Structure of the catalyst interfaceThe Mn2O3/SMO catalyst was prepared by the nonstoichiometric exsolution method. During a high-temperature calcination (800 °C), a surplus of MnOx was accumulated on the mullite (SMO) surface and transitioned into a stable Mn2O3 phase. Pristine SMO, Mn2O3, and the co-impregnated sample of Mn2O3 on SMO (Mn2O3-SMO) were synthesized and calcined at 800 °C16,17. The X-ray diffraction (XRD) results depicted confirm that the Mn2O3/SMO possesses a mullite structure whilst further containing an exsolved-Mn2O3 phase at 32.9° (Fig. 1a). The main diffraction peak (211) of Mn2O3/SMO exhibits an obvious shift compared to that of SMO and Mn2O3-SMO (Supplementary Fig. 1), suggesting the lattice contraction occurs on the mullite support of Mn2O3/SMO. The refined XRD results of SMO and Mn2O3/SMO (Supplementary Fig. 2 and Supplementary Table 1) also demonstrate that the lattice constant and unit cell volume parameter of SMO change from a = 7.44 Å (V = 363.53 Å) to a = 7.43 Å of Mn2O3/SMO (V = 362.6 Å). Such a lattice contraction is mainly attributed to the produced oxygen vacancies and interface interactions.Fig. 1: Surface and interface electronic structure characterizations.a The X-ray diffraction patterns; the high-angle annular dark-field scanning transmission electron microscopies (HAADF-STEM) of (b) SmMn2O5 (SMO) and (c) Mn2O3/SmMn2O5 (Mn2O3/SMO; A: SMO surface; B: interface; C: Mn2O3); (d) the normalized Mn L-edges electron energy loss spectroscopies (EELS) of SMO, Mn2O3-SmMn2O5 (Mn2O3-SMO), and Mn2O3/SMO at the position of A, B, and C; (e) Schematic illustration of the interface interactions on Mn2O3/SMO; (f) the R-space Fourier-transformed FT (k3χ(k)) of Mn K-edge (extended x-ray absorption fine structure) EXAFS spectra, (g) Sm L3-edge (X-ray absorption near edge structure) XANES spectra, (h) Sm 3d XPS, and i the normalized O K-edge XANES spectra of SMO, Mn2O3-SMO, and Mn2O3/SMO.The interface atomic structure was imaged utilizing high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM), revealing that the spacings of the lattice fringe on the SMO sample are 3.46 and 2.93 Å, assignable to the (102) and (200) facets of mullite, respectively (Fig. 1b and Supplementary Figs. 3–4). It is interesting to note that the spacing on the Mn2O3/SMO is 3.63 Å, assignable to the (020) facet of mullite (Fig. 1b, c), coherently aligning with the peak shift in the XRD results. The spacings of the lattice fringe on the exsolved oxide region can be allocated to the (2̅00) and (222) facets of the Mn2O3 phase. The electron energy loss spectroscopy (EELS) reveals intricate details about the local electronic structure of catalysts. Mn L-edge EELS spectra (Fig. 1d) show that the Mn2O3/SMO contains more Mn3+ cations and lower shifts for peak positions than does Mn2O3-SMO at the interface and mullite regions, but a single SMO surface exists with the same Mn valence states and peak positions with mullite regions of Mn2O3-SMO. These results directly imply that the exsolved Mn2O3/SMO possesses strong interface interactions while there is a negligible interfacial interaction on Mn2O3-SMO. The Mn L-edge EELS results highlight that the conventional impregnation method can only present a slight impact on the electronic structure between Mn2O3 and SMO. However, the interface between exsolved-Mn2O3 and SMO of Mn2O3/SMO presents an important electron enrichment on Mn atoms, which is a prerequisite for generating a robust electron transfer at the interface. Furthermore, O K-edge EELS spectra reveal that the Mn-O bond covalency at the SMO surface of exsolved Mn2O3/SMO is stronger than mullite and the deposited Mn2O3-SMO due to the higher energy position for the adsorption peak of the Mn-O bond (Supplementary Fig. 5). While both SMO and Mn2O3-SMO show the same covalency of Mn-O bond because of the same peak position. These results demonstrate the interface interactions and the activation of lattice oxygen in exsolved Mn2O3/SMO. A schematic representation of the electron transfer at the interface (Fig. 1e) reveals that the electron transfer from exsolved Mn2O3 to SMO activates the Olatt sites of Mn2O3/SMO.The H2 temperature programmed reduction (H2-TPR) profiles (Supplementary Fig. 6) infer that the Mn4+ of Mn2O3/SMO exhibits a slightly lower reduction peak at 410 °C than the other samples, suggesting its high reducibility. The O2 temperature programmed desorption (O2-TPD) profiles (Supplementary Fig. 7) manifest that Mn2O3/SMO possesses more adsorbed oxygen species below 180 °C and the considerable amount of active surface Olatt occurring at 180–300 °C. The NO temperature programmed desorption (NO-TPD) profiles (Supplementary Fig. 8) display a lower temperature for NOx desorption at 150 °C on Mn2O3/SMO. The above results confirm that the interface between exsolved-Mn2O3 and the mullite support of Mn2O3/SMO can improve the reducibility of surface Mn and the number of Olatt, enhancing the NOx desorption ability at low temperatures.Mn K-edge X-ray absorption near edge structure (XANES) and extended X-ray absorption fine (EXAFS) spectroscopies reveal that the Mn2O3/SMO possesses the same average valence state of bulk Mn atoms with Mn2O3-SMO due to the similar adsorbed edge position and exhibits a shorter Mn–O bond length compared to Mn2O3-SMO and SMO (Fig. 1f and Supplementary Figs. 9, 10). This result demonstrates the important effect of the exsolved interface on tuning the Mn coordination environment and surface electronic state because the average valence of the bulk Mn atom has little relevance with the interface. Furthermore, the Sm L3-edge XANES spectra (Fig. 1g) demonstrate that Mn2O3/SMO possesses a weaker intensity of white line peak than Mn2O3-SMO and SMO, resulting in an increased number of electrons at the Sm sites. The consequential electron transfer and ensuing modifications are further confirmed by a shift in the Sm 3d spectra of X-ray photoelectron spectroscopy (XPS) (Fig. 1h). With respect to the Mn L-edge XAS spectra (Supplementary Fig. 11), Mn2O3/SMO presents a lower intensity of white line peak and a diminished peak intensity ratio of L3/L2 compared to SMO and Mn2O3-SMO, suggesting that the Mn2O3/SMO hosts surplus electrons in the Mn 3d orbitals. An exploration of O K-edge XAS spectra on Mn2O3/SMO indicates notable variations in Fig. 1i and Supplementary Fig. 12) a stronger white line peak; 2) more hybrid orbitals of Mn 3d and O 2p; 3) a higher peak intensity ratio of α/ γ obtained on Mn2O3/SMO, where the peaks α and γ represent the Mn 3d eg/t2g and eg orbitals, respectively36,37. These Mn L-edge and O K-edge soft-XAS spectra analyses agree well with the Mn L-edge and O K-edge EELS spectra in surface and interface electronic structure of these samples. The EPR spectra further manifest that the exsolved Mn2O3/SMO possesses the most oxygen vacancies among these samples, which stems from the strong interface interactions (Supplementary Fig. 13).These disparities indicate that Mn2O3/SMO displays a higher covalency of the Mn-O bond because of the strong orbital hybridizations of Mn 3d and O 2p. These observations are complemented by a comprehensive analysis to gain insights into the electronic structures at the interface between exsolved-M2O3 and SMO. Fundamentally, this interface facilitates effective electron transfer from the exsolved-Mn2O3 to SMO, enhancing the covalency of the Mn-O bond at the Olatt adjacent to the Mn3+ cations (Fig. 1e)16,17,38.NO oxidation performanceThe NO conversions of Pt/Al2O3, SMO, Mn2O3/SMO, Mn2O3, and Mn2O3-SMO catalysts are conducted to analyze the catalytic performance (Fig. 2a). Distinctly paramount among these, the Mn2O3/SMO exhibits superior NO conversion (93%) at 270 °C. The T80 (the onsite temperature when NO conversion is higher than 80%) of Mn2O3/SMO (235 °C) is also 50 and 100 °C lower than that of the Pt/Al2O3 and Mn2O3-SMO samples. The activity profile of Mn2O3-SMO is quite similar to that of SMO. This finding underscores the critical roles of the exsolved interface, demonstrating that the exsolved Mn2O3 (Mn2O3/SMO) rather than the impregnated Mn2O3 (Mn2O3-SMO) improves the NO oxidation activity of SMO. To better understand the impact of exsolved Mn2O3 on NO oxidation, we have synthesized a variety of exsolved Mn2O3-supported samples with diverse Mn2O3 content. Incrementing the Mn2O3 content does not contribute much to the activity of NO oxidation (Supplementary Fig. 14). Combined with the results of structure characterizations, the active sites in Mn2O3/SMO are mainly at the surface/interface of the SMO and the key role of the Mn2O3 species is to tune the surface/interface electronic structure of the SMO. It is also discovered that Mn2O3/SMO maintains the greatest activity across the component range because of the higher interface oxygen defect concentrations. This provides a reasonable deduction that the main active sites of Mn2O3/SMO are located primarily at the interface instead of surface Mn2O3.Fig. 2: Oxidation activity of diesel exhaust gas.a NO oxidation activity of Pt/Al2O3, SmMn2O5 (SMO), Mn2O3, Mn2O3/SmMn2O5 (Mn2O3/SMO), and Mn2O3-SmMn2O5 (Mn2O3-SMO). b Plot of maximum conversion for NO oxidation vs. temperature for Mn2O3/SMO, mullite-based catalysts, perovskite-based catalysts, and noble metal catalysts. c Diesel oxidation catalyst (DOC) performance of SMO, Mn2O3/SMO, and Pt/Al2O3. d DOC stability of Mn2O3/SMO for long-term oxidation reaction at 300 °C. Reaction conditions: 500 ppm NO, 1% CO, 2000 ppm C3H6, 10% O2, 5 ± 1% H2O, and balance N2.In order to obtain a comparative measurement of activity, a graph representing the correlation between maximum NO conversion and temperatures for previously reported catalysts (with/without noble metals) is summarized (Fig. 2b). Mn2O3/SMO still manifests a supreme performance as it functions at a substantially lower reaction temperature and demonstrates a higher maximum NO conversion compared to both noble metal39,40,41,42,43,44,45 and transition metal-oxide (perovskite3,34,46 and mullite2,17,27,30,47) counterparts. In addition, we also tested the DOC performance in terms of CO, C3H6, and NO subjected to laboratory-simulated diesel exhaust conditions (Fig. 2c). Mn2O3/SMO still exhibits superior activities in the oxidations of CO, C3H6, and NO as compared to Pt/Al2O3 and Mn2O3-SMO17. Fig. 2d shows the higher durability and water vapor resistance (Supplementary Fig. 15) of Mn2O3/SMO in 50 h, as well as excellent thermal stability at 800 °C aging for 10 h (Supplementary Fig. 16). We have also coated the slurry of Mn2O3/SMO on a honeycomb Cordierite-support and tested its DOC performance (Supplementary Fig. 17). The DOC performance of the honeycomb samples can meet the DOC standards of China Environmental Protection Industry Association (T/CAEPI 12.1-2017). This reinforces the superior robustness and functionality of Mn2O3/SMO, as well as the simple preparation method, making it one a promising candidate as a DOC catalyst.To further align with realistic operational conditions in diesel vehicles, a CO2 concentration of around 8% and H2O at around 8% as well as thermal aging resistance are necessary to pretreat the catalyst before evaluating the catalytic performance of SMO and Mn2O3/SMO (Supplementary Fig. 18)48. Mn2O3/SMO displays better catalytic activity for NO oxidation (71.8% conversion at 325 °C), CO oxidation (100% conversion at 300 °C), C3H6 oxidation (100% conversion at 300 °C) than SMO (100% CO and 99.4% C3H6 conversions at 300 °C and 58.1% NO conversion at 350 °C). The tested data demonstrate the application prospect of the exsolved Mn2O3/SMO catalyst for exhaust oxidation treatment. In addition, we also measured the DOC durability of the exsolved Mn2O3/SMO catalyst for 64 h under the conditions of 8% CO2 and 8% H2O (Supplementary Fig. 19). 8% water vapor can decrease the catalytic oxidation property (56.9% NO conversion), which stems from the competitive adsorption of water vapor at the active sites. When water vapor is removed from reaction system, the catalytic activity of Mn2O3/SMO becomes better (73% NO conversion) and keeps the long-term durability for 15 h. Then, the 8% water vapor is pumped into the reactor again, the catalytic performance is the same as for the last water vapor condition (56.9% NO conversion). After removing the water vapor, the catalytic activity rapidly restores to the original level before secondly adding water vapor (73% NO conversion) and the system remains unchanged for the next 10 h. The initial tests of 20 h under NO, CO, C3H6, O2, H2O, CO2, and N2 belong to the pre-treatment process of catalysts. According to the standard practices of DOCs, these catalysts are aged at 750 °C for 50 h with 10% concentration of water vapor. After pretreatment, the catalytic oxidation reaction measurements of these aged catalysts are further conducted under the conditions of a large amount of water vapor (≥8%) and CO2 (≥8%) (Supplementary Fig. 20). The exsolved Mn2O3/SMO shows the higher performance for NO, CO, and C3H6 oxidation than the SMO and commercial Pt/Al2O3. At 325 °C, CO, NO, and C3H6 conversions of the Mn2O3/SMO are 100%, 62.6%, and 100%, respectively.Dynamic evolutions of Olatt-Mn3+ sitesIn order to elucidate the dynamic evolutions of the valence state of Mn and reactivity of Olatt in NO oxidation, NAP-XPS spectra of Mn2O3/SMO and SMO were employed. The O 1 s and Mn 2p XPS spectra of Mn2O3/SMO display significant peak shifts to higher binding energy in O 1 s and to lower binding energy in Mn 2p under both NO and NO + O2 atmospheres at 200 and 300 °C (Supplementary Fig. 21). In comparison, the SMO shows trivial changes in NO adsorption and oxidation processes. To further investigate the dynamic changes, we derived the variation curves of the valence state of Mn and Olatt content versus the reaction conditions by calculating the fitting data of Mn 2p and O 1 s XPS spectra (Supplementary Figs. 22–41). There is a slight decrease in the average valence state of Mn (Fig. 3a) and a consequent increase of adsorbed oxygen content (Supplementary Fig. 42) on the SMO surface at 200 °C, however, the lattice oxygen position and Olatt/(Sm+Mn) does not change during the process (Fig. 3a and Supplementary Fig. 43). The results suggest that the surface Olatt species on SMO have poor reactivity in NO oxidation and that NO gas mainly adsorbs at the SMO surface at low temperature. When O2 is also introduced at 200 °C, a weak increase in the valence state of Mn is observed, which can be attributed to the partial oxidation of NO and stored on SMO at Mn sites. The position of Olatt still fails to respond to NO and O2 even at 300 °C. The results suggest that traditional Mn sites are only active, the Olatt sites of SMO hardly participate in NO adsorption and oxidation reaction below 300 °C. As for Mn2O3/SMO, when NO is introduced, the position of Olatt slightly shifts to lower binding energy and there is a decrease in the ratio of Olatt/Oads and Olatt/(Sm+Mn); at the same time, the valence state of Mn also decreases (Fig. 3a). The results indicate that the Olatt sites considerably react with NO even at 200 °C, which leads to the electron transfer from NO to Mn via the covalent bond of Olatt-Mn3+. When O2 is also introduced at 200 °C, the valence state of Mn is even higher than that under vacuum conditions (surface oxidation process), and the position of Olatt moves back to its original binding energy, suggesting that the reacted Olatt-Mn3+ groups undergo a regeneration process by the O2 dissociation at oxygen vacancies (Supplementary Fig. 43).Fig. 3: Dynamic structure evolutions of lattice O and metal Mn sites.a Dynamic evolutions of Mn valence state in NAP-XPS results, Olatt/(Sm+Mn) ratio and lattice oxygen (Olatt) content of Mn2O3/SmMn2O5 (Mn2O3/SMO) and SmMn2O5 (SMO) in NO oxidation; dynamic evolutions of covalency of the Mn-O bond and Mn 3d orbitals at the Mn sites of Mn2O3/SMO and SMO in NAP-XAS results of (b) AEY and (c) TEY modes. These error bars are mainly derived from fitting, reading, and random errors of experimental data.On increasing the temperature, the reactivity of Olatt also increases because of increased changes in NO and NO + O2 atmospheres. One should note that there is an imbalance between the Mn valence state and the Olatt quantity in the regeneration process, namely that Mn in Mn2O3/SMO has a higher valence state and Olatt has a lower position, whereas the oxidation state of the Mn atoms in SMO increases and the Olatt position remains unchanged. This can be due to the enhancement of oxygen adsorption, dissociation, and exchange. At 300 °C, the consumption rate of Olatt on Mn2O3/SMO is even larger than the compensation rate. The dynamic evolutions from NAP-XPS confirm that the Olatt sites on Mn2O3/SMO act as the primary active sites in NO oxidation. Moreover, we also directly observed that the surface Sm content (Sm/Mn ratio) in SMO increases when increasing the reaction temperature, whereas the surface Sm content in Mn2O3/SMO decreases (Supplementary Fig. 44). Large-ion-radius Sm cations on Mn2O3/SMO contribute to preserve its mullite crystal structure and tune the interface interactions, but they are the inactive as metal sites in NO oxidation (Supplementary Figs. 45, 46). Thus, the SMO catalyst shows poor durability during the catalytic oxidation process due to surface Sm enrichment, whereas the Mn2O3/SMO catalyst possesses excellent durability owing to surface Mn enrichment.We further summarize the transfer process of oxygen species on the Mn2O3/SMO surface in the cooperative MvK mechanism. Traditional SMO catalysts mainly comply with the E-R mechanism, where adsorbed oxygen species at the Mn sites react with NO gas and are converted into adsorbed nitrates. Differently, for Mn2O3/SMO there exists a strong electron transfer between two phases, which efficiently activates the surface Olatt sites. Mn2O3/SMO has obvious changes in lattice oxygen and adsorbed oxygen species at 300 °C (Fig. 3a and Supplementary Figs. 42, 43). Under a NO atmosphere, NO gas reacts with lattice oxygen into adsorbed nitrate and nitrite species, where lattice oxygen is consumed and are not compensated. The Mn valence state in Mn2O3/SMO also has a distinct decrease. When NO and O2 are introduced, lattice oxygen is dynamically consumed and regenerated. The lattice oxygen position does not restore the original peak position, but the valence state of the Mn atoms has a strong increase (Fig. 3a). These results indicate that the lattice oxygen consumption rate of Mn2O3/SMO is faster than regeneration rate at 300 °C. Therefore, Mn2O3/SMO complies with the dominated MvK mechanism and to a lesser degree the E-R mechanism. The oxygen transfer process with the Mn2O3/SMO catalyst mainly occurs at the Mn sites and oxygen vacancies. Introduction of oxygen makes nitrites and nitrates easily desorbed from the surface due to the oxidative environment.As a complementary technology to NAP-XPS spectra (1–4 nm), NAP-XAS spectra usually provide a shallower surface probe depth (1-3 nm in Auger electron yield (AEY) mode). It can be used to identify the chemical descriptors such as covalency of metal-oxygen bond and electron density by examining the electron transition probability at various orbitals under different reaction conditions49,50. To monitor the dynamic evolutions on surface electronic structure at Olatt and Mn sites, we employed NAP-XAS spectra of the O K-edge and Mn L-edge using the total electron yield (TEY) and AEY modes (Supplementary Figs. 47–54), respectively. As known, the AEY mode is more surface-sensitive than the TEY mode, and it includes more detailed covalency information of the metal-oxide bond, while the TEY mode includes information about the O2 adsorption and electron density of cations37. The covalency of the Mn-O bond is confirmed by the peak of Mn 3d-O 2p hybrid orbitals at the O K-edge at around 529 eV and the peak distance (ΔE) for the Mn L3-edge in 641–644 eV. In the AEY mode, the covalency of the Mn-O bond on SMO (Fig. 3b) is inactive under both reduced and oxidized atmospheres. In contrast, Mn2O3/SMO exhibits a distinct enhancement in the covalency of the Mn-O bond under a reduced atmosphere, and an additional shoulder peak in the Mn L3-edge spectra at around 640 eV occurred (Supplementary Fig. 52). The results suggest that surface Olatt shows high activity in the process of bonding with H* radicals and reduction of Mn sites, leading to the increase of Mn 3d empty orbitals. When O2 was introduced, it could adsorb at Mn sites and partially heal the oxygen vacancies at 150 °C, resulting in the decrease of covalency of the Mn-O bond. One should note that the position of Olatt under a H2 atmosphere at 300 °C is still lower than that at 150 °C, which can be due to the influence of temperature on the Olatt migration. Under oxidized conditions at 300 °C, O2 can almost fill nearly all the oxygen vacancies, restoring the pristine covalency of the Mn-O bond under vacuum.The TEY mode of O K-edge and Mn L-edge XAS spectra is used to describe the O2 adsorption and electron density (Fig. 3c). The O2 adsorption intensity is calculated using the peak intensity ratios of the first two peaks, α and β, in the O K-edge XAS spectra, while the electron density is calculated using the peak intensity ratios of two peaks, A and B, in the Mn L3-edge. The electron density of SMO at the Mn sites slightly changes under redox cycles, while O2 can adsorb and dissociate at the Mn sites of SMO, leading to a decrease in the electron density of the Mn sites. Whereas Mn2O3/SMO exhibits a considerable increase in electron density of Mn sites under an H2 atmosphere, especially at 150 °C, and the behavior of O2 adsorption and electron density of Mn sites under O2 atmosphere at 300 °C are similar to SMO. The results suggest that the Mn sites of Mn2O3/SMO are still less active at 300 °C, consistent with the H2-TPR results. The time-resolved MS signals of H2O also confirm the redox cycles (Supplementary Fig. 55). It is found that the active sites on Mn2O3/SMO are dominated by Olatt sites, whereas on SMO, it is the traditional Mn sites. The active order shows accordance with that in NO oxidation at 150 and 300 °C. The above NAP-XPS and NAP-XAS results directly demonstrate the temperature-dependent changes in the valence state of Mn and the covalence of the Mn-O bond and further validate the roles of Olatt at the interface between exsolved-Mn2O3 and mullite support.Key intermediate species in NO oxidationTo gain insights into intermediates in NO oxidation, in-situ diffuse reflectance infrared Fourier transformations spectroscopy (DRIFTS) spectra were recorded for Mn2O3/SMO and SMO under NO and NO + O2 conditions (Fig. 4a–c and Supplementary Fig. 56). The peak intensity of the IR spectra was converted into Kubelka-Munk units to analyze the concentration variations of intermediates. The observed bands at 1235, 1270, and 1570 cm−1 correspond to the bridged, monodentate, and bidentate nitrate species, respectively16,17. In addition, the band at 1615 cm−1 is assigned to the weakly adsorbed NO2 at Olatt sites (Olatt-NO2) as the nitrite species5,10. Mn2O3/SMO exhibits a stronger IR peak signal for nitrate species compared to SMO under NO + O2 at 200 °C (Fig. 4a), indicating the superior NO adsorption ability of Mn2O3/SMO. Mn2O3/SMO possesses more monodentate nitrates than SMO. In addition, we observed an increase in the amount of adsorbed NO2 (1615 cm−1, nitrite species) at active Olatt sites of Mn2O3/SMO. Desorption of nitrites and monodentate nitrates to NO2 is known to be more facile than the bridged nitrate17. Mn2O3/SMO exhibits rapid NO oxidation activation on the surface due to the larger peak intensity variations compared to SMO (Fig. 4b-c). When O2 is introduced, the amount of monodentate nitrate in Mn2O3/SMO distinctly decreases and the bidentate and bridged nitrate species show slight decreases. However, the adsorbed NO2 at Olatt sites (nitrite species) of Mn2O3/SMO increased17,34. The result demonstrates the transformation from the nitrates to nitrites during oxygen-rich NO oxidation.Fig. 4: Active sites and reaction mechanisms in NO oxidation.a In-situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) of Mn2O3/SmMn2O5 (Mn2O3/SMO) and SmMn2O5 (SMO) under NO + O2 at 200 °C. Time-resolved in-situ DRIFTS spectra of the (b) SMO and (c) Mn2O3/SMO under NO gas at 200 °C. d Time-resolved isotopic 36O2 exchange reaction at 300 °C, (e) time-resolved isotopic 36O2 labeled online MS for CO oxidation at 150 °C, (f) time-resolved isotopic 36O2 labeled online MS for NO oxidation at 300 °C.The N 1s NAP-XPS spectra (Supplementary Figs. 57, 58) confirm that NO gas preferred to adsorb at the Olatt sites of Mn2O3/SMO, and its intermediates contain extra Osite-NO and Osite-NO2 species compared to traditional SMO. The adsorbed NO on Olatt can transform into nitrite species and oxygen vacancy (-Olatt-NO2-Mn-Ov), and the chemical bond for Olatt-N can be broken into NO2. The SMO catalyst mainly complies with the E-R mechanism, where adsorbed oxygen species react with NO gas into nitrate species, such as monodentate, bidentate, bridged nitrates. Different from SMO, Mn2O3/SMO depends on the dominated MvK mechanism and to a lesser extent the E-R mechanism. NO gas preferentially adsorbs on lattice oxygen sites and produces the extra nitrite species (Osite-NO2). The nitrites on the lattice oxygen can more easily desorb from the catalyst surface compared to traditional nitrate species. These results highlight that the higher NO oxidation activity of Mn2O3/SMO stems from the quick decomposition or desorbed process of nitrite species and monodentate nitrates to NO2.To clarify the water effect mechanism on mullite oxides in NO oxidation, we investigated in-situ DRIFTS spectra under the atmosphere of NO + O2 + H2O at 200 °C (Supplementary Fig. 59). The intermediates formed over the SMO and Mn2O3/SMO catalysts are described in Supplementary Table 2. The IR vibration intensity of water over the SMO catalyst (water at 1625 cm−1 and hydroxyl in 3400–3700 cm−1) is far stronger than that over Mn2O3/SMO (1647 cm−1 and 3400–3700 cm−1), indicating that the SMO surface is covered by water molecules and a stronger competitive adsorption exists with reactant gases than does for Mn2O3/SMO (Supplementary Fig. 59). The IR peak at 1625 cm−1 can be attributed to absorbed nitrite species (Supplementary Table 2). If it is nitrite species, this will confirm that water is mainly adsorbed at the Mn active sites of the SMO catalyst and the competitive adsorption makes the NO oxidation comply more with the MvK mechanism at the Olatt sites. Therefore, the water poisoning mainly originates from the competitive adsorption and the desorbed NO3− species binding with water. The observed monodentate nitrate and nitrite species demonstrate the E-R mechanism at the Mn site and cooperative MvK mechanism at Olatt site in NO oxidation, respectively. These two reaction mechanisms are negatively affected by water vapor through the competitive adsorption and desorbed nitrates.DFT calculations were conducted to further elucidate the impacts of Olatt sites in NO oxidation and NO2 desorption (Supplementary Fig. 60). The models are constructed based on the experimental results and previous reports10,27,33. The O-Mn4+ of the SMO model exhibits a higher energy barrier of 2.79 eV in NO oxidation (E1) at the Olatt sites. However, the O-Mn3+ of the SMO model in E1 is lower (1.45 eV), whereas the O-Mn3+ of the Mn2O3/SMO model is even lower at 0.56 eV. The results confirm that O-Mn3+ of Mn2O3/SMO is the major active site in NO oxidation. Oxygen vacancy formation energies of interface Mn3+-Osite in Mn2O3/SMO, Mn3+-Osite in SMO, and Mn4+-Osite in SMO also agree with the above calculations, where they are ~1.1 eV, ~1.8 eV and ~2.7 eV, respectively. In addition, the NO2 desorption (Olatt-NO2 species) on the O-Mn3+ of Mn2O3/SMO and SMO are exothermic processes (−0.68 and −0.50 eV), while it is an endothermic process (0.11 eV) on the O-Mn4+ of SMO. The results indicate that NO2 is more easily desorbed from the O-Mn3+ sites, which is aligned well with the in-situ spectroscopy results.The time-resolved isotopic 36O2 exchange experiments are measured to demonstrate the activity of lattice oxygen (Fig. 4d). Each sample was pretreated at 300 °C for 3 h under an Ar atmosphere to decrease surface adsorbed oxygen. It was revealed that Mn2O3/SMO exhibits higher Olatt activation and dissociation abilities compared to SMO because of the stronger 34O2 signals of Mn2O3/SMO versus those of SMO. To investigate the activity of Olatt sites, we use CO and isotopic 36O2 as the probe molecules at 150 °C. To further investigate the activation and dissociation ability of molecular oxygen, the transient oxygen exchange test was conducted at different temperatures. Wachsman et al. proposed that the exchange of lattice oxygen (34O2) on non-stoichiometric metal oxides is usually better at oxygen dissociation when using 1:1 isotopic mixture51,52. By analyzing the concentration variations of oxygen exchange (Supplementary Figs. 61, 62), we demonstrate that the oxygen exchange rate of Mn2O3/SMO is distinctly higher than that of SMO at the same temperature. At 300 °C, the Mn2O3/SMO shows the better oxygen activation ability, while the oxygen exchange rate of SMO is very low. The results are consistent with the NAP-XPS spectra and previous work on perovskites51,52. Furthermore, we also calculated the apparent activation energy of O2 for the two catalysts by the slope of Arrhenius plot. The apparent oxygen activation energy of SMO and Mn2O3/SMO catalysts are 38.8 and 44.8 kJ/mol, respectively.Both the MvK and E-R mechanisms exist in the Mn2O3/SMO catalyst (Fig. 4e), as indicated by the simultaneous appearance of 44CO2 and 46CO2 signals. Interestingly, Mn2O3/SMO exhibits an earlier crossing time (570 s) in the 44CO2 and 46CO2 signals compared to SMO (870 s), with a stronger initial intensity of 44CO2. This suggests that the consumption rate of Olatt with CO in Mn2O3/SMO is much higher than that in SMO. On decreasing the temperature to 100 °C (Supplementary Fig. 63), the Olatt activity in Mn2O3/SMO weakens, and no crossing point is observed during the 2400 s. However, the signal of 44CO2 is still stronger than that of 46CO2. Therefore, the CO oxidation on Mn2O3/SMO mainly follows the MvK mechanism and Olatt is the important active site. Furthermore, the time-resolved isotopic 36O2 signal at 300 °C (Fig. 4f) shows that the Mn2O3/SMO exhibits a much stronger 46NO2 signal compared to SMO, while the intensities of 48NO2 signal are nearly the same, indicating that the reaction rate at the Mn sites of Mn2O3/SMO is low and similar to that of SMO (the E-R mechanism). In contrast, the reaction rate at the Olatt sites is significantly improved on Mn2O3/SMO, and the NO oxidation pathway is determined by the MvK mechanism.The reaction orders of Mn2O3/SMO and SMO were calculated to achieve strict kinetic control (Supplementary Table 3). The reaction orders of SMO for NO and O2 are 0.61 and 0.59, respectively, while a noticeable decrease of Mn2O3/SMO is observed (0.55 to NO and 0.47 to O2). As known, lower reaction orders to O2 typically indicate that adsorbed oxygen can be easily activated at a catalyst surface with oxygen vacancies53. That is, decomposition of O2 is more facile at the surface of Mn2O3/SMO. In addition, the decrease in NO reaction orders can be attributed to the strong competitive adsorption of O2 to NO. As demonstrated through in-situ spectroscopy results, the Mn2O3/SMO catalyst presents both higher covalency of Mn–O bond and electron density of Mn sites, which leads to the generation of more Olatt sites. These Olatt sites facilitate the activities of NO oxidation dominated by the cooperative MvK mechanism.
