RH. e O3 conversion on Meso-MnO@C-2G and Meso-MnO at 65% RH over 100 h (inset is the water contact angel of Meso-MnO@C-2G). f O3 conversion on Meso-MnO@C-2G under alternate RH (30% RH and 90% RH).

Full size image

To further confirm the necessity of the existence of Lewis acidic and basic sites of Meso-MnO@C-2G toward ozone decomposition, the catalytic reaction was performed in the presence of other Lewis acid or base by addition of trace amount of Lewis base pyridine and Lewis acid pyrrole to quench the acidic and basic sites, respectively. As shown in Fig. 3b, the blockage of surface Lewis acidic leads to activity decreased of 30% and quenching the basic sites leads to the activity decrease of 42%, where the sum of them is 72%, which the results is similar to the difference in catalytic activity between the LPs rich Meso-MnO@C-2G and the LPs deficient Com-MnO/C (23.3%) (Fig. 3a), revealing that the coexistence of LA and LB sites on the surface of Meso-MnO@C-2G are essential to achieve high activity toward ozone decomposition. As depicted in Supplementary Fig. 17, no changes can be observed after pyridine adsorption (blockage of the surface LA) on MnO2-0G, while the activity decreased from 67.8% to 40.2% after pyrrole adsorption (quenching the LB). The effect of quenching experiment on the performance of MnO2-0G was significantly weaker than that of Meso-MnO@C, revealing the presence of VMn induced the formation of LA and LB sites. It is thus speculated that the coupling between the LA and LB sites enables Meso-MnO@C-2G to continuously decompose ozone with refreshed active sites. Moreover, the superiority of this work over previously reported studies4,10,13,33,34,35,36,37,38 was demonstrated in Fig. 3c and Supplementary Table 3.

Water-induced resistance is one of the challenges for the development of catalytic O3 decomposition30. The presence of water vapor tends to occupy the active sites, thus reducing the O3 decomposition activity of catalysts. Under the condition of 65% RH (Fig. 3d), the O3 conversion efficiency of Meso-MnO@C-2G is stabled at 100% after 2 h reaction, while the efficiency dramatically drop to 42.8%, 52.2%, 57.4%, 62.7%, and 6.27 % for MnO2-0G, Meso-MnO, Meso-MnO+C, and Com-MnO, respectively. Even the reaction time prolonged to 100 h under RH of 65%, the O3 conversion over Meso-MnO@C -2G can still be maintained at 99.5% (Fig. 3e). Notably, when the RH shifts from 90% to 30%, the O3 decomposition efficiency can be rapidly recovered to 100% over the Meso-MnO@C-2G (Fig. 3f), indicating the good environmental adaptability. The intensity of the XRD diffraction peaks of Meso-MnO@C used under humid environment is similar with that of the fresh material and the valence state of Mn in the XPS spectra of Meso-MnO@C-2G slightly changed (Supplementary Fig. 18 and Supplementary Fig. 19). Moreover, the superiority of this work under humid conditions over previously reported studies was demonstrated in Supplementary Table 4.

The affinity of the Meso-MnO@C-2G for water was identified by water contact angle measurement. The Meso-MnO@C-2G has a 135° water contact angle of a water droplet (Fig. 3e), while the Meso-MnO that absence of carbon demonstrates a water contact angle of 37˚ (Supplementary Fig. 20), revealing the hydrophobic nature of Meso-MnO@C-2G originates from the carbon species. To understand the H2O accumulation behavior on the catalyst surface, H2O-TPD is studied. The water desorption peaks of Com-MnO appears at 154 °C and 410 °C, corresponding to the physically adsorbed water and chemically adsorbed water (Supplementary Fig. 11d), respectively. However, the water desorption only appears at 154 °C in Meso-MnO@C-2G, revealing only the presence of physically adsorbed water on Meso-MnO@C-2G. The hydrophobic carbon species promote the water resistance of Meso-MnO@C-2G by hindering the chemisorption of water vapor on the catalyst surface, which decreases the competitive adsorption between water vapor and O3.

To evaluate the application potential of the Meso-MnO@C-2G catalyst for air purification, O3 and volatile organic compounds (CH3SH) at an initial concentration of 50 and 40 ppm, respectively, were mixed under a WHSV of 120,000 mL⋅g−1 h−1 for simultaneous catalytic destruction test. As shown in the Fig. 4a, the Meso-MnO@C-2G exhibits not only 100% removal of O3, but also a highly stable removal efficiency of 90% for CH3SH, significantly exceeding that of bare O3 (29.7%), indicating the high activity of Meso-MnO@C-2G to eliminate both VOCs and O3 at room temperature. Moreover, to assess the practical application of Meso-MnO@C-2G toward atmospheric O3, a 0.216 m3 indoor chamber was assembled to evaluate the catalytic performance (Fig. 4b and Supplementary Fig. 21). Meso-MnO@C-2G exhibits a high O3 removal ratio of 100% under initial O3 concentrations of about 20 ppm and reach the safe O3 concentration (100 μg m−3, 50.9 ppb) recommended by the Global Air Quality Guidelines (AQGs) of the WHO within 80 min (Fig. 4c). However, O3 were decomposed slowly over Com-MnO, even after 300 min, the O3 concentration still maintain at 4.11 ppm. The above experimental results show that the Meso-MnO@C-2G catalyst realized a rapid reaction with strong redox activity and excellent water resistance, representing great application potential in O3 elimination and complex air pollution purification.

Fig. 4: The evaluation of the application potential of the Meso-MnO@C-2G for air purification.
figure 4

a The performance of synergistic O3 decomposition and VOCs removal over Meso-MnO@C-2G. b The schematic diagram of reaction chamber. c Ozone decomposition over catalyst in chamber, inset is the photograph of catalyst deposit on glass films.

Full size image

Reaction mechanism

The above experiments strongly suggest that the coexistence of Lewis acidic and basic sites plays extremely important roles in ozone decomposition. To further understand the specific role of LPs on ozone decomposition, we performed systematic in situ DRIFTs, in situ Raman and DFT calculation. Manganese oxides -based catalysts always bear the accumulation of the peroxide intermediates (O2−/O22−) during ozonolysis due to the frustrated electron transfer from peroxide to catalyst (Eq. 5), the accumulated O2−/O22− may be difficult to dissociate from active sites, resulting in the stop of subsequent reactions. The reactants evolution over Meso-MnO@C-2G were analyzed by in situ DRIFTs and in situ Raman spectroscopy. Figure 5a displayed the in situ DRIFTs spectroscopy of Meso-MnO@C-2G and Com-MnO. The peak of adsorbed O3 (935 cm−1 and 881 cm−1) can be observed on Meso-MnO@C-2G and Com-MnO. The peak at 1380 cm−1 associated to the reaction intermediate O2−/O22− accumulation on catalyst surface. Notably, the peak intensity of O2−/O22− on Meso-MnO@C-2G are significantly weaker than that on Com-MnO, indicating that the inhibited accumulation of O2−/O22− on Meso-MnO@C-2G. The spectra of in situ Raman are shown in Supplementary Fig. 22a, the weak peak signal of O2−/O22− at 874 cm−1 on Meso-MnO@C-2G was observed. Furthermore, the dissociation of surface oxygen species after reaction over different samples was tested by FTIR of samples after long-term reactions (Supplementary Fig. 22b). Even though the after reaction for 50 h, the signal of O2−/O22− in Meso-MnO@C-2G is much weaker than that of Meso-MnO and MnO2 after reaction for 2 h, suggesting that the O2−/O22− accumulation is impeded.

Fig. 5: Mechanistic investigation of ozone decomposition on Meso-MnO@C-2G.
figure 5

a In situ DRIFTs spectra of Meso-MnO@C-2G and Com-MnO with 60 min continuous O3 flow passing at room temperature (25 °C) under RH of 0%. b Structure diagram and electron transfer during O3 adsorption over VMn-MnO and MnO, respectively. c Reaction pathways of O3 decomposition and the related electron donation and abstraction process on the Meso-MnO@C-2G surface.

Full size image

To in-depth explore the role of LP sites in the reaction process, DFT calculations of electron donation and abstraction process and the reaction barrier of O3 decomposition on VMn-MnO and MnO models were investigated based on the experimental results. The equations for calculation in each step are shown in Supplementary Text 6. The Bader charge calculations show that ozone molecule is anchored on the LA sites (Mn nearby VMn) in VMn containing MnO and the electron transfer from nearby LB sites to ozone molecule (Fig. 5b and Fig. 5c). Then, the O-O bond cleavage of O3 occurs to generate *O on surface with electron donating from VMn-MnO to O-O bond (0.26e per VMn in MnO) and eliminating a O2 molecule. The amount of electron donating from VMn-MnO to O3 is higher than that of MnO with absence of VMn (0.08e) (Supplementary Fig. 24). Afterwards, anther O3 molecule adsorbs on catalyst and reacted with *O to produce *O2 intermediates with electron donating from VMn-MnO to *O (0.18e per VMn in MnO). The electron donation of VMn-MnO is higher than that of MnO with absence of VMn(0.02e). Subsequently, *O2 deliver electron to catalyst and desorbed from catalyst surface with eliminating a O2 molecule. The high energy barrier of desorption *O2 is regarded as the rate-limiting step in O3 decomposition. The amount of electron abstracting from *O2 to VMn-MnO (0.14e per *O2) is higher than that of MnO with absence of VMn (0.09e), and the desorption energy of *O2 on VMn-MnO and MnO without VMn are 0.93 and 3.93 eV, respectively, indicating the presence of VMn in MnO could dramatically facilitate the O2* dissociation. The electron transfer calculations reveal that the presence of VMn-based LP on Meso-MnO@C-2G could deliver a higher electron-donating and abstracting ability during O3 decomposition. Furthermore, the electron mobility in the catalysts was evaluated by CV curves. As shown in Supplementary Fig. 25, The current of Meso-MnO@C-2G is much higher than that of Com-MnO, proving the stronger electron transfer ability of Meso-MnO@C-2G. The results of in situ DRIFT and Raman are consistent with the DFT calculation results, confirmed the cooperation between VMn-based Lewis pairs for enhancing the stability and activity of Meso-MnO@C-2G.

Based on the above discussions and reported O3 conversion mechanism39, here, we proposed the O3 decomposition mechanism on Meso-MnO@C-2G (Fig. 6). O3 molecules are captured by the Lewis acid sites (VMn-Mn) in Meso-MnO@C-2G, electrons are transferred from electron-rich LB sites in Meso-MnO@C-2G (the space between O atoms and VMn) to O3 molecules, resulting in the formation of O−/O2− and the release of O2. Then, O−/O2− react with another O3 molecule to form O2–/O22− with consuming the electrons donating by LB sites in Meso-MnO@C-2G. Finally, the O2−/O22− transfers electrons to the charge deficiency sites (LA) and then dissociation to O2. Meanwhile, the electron transition from LA to LB. The Meso-MnO@C-2G shows rapid electron shuttle for accelerating the ozone decomposition, as well as the hydrophobic carbon layer inhibits the H2O accumulation on the catalyst surface, hence the excellent O3 decomposition performance was realized.

Fig. 6: The schematic illustration of O3 decomposition over Meso-MnO@C-2G.
figure 6

Proposed reaction pathway for the O3 decomposition catalyzed at the LA-LB sites on Meso-MnO@C-2G.

Full size image

Discussion

In summary, we have developed a carbon-containing mesocrystalline MnO catalyst, Meso-MnO@C-2G, which features abundant VMn-based Lewis pairs and delivers outstanding activity for O3 catalytic decomposition with impressive stability across a wide humidity range. The Meso-MnO@C-2G catalyst has demonstrated excellent performance in O3 elimination, as shown in tests within a 0.216 m3 chamber and under complex air pollution conditions. The high density of VMn is crucial for creating surface Lewis acidic (VMn-Mn) and basic sites (electron-reconstructed O atoms near VMn), which enhance electron transfer, improve ozone affinity, and lower the energy barrier for converting intermediate oxygen to O2. Additionally, the hydrophobic carbon layer protects the catalyst’s active sites from deactivation due to water adsorption and degradation during the O3 decomposition process. This work offers a novel approach for developing catalysts with enhanced activity, stability and water resistance for efficient air purification.

Methods

The material, instruments involved in this work are discussed in the Supplementary Information (Supplementary Text 1 and 2). The synthesis of catalyst and computational details are descripted in the Supplementary Information (Supplementary Text 3 and 4).

Catalytic activity tests

The activity of the catalysts for O3 decomposition were measured in a continuous flow membrane reactor at room temperature. The simultaneous catalytic removal of CH3SH and O3 was carried out in a continuous-flow column reactor (150 mL) that equipped with a CH3SH detector (DM-400IS, Detcon, USA), and 40 mg catalyst was added into the column reactor. The details are provided in Supporting Information. The details are descripted in the Supplementary Information (Supplementary Text 5).