Tetrapropylammonium bromide (TPABr, 98%), silica sol (20%-40%), sodium aluminate (NaAlO₂, analytical reagent), and sodium hydroxide (NaOH, 96%) were purchased from Shanghai Macklin Biochemical Co., Ltd. Hexachloroiridic acid hydrate (H₂Cl₆Ir·xH₂O, 36%), ammonium molybdate [(NH₄)₆Mo₇O₂₄·4H₂O], ethylenediamine (EDA, 99.5%), and tetraethylenepentamine (TEPA, technical grade‌) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. Unless otherwise specified, all chemicals were of analytical grade and used as received without further purification.

TPABr, silica sol, NaAlO₂, and NaOH were used for preparing ZSM-5 (hereinafter referred to as Z5). An improved ligand-protected one-step synthesis strategy was used to obtain ZSM-5-encaged Ir-Mo bimetallic clusters, denoted as Ir-Mo@Z5. Ir-Mo/Z5 catalysts were prepared by the impregnation method with the same metal loading for comparison. The symbols “@” and “/” are used to distinguish catalysts prepared via encapsulation within zeolite frameworks and surface-supported configurations, respectively. Before evaluating catalytic performance, the synthesized Ir-Mo@Z5 and Ir-Mo/Z5 catalysts were pretreated in 5% hydrogen/argon (H₂/Ar) at 200 °C for 30 min. The detailed synthetic method is provided in the Supplementary Materials. A schematic diagram of Ir-Mo@Z5 catalyst preparation is shown in Supplementary Figure 1.

The activity measurements of CO-SCR at atmospheric pressure were conducted in a quartz tubular continuous-flow reactor (inner diameter = 6 mm) using 40-60 mesh catalyst particles. The total flow rate of the simulated reactant gas was 100 mL⋅min^-1 with a gas hourly space velocity (GHSV) of 16,000 h^-1. The gas composition contained 400 ppm NO, 8,000 ppm CO, 5% O₂, 200 ppm SO₂, and N₂ as the balance gas. Concentrations of CO and NO_x were analyzed in real time using an online Fourier transform infrared (FTIR) spectrometer equipped with a deuterated triglycine sulfate (DTGS) detector. The conversion and selectivity were calculated as below^[20]:

(1) $$ \mathrm{X}_{\mathrm{NO}_{\mathrm{x}}}=\left[\frac{\mathrm{NO}(\text{in})-\mathrm{NO}(\text{out})-\mathrm{N}_{2}\mathrm{O}(\text{out})-\mathrm{NO}_{2}(\text{out})}{\mathrm{NO}(\text{in})}\right] \times 100\%, \\ $$

(2) $$ \mathrm{X}_{\mathrm{NO}}=\left[\frac{\mathrm{NO}(\text{in})-\mathrm{NO}(\text{out})}{\mathrm{NO}(\text{in})}\right] \times 100\%, \\ $$

(3) $$ \mathrm{~X}_{\mathrm{CO}}=\left[\frac{\mathrm{CO}(\text{in})-\mathrm{CO}(\text{out})}{\mathrm{CO}(\text{in})}\right] \times 100\%, \\ $$

(4) $$ \mathrm{~S}_{\mathrm{N}_{2}}=\left[\frac{\mathrm{NO}(\text{in})-\mathrm{NO}(\text{out})- \mathrm{2N}_{2}\mathrm{O}(\text{out})-\mathrm{NO}_{2}(\text{out})}{\mathrm{NO}(\text{in})-\mathrm{NO}(\text{out})}\right] \times 100\%, \\ $$

(5) $$ \mathrm{~S}_{\mathrm{N}_{2}\mathrm{O}}=\left[\frac{\mathrm{2N}_{2}\mathrm{O}(\text{out})}{\mathrm{NO}(\text{in})-\mathrm{NO}(\text{out})}\right] \times 100\%, \\ $$

(6) $$ \mathrm{~S}_{\mathrm{NO}_{2}}=\left[\frac{\mathrm{NO}_{2}(\text{out})}{\mathrm{NO}(\text{in})-\mathrm{NO}(\text{out})}\right] \times 100\%. $$

Detailed information on catalyst characterizations is provided in the Supplementary Materials.

The CO-SCR performance of Ir-Mo@Z5 and Ir-Mo/Z5 catalysts under typical industrial flue gas conditions was evaluated within the temperature range of 150-400 °C. The results are presented in Figure 1. As shown in Figure 1A, the NO conversion of both samples follows a volcano curve, with or without SO₂ introduction, consistent with recent reports in the literature^[13,21]. It is worth noting that after SO₂ injection, the NO conversion of Ir-Mo/Z5 increased significantly, whereas that of Ir-Mo@Z5 decreased slightly. Comparison of the NO conversion of Ir-Mo@Z5 with that of several high-performance bimetallic catalysts reported in the literature [Supplementary Table 1] shows that Ir-Mo@Z5 exhibits superior catalytic performance under comparable reaction conditions, highlighting its advantages and potential applications in CO-SCR systems. To more accurately reflect the actual efficiency of NO reduction to N₂ and eliminate the interference from byproducts such as NO₂, NO_x conversion is presented in Figure 1B. Unlike NO conversion, NO_x conversion is synchronously promoted for both samples after SO₂ injection. The NO_x conversion of Ir-Mo@Z5 was higher than that of Ir-Mo/Z5 across the entire temperature range, reaching a maximum of 62.4%. Subsequently, the long-term catalytic stability of Ir-Mo@Z5 and Ir-Mo/Z5 was further evaluated. As shown in Supplementary Figure 2, both catalysts exhibit excellent durability during continuous reaction, demonstrating their promising potential for practical applications. Figure 1C confirms that CO achieves nearly complete conversion at the temperature where NO or NO_x conversion reaches a maximum over both catalysts, indicating that the volcano-shaped NO or NO_x conversion profile originates from preferential oxidation of CO by O₂ at elevated temperatures. In addition, Figure 1D indicates that the SO₂ injection effectively improves N₂ selectivity, and Ir-Mo@Z5 exhibits superior N₂ selectivity compared to Ir-Mo/Z5. Supplementary Figure 3 illustrates the effect of O₂ on CO-SCR performance at 275 °C. Firstly, both catalysts exhibit reduced NO_x conversion with increasing O₂ concentration. However, Ir-Mo@Z5 maintains higher activity in the absence of the SO₂ promotional effect, indicating better O₂ tolerance.

These results indicate that although SO₂ exhibits similar promotional effects on NO_x conversion over the two catalysts, the underlying mechanisms differ. Undoubtedly, SO₂ effectively improves the selective reduction of NO to N₂. As we previously reported^[22], this is attributed to the regulatory effect of SO₂ on the “ineffective reaction pathway” in the presence of O₂ with nitrate as an intermediate. More importantly, SO₂ also promotes NO adsorption on catalysts prepared by the conventional impregnation method.

Supplementary Figure 4 shows the X-ray diffraction (XRD) patterns of Ir-Mo@Z5 and Ir-Mo/Z5 catalysts. The obtained catalysts exhibited a well-ordered MFI-type zeolite structure of Z5 (PDF#44-0003)^[23,24]. No distinct characteristic peaks corresponding to Ir or Mo species were detected due to the low metal loading and high dispersion^[25]. The results demonstrate that the addition of EDA, TEPA, and metal precursors into the synthetic gel had a negligible effect on the crystallization of the Z5 support. N₂ adsorption-desorption isotherms of both Ir-Mo@Z5 and Ir-Mo/Z5 catalysts exhibit typical Type I behavior [Supplementary Figure 5], indicating a microporous structure with a dominant pore size of ~0.55 nm [Supplementary Figure 6]. As summarized in Supplementary Table 2, both catalysts maintain relatively high specific surface areas (~280 m²g^-1), which provide abundant sites for the adsorption of CO and NO molecules. Inductively coupled plasma (ICP) analysis indicates that the Si/Al ratios of both catalysts are approximately 25. The results also reveal that the loadings of Ir and Mo are ~0.4 and 0.85 wt.%, respectively, confirming successful incorporation of Ir and Mo into the Z5 support.

According to the transmission electron microscopy (TEM) images of the catalysts [Figure 2], the well-dispersed Ir species in Ir-Mo@Z5 exhibited an average size of 1.0 nm through the one-step synthesis method. The observation that these clusters are larger than the micropore channels in Z5 (0.55 nm) can be attributed to partial collapse of neighboring micropores during metal growth^[26]. In other words, the clusters are likely confined within zeolite crystals rather than within the microporous channels^[27]. Combined with energy-dispersive X-ray spectroscopy (EDS) mapping results, the Mo signal completely overlaps with that of Ir, suggesting the formation of Ir-Mo bimetallic clusters^[23]. In addition, Ir-Mo bimetallic clusters show clear interplanar lattice fringes of 0.23 nm, which may correspond to the Ir (111) facet, implying the favorable dispersion of Mo species on the Ir surface. Previous studies have found that the fraction of bimetallic clusters in zeolite-supported catalysts gradually increases with prolonged hydrogen (H₂) reduction time^[12,28], confirming that H₂ pretreatment is beneficial for the formation of Ir-Mo bimetallic interactions. In contrast, larger bimetallic clusters (with an average size of ~8.0 nm) were unevenly distributed on the zeolite’s external surface in catalysts prepared by the conventional impregnation method.

The Ir 4f X-ray photoelectron spectroscopy (XPS) spectra of Ir-Mo@Z5 and Ir-Mo/Z5 catalysts are shown in Figure 3. In Figure 3A, the Ir 4f_7/2 peak centered at 61.20 eV and 62.20 eV observed for Ir-Mo@Z5, which were assigned to metallic Ir⁰ and oxidized Ir^δ+, respectively^[29,30]. After peak fitting, it was found that the proportion of Ir⁰ in Ir-Mo@Z5 was 52.77%, which is significantly lower than that in Ir⁰ on Ir-Mo/Z5 (61.54%) [Figure 3B]. Therefore, we confirm that the ratio of Ir⁰ to Ir^δ+ can be tuned by encapsulating bimetallic clusters within the zeolite micropores, thereby effectively regulating their redox properties. This is consistent with our previous finding on Ir clusters encapsulated within zeolite micropores^[11]. Furthermore, NO molecule dissociation can be enhanced by electron-rich Ir⁰ species through the electron donation from the metal center into the empty π* orbital of NO^[3], which is crucial for low-temperature catalytic performance. However, oxidized Ir^δ+ species can promote selective NO adsorption in the presence of O₂ and even help maintain a balance between CO oxidation and NO reduction reactions in the presence of SO₂^[11,31]. Moreover, the Ir 4f XPS signal intensity of Ir-Mo@Z5 is lower than that of Ir-Mo/Z5, indicating fewer metal species are dispersed on the external surface of Ir-Mo@Z5, which confirms that Ir-Mo bimetallic clusters are mainly confined within zeolite crystals via the one-step synthesis method. As shown in Supplementary Figure 7, compared with Ir@Z5, the binding energy of Ir 4f_7/2 in Mo-modified Ir-Mo bimetallic cluster catalysts exhibits an obvious redshift of 0.45 eV toward lower binding energy. Given Ir’s higher electronegativity compared to Mo, Mo tends to donate electrons in bimetallic clusters. After H₂ pretreatment at 200 °C, Mo species in Ir-Mo@Z5 and Ir-Mo/Z5 serve as electron donors, transferring partial electrons to Ir, thereby increasing the electron density at Ir sites. The electron-rich Ir⁰ can elongate the N–O bond and facilitate NO dissociation.

Temperature-programmed experiments were conducted to investigate the adsorption and desorption behaviors of NO and CO on the two catalysts. As shown in Figure 4A and B, two N₂O desorption peaks were observed for the Z5 support at 335 and 650 °C in the NO-temperature-programmed desorption (NO-TPD) and N₂O-TPD profiles. This indicates that all NO molecules adsorbed on the Z5 support undergo a coupling-dissociation process^[4,32]. The ONNO intermediate formed through coupling is prone to N–O bond cleavage, which is completed at temperatures below 335 °C. The NO-TPD results for the Ir-Mo@Z5 and Ir-Mo/Z5 catalysts also show NO desorption peaks. According to previous reports^[13], these NO species are likely adsorbed and decomposed into nitrates. Compared with the three N₂O desorption peaks observed in the N₂O-TPD profiles (175, 335, and 530 °C), only two desorption peaks at 335 and 530 °C were observed in the NO-TPD profile of Ir-Mo/Z5. This indicates that the dissociation temperature of ONNO on Ir-Mo/Z5 is between 175 and 335 °C, whereas that on Ir-Mo@Z5 is between 320 and 385 °C. These results confirm that electron-rich Ir⁰ species dominate the ONNO dissociation. Increasing the content of metallic Ir⁰ species in the catalyst is beneficial for intermediate conversion.

As shown in Figure 4C, carbon dioxide (CO₂) is the only desorption product observed for the two catalysts and the Z5 support during programmed heating (after CO saturation); no significant CO desorption peak is detected, which is typically associated with the reaction between CO and lattice oxygen species. Compared with the CO₂ desorption peak observed on the Z5 surface at 650 °C, two additional desorption peaks appeared at lower temperatures for the Ir-Mo@Z5 and Ir-Mo/Z5 catalysts. Among them, the two CO₂ desorption peaks of Ir-Mo/Z5 occur at lower temperatures than those of Ir-Mo@Z5, indicating that the lattice oxygen activity of Ir-Mo/Z5 is superior to that of Ir-Mo@Z5. We propose that electrons are transferred from coordinated oxygen species to the Ir species, reducing the oxidized Ir^δ+ species on Ir-Mo/Z5 to metallic Ir⁰. This process facilitates oxygen-ion activation and vacancy formation to maintain charge balance^[4,33], thereby explaining the lower activation temperature for CO oxidation on Ir-Mo/Z5. In addition, the simultaneous formation of CO₂ and H₂ at 370 °C on Ir-Mo/Z5 is observed, originating from the reaction between CO and surface -OH species (2CO + 2OH → 2CO₂ + H₂)^[34,35]. According to our previous report^[11], Ir species loaded on the surface of Z5 usually occupy surface -OH groups^[36], especially bridged silanol-aluminum groups (Si-OH-Al). Therefore, the larger bimetallic clusters on Ir-Mo/Z5 obtained via the conventional impregnation method imply fewer occupied -OH groups, which may be one of the reasons for the enhanced CO oxidation activity. Furthermore, Figure 4D shows the CO-TPR results. In general, the CO₂ desorption temperature corresponding to the reaction between gas-phase CO and lattice oxygen species is always higher than that for adsorbed CO on the two catalysts and the Z5 support, indicating that the Langmuir-Hinshelwood (L-H) mechanism occurs more readily at lower temperatures.

To investigate the effects of O₂ and SO₂ on reaction intermediates, NO, NO + O_2, or NO + O₂ + SO₂ were introduced over various catalysts at 275 °C. According to Wang et al.^[37], the promotional effects of O₂ and SO₂ on the performance of Ir-based CO-SCR catalysts may be related to surface -OH groups. Once NO-adsorbed species occupy a portion of the catalyst surface -OH groups, the intensities of the negative -OH bands would decrease because fewer adsorption sites remain available for further adsorption. Figure 5A and B shows changes in the in situ DRIFTS spectra over the range of 3,100-3,900 cm^-1 under different adsorption conditions. The absorption bands observed at 3,678 and 3,587 cm^-1 are attributed to the stretching vibrations of isolated silanol groups (Si-OH) and the bridged silanol-aluminum groups (Si-OH-Al), respectively. After NO introduction, the peak areas of Si-OH and Si-OH-Al decrease to different extents. To quantify the variations in the two -OH groups, Gaussian fitting was applied to deconvolute the infrared absorption bands, and the results are summarized in Figure 5C. The peak area of Si-OH-Al decreases more significantly after NO introduction, showing a reduction more than twice that of Si-OH on both Ir-Mo@Z5 and Ir-Mo/Z5 catalysts. This indicates that NO is more readily adsorbed on Si-OH-Al sites. However, once NO and O₂ were introduced simultaneously, the extent of Si-OH-Al consumption was significantly reduced, supporting the competitive adsorption between NO and O₂. Interestingly, the -OH groups on the two catalysts exhibited distinctly different trends after further SO₂ introduction. In general, the two types of -OH groups on the Ir-Mo/Z5 catalyst were significantly consumed, whereas those on the Ir-Mo@Z5 catalyst were largely preserved after SO₂ introduction. Wang et al.^[37] reported that SO₂ is first oxidized to SO₃ on the Ir surface and subsequently forms sulfate species on -OH sites, thereby suppressing a series of side reactions in the presence of O₂. As shown in the SO₂-TPD profiles in Supplementary Figure 8, the SO₂ desorption amount is higher on supports with lower Si/Al ratios, providing direct evidence for SO₂ adsorption and activation by -OH groups. Our previous studies^[22] also showed that SO₂ oxidation at Ir (111) sites is beneficial for suppressing the “ineffective reaction pathway” mediated by nitrate species in the presence of O₂. As mentioned earlier, more oxidized Ir^δ+ species on Ir-Mo/Z5 are readily reduced to metallic Ir⁰, facilitating oxygen ion activation and vacancy formation for charge balance maintenance, thereby resulting in higher lattice oxygen activity on Ir-Mo/Z5 than on Ir-Mo@Z5. Therefore, SO₂ is more readily oxidized on the Ir-Mo/Z5 catalyst, leading to greater consumption of -OH groups. These results clearly explain why the promotional effect of SO₂ is more pronounced over Ir-Mo/Z5 than over Ir-Mo@Z5, as shown in Figure 1 and Supplementary Figure 3.

The activity tests revealed that the mechanism by which SO₂ influences CO-SCR performance depends on the catalyst preparation method. The NO_x conversion over Ir-Mo@Z5 is higher than that over Ir-Mo/Z5 across the entire test temperature range, reaching a maximum of 62.4%. The one-step synthesis method yields small, uniform Ir-Mo bimetallic species confined within the zeolite interior, whereas the conventional impregnation method produces larger, unevenly distributed particles on the external surface. The formation of well-dispersed bimetallic structures is facilitated by appropriate reduction treatment. The confinement effect of zeolite micropores regulates the electronic state of the active sites, which is crucial for the adsorption and dissociation of NO molecules. Meanwhile, particles on the external surface favor surface oxygen activation and reactions with surface hydroxyl groups, thereby enhancing CO oxidation. These results demonstrate that the spatial distribution and electronic properties of the active sites can be rationally regulated through synthetic strategies, thereby governing the adsorption and reaction pathways of NO and CO.

To investigate the dynamic evolution of adsorbed species on the two catalysts, NO was first pre-adsorbed on the catalyst surface until saturation, followed by CO introduction to initiate the transient reaction [Supplementary Figure 9A and B]. Prominent characteristic adsorption peaks corresponding to ONNO (1,720-1,724 cm^-1) and N₂O₄ (1,683 cm^-1) were detected on both Ir-Mo@Z5 and Ir-Mo/Z5, with dimeric ONNO species serving as the dominant adsorbed intermediates^[11]. After CO introduction, the intensity of the ONNO peak gradually decreased over time on both catalysts, confirming the synergistic catalytic effect of Ir and Mo sites in promoting ONNO dissociation. Notably, distinct adsorption peaks assigned to CO (2,057-2,063 cm^-1) and NO (1,830 cm^-1) adsorbed on metallic Ir⁰ were observed over Ir-Mo/Z5, confirming the presence of Ir⁰ species on this catalyst. In contrast, no obvious adsorption peaks associated with Ir⁰ were detected on Ir-Mo@Z5. This strongly indicates that the one-step synthesis strategy effectively enhances the interaction between Ir and Mo within sub-nanoclusters and suppresses excessive reduction and aggregation of metallic species. To further explore the effect of O₂ on the CO-SCR reaction pathway, in situ DRIFTS experiments were conducted by introducing CO following the pre-adsorption of NO + O₂ [Supplementary Figure 9C and D]. Under aerobic conditions, the proportion of N₂O₄ species on Ir-Mo/Z5 increased markedly, whereas the formation of ONNO, a key low-temperature active intermediate, was significantly inhibited. By contrast, ONNO remained the dominant NO adsorption species on Ir-Mo@Z5. This comparative result reveals that the strong bimetallic interaction within Ir-Mo sub-nanoclusters efficiently modulates the electronic structure of active sites and facilitates the directional conversion of NO into ONNO intermediates in the presence of O₂. The superior capability of Ir-Mo@Z5 to stably maintain ONNO formation and accelerate its dissociation and transformation under aerobic conditions accounts for its outstanding catalytic activity in the aerobic CO-SCR reactions. For both catalysts, NO first couples to form ONNO species, which subsequently dissociate. The difference in the ONNO dissociation temperature between the two catalysts is primarily determined by the electron-rich metallic Ir⁰ species, and a higher Ir⁰ content facilitates intermediate conversion. For CO activation, it mainly reacts with lattice oxygen species to form CO₂. The catalyst prepared via the conventional impregnation method exhibits higher lattice-oxygen activity and, therefore, a lower CO oxidation temperature. This catalyst also provides more available surface hydroxyl groups, which further participate in CO oxidation.

In situ DRIFTS results indicate that although SO₂ exhibits similar promotional effects on NO_x conversion over the two catalysts, the underlying mechanisms differ. Undoubtedly, SO₂ promotes the actual efficiency of NO reduction to N₂. As we previously reported^[22], this is attributed to the regulatory effect of SO₂ on the “ineffective reaction pathway” in the presence of O₂ with nitrate species as intermediates. More importantly, SO₂ effectively promotes NO adsorption onto catalysts prepared by conventional impregnation.