Ti doping-induced strain engineering boosts low-temperature NH₃-SCR deNO_x performance and H₂O/SO₂ resistance of Ce₈MnO_x

Abstract

H₂O and SO₂ poisoning of cerium-manganese oxide catalysts remains a critical challenge for low-temperature selective catalytic reduction of nitrogen oxides by NH₃. Here we show that Ti doping of CeO₂ induces lattice contraction and structural distortion, promoting the generation of oxygen vacancies, active oxygen species, and stable Ce-O-Ti interfaces. These strain-engineered structural modifications enhances surface Lewis acidity and triggers a downshift of the d-band center, collectively improving low-temperature catalytic activity and resistance to H₂O/SO₂ poisoning. The optimized Ce₈MnTi_2.5O_x catalyst achieves over 90% nitrogen oxides removal across a 150–335 °C temperature range. Under 5 vol.% water and 100 ppm SO₂, it maintains over 98% efficiency for 12 h at 210 °C. Furthermore, during stepwise heating from 150 °C to 240 °C with 5 vol.% H₂O and 200 ppm SO₂, initial activity suppression below 180 °C is fully reversed at or above 210 °C, demonstrating temperature-driven regeneration. This study introduces a strain-engineering strategy for designing robust catalysts for low-temperature nitrogen oxides removal.

This study demonstrates that Ti-induced strain engineering creates robust cerium–manganese oxide catalysts with enhanced low-temperature nitrogen oxide removal, strong resistance to H₂O and SO₂ poisoning, and temperature-driven self-regeneration.

Introduction

Nitrogen oxides (NO_x), as critical precursors to PM_2.5, ozone, and acid rain, pose a serious threat to air quality and public health¹. Consequently, reducing NO_x emissions has become a critical mandate for sustainable industrial development, with high-performance selective catalytic reduction (SCR) catalysts serving as the technological backbone. While ultra-low NO_x emission standards have been successfully implemented in coal-fired power plants, non-electric industries, such as cement and coking, now dominate NO_x emissions, shifting the focus of air pollution control toward these sectors. Flue gas from non-electric industries presents unique challenges: low operating temperatures (150–300 °C), high moisture content (≥5%), and elevated SO₂ concentrations (>400 mg/Nm³)². Currently, commercial low-temperature deNO_x catalysts are primarily vanadium-based with high V₂O₅ content. However, V₂O₅ is highly toxic³, causes secondary pollution, and poses hazardous waste disposal issues. Therefore, the development of eco-friendly low-temperature catalysts with a broad active temperature window (150–300 °C) and exceptional H₂O/SO₂ resistance has become an urgent priority for NO_x abatement in complex flue gas environments⁴.

Current research on low-temperature deNO_x catalysts primarily focuses on ion-exchanged zeolites and mixed metal oxides⁵. Among ion-exchanged zeolites, small-pore zeolites with CHA topology⁶, such as Cu-SSZ-13⁷ and Cu-SAPO-34⁸, exhibit broad active temperature windows (150–500 °C) and superior hydrothermal stability. However, their performance deteriorates in dust-laden, humid, and sulfur-rich atmospheres due to ammonium sulfate deposition, metal sulfation, and pore blockage, rendering them unsuitable for such complex flue gas conditions⁹. In contrast, MnO_x-based materials have garnered significant attention for their exceptional low-temperature redox properties and high intrinsic deNO_x activity. Yet, industrial deployment remains limited due to challenges such as MnO_x water-solubility, poor H₂O/SO₂ tolerance, and low N₂ selectivity^10,11. To address these limitations, researchers have explored the incorporation of promoters, including Ce¹², Co¹³, Fe¹⁴, Sm¹⁵, and Eu¹⁶, to enhance low-temperature activity and poisoning resistance, with Ce emerging as the most effective additive¹⁷. Despite these advancements, Mn-Ce catalysts remain highly susceptible to chemical deactivation via metal sulfates formation under sulfur-rich, low-temperature conditions—a persistence bottleneck that continues to impede industrial application¹⁸.

To enhance the stability of Mn-Ce catalysts under harsh flue gas conditions, recent research has concentrated on three key strategies: morphological control, promoter doping, and support optimization. Morphological control, such as constructing layered¹⁹ or core-shell architectures²⁰, can effectively limit SO₂ penetration and facilitate the decomposition of reaction intermediates, thereby improving resistance at lower temperatures and moderate SO₂ levels. However, their structural protection remains inadequate under the high-sulfur and high-humidity conditions typical of industrial flue gas. Promoter modification offers another pathway: incorporating electronic promoters such as Sm²¹ and Fe²² can tune the electronic structure, surface acidity, and oxygen species activity. These adjustments suppress SO₂ over-oxidation, enhance NH₃ adsorption, and mitigate competitive H₂O adsorption, collectively improving catalyst resilience. Additionally, advanced support designs, such as 3D-ordered macroporous (3DOM) structures²³ and carbon nanotube (CNT) loading²⁴, further optimize textural properties, increase active site dispersion, and boost mass transfer efficiency. Despite these innovations, achieving long-term stability under real-world conditions (150–300 °C, H₂O ≥ 5%, SO₂ > 400 mg/Nm³) remains an unresolved challenge, underscoring the urgent need for breakthrough solutions.

Building on these challenges, this study proposes a strategy to overcome the persistent limitations of Mn–Ce catalysts under harsh flue gas conditions. To further enhance H₂O/SO₂ poisoning resistance, we focus on Ce₈MnO_x, a catalyst known for its high low-temperature deNO_x activity²⁵. By employing Ti-doping to trigger strain engineering with hierarchical coupling of structure-defect-electron-activity-resistance, we activate a series of synergistic effects that favor NH₃-SCR performance and anti-poisoning capability. The ionic radius mismatch between Ti and the host lattice induces lattice contraction and distortion in CeO₂, which promotes the formation of abundant oxygen vacancies, constructs a highly synergistic Ce-O-Ti interface, enhances surface acidity, and shifts the d-band center downward. Collectively, these modifications boost low-temperature activity while mitigating the detrimental adsorption of H₂O and SO₂. In this work, a series of Ti-doped Ce₈MnO_x catalysts was synthesized via a solvothermal method. Using a comprehensive suite of characterization techniques—including HR-TEM, XRD, Raman, NH₃-TPD, H₂-TPR, XPS, TG, FT-IR, and in situ DRIFTS—combined with theoretical calculations, we systematically investigated the influence of Ti-doping on catalyst performance and mechanism. This study provides new insights into designing high-performance catalysts for low-temperature NO_x removal in complex flue gas environments, addressing a critical gap in industrial air pollution control.

Results and discussion

Microstructure and physicochemical properties of Ce₈MnTi_aO_x

To validate the design rationale of optimizing catalytic performance through Ti-induced strain engineering, we synthesized a series of Ce₈MnTi_aO_x catalysts with varying Ti contents using a solvothermal method. Detailed synthesis protocols, performance evaluation metrics, and characterization techniques are provided in the Supplementary Information.

SEM and TEM analyses reveal that the as-prepared catalysts consist of agglomerated nanoparticles with grain sizes ranging from 4 to 10 nm (Fig. 1a–c). High-resolution TEM images (Fig. 1e–h) and the selected area electron diffraction (SAED) pattern (Fig. 1i) display lattice fringes with interplanar spacings of approximately 0.311 and 0.191 nm, corresponding to the (111) and (220) planes of CeO₂, respectively. Additionally, lattice fringes of 0.248 nm are observed, assigned to the (211) plane of Mn₃O₄. Notably, the spacing associated with the CeO₂ (111) plane is reduced compared to that of pristine CeO₂. This suggests that Ti-doping induces lattice contraction-a key indicator of strain engineering. Furthermore, energy-dispersive X-ray spectroscopy (EDS) mapping (Fig. 1d) confirms the homogeneous distribution of Ce, Mn, and Ti elements throughout the catalyst structure, reinforcing the successful incorporation of Ti into the lattice.

Fig. 1. Morphological and structural characterization of Ti-doped Ce₈MnO_x catalysts.

a, b SEM images showing nanoparticle agglomerates. c Bright-Field(BF)-TEM image. d EDS elemental mapping confirming uniform Ce, Mn, and Ti distribution. e–h HR-TEM images revealing lattice fringes and contraction of CeO₂ (111) planes indicative of Ti-induced strain. i SAED pattern of Ce₈MnTi_2.5O_x confirming crystalline phases. The original unedited TEM images are shown in Fig. S2.

XRD analysis (Fig. 2a) confirms that all samples exhibit characteristic diffraction peaks of cubic fluorite CeO₂ (PDF No. 75-0076). No distinct crystalline phases of Mn or Ti oxides are detected, indicating that Mn and Ti species are either highly dispersed as nanocrystals, present in amorphous states²⁶, or incorporated into the CeO₂ lattice. Specifically, as the Ti/Mn molar ratio increases from 1.5 to 2.5, the diffraction peak corresponding to the CeO₂ (111) plane shifts toward higher angles, providing direct evidence of lattice contraction²⁷. Furthermore, the fitting results of the EXAFS fitting parameters at the Ce-L3edge (Table S2), show a clear contraction of the Ce-O bond length upon Ti incorporation (from 2.38 Å in Ce₈MnO_x to 2.30 Å in Ce₈MnTi_2.5O_x). At a Ti/Mn ratio of 3.5, the 2θ position stabilizes, suggesting saturation of Ti-doping. Considering that Ti⁴⁺ shares the same valence state but a smaller ionic radius than Ce^4+25,28, these observations strongly confirm the successful incorporation of Ti⁴⁺ into the CeO₂ lattice, forming a solid solution.

Fig. 2. Structural and textural characterization of Ti-doped Ce₈MnO_x catalysts.

a XRD patterns of Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x powders. The inset shows the enlarged view of the selected region (25°–32°), highlighting the subtle structural variations/peak shifts among the samples. b Raman spectra of Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x catalysts. The boxed regions labeled I and II highlight the characteristic Raman vibrational modes, where region I corresponds to Mn₃O₄ and region II corresponds to CeO₂. c Pore volume and pore size distribution diagrams of Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x catalysts. d N₂ adsorption-desorption isotherm curves for Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x catalysts. Source data are provided as a Source Data file.

Raman spectroscopy (Fig. 2b) provides insights into the microstructural evolution. The band near 275 cm⁻¹ in Ce₈MnO_x corresponds to Mn₃O₄²⁹, associated with MnO₆ octahedral vibrations³⁰ or O–Mn–O bending modes³¹. With increasing Ti content, this peak gradually diminishes, indicating that Ti⁴⁺ incorporation weakens the Mn-related vibrational response. Meanwhile, the F_2g mode of CeO₂ at ~465 cm^−1,32 exhibits decreased intensity and increased FWHM as the Ti/Mn molar ratio increases from 1.5 to 2.5, suggesting the formation of Ce–O–Ti linkages and suppressed crystallite growth³³. A pronounced blue shift of the F_2g band, particularly for Ce₈MnTi_2.5O_x, further confirms lattice contraction and distortion induced by Ti doping^34,35. Such lattice distortion enhances defect-related structural disorder and is accompanied by an intensified band at ~610 cm⁻¹, commonly associated with vacancy-induced vibrational features³⁶.

However, EPR results reveal that the signal intensity of paramagnetic oxygen vacancy centers in Ce₈MnTi_2.5O_x is lower than that in Ce₈MnTi_3.5O_x (Fig. S5). This discrepancy indicates that Raman and EPR probe different types of defect states. Specifically, while Raman reflects overall lattice distortion and vacancy-associated structural disorder, EPR selectively detects isolated paramagnetic oxygen vacancies³⁷. Therefore, the enhanced Raman defect signal at moderate Ti doping (x = 2.5) is more likely related to increased lattice distortion and possibly the formation of vacancy clusters or non-paramagnetic defects, whereas higher Ti content (x = 3.5) favors the generation of more EPR-active isolated vacancies. At excessive Ti loading, the F_2g red shift and reduced Raman defect intensity suggest partial strain relaxation and possible formation of Ti-rich amorphous domains, further supporting a redistribution of oxygen vacancy states rather than a simple change in total vacancy concentration.

BET analysis (Table 1) reveals that compared with Ce₈MnO_x, Ce₈MnTi_aO_x exhibits a significant increase in BET specific surface area, average pore diameter, and pore volume. Furthermore, specific surface area, average pore diameter, and pore volume of Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x all decrease progressively with increasing Ti content. This trend is attributed to Ti-doping, which inhibits CeO₂ grain growth and promotes denser grain packing. Pore size distribution analysis (Fig. 2c) further confirms a marked reduction in mesopore volume upon Ti incorporation. Meanwhile, N₂ adsorption-desorption isotherms (Fig. 2d) for all samples exhibit Type IV profiles with H₃-type hysteresis loops, indicative of slit-like pores formed by the nanoparticle agglomeration³⁸. These textural changes, coupled with structural distortions observed in XRD and Raman analyses, reinforce the role of Ti-doping in tailoring both the microstructure and porosity—key factors for enhancing catalytic performance under harsh conditions.

Table 1.

BET specific surface area, average pore size, and pore volume of Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x

Catalyst	BET surface area (m2 g−1)	Average pore diameter (nm)	Pore volume (cm3 g−1)
Ce8MnOx	144.90	4.82	0.17
Ce8MnTi1.5Ox	160.20	16.03	0.59
Ce8MnTi2.5Ox	152.80	14.91	0.51
Ce8MnTi3.5Ox	130.70	14.67	0.43

The redox properties of Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x were examined using H₂-TPR. As shown in Fig. 3a, all samples exhibit low-temperature reduction peaks (300–500 °C) corresponding to the reduction of surface Ce⁴⁺ to Ce^3+39,40. In contrast, high-temperature peaks (700–750 °C) are associated with deep reduction of lattice oxygen in bulk CeO₂⁴¹. Notably, the undoped Ce₈MnO_x catalyst shows the lowest surface Ce⁴⁺ reduction peak temperature at 378.3 °C. With increasing Ti/Mn molar ratio (1.5 → 3.5), the Ce⁴⁺ reduction peak shifts toward higher temperatures, indicating that Ti-doping reduces the Ce lattice content and induces lattice contraction and distortion, thereby stabilizing Ce–O bonds. A broad and low-intensity hydrogen consumption feature observed in the range of 200–350 °C can be attributed to the reduction of surface-dispersed Mn⁴⁺ species, confirming the redox-active nature of the Mn component even at low concentrations.

Fig. 3. Redox properties and acidity of the catalysts.

a H₂-TPR profiles of Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x catalysts. b NH₃-TPD Profiles of Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x catalysts. c Quantitative distribution of weak, medium, and strong acid sites for Ce₈MnO_x, Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x catalysts. d Ce-Ti oxides of Tanabe’s rule. Source data are provided as a Source Data file.

Surface acidity was further analyzed by NH₃-TPD (Fig. 3b), revealing weak (<300 °C), medium (~450 °C), and strong (>550 °C) acid sites across all catalysts⁴². Notably, the undoped Ce₈MnO_x catalyst exhibits the lowest total acidity among all samples. As the Ti/Mn ratio increases, the total acidity progressively increases (Fig. 3c). This trend may result from electron transfer from Ce to Ti due to Ti’s higher electronegativity, which enhances the electropositivity of Ce sites and strengthens surface Lewis acidity. Moreover, Ti incorporation promotes the formation of Ce-O-Ti interfaces. According to Tanabe’s rules for solid acidity⁴², calculations indicate an excess positive surface charge on Ce₈MnTi_2.5O_x, manifesting as Lewis acidity and NH₃ adsorption (Fig. 3d). This hypothesis warrants further validation through DFT calculations.

Regulation of the electronic properties of CeO₂ by Ti doping

To elucidate the mechanisms by which Ti doping induces lattice contraction, charge redistribution, and their subsequent impacts on surface acidity and reactant adsorption, Density Functional Theory (DFT) calculations were performed to probe the atomic and electronic structure of CeO₂. Models of pristine and Ti-doped CeO₂ were constructed. As shown in Fig. 4a, b, the lattice parameter decreases from 5.490 to 5.454 Å upon Ti⁴⁺ incorporation, providing direct theoretical evidence that substitution of Ce⁴⁺ by smaller Ti⁴⁺ triggers lattice contraction. This result aligns well with XRD and Raman trends and represents the structural basis for subsequent electronic modulation.

Fig. 4. DFT calculations of structural, electronic, and adsorption properties.

a–c Charge distribution structures of pristine CeO₂, Ti-doped CeO₂, and CeO₂ under compressive lattice strain. d–f Density of states (DOS) for the same three models. g, h Calculated NH₃ adsorption energies on CeO₂ (111) and Ti-doped CeO₂ (111) surfaces. i, j Differential charge density maps illustrating electron transfer during NH₃ adsorption. Source data are provided as a Source Data file.

Mulliken population analysis (Fig. 4a, b) indicates that Ti doping increases the charge on Ce sites from (≈1.270 → 1.350) and slightly reduces the charge on some O sites (−0.630 → −0.620). This confirms that Ti, with higher electronegativity, draws electrons from Ce, enhancing Ce’s electropositivity and strengthening Lewis acidity—consistent with NH₃-TPD observations. Surface adsorption calculations (Fig. 4g, h) further show that Ti-doped CeO₂ (111) exhibits higher NH₃ adsorption energy. Differential charge density maps (Fig. 4i, j) visually confirm electron transfer from NH₃ lone pairs to Ce sites, demonstrating that Ti-induced electronic reconfiguration reinforces Lewis acidity and NH₃ adsorption.

Density of states (DOS) analysis (Fig. 4d, e) reveals a significant downshift of the d-band center after Ti doping. To decouple lattice strain from electronic effects, a CeO₂ model with pure compressive strain (no Ti) was examined (Fig. 4c, f). Under strain alone, Ce charge slightly decreases (1.270 → 1.260), and the d-band center shifts upward—opposite to Ti doping effects. This comparison shows that electron redistribution, not lattice strain, drives the d-band downshift. Since the d-band center governs adsorbate–surface orbital coupling⁴³, its downshift of the d-band center signifies a decrease in the average energy of the metal d-orbitals, its downshift lowers the average energy of Ce d-orbitals, weakening orbital hybridization and chemical bonding strength^44,45.

This framework explains the observed phenomena: (i) charge transfer enhances Lewis acidity, promoting NH₃ adsorption and activation for improved deNO_x activity; (ii) the d-band downshift likely reduces adsorption of poisons such as H₂O and SO₂, providing a robust mechanism for enhanced tolerance.

NH₃-SCR catalytic performance of Ce₈MnTi_aO_x catalysts

To evaluate the NH₃-SCR deNO_x performance of Ce₈MnTi_aO_x, a series of catalysts with varying Ti contents was tested (Fig. 5a), and the results are summarized in Fig. 5b–i. As shown in Fig. 5b, Ce₈MnO_x exhibits an activity profile characterized by an initial increase, stabilization, and subsequent decline across 114–375 °C. Specifically, NO_x conversion rises from ~90% at 114 °C to 100% between 225 and 275 °C, followed by a decrease beyond 275 °C due to non-selective NH₃ oxidation. Ti incorporation enhances low-temperature activity (150–218 °C) and significantly broadens the high-temperature window (>275 °C). Based on DFT and NH₃-TPD results, this improvement is attributed to Ti-induced electronic reconfiguration, which strengthens Lewis acidity and facilitates NH₃ adsorption and dissociation²⁵. However, excessive Ti loading reduces low-temperature activity—particularly at Ti/Mn ratios of 0.5–2—due to decreased surface area. At a Ti/Mn ratio of 2.5, optimal catalytic performance is achieved, with Ce₈MnTi_2.5O_x maintaining >90% NO_x conversion over a wide temperature window of 150–335 °C, which can be attributed to a balanced interplay among oxygen vacancy concentration, surface acidity, and surface area. In fact, the high-temperature activity of the catalyst exhibits a moderate positive correlation with the concentration of oxygen vacancies (Fig. 5c). This can be attributed to the fact that during the formation of oxygen vacancies in CeO₂, the exposed Ce³⁺ species on the surface act as Lewis acid sites, which enhance the acidity of the catalyst and thereby improve its high-temperature activity⁴⁶.

Fig. 5. Catalyst performance.

a Schematic and photograph of the catalytic testing setup. b NO_x conversion profiles of Ce₈MnO_x and Ti-doped samples as a function of temperature. The horizontal dashed line indicates the 90% NO_x conversion level. c The relationship between the oxygen vacancy intensity and the rate of NO conversion at 378 °C. The detailed fitting parameters can be found in the Methods section. d DeNO_x activity of Ce₈MnTi_aO_x under 150 °C with 5vol.% H₂O and 100 ppm SO₂. Initial deNO_x activity: Ce₈MnTi_0.5O_x (96.3%), Ce₈MnTi₁O_x (98.1%), Ce₈MnTi_1.5O_x (97.0%), Ce₈MnTi₂O_x (98.5%), Ce₈MnTi_2.5O_x (96.3%), Ce₈MnTi₃O_x (91.7%), Ce₈MnTi_3.5O_x (92.5%). e Comparison of fresh and post-calcination activity for representative catalysts. The horizontal dashed line indicates the 90% NO_x conversion level. f Operational stability of Ce₈MnTi_2.5O_x and Ce₈MnO_x during deNO_x under varying temperatures in high-humidity sulfur-containing atmospheres. Initial deNO_x activity: Ce₈MnO_x-180 °C (94.8%), Ce₈MnTi_2.5O_x-180 °C (98.4%), Ce₈MnTi_2.5O_x-210 °C (99.8%), Ce₈MnTi_2.5O_x-240 °C (100%). g Effect of temperature rise on the long-term SO₂ and H₂O resistance of Ce₈MnTi_2.5O_x catalysts. (Reaction condition = 1000 ppm NO, 1000 ppm NH₃, 10% O₂, N₂ as balance, and GHSV = 30,000 h⁻¹). h Recovery of Ce₈MnTi_2.5O_x activity during stepwise heating in a high-humidity, sulfur-rich atmosphere. i Activity recovery of Ce₈MnTi_2.5O_x catalyst via stepwise temperature cycling under SO₂ and H₂O atmosphere. Reaction condition = 1000 ppm NO, 1000 ppm NH₃, 10% O₂, N₂ as balance, and GHSV = 20,000 h⁻¹. Source data are provided as a Source Data file.

The H₂O/SO₂ resistance of the catalyst is critical for practical application in complex flue gas environments. Under 5 vol.% H₂O and 100 ppm SO₂ at 150 °C, all catalysts experienced significant deactivation (Fig. 5d), with recovery correlated to Ti content. For Ti/Mn ratios of 0.5–2.5, NOₓ conversion dropped sharply within the first hour, then gradually declined to ~60% over 12 h. After removing H₂O and SO₂, activity recovered to 70–80%. For higher Ti/Mn ratios (3–3.5), initial deactivation was slower, but conversion eventually fell to ~50%, recovering to 70–75% post-poison removal. This deactivation is primarily due to physical blockage of active sites and mass-transfer resistance from ammonium sulfate accumulation at low temperatures⁴⁷. Secondary calcination at 450 °C restored full activity for Ce₈MnTi_1.5O_x, Ce₈MnTi_2.5O_x, and Ce₈MnTi_3.5O_x (Fig. 5e), even slightly improving performance at 300–375 °C. This confirms that deactivation is reversible, as ammonium sulfates decompose at elevated temperatures, re-exposing active sites. Long-term tests at higher temperatures (Fig. 5f) further validated this mechanism: Ce₈MnTi_2.5O_x maintained >90% NO_x conversion for 12 h at 180 °C under poisoning conditions and fully recovered afterward. At 210 and 240 °C, activity remained virtually unaffected (98–100% conversion). Given that Mn, Ce, and Ti sulfates decompose above 600 °C^48,49, irreversible sulfation of the metal oxides can be excluded. As shown in the Fig. 5g, the Ce₈MnTi_2.5O_x catalyst still exhibited excellent stability under these harsh conditions: After introducing 100 ppm SO₂ and 10 vol.% H₂O at 210 °C, the NO_x conversion slightly decreased from the initial ~98% to ~90% within the first 1 h, and then remained remarkably stable at ~90% for the subsequent 23 h without any further deactivation. When the temperature was temporarily increased to 240 °C, the NO_x conversion rapidly recovered to 100%, indicating that the slight deactivation was caused by the deposition of ammonium sulfate salts rather than irreversible sulfation of active sites. When the temperature was returned to 210 °C, the NO_x conversion recovered to ~90%, consistent with the stable value before the temperature increase. After cutting off SO₂ and H₂O, the NO_x conversion completely recovered to the initial 100%.

DFT calculations provide a mechanistic explanation for the superior H₂O/SO₂ resistance of Ti-doped Ce₈MnTi_2.5O_x. Ti-induced electron redistribution drives a downshift of the CeO₂ d-band center, weakening competitive adsorption of H₂O and SO₂ and enhancing intrinsic tolerance. This electronic effect, combined with reversible physical coverage, explains the catalyst’s excellent online thermal regeneration capability. Dynamic stepwise heating experiments (150 °C → 180 °C → 210 °C → 240 °C) further demonstrated (Fig. 5h) that even under 200 ppm SO₂, activity—initially suppressed by high humidity and sulfur—gradually recovered with increasing temperature. Above 210 °C, Ce₈MnTi_2.5O_x achieved rapid, complete online recovery to fresh catalyst levels. However, these single-direction heating experiments cannot confirm whether the active sites were fully restored. To resolve this ambiguity and provide definitive evidence for the reversible nature of deactivation, we performed a complete temperature cycling experiment incorporating a controlled cooling step following sulfate decomposition at 240 °C (Fig. 5i). Specifically, after maintaining full NO_x conversion at 240 °C under SO₂ and H₂O atmosphere, the reaction temperature was gradually decreased stepwise back to 180 °C. Remarkably, the NO_x conversion at 180 °C after the high-temperature treatment remained nearly identical to that of the fresh catalyst at the same temperature. Subsequent reheating to 210 °C immediately restored NO_x conversion to the fresh catalyst level. This fully reversible activity profile unequivocally demonstrates that no irreversible sulfation of active Ce or Mn sites occurred during the entire poisoning process. Instead, deactivation was solely caused by the reversible deposition of ammonium sulfate species, which can be completely decomposed and removed by heating to ≥ 210 °C. This adaptability to fluctuating flue gas temperatures eliminates the need for shutdown and offline regeneration required by commercial catalysts, underscoring its significant practical value.

Thermal regeneration mechanism and reaction pathway of the Ce₈MnTi_aO_x catalyst

To elucidate the deactivation and regeneration mechanisms of Ce₈MnTi_2.5O_x under H₂O/SO₂ atmospheres, thermogravimetric (TG) analysis was performed on samples after reaction at 150 °C and 240 °C (Fig. 6a). The poisoned catalysts exhibited weight loss of 4.5% (150 °C) and 3.0% (240 °C), indicating greater accumulation of physisorbed water and deposited ammonium sulfates at lower temperatures. Stepwise mass-loss analysis shows that 30–200 °C corresponds to evaporation of physisorbed water⁵⁰, while the 200–450 °C range reflects decomposition of ammonium sulfates and release of chemisorbed water⁵¹; mass loss stabilizes above 450 °C. These results indicate that low-temperature activity decline originates from competitive water adsorption and physical coverage by ammonium sulfates.

Fig. 6. Characterization of surface properties and reaction behavior of Ce₈MnTi_2.5O_x catalysts.

a Thermogravimetric (TG) and derivative thermogravimetric (DTG) curves of Ce₈MnTi_2.5O_x catalysts after reaction at 150 °C and 240 °C. b O 1s XPS spectra of Ce₈MnO_x, Ce₈MnTi_2.5O_x and Ce₈MnTi_2.5O_x-150. c Ce 3 d spectra of the samples, with the surface atomic ratios of Ce³⁺/Ce⁴⁺ calculated from the integrated peak areas. The eight peaks are labeled as v, u, v’, u’, v”, u”, v”’, and u”’. Peaks labeled as v’ and u’ (in red) are the characteristic peaks of the Ce³⁺ oxidation state, while the remaining peaks represent the Ce⁴⁺ oxidation state. d Mn 2p spectra of the samples, with the Mn³⁺/Mn⁴⁺ ratios determined from the peak area analysis. e Ti 2p spectra of the samples. The detailed fitting parameters, peak area values, and the calculation methodology for the metal species ratios are provided in the Supplementary Information. f FT-IR spectra of Ce₈MnTi_2.5O_x under sulfide poisoning at different temperatures. g, h In situ DRIFTS spectrum of Ce₈MnTi_2.5O_x catalyst after reaction at 150 and 240 °C. Source data are provided as a Source Data file.

Comparative XPS analysis (Fig. 6b–e and Table 2) reveals the evolution of surface chemical states of Ce₈MnTi_2.5O_x before and after exposure to H₂O/SO₂. O1s fitting shows that the fraction of surface-adsorbed oxygen (O_α) at ~531.4 eV increases from 30.2% to 36.4% after poisoning, accompanied by a decrease in lattice oxygen (O_β) at ~529.2 eV, likely associated with sulfur-containing surface species (e.g., sulfites, ammonium bisulfates)⁵². Ce3 d and Mn 2p spectra indicate changes at the redox-active sites: the Ce³⁺/(Ce³⁺+Ce⁴⁺) ratio decreases from 21.2% to 18.2%, and Mn⁴⁺/(Mn²⁺ + Mn³⁺ + Mn⁴⁺) decreases from 19.3% to 16.8%, with a concomitant increase in Mn²⁺. These changes suggest inhibition of the Ce³⁺ + Mn⁴⁺ ↔ Ce⁴⁺ + Mn³⁺ redox cycle⁵³, primarily due to physical coverage of active sites by ammonium sulfate deposits⁵⁴, which hinder contact between Mn centers and gas-phase O₂, delaying reoxidation of Mn²⁺ and thereby interrupting the catalytic cycle. Notably, Ti doping induces CeO₂ lattice distortion and promotes oxygen-vacancy formation, yielding a higher O_α fraction in Ce₈MnTi_2.5O_x than in Ce₈MnO_x. These abundant active oxygen species can efficiently activate gas-phase O₂, enhance NO_x conversion, and compete with ammonium sulfates for surface sites to facilitate their decomposition, thereby imparting thermal regeneration capability.

Table 2.

Relative concentration ratios of O, Ce, Mn, and Ti ions on the surface of Ce₈MnTi_2.5O_x and Ce₈MnTi_2.5O_x-150 catalysts

Catalyst	Mn4+ (%)	Mn3+ (%)	Mn2+ (%)	Ce4+ (%)	Ce3+ (%)	Ti4+ (%)	Oα (%)
Ce8MnOx	13.1	26.6	60.3	76.4	23.6	-	20.7
Ce8MnTi2.5Ox	19.3	27.7	53	78.8	21.2	100	30.2
Ce8MnTi2.5Ox-150	16.8	25.9	57.3	81.8	18.2	100	36.4

The detailed values of the binding energies and peak areas are listed in Table S4.

Further comparison of the Ti 2p XPS spectra of Ce₈MnTi_2.5O_x after long-term H₂O/SO₂ exposure at 150 °C (Ce₈MnTi_2.5O_x-150) with the fresh catalyst shows that Ti 2p_3/2 (457.9 and 458.7 eV) and Ti 2p_1/2 (463.6 and 463.9 eV) shift to higher binding energy, indicating lower local electron density around Ti atoms. This shift is attributed to the electron-withdrawing influence of negatively charged sulfur/oxygen-containing deposits, which increase core-level binding energies⁵⁵. Because SO₂ chemisorption on metal centers is electrophilic and requires electron donation from the metal⁵⁶, the cooperative valence changes observed for Ce and Mn—paired with an essentially unchanged Ti valence—suggest that under reaction conditions, charge transfer occurs preferentially between Ce and Mn, while significant electron transfer from metal sites to SO₂ is unfavorable. Consequently, formation of stable, chemically adsorbed metal sulfates is unlikely. The activity deterioration at 150 °C is thus primarily attributed to physical coverage by ammonium sulfate deposits, which impede mass transfer and, more critically, suppress Mn-centered redox cycling.

To probe the chemical nature and temperature-dependent evolution of surface deposits under H₂O/SO₂, FT-IR spectra were collected for samples reacted at 150, 180, 210, and 240 °C in 100 ppm SO₂ and 5 vol.% H₂O (Fig. 6f). A band at 1310 cm⁻¹ assigned to NH₃ adsorption^57,58 remains relatively constant from 150–240 °C, consistent with NH₃-TPD evidence for strong surface acidity with minimal temperature dependence. The 1530 cm⁻¹ band corresponds to bridged/bidentate nitrate species—key NH₃-SCR intermediates⁵⁹, and the ~1640 cm⁻¹ band arises from δ(H–O–H) bending of adsorbed water⁶⁰. At 150 °C, weak peaks at 893 and 869 cm⁻¹ indicate HSO₄^- (ammonium bisulfate)⁶¹, while features at 1020 and 1066 cm^-1 correspond to SO₃^2- (ammonium sulfite)²⁵. These findings show that small amounts of ABS and ammonium sulfite accumulate under moisture/sulfur co-feeding at 150 °C, driving the activity decline at low temperature.

With increasing temperature (150 → 240 °C), the SO₃²⁻ and HSO₄⁻ bands gradually weaken and disappear by 180 °C, demonstrating temperature-promoted regeneration of Ce₈MnTi_2.5O_x. Concurrently, sub-800 cm⁻¹ bands assigned to Mn–O/Ce–O stretching intensify with temperature^62,63, indicating that thermal decomposition of ammonium sulfite/ABS overlayers re-exposes metal-oxide active sites. These FT-IR results confirm that deactivation in sulfur/moisture atmospheres is primarily reversible physical poisoning; deposited ammonium salts decompose rapidly above ~180 °C, restoring activity to fresh-catalyst levels.

To clarify NH₃-SCR pathways and the anti-poisoning online thermal regeneration mechanism under H₂O/SO₂, in situ DRIFTS was conducted at 150 and 240 °C (Fig. 6g, h) under 1000 ppm NH₃ + 1000 ppm NO + 10% O₂ + 5 vol.% H₂O + 100 ppm SO₂. Peaks at 1558/1561 cm⁻¹ (coordinated NH₃ on Lewis acid sites) and 1315/1319 cm⁻¹ (NH₄⁺ on Brønsted acid sites) are observed⁶⁴, with intensities increasing over time, indicating progressive NH₃ adsorption and activation on both acid site types. The persistence and growth of these features—particularly at 240 °C—support a mechanism in which Ti-induced Lewis acidity and accessible Brønsted sites jointly sustain SCR activity while temperature facilitates removal of sulfate overlayers, enabling online regeneration.

Further analysis of reaction pathways at different temperatures reveals distinct mechanistic shifts. At 150 °C, the intensities of bridged nitrates (1623 cm⁻¹) and bidentate nitrates (1540 cm⁻¹) increase over time⁶⁵, while monodentate nitrates (1377, 1402, 1441, 1489 cm⁻¹) gradually weaken^66,67. This indicates that active NH₃ species react with monodentate nitrates via the Langmuir–Hinshelwood (L–H) mechanism^25,68. At 240 °C, the intensities of monodentate nitrates (1377 cm⁻¹), monodentate nitrites (1397, 1446 cm⁻¹)⁶⁹, and bidentate nitrates (1023, 1519, 1588–1618 cm⁻¹) all increase^66,70, suggesting that surface nitrates do not participate in the reaction. Instead, NO reacts directly with adsorbed NH₃, following the Eley–Rideal (E–R) mechanism. Thus, under H₂O/SO₂ conditions, NH₃-SCR over Ce₈MnTi_2.5O_x is dominated by L–H at low temperature and E–R at high temperature. To probe the thermal regeneration mechanism against water and sulfur, attention was given to ammonium salt peaks (1036, 1136, 1189, 1020–1134, 1179 cm⁻¹)⁷¹. In the Ce₈MnTi_2.5O_x-150 sample, these peaks intensify and accumulate over time, indicating continuous deposition. Conversely, in Ce₈MnTi_2.5O_x-240, the same peaks weaken, confirming dynamic decomposition of deposited salts at elevated temperature. This progressive removal re-exposes active sites, enabling efficient and stable deNO_x activity in sulfur- and moisture-rich atmospheres.

To further compare the competitive adsorption of NH₃, H₂O, and SO₂ species on the catalyst surface, the evolution of surface adsorbed species during this process was investigated via stepwise introduction of NH₃, NO + O₂, H₂O, and SO₂ (Fig. S9). The experimental procedure was as follows: pure NH₃ was introduced for 15 min (0–15 min) to achieve adsorption saturation; at 15 min, the gas stream was switched to NH₃ + NO + O₂ mixed atmosphere for 20 min of reaction; subsequently, H₂O was introduced to form NH₃ + NO + O₂ + H₂O atmosphere for another 20 min of reaction; finally, SO₂ was introduced to evaluate the sulfur resistance performance.

During the pure NH₃ adsorption process, the peak appearing at 1350 cm⁻¹ and gradually shifting to 1375 cm⁻¹ was assigned to -NH₂ species adsorbed on Lewis acid sites⁷², while the peak at 1558 cm⁻¹ was attributed to NH₃ coordinated to Lewis acid sites. The intensities of both peaks increased continuously with adsorption time, indicating that NH₃ molecules could be stably adsorbed on the catalyst surface and gradually reach saturation.

Upon introduction of NO + O₂, the intensities of NH₃ adsorption peaks at 1350 and 1558 cm⁻¹ continued to rise, suggesting that the adsorption rate of NH₃ was still higher than its reaction consumption rate with NO_x at this stage. Meanwhile, a series of new characteristic peaks emerged in the 1000–1500 cm⁻¹ region: the peak at 1016 cm⁻¹ was assigned to bidentate nitrate species⁷³, the peak at 1173 cm⁻¹ to surface-adsorbed NO^- species⁷⁴, and the peak at 1258 cm⁻¹ to bridged nitrate species²⁵. The appearance of these nitrate species indicated that NO was oxidized to NO₂ by surface active oxygen species in the presence of O₂, and further combined with surface sites to form adsorbed nitrates, which gradually accumulated on the surface with prolonged reaction time.

After introducing H₂O, a peak at 3545 cm⁻¹ assigned to surface-adsorbed hydroxyl groups was observed⁷⁵. In addition, the peaks attributed to NH₃ adsorption were further enhanced. This was because the dissociation of H₂O on the catalyst surface generated hydroxyl groups, which promoted the formation of Brønsted acid sites and thus significantly enhanced the NH₃ adsorption capacity⁷⁶. Meanwhile, the adsorption peak intensities of NO^- species, bridged nitrates, and bidentate nitrates further increased and reached their maximum values. This was attributed to the fact that the abundant surface -OH groups generated by H₂O dissociation greatly promoted the NO oxidation process and the formation of nitrate species. At this point, the formation rate of nitrates was much higher than their reaction consumption rate with NH₄⁺, leading to a significant accumulation of surface nitrate concentration.

Upon introduction of SO₂, the intensities of NH₃ adsorption peaks at 1350 and 1558 cm⁻¹ continued to increase sharply. This was due to the reaction of SO₂ with surface-adsorbed NH₃ and H₂O to form sulfur-containing ammonium salt species such as ammonium bisulfate and ammonium sulfate. These species themselves possess strong acidity and can adsorb more NH₃, resulting in the significant increase in peak intensities here. In contrast, the peak intensity of NO^- species at 1173 cm⁻¹ decreased significantly, because SO₂ had a stronger competitive adsorption capacity on the catalyst surface and preferentially occupied the adsorption sites of NO^- species. However, the peak intensities of bidentate and bridged nitrate species were further enhanced instead. This was because these two types of nitrate species were more stable, and the competitive adsorption of SO₂ forced the adsorption of NO_x to shift from vulnerable adsorption states to the formation of relatively stable bridged and bidentate nitrates that were less susceptible to SO₂ interference⁷⁷.

In this study, Ti-doped Ce₈MnO_x catalysts (Ce₈MnTi_aO_x) were synthesized via a solvothermal method to investigate the effect of Ti incorporation on low-temperature NH₃-SCR activity and resistance to H₂O/SO₂ poisoning. The optimized Ce₈MnTi_2.5O_x catalyst demonstrated enhanced performance, achieving >90% NOₓ conversion across 150–335 °C. Under harsh conditions (100 ppm SO₂ and 5 vol.% H₂O at 180 °C), activity remained above 90% for 12 h and fully recovered after poison removal. At 210 °C under 200 ppm SO₂, the catalyst maintained full activity, confirming its online, temperature-driven regeneration capability.

Mechanistic analysis revealed that Ti incorporation into the CeO₂ lattice induces lattice contraction and distortion, generating abundant oxygen vacancies and forming Ce–O–Ti interfaces with strong synergistic effects. These structural changes enhance Lewis acidity and facilitate NH₃ adsorption and activation. DFT calculations further show that Ti doping triggers electronic reconfiguration and a downshift of the d-band center, which not only promotes NH₃ activation but also suppresses competitive adsorption of H₂O and SO₂. TG and FT-IR analyses confirm that low-temperature deactivation arises from reversible physical deposition of ammonium sulfates and water adsorption, rather than irreversible metal sulfation. Above 210 °C, these deposits decompose rapidly, re-exposing active sites and enabling autonomous regeneration.

In summary, this work establishes a multi-level strain-engineering strategy—coupling structure, defects, electronic modulation, catalytic activity, and poisoning resistance—to design efficient, low-temperature NH₃-SCR catalysts with robust tolerance to water and sulfur. These findings provide a valuable blueprint for developing next-generation deNO_x catalysts for real-world flue gas applications.

Methods

Materials and reagents

All chemicals used in this study were of analytical grade and used as received without further purification. Cerium(III) nitrate hexahydrate (Ce(NO₃)₃·6H₂O, Aladdin, ≥99.95%), manganese acetate tetrahydrate (MnC₄H₆O₄·4H₂O, Hushi, ≥99.0%), absolute ethanol (Yasheng, ≥99.7%), acetic acid (C₂H₄O₂, Hushi, ≥99.5%), and titanium(IV) butoxide (C₁₆H₃₆O₄Ti, Hushi, ≥98.0%) were utilized for the catalyst synthesis.

Catalyst preparation

The Ce₈MnTi_aO_x catalysts were synthesized via a solvothermal method. Initially, 64 mL of absolute ethanol was measured into a 100 mL beaker, followed by the addition of 48 mmol Ce(NO₃)₃·6H₂O and 6 mmol MnC₄H₆O₄·4H₂O into the solvent. The mixture was stirred for 60 min to obtain a transparent and homogeneous solution. Subsequently, 10 mL of C₂H₄O₂ was added to the above solution under continuous stirring for 10 min until uniformity was achieved. Different amounts of C₁₆H₃₆O₄Ti were then introduced into the mixture, which was stirred for an additional 20 min. The resulting mixture was transferred into a 100 mL Teflon-lined autoclave and maintained at 180 °C for 12 h. After reaction, the precipitate was collected by centrifugation and washed several times with absolute ethanol. The washed precipitate was dried at 80 °C for 12 h in a vacuum oven. The dried sample was subsequently calcined in a muffle furnace at 450 °C for 2 h under static air atmosphere, with a heating rate of 2 °C/min, yielding the final Ce₈MnTi_aO_x catalysts (where “a” = 0, 0.5, 1, 1.5, 2, 2.5, 3, and 3.5). For comparison, the Ce₈MnO_x sample was prepared following the identical procedure, except that only Ce(NO₃)₃·6H₂O and MnC₄H₆O₄·4H₂O were added during the synthesis.

The Ce₈MnTi_aO_x catalyst was subjected to a water-sulfur resistance test by passing 100 ppm SO₂ and 5 vol.% H₂O through it. The resulting catalyst was designated as Ce₈MnTi_aO_x-T (where “T” represents the water-sulfur resistance test temperature).

To prepare the Ce₈MnTi_2.5O_x-T catalyst after thermal regeneration, the catalyst was placed in a muffle furnace and heated at a rate of 2 °C/min to 450 °C under an air atmosphere. It was then held at this temperature for 2 h. The resulting catalyst was designated as Ce₈MnTi_2.5O_x-T-450.

NH₃ selective catalytic reduction of NO_x test

The NH₃-SCR reaction for NO_x removal over the samples was conducted in a laboratory fixed-bed reactor, as shown in Fig. S1. A 1.8 mL sample with a particle size of 20 ~ 40 mesh was placed into a quartz reaction tube with an inner diameter of 8 mm. To simulate the complex flue gas composition of ultra-low-temperature, high humidity, and SO₂-containing conditions, the inlet gas components and their concentrations were configured as follows: [NO]_in: 1000 ppm, [NH₃]_in: 1000 ppm, [O₂]_in: 10 vol.%, [SO₂]_in: 100 ppm (when used), [H₂O]_in: 5 vol.% (when used), N₂ was used as the balance gas, and the gas hourly space velocity (GHSV) of the reaction was set to 20,000 h⁻¹. A flue gas analyzer (MRU VarioPlus, Germany) was used to online detect the concentrations of NO, NO₂, and NO_x at the inlet and outlet of the reactor. Data were collected and recorded after the reaction stabilized.

The NO_x conversion rate of the catalyst during the NH₃-SCR deNO_x reaction was calculated by equation (1):

$${\text{NO}}^{\mathrm{x}}\mathrm{conversion}\left(\%\right)=\frac{{\left[{\text{NO}}^{\mathrm{x}}\right]}_{\text{in}}-{\left[{\text{NO}}^{\mathrm{x}}\right]}_{\text{out}}}{{\left[{\text{NO}}^{\mathrm{x}}\right]}_{\text{in}}}\times 100\%$$ 1

The flue gas analyzer (MRU VarioPlus, Germany) was periodically calibrated by the manufacturer, ensuring a detection accuracy of up to 1 ppm. During the catalytic tests, the main source of error arises from slight fluctuations in the gas flow. Although a gas mixing tank was employed to ensure sufficient mixing of all gas streams prior to entering the fixed-bed quartz reactor, a fluctuation of NO_x concentration within 5 ppm was observed under practical operation conditions. Considering an initial NO_x concentration of 1000 ppm, the corresponding uncertainty in NO_x conversion is estimated to be within ±0.5%. Given the relatively high gas hourly space velocity and the limited conversion uncertainty (±0.5%), the reaction can be reasonably considered to proceed under kinetically controlled conditions.

The concentrations of NO_x, NH₃, and N₂O at the inlet and outlet of the reactor were measured by an FTIR spectrometer Protea atmos-FIRt (AFS-B2T-C-1911). The N₂ selectivity were calculated by Eq. (2):

$${\text{N}}_2\text{selectivity}\left(\%\right)=\left(1-\frac{2{\left[{\text{N}}_2\text{O}\right]}_{\text{out}}}{{\left[\text{N}{\text{O}}_{\text{x}}\right]}_{\text{in}}+{\left[\text{N}{\text{H}}_3\right]}_{\text{in}}-{\left[{\text{NO}}_{\text{x}}\right]}_{\text{out}}-{\left[{\text{NH}}_3\right]}_{\text{out}}}\right)\times 100\%$$ 2

Linear regression of the oxygen vacancy intensity and the rate of NO conversion

A simple linear regression was performed to analyze the relationship between the oxygen vacancy intensity and the rate of NO conversion at 378 °C, as shown in Fig. 5c. The coefficient of determination (R²) was 0.81258. The standard error of the intercept and the slope were 0.09182 and 0.14128, respectively.

Statistics and reproducibility

All experiments described in this study were performed at least three times independently with similar results. Representative data, including micrographs, were selected from these independent experiments to ensure the consistency and reproducibility of the reported findings.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Yuesong Shen (Y.S.S.) conceived and supervised the project. Y.S.S. and Xuewen Guo (X.W.G.) contributed to the methodology. X.W.G. and Yongji Hu (Y.J.H.) wrote the original draft. X.W.G. performed data curation and validation. Y.J.H. conducted the investigation and software analysis. Xiaoyi Wu (X.Y.W.) and Y.S.S. reviewed and edited the manuscript. Y.S.S. acquired the funding and provided resources. All authors discussed the results and contributed to the final manuscript. Y.S.S. is the corresponding and submitting author.

Peer review

Peer review information

Nature Communications thanks Roger Glaeser, who co-reviewed with Michael Liebau, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

The work in Nanjing was financially supported by the National Key Research and Development Plan Subject (2021YFB3500603), the State Key Laboratory of Materials-Oriented Chemical Engineering (SKL-MCE-25A08), the Key Research and Development Plan of Jiangsu Province (Social Development, BE2021713), the Six Talent Peaks Project of Jiangsu Province (JNHB-044), the Qinglan Project of Jiangsu Province, the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

Data availability

All data generated or analyzed during this study are included in this published article (and its supplementary information files). Source data are provided with this paper. The DFT calculation of structural data of Figs. 4a– c g, h and S3a, b are included in Supplementary Dataset 1. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76719-6.