Efficient and durable light-alkane oxidation over sintered Pt catalysts Authors: Xuan Tang, Yang You, Lei Ying, Beien Zhu, Wangcheng Zhan, Yun Guo, Sheng Dai Published: 2026 ABSTRACT Nanoparticle sintering is typically regarded as a deactivation mechanism for supported metal catalysts, motivating efforts to maximize metal dispersion. Here we demonstrate the opposite trend for platinum catalysts in light-alkane oxidation. Intentionally pre-sintered platinum particles, tens of nanometers in size and supported on thermally stable magnesium aluminate, show higher activity than highly dispersed platinum species for propane oxidation. Theory-guided adsorption calculations suggest that oxygen-resistant metallic facets prevalent on large platinum particles mitigate oxygen poisoning, a concept validated by controlled calcination. The sintered catalyst delivers a turnover frequency of 0.1625 s−1 at 220 °C, 116-fold higher than platinum nanoclusters, and lowers the temperature for 90% propane conversion from 360 to 245 °C. It remains stable for 48 hours in dry and water-containing feeds and after hydrothermal aging. Mechanistic analyses reveal that low-index metallic facets sustain propane activation, while oxygen-rich feeds drive reconstruction toward less active stepped surfaces. Introduction Light alkanes are major constituents of liquefied petroleum gas (LPG) and also relevant pollutants in automotive exhaust and volatile organic compounds (VOCs)1. Low-temperature catalytic complete oxidation of light alkanes is an effective route for gas-turbine power generation and emission abatement2,3. Among them, propane oxidation is particularly challenging because of propane’s intrinsic structural stability4. Developing catalysts that are simultaneously active and durable for propane oxidation is therefore crucial for advancing toward cleaner and more efficient oxidation/abatement technologies5–7. Supported Pt catalysts represent one of the most effective systems for this purpose8–10, since C–H bond activation in propane oxidation predominantly occurs at Pt sites11–13. The geometric and electronic properties of Pt species jointly determine catalytic performance and cost efficiency. Consistent with the common belief that sintering causes deactivation14–16, most prior work has focused on maximizing Pt dispersion and implementing anti-sintering strategies to increase metal utilization for oxidation reaction17–21. Notably, particle downsizing not only reshapes geometry but also changes electronic structure via finite-size effects and metal-support charge transfer22–25. For example, smaller Pt species often become less electron-rich since electron transfer from Pt to oxide supports renders Pt increasingly electron-deficient26. However, efficient propane activation requires electron back-donation from Pt d orbitals into the antibonding orbitals of C–H bonds, a process favored on electron-rich Pt sites27. Moreover, under oxidative conditions, strong oxygen adsorption can inhibit propane adsorption and activation, leading to oxygen poisoning of Pt surfaces (conceptually analogous to CO poisoning). While highly dispersed Pt species maximize surface exposure, their abundant undercoordinated sites and finite-size effects can strengthen oxygen binding, promote surface oxidation, and reduce the availability of metallic Pt ensembles required for propane adsorption and C–H activation, thereby paradoxically limiting their intrinsic activity in complete propane oxidation28. This creates a challenge in designing efficient and durable Pt-based catalysts. Herein, we propose an unconventional design strategy that employs sintered Pt catalysts to create extended low-index Pt surfaces, thereby lowering oxygen coverage and enhancing propane oxidation performance. Guided by theoretical calculation, we fabricate sintered Pt particles supported on MgAl2O4 (Pt/MAO) via controlled thermal annealing and investigate their catalytic performance and dynamic surface evolution during propane oxidation. Remarkably, the sintered Pt/MAO catalyst exhibits significant improvements in activity, water resistance, and durability, particularly, achieving a turnover frequency (TOF) up to 116 times higher than that of Pt nanoclusters. The Pt/MAO catalyst pre-sintered at 800 °C inherently maintains excellent stability under harsh conditions, including hydrothermal aging and prolonged operation. Through combined in situ transmission electron microscopy (TEM) characterization and theoretical analysis, we demonstrate that the extended low-index facets on sintered Pt particles enhance oxidation resistance under stoichiometric propane oxidation conditions: they weaken oxygen adsorption and preserve a metallic surface, synergistically ensuring high activity and stability. Finally, this sintering strategy can be applied to upgrade commercial Pt/γ-Al2O3 catalysts and extended to other light-alkane oxidations, such as ethane and n-butane, demonstrating its broad applicability. Results Oxygen-coverage effects on propane adsorption and activation Previous theoretical studies have highlighted the pivotal role of propane (C3H8) adsorption in the propane oxidation reaction, as well as the influence of oxygen-induced inhibition28–30. To gain deeper insights, we performed density functional theory (DFT) calculations to quantify the adsorption energetics of C3H8 on Pt surfaces under different oxygen coverages (see Table 1; additional calculation details are provided in the Supplementary Information). The corresponding optimized adsorption structures are displayed in Supplementary Figs. 1–5. The results show that pre-adsorbed oxygen generally weakens propane adsorption, with C3H8 adsorption energies becoming markedly less exothermic on O-precovered Pt surfaces, consistent with the oxygen-inhibition effect reported for propane oxidation31.Table 1Adsorption energies of C3H8 and O on representative Pt surfaces and the Pt38 clusterAdsorption Energy (eV)Pt(100)Pt(111)Pt(211)Pt38\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{\mathrm{O}}$$\end{document}EadsO on clean model−1.0356−0.9856−1.27−1.36\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EadsC3H8 on clean model−0.64−0.62−0.53−0.35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EadsC3H8 on moderate O-precovered surface−0.45−0.39−0.26−0.34\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EadsC3H8 on high O-precovered surface−0.34−0.39−0.34−0.33For extended Pt slabs, the moderate and high O-precovered models correspond to 0.5 ML and 1.0 ML O*, respectively. For the Pt38 cluster, the moderate and high O-precovered models correspond to 6 and 12 pre-adsorbed O atoms, respectively. Table 1Adsorption energies of C3H8 and O on representative Pt surfaces and the Pt38 clusterAdsorption Energy (eV)Pt(100)Pt(111)Pt(211)Pt38\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{\mathrm{O}}$$\end{document}EadsO on clean model−1.0356−0.9856−1.27−1.36\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EadsC3H8 on clean model−0.64−0.62−0.53−0.35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EadsC3H8 on moderate O-precovered surface−0.45−0.39−0.26−0.34\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EadsC3H8 on high O-precovered surface−0.34−0.39−0.34−0.33For extended Pt slabs, the moderate and high O-precovered models correspond to 0.5 ML and 1.0 ML O*, respectively. For the Pt38 cluster, the moderate and high O-precovered models correspond to 6 and 12 pre-adsorbed O atoms, respectively. Adsorption energies of C3H8 and O on representative Pt surfaces and the Pt38 cluster For extended Pt slabs, the moderate and high O-precovered models correspond to 0.5 ML and 1.0 ML O*, respectively. For the Pt38 cluster, the moderate and high O-precovered models correspond to 6 and 12 pre-adsorbed O atoms, respectively. Importantly, the calculations further reveal a pronounced facet- and size-dependence in oxygen binding. For example, high-index stepped Pt(211) binds oxygen more strongly (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{\mathrm{O}}$$\end{document}EadsO = −1.27 eV) than low-index facets such as Pt(100) and Pt(111) (−1.03 and −0.98 eV, respectively). Moreover, a Pt38 nanocluster, representative of ~1.0 nm Pt species, exhibits even stronger oxygen adsorption (−1.36 eV), consistent with finite-size effects reported in the literature32. The stronger oxygen affinity on Pt stepped sites and nanoclusters favors high oxygen surface coverages and facilitates partial oxidation, which in turn suppresses the availability of metallic Pt required for effective C3H8 adsorption and C–H activation. We further calculated the activation barriers for the first C–H bond cleavage of propane on representative Pt models (Supplementary Table 1). On the clean models, the calculated barriers are 0.33 eV on Pt(100), 0.57 eV on Pt(111), 0.26 eV on Pt(211), and 0.24 eV on Pt38 (Supplementary Fig. 6), indicating that undercoordinated Pt sites are intrinsically favorable for initial C–H activation33. However, this kinetic advantage is strongly affected by oxygen precoverage. On the high O-precovered models: the barrier increases to 0.56 eV on high O-precovered Pt(211) and to 1.07 eV on Pt38, whereas Pt(100) shows only a moderate increase from 0.33 to 0.39 eV (Supplementary Fig. 7). These results indicate that oxygen occupation particularly suppresses the reactivity of undercoordinated Pt sites. In addition, our Wulff construction analysis reveals that larger Pt particles (> 20 nm) possess a significantly higher proportion of low-index terrace sites, dominated by (100) and (111) facets, while the fraction of high-index sites such as edges and corners markedly decreases (Supplementary Fig. 8). This structural feature leads to weaker oxygen binding on such terraces, which helps preserve an oxygen-resistant, metallic Pt surface, as supported by our DFT calculations. Such a favorable surface configuration promotes stronger propane adsorption and more efficient activation, thereby providing a basis for enhanced catalytic performance in propane oxidation. Thus, we consider that large Pt particles are likely to outperform highly dispersed Pt species in this reaction, which is in contrast with the conventional understanding. Synthesis of sintered Pt catalysts We intentionally synthesized larger Pt nanoparticle catalysts using controlled thermal annealing. The spinel MgAl2O4 (MAO) was selected as the support owing to its chemical stability under high-reaction temperature conditions. The MAO support was first prepared via the solvothermal hydrolysis of aluminum isopropoxide with magnesium nitrate hexahydrate. Then, the Pt(NO3)2 precursor was deposited on MAO using the incipient wetness impregnation method with a nominal Pt loading of 1.0 wt.%. Subsequently, the Pt/MAO catalyst was calcined in air at 500 °C for 5 h to remove any remaining ligands, denoted as Ptn/MAO (Pt nanoclusters/nanoparticles). Notably, the Ptn/MAO catalyst was further calcined at 650 °C and 800 °C for 12 h, respectively, to obtain Ptp/MAO (Pt particles) and Pts/MAO (sintered particles), as illustrated in Fig. 1a. The actual Pt loadings of all three Pt/MAO catalysts were very close, ranging from 0.87 to 0.88 wt.% (Supplementary Table 2).Fig. 1Structure characterization of Pt/MAO catalysts.a Illustration of intentional sintering of Pt/MAO catalyst. b–j Representative HAADF-STEM images of b, c Ptn/MAO, e, f Ptp/MAO, and h, i Pts/MAO. Yellow circles identify the Pt nanoclusters. Pt size statistics of d Ptn/MAO, g Ptp/MAO, and j Pts/MAO. k XRD patterns and l enlarged XRD region in k of pure MAO and Pt/MAO catalysts. Structure characterization of Pt/MAO catalysts. a Illustration of intentional sintering of Pt/MAO catalyst. b–j Representative HAADF-STEM images of b, c Ptn/MAO, e, f Ptp/MAO, and h, i Pts/MAO. Yellow circles identify the Pt nanoclusters. Pt size statistics of d Ptn/MAO, g Ptp/MAO, and j Pts/MAO. k XRD patterns and l enlarged XRD region in k of pure MAO and Pt/MAO catalysts. The morphology of the Pt/MAO catalysts under different annealing temperatures was first investigated using aberration-corrected scanning transmission electron microscopy (AC-STEM). Figure 1b–j presents representative high-angle annular dark field-STEM (HAADF-STEM) images of these catalysts. In the pristine Ptn/MAO catalyst, highly dispersed Pt nanoclusters/nanoparticles were predominant (Fig. 1b, c), exhibiting an average size of 1.0 ± 0.4 nm (Fig. 1d). Upon additional annealing under 650 °C, the initially dispersed Pt clusters clearly agglomerated into larger Pt nanoparticles (Fig. 1e, f). Accordingly, a mixture of residual Pt clusters with an average size of 0.9 ± 0.2 nm (Supplementary Fig. 9) and agglomerated Pt nanoparticles with an average size of 32.8 ± 11.3 nm (Fig. 1g) were observed in Ptp/MAO. A significant morphology transformation of Pt occurred in the Pts/MAO catalyst under severe thermal treatment at 800 °C. The HAADF-STEM images revealed extensive Pt sintering (Fig. 1h, i), resulting in particles with a broad size distribution ranging from 15 to 90 nm (Fig. 1j) and the complete disappearance of the Pt cluster (Supplementary Fig. 10). Notably, the sintered Pt particles exhibited well-defined crystalline facets (Fig. 1i), suggesting substantial faceting under high-temperature conditions. The crystalline structures of the Pt/MAO catalysts and the MAO support were also characterized by X-ray diffraction (XRD), as shown in Fig. 1k. The Ptn/MAO catalyst exhibits no detectable Pt diffraction features, confirming the high dispersion of Pt on the MAO support. In contrast, both Ptp/MAO and Pts/MAO catalysts show distinct diffraction peaks corresponding to metallic Pt (JCPDS 04-0802) (Fig. 1l), evidencing thermal-induced agglomeration into crystalline Pt nanoparticles at 650 and 800 °C, respectively. Notably, the intensity of the metallic Pt (111) peak at 39.7° significantly increases as the calcination temperature rises from 650 to 800 °C, showing the progressive growth of Pt particles at elevated temperatures. This trend is consistent with the decreasing Pt dispersion measured by CO pulse chemisorption (74%, 38%, and 6% for Ptn/MAO, Ptp/MAO, and Pts/MAO, respectively; Supplementary Table 2) and with the AC-STEM observations. In addition, we also performed in situ CO probe molecule diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) measurements at room temperature on Ptn/MAO and Pts/MAO, and the results are shown in Supplementary Fig. 11. For Ptn/MAO, no distinct CO chemisorption bands associated with Pt species were observed, which is consistent with the weak CO adsorption ability of the highly oxidized Pt species in this catalyst. In contrast, Pts/MAO exhibits two clear CO stretching bands at 2098 and 2063 cm−1, which can be assigned to linearly adsorbed CO on well-coordinated Pt sites on low-index facets and under-coordinated Pt sites, respectively. These features are consistent with the presence of large metallic Pt particles34. Catalytic performance The catalytic performance of the Pt/MAO catalysts for propane oxidation was evaluated under stoichiometric conditions (Supplementary Fig. 12), serving as a model reaction for light alkane oxidation in gasoline vehicle emission control35,36. Figure 2a displays the third-run light-off curves of the three Pt/MAO catalysts, where the activity reached a steady state after the second run. Ptn/MAO showed propane oxidation activity with a T90 (the temperature at 90% conversion of propane) of approximately 360 °C. In contrast, higher temperature calcination significantly enhanced the catalytic performance of the Pt/MAO catalysts for propane complete oxidation. Both Ptp/MAO and Pts/MAO exhibited superior catalytic activity, representing T90 values of 290 °C and 245 °C, respectively. The sintered catalyst also markedly improves the Pt mass activity for propane oxidation. As shown in Fig. 2b, at 220 °C, Pts/MAO exhibits a reaction rate of 51.6 µmol gPt−1 s−1, ~10 times higher than Ptn/MAO (5.3 µmol gPt−1 s−1) and ~2 times higher than Ptp/MAO (27.8 µmol gPt−1 s−1). Notably, although sintering greatly decreases the Pt dispersion, the Pts/MAO catalyst still exhibits the highest Pt-mass-normalized reaction rate among all the catalysts studied. This result demonstrates that, in this system, the intrinsic activity enhancement of sintered Pt particles outweighs the loss of exposed Pt surface atoms caused by sintering. Consistently, the turnover frequencies (TOFs) of Pts/MAO reach 0.1625 s−1 at 220 °C (Fig. 2c), representing a 116-fold increase over Ptn/MAO (0.0014 s−1) and a 12-fold increase over Ptp/MAO (0.0141 s−1). To the best of our knowledge, this exceptional performance of the sintered Pts/MAO catalyst also surpasses that of most high-performance Pt-based catalysts reported recently in the literature (Fig. 2d, Supplementary Table 3), indicating remarkably superior site efficiency.Fig. 2Catalytic performance of Pt/MAO catalysts for propane oxidation.a Propane oxidation activity, b reaction rate at 220 °C, and c TOFs at 220 °C of Ptn/MAO, Ptp/MAO, and Pts/MAO catalysts. d Comparison of the TOF of supported Pt catalysts in the literature recently for the complete propane oxidation reaction. e The stability test of Ptn/MAO and Pts/MAO catalysts at 250 °C under dry and 4% H2O conditions. f Propane oxidation activity of Pts/MAO catalyst after hydrothermal aging at 750 °C for 6 h. Reaction conditions: 0.2 vol% C3H8, 1 vol% O2, and N2 balance; WHSV = 30,000 mL g−1 h−1. Catalytic performance of Pt/MAO catalysts for propane oxidation. a Propane oxidation activity, b reaction rate at 220 °C, and c TOFs at 220 °C of Ptn/MAO, Ptp/MAO, and Pts/MAO catalysts. d Comparison of the TOF of supported Pt catalysts in the literature recently for the complete propane oxidation reaction. e The stability test of Ptn/MAO and Pts/MAO catalysts at 250 °C under dry and 4% H2O conditions. f Propane oxidation activity of Pts/MAO catalyst after hydrothermal aging at 750 °C for 6 h. Reaction conditions: 0.2 vol% C3H8, 1 vol% O2, and N2 balance; WHSV = 30,000 mL g−1 h−1. Long-term durability tests were conducted by running the propane oxidation reaction at 250 °C. Remarkably, Pts/MAO maintained stable propane conversion over 48 h under dry conditions without any degradation (Fig. 2e), a robustness attributed to the large, pre-sintered Pt particles. To assess practical applicability, we introduced 4% H2O into the reaction stream, mimicking humid conditions. The presence of H2O did not influence the activity and stability of Pts/MAO; however, in contrast, it did reduce the activity of Ptn/MAO for propane oxidation. This comparison highlights the strong resistance of sintered Pt particles to steam during propane oxidation. To further challenge the catalyst durability, we subjected Pts/MAO to harsh hydrothermal aging (10% H2O in air flowed at 750 °C for 6 h). Impressively, the catalyst retained nearly identical propane conversion activity after treatment, demonstrating exceptional stability under extreme conditions (Fig. 2f). We further calcined the Pt/MAO catalyst at 1000 °C, and the resulting sample is denoted as Pts/MAO−1000. As shown in Supplementary Fig. 13, Pts/MAO−1000 exhibits nearly the same propane oxidation activity as Pts/MAO. This result shows that no further activity enhancement is achieved upon increasing the calcination temperature from 800 to 1000 °C, suggesting that the activity has already reached a plateau at 800 °C. Moreover, the fact that the catalyst still maintains comparable activity even after calcination at 1000 °C further highlights the exceptional thermal stability of the sintered Pt catalyst. Reaction kinetics were probed by measuring the dependence of reaction rates on oxygen and propane partial pressures at 220 °C for all Pt/MAO catalysts. As shown in Fig. 3a, all catalysts exhibit positive apparent reaction orders with respect to propane, whereas negative apparent reaction orders are observed with respect to oxygen (Fig. 3b). This kinetic feature is consistent with competitive adsorption between propane and oxygen species on Pt surfaces, where oxygen species occupy Pt sites and inhibit propane adsorption/activation during propane oxidation8,9. Notably, elevating the calcination temperature from 500 °C to 800 °C induced significant changes in the rate orders. The rate order of propane increased from 0.87 (Ptn/MAO) to 1.86 (Pts/MAO), while the rate order of oxygen decreased from -0.53 (Ptn/MAO) to -1.64 (Pts/MAO) across the Pt/MAO catalysts, as illustrated in Fig. 3a, b. These kinetic trends suggest that the reaction over sintered Pt catalysts becomes more sensitive to propane partial pressure and more strongly inhibited by increasing oxygen partial pressure.Fig. 3Adsorption and activation capabilities of Pt/MAO catalysts for propane oxidation.Rate orders for a propane and b oxygen in propane oxidation at 220 °C. c H2-TPR patterns of MAO, Ptn/MAO, Ptp/MAO, and Pts/MAO catalysts. d C3H8-TPD patterns of MAO, Ptn/MAO, and Pts/MAO catalysts. e O2-TPD patterns of MAO, Ptn/MAO, Ptp/MAO, and Pts/MAO catalysts. Adsorption and activation capabilities of Pt/MAO catalysts for propane oxidation. Rate orders for a propane and b oxygen in propane oxidation at 220 °C. c H2-TPR patterns of MAO, Ptn/MAO, Ptp/MAO, and Pts/MAO catalysts. d C3H8-TPD patterns of MAO, Ptn/MAO, and Pts/MAO catalysts. e O2-TPD patterns of MAO, Ptn/MAO, Ptp/MAO, and Pts/MAO catalysts. After the propane oxidation reaction, the Pts/MAO-used catalyst retains distinct diffraction peaks of metallic Pt, including the obvious Pt(111) reflection at 39.7° (Supplementary Fig. 14). The Ptn/MAO-used catalyst still shows no detectable Pt diffraction features. These results are further supported by AC-STEM observations (Supplementary Figs. 15, 16), which show that highly dispersed Pt nanoclusters/nanoparticles remain predominant in the used Ptn/MAO catalyst, with an average size of 1.3 ± 0.4 nm, while the used Pts/MAO catalyst still exhibits extensively sintered Pt particles with a broad size distribution of 15–90 nm, similar to those of the corresponding fresh catalysts. It is concluded that the sintered metallic Pt particles in Pts/MAO are preserved, whereas such large Pt particles cannot be obtained through prolonged propane reaction over highly dispersed Pt catalysts. Structure-activity relationship The chemical states of Pt species in the Pt/MAO catalysts were investigated using hydrogen temperature-programmed reduction (H2-TPR, Fig. 3c), selected in preference to X-ray photoelectron spectroscopy (XPS) because of the significant overlap between the Pt and support signals in the XPS (Pt 4f with Al 2p, and Pt 4d with Mg KLL, Supplementary Figs. 17, 18). The H2-TPR profiles reveal a distinct thermal evolution of Pt-related oxygen-containing species. Ptn/MAO exhibits two broad low-temperature reduction features centered at approximately 110 and 160 °C, together with a pronounced high-temperature peak around 450 °C. Ptp/MAO shows a much weaker overall reduction response (the total reduction peak area decreased by 56%, Supplementary Table 4), with a broad low-temperature feature at around 170 °C and weak peaks near 410 and 485 °C. The low-temperature features are mainly associated with the reduction of relatively weakly bound PtOx species, whereas the higher-temperature peaks are more reasonably attributed to strongly interacting interfacial Pt-O-support species and/or other Pt-related oxygen-containing species with enhanced resistance to reduction26. In contrast, no detectable reduction peaks are observed for Pts/MAO, consistent with the predominance of metallic Pt nanoparticles formed after calcination at 800 °C. To further investigate the adsorption abilities of propane and oxygen on Pt/MAO catalysts, temperature-programmed desorption experiments for propane (C3H8-TPD) and oxygen (O2-TPD) were performed. As shown in Fig. 3d, the C3H8-TPD results reveal that Pts/MAO exhibits a significantly higher propane desorption amount (15 μmol g−1) than Ptn/MAO (7 μmol g−1). After subtracting the propane desorption contribution of the MAO support and normalizing by the number of exposed Pt surface sites estimated from CO chemisorption-derived Pt dispersion, the Pt-site-normalized propane desorption amount for Pts/MAO (3.74 μmolC3H8 μmolsurface Pt−1) remains much higher than that for Ptn/MAO (0.06 μmolC3H8 μmolsurface Pt−1). These results further indicate that sintered Pt particles possess enhanced propane adsorption capability. In contrast, the O2-TPD results (Fig. 3e, Supplementary Table 5) clearly revealed a dramatic reduction in the total desorption peak area for Pts/MAO, which decreased to 1/4000 of that observed for Ptn/MAO. This evidence suggests a significantly lower oxygen adsorption capacity of sintered Pt particles compared to their highly dispersed counterparts. Notably, only the Ptn/MAO catalyst exhibited a distinct O2 desorption peak at 700 °C, reflecting a strong binding between O and Pt clusters37. The aforementioned results reveal a competitive relationship between the adsorption and activation of oxygen and propane on the Pt surface. Specifically, the strong oxygen adsorption ability suppresses the adsorption and activation of propane, ultimately influencing the overall catalytic reactivity of Pt, since propane activation is the rate-limiting step in propane oxidation9. It can be inferred that the prevalence of low-index facets on larger Pt particles with stable metallic states is favorable for propane oxidation. These experimental results show good consistency with our DFT calculation (Table 1). Since the surface energies of different Pt facets are strongly influenced by oxygen and propane coverage, the reaction atmosphere can alter the facet configurations, potentially inducing surface reconstruction of the catalyst and consequently affecting its catalytic performance38–40. To examine this effect systematically, we further conducted catalytic performance tests under different O2/C3H8 ratios (Supplementary Figs. 19–21). All Pt/MAO catalysts showed pronounced activity deterioration with increasing O2/C3H8 ratios. Particularly, the Pts/MAO catalyst exhibited a 90% reduction in TOF, from 0.1625 s−1 at an O2/C3H8 ratio of 5:1 to 0.0196 s−1 at an O2/C3H8 ratio of 40:1 (Supplementary Fig. 20), highlighting the critical sensitivity of propane oxidation activity to oxygen partial pressure on Pt surfaces, consistent with the rate-order results (Fig. 3a, b). This dramatic activity suppression suggests that while sintered Pt catalysts are intrinsically active for propane oxidation, their performance is strongly modulated by the reaction atmosphere through potential surface reconstruction and competitive adsorption effects41. To elucidate the dynamic surface evolution of sintered Pt particles under propane oxidation conditions, in situ atmospheric STEM experiments were conducted on Pts/MAO by using a windowed gas cell technique42,43. The as-prepared Pts/MAO catalyst was initially heated to 300 °C under the vacuum condition (Condition 1). The cell was then filled with flowing gases (balanced with N2) at 1.0 bar, and the Pt surface configuration was examined sequentially under three environments: (i) 1% C3H8, (ii) stoichiometric reaction conditions (1% C3H8 + 5% O2; O2/C3H8 = 5:1), and (iii) oxygen-rich conditions (0.5% C3H8 + 20% O2; O2/C3H8 = 40:1). For detailed analysis, a representative 30-nm Pt particle oriented along the [001] zone axis was selected for observation (Supplementary Fig. 22). As depicted in Fig. 4a1, the initially sintered Pt particle exhibited large facets and small rounded edges at the intersection corners under vacuum at 300 °C. The corresponding enlarged HAADF-STEM image (Fig. 4b1) reveals that these rounded corners consist of small, high-index steps adjacent to a (100) facet. These rounded features likely resulted from the adsorption of residual oxygen during prior air exposure. Upon introducing 1% C3H8, these rounded corners underwent obvious faceting and became more flattened (Fig. 4a2). This transformation suggests that propane reacts with and partially removes the pre-adsorbed oxygen, thereby lowering the oxygen coverage and driving the surface toward well-defined low-index facets. The atomic-scale HAADF-STEM image (Fig. 4b2) shows that the high-index steps have reconstructed into smooth (100) and (110) facets.Fig. 4Mechanistic analysis of surface reconstruction of Pt/MAO in response to different propane oxidation reaction conditions.Aberration-corrected a BF-STEM and b atomic-scale HAADF-STEM images of a 30-nm Pt particle taken along the [001] zone axis in Pts/MAO catalyst. Condition 1: at 300 °C under vacuum for 1 h; Condition 2: at 300 °C in 1 bar of 1% C3H8 in N2 for 1 h; Condition 3: at 300 °C in 1 bar of 1% C3H8 and 5% O2 in N2 for 1 h; Condition 4: at 300 °C in 1 bar of 0.5% C3H8 and 20% O2 in N2 for 1 h. c Facet fractions as a function of O coverage. Typical structures corresponding to different O coverage are shown at the top. Mechanistic analysis of surface reconstruction of Pt/MAO in response to different propane oxidation reaction conditions. Aberration-corrected a BF-STEM and b atomic-scale HAADF-STEM images of a 30-nm Pt particle taken along the [001] zone axis in Pts/MAO catalyst. Condition 1: at 300 °C under vacuum for 1 h; Condition 2: at 300 °C in 1 bar of 1% C3H8 in N2 for 1 h; Condition 3: at 300 °C in 1 bar of 1% C3H8 and 5% O2 in N2 for 1 h; Condition 4: at 300 °C in 1 bar of 0.5% C3H8 and 20% O2 in N2 for 1 h. c Facet fractions as a function of O coverage. Typical structures corresponding to different O coverage are shown at the top. Upon switching to the stoichiometric reactant mixture (O2/C3H8 = 5:1), which corresponds to high intrinsic activity of Pts/MAO for propane oxidation, the particle maintained a faceted morphology similar to that observed under the C3H8-only feed, with large facets and flattened edges preserved (Fig. 4a3). Atomic-scale HAADF-STEM image (Fig. 4b3) confirms that the exposed (100) and (110) facets remained stable under the stoichiometric condition. This evidence shows that under stoichiometric reaction conditions, O2 and C3H8 can undergo continuous adsorption and reaction on the surface without causing severe oxygen poisoning. In contrast, when the O2/C3H8 ratio was elevated to 40:1 (Fig. 4a4), the particle morphology reverted to rounded edges and corners, indicating strong oxygen adsorption at these regions, and corresponds to a lower catalytic activity due to excessive oxygen. Fig. 4b4 reveals that the previously observed low-index facets have reconstructed into high-index stepped sites, accounting for the observed morphological rounding. These in situ STEM observations demonstrate a clear correlation between surface reconstructions and the reaction environments. At lower O2/C3H8 ratios, larger Pt particles preferentially expose low-index facets that enhance C3H8 oxidation activity. Conversely, elevated O2 concentrations induce Pt surface reconstruction to high-index stepped configurations, concomitant with activity deterioration. To further corroborate the observed atmosphere-dependent Pt surface reconstruction, we performed additional DFT calculations. Using the multiscale structure reconstruction (MSR) model embedded in the Multiscale Operando simulation package (MOSP)44, we simulated the equilibrium configurations of Pt particles under varying oxygen coverages (see Supplementary Table 6 for computational parameters). As shown in Fig. 4c, our DFT calculations reveal a clear structural evolution of Pt surfaces with increasing oxygen coverage: as oxygen coverage rises, low-index facets progressively diminish while high-index (211) stepped facets become increasingly dominant. Notably, when O coverage surpasses the critical threshold of 0.42 ML, the Pt surface undergoes complete reconstruction to predominantly expose (211) facets, which exhibit both the stronger oxygen binding affinity and weaker propane adsorption capability (Table 1). These coverage-dependent structural changes explain our in situ STEM observations that higher O2/C3H8 ratios lead to increased oxygen coverage, driving surface reconstruction toward small high-index steps, for example, (211) facets, and, in turn, significantly reduce propane oxidation activity. This mechanism highlights how oxygen-induced surface reconstruction critically governs catalytic performance in propane oxidation. Through a comprehensive investigation combining in situ STEM, kinetic measurements, and DFT calculations, we have concluded that the low-index facets of sintered Pt particles with stable metallic states serve as the dominant active sites for complete propane oxidation under stoichiometric reaction conditions. These well-defined Pt surfaces exhibit enhanced propane adsorption and activation capabilities while maintaining an optimal balance in oxygen interactions that effectively suppresses oxygen poisoning. In contrast, smaller Pt clusters suffer from catalytic deactivation due to their inherently higher density of undercoordinated edge and vertex sites with electron-deficient states, and the pronounced finite-size effect that promotes excessive oxygen adsorption. In addition, these results also indicate that the catalytic performance of sintered Pt catalysts is highly sensitive to the reaction atmosphere. By increasing the propane-to-oxygen ratio, the effective oxygen coverage on Pt can be reduced, thereby suppressing oxygen poisoning and stabilizing the low-index metallic facets responsible for the high intrinsic activity (Supplementary Figs. 19–21). This finding suggests that, for oxidation processes in which operating parameters can be adjusted, apparent sintering-induced deactivation may be mitigated not only through structural stabilization of catalysts but also through rational optimization of reaction conditions, which can help recover the active surface state and improve catalytic performance. Generality of the sintering strategy The versatility of this sintering strategy is further demonstrated by its successful extension to other light-alkane oxidations, where substantial activity enhancements are observed for both ethane and n-butane. For ethane oxidation, the sintered Pts/MAO catalyst lowers T90 by 101 °C (from 434 to 333 °C) (Fig. 5a) and increases the TOF by >25-fold, from 0.5 to 13.0 s−1 at 240 °C (Supplementary Fig. 23), relative to Ptn/MAO. For n-butane oxidation, Pts/MAO reduces T90 by 71 °C (from 286 to 215 °C) (Fig. 5b) and enhances the TOF by >22-fold, from 0.8 to 17.9 s−1 at 150 °C (Supplementary Fig. 24), compared with Ptn/MAO. Collectively, these results establish sintered Pt catalysts as a promising and broadly applicable platform for industrially relevant light-alkane oxidation reactions.Fig. 5Generality of the sintering strategy across light-alkane oxidation reactions and catalyst supports.a Ethane oxidation activity of Ptn/MAO and Pts/MAO catalysts. Reaction conditions: 0.2 vol% C2H6, 0.7 vol% O2, and N2 balance; WHSV = 30,000 mL g−1 h−1. b N-butane oxidation activity of Ptn/MAO and Pts/MAO catalysts. Reaction conditions: 0.2 vol% n-C4H10, 1.3 vol% O2, and N2 balance, WHSV = 30,000 mL g−1 h−1. c XRD patterns and d propane oxidation activity of Ptn/γ-Al2O3 and Pts/γ-Al2O3 catalysts. Reaction conditions: 0.2 vol% C3H8, 1 vol% O2, and N2 balance; WHSV = 30,000 mL g−1 h−1. Generality of the sintering strategy across light-alkane oxidation reactions and catalyst supports. a Ethane oxidation activity of Ptn/MAO and Pts/MAO catalysts. Reaction conditions: 0.2 vol% C2H6, 0.7 vol% O2, and N2 balance; WHSV = 30,000 mL g−1 h−1. b N-butane oxidation activity of Ptn/MAO and Pts/MAO catalysts. Reaction conditions: 0.2 vol% n-C4H10, 1.3 vol% O2, and N2 balance, WHSV = 30,000 mL g−1 h−1. c XRD patterns and d propane oxidation activity of Ptn/γ-Al2O3 and Pts/γ-Al2O3 catalysts. Reaction conditions: 0.2 vol% C3H8, 1 vol% O2, and N2 balance; WHSV = 30,000 mL g−1 h−1. Furthermore, we demonstrate that our sintering strategy presents a facile yet effective approach to enhance the performance of commercial Pt catalysts for propane oxidation. Using a conventional γ-Al2O3-supported Pt catalyst as an example, we observed significant improvements after high-temperature sintering treatment. XRD analysis (Fig. 5c) confirms the successful formation of sintered metallic Pt particles in the 800 °C-treated γ-Al2O3-supported Pt catalyst (Pts/γ-Al2O3), in contrast to the 500 °C-treated counterpart (Ptn/γ-Al2O3). The sintered Pts/γ-Al2O3 catalyst demonstrates remarkable enhancement in propane oxidation activity under stoichiometric conditions, achieving a 125 °C reduction in T90 (from 373 °C to 248 °C) compared to Ptn/γ-Al2O3 catalyst (Fig. 5d). These results validate the generality of the sintering strategy and highlight its practical potential for industrial propane oxidation and underscore its promise as a simple thermal pretreatment to improve the activity of practical Pt-based oxidation catalysts. Discussion In summary, this study demonstrates that the sintered Pt particles with sizes in the dozens of nanometers exhibit superior activity, water resistance, and stability for propane complete oxidation, compared to small Pt clusters. Through in situ experiments and theoretical calculations, it is disclosed that sintered Pt particles preferentially expose low-index metallic surfaces under stoichiometric reaction conditions. These surfaces facilitate exceptional propane adsorption and activation while effectively suppressing oxygen poisoning. The intrinsic activity of these low-index Pt surfaces with stable metallic states is at least two orders of magnitude higher than that of highly dispersed Pt-based catalysts, enabling exceptional adsorption and activation of C3H8 while effectively mitigating the adverse effects of oxygen poisoning. Moreover, the sintered Pt catalysts, prepared through high-temperature annealing at 800 °C, exhibit inherent thermal stability and outstanding durability under harsh conditions, including hydrothermal aging and long-term reactions. This facile sintering strategy can also improve the performance of commercial Pt/γ-Al2O3 catalysts for propane oxidation and can potentially be extended to other light alkane oxidation reactions. Our findings present an unconventional yet effective approach to improving catalytic activity and stability in certain reactions by utilizing sintered particles. The sintering strategy holds promise in other heterogeneous catalysts where resistance to oxidation or oxygen poisoning is critical for achieving high performance. Methods Synthesis of MgAl2O4 (MAO) A total of 8.170 g aluminum isopropoxide and 5.128 g magnesium nitrate hexahydrate were added to a 120 mL mixed solution (denatured ethyl alcohol and concentrated nitric acid with the volume ratio 7:1) in a 400 mL beaker with sealing treatment, then the beaker was transferred to the magnetic stirring table and stirred for 1 h until the suspension was dissolved. This mixed solution was transferred into a 200 mL Teflon-lined stainless-steel autoclave and solvothermal treated at 150 °C for 12 h. The formed solid was dried at 90 °C overnight, followed by calcining in static air at 800 °C for 5 h. All the above reagents were purchased by Sinopharm Chemical Reagent Co., Ltd (China). Synthesis of Pt/MAO catalysts The Pt/MAO catalyst was prepared by the wet impregnation method. Firstly, the target content of Pt(NO3)2 solution (purchased by Sino-Platinum Metals Co., Ltd) was added dropwise to MAO powder, and this slurry was stirred for 15 min with ultrasonic wave assistance. Following this, the slurry was dried at 80 °C overnight and calcined in static air at 500 °C, obtaining a catalyst denoted with Ptn/MAO. Next, Ptn/MAO was further calcined at 650 °C and 800 °C for 12 h to acquire Ptp/MAO and Pts/MAO catalysts, respectively. Characterization Powder X-ray diffraction (XRD) patterns of catalysts were carried out on a Bruker D8FOCUS diffractometer with Cu Kα radiation (λ = 1.5418 Å) operated at 40 kV and 40 mA. The surface areas of catalysts were measured on a Micromeritics ASAP 2400 instrument by the N2 adsorption at 77 K and calculated by the Brunauer-Emmett-Teller (BET) method. The catalysts were degassed at 453 K for 12 h before measurement. The actual Pt contents were measured by inductively coupled plasma optical emission spectroscopy (ICP-OES). Aberration-corrected scanning transmission electron microscopy (AC-STEM) was operated on a ThermoFisher Themis Z transmission electron microscope equipped with two aberration correctors, operated at 300 kV with a convergence semi-angle of 17.8 mrad. The microscope was aligned using a standard gold sample before observations. High-angle annular dark-field (HAADF)-STEM images were recorded using a camera length of 115 mm on HAADF detectors, and the inner- and outer collection angles of 47 and 200 mrad, respectively. The acquired images were processed by the Gaussian blur filter. The in situ atmospheric STEM experiment was performed using a Climate S3 in situ transmission electron microscope holder (DENS Solutions Company), allowing for in situ STEM observation under atmospheric pressure and high temperature. The screen current was adjusted to 10.0 pA to avoid electron beam effects. In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) of the CO adsorption experiment was recorded by a Nicolet Model Nexus 670 FT-IR spectrometer equipped with an MCT detector cooled by liquid N2. For the CO-DRIFTS, the sample was degassed in pure Ar (40 mL min−1) at 300 °C for 30 min. After cooling to room temperature, the Ar was switched to 1 vol% CO/Ar (40 mL min−1). After adsorbing CO for 20 min, the sample was purged with pure Ar (40 mL min−1). Temperature-programmed H2 reduction (H2-TPR), temperature-programmed desorption of oxygen (O2-TPD), and CO pulse chemisorption were performed on the Micromeritic Autochem II 2920 instrument. For the H2-TPR experiment, the 100 mg sample was heated from room temperature to 600 °C at a rate of 10 °C min−1 under the 10 vol% H2/N2 (40 mL min−1). The hydrogen consumption was measured by a thermal conductivity detector (TCD). For the O2-TPD experiment, the 50 mg sample was pretreated with 2 vol% O2/Ar (40 mL min−1) at 400 °C for 0.5 h. Then, the sample was cooled to room temperature. After being purged by He for 0.5 h, the sample was heated from 50 to 800 °C at a rate of 10 °C min−1. The O2 signal was recorded with Hiden VII mass spectra. CO pulse chemical adsorption was performed to obtain the Pt dispersion. The sample was pretreated by 5 vol% H2/N2 (40 mL min−1) at 200 °C for 0.5 h. Then, the sample was purged using purge Ar at 200 °C for 0.5 h. After cooling to the temperature, the sample was kept in the atmosphere of purged Ar for 0.5 h at room temperature. The 1 vol% CO + 2 vol% Ar + 97 vol% He gas was injected into the chamber by pulse, and the CO signal was recorded by Hiden VII mass spectra. Assuming the CO/Pt = 1, the Pt dispersion can be calculated. The C3H8-TPD experiment was conducted by the FID detector with a methanizer. Typically, the 100 mg sample was taken in a quartz tube and heated up to 500 °C for 1 h under pure N2 (50 mL min−1). Then the sample was cooled down to room temperature under N2 atmosphere, and then the gas was switched to a mixture of 0.2 vol.% C3H8 + 99.8 vol.% N2 (50 mL min−1) and kept for 30 min. Then the gas was switched to N2 (50 mL min−1), and the temperature was raised to 50 °C and kept for 2 h. Afterward, the sample temperature was raised to 500 °C at a rate of 10 °C min−1 and kept for 1 h. The FID detector was used to record the C3H8 concentrations in the outlet gas. The quantitative formula is1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{Total} \, {Desorption} \, {Amounts}\, (\mathrm{mol})={\int }_{{\mathrm{t}}_{0}}^{{\mathrm{t}}_{1}}{\rm{CV}}\,{\rm{dt}}$$\end{document}TotalDesorptionAmounts(mol)=∫t0 t1CVdtwhere C is the C3H8 concentration in the outlet gas, t is the time (in sec), and V is the outlet gas flow rate. Evaluation of catalytic performance The catalytic performance of the catalysts for C3H8 oxidation was tested in a fixed-bed reactor, including 100 mg of the catalysts. The typical feed gas consisted of 0.2 vol% C3H8, 1 vol% O2, and N2 balance with a total flow of 50 mL min−1 (weight hourly space velocity, WHSV = 30,000 mL g−1 h−1). The temperature of the reaction bed was ramped to 480 °C with a speed of 3 °C min−1. The reactant was monitored by an online chromatograph (GC-9790). The selectivity of CO2 was close to 100% over the catalysts under reaction conditions, and the other possible products were under the detection limit of the GC. The C3H8 conversion (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${X}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}XC3H8) can be calculated by the following equation2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{X}}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}=\frac{{[{\mathrm{C}}_{3}{\mathrm{H}}_{8}]}_{\mathrm{in}}-{[{\mathrm{C}}_{3}{\mathrm{H}}_{8}]}_{\mathrm{out}}}{{[{\mathrm{C}}_{3}{\mathrm{H}}_{8}]}_{\mathrm{in}}}$$\end{document}XC3H8=[C3H8]in−[C3H8]out[C3H8]inwhere [C3H8]in and [C3H8]out are the C3H8 concentrations in the inlet and outlet gas, respectively. Measurement of reaction kinetics Before kinetic testing, every catalyst has been pretreated by the cycle test to obtain stable and reliable data. Kinetic data were measured by adjusting the WHSV and feed gas composition. The propane conversion was restricted in a range from 4% to 15% by controlling WHSV. The specific rate, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${r}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}rC3H8 (µmolpropane gPt−1 s−1), was calculated by the equation3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{r}}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}=\frac{{{X}}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}{\rm{\cdot }}{{C}}_{{{C}}_{3}{\mathrm{H}}_{8}}}{{\mathrm{g}}_{\mathrm{Pt}}}$$\end{document}rC3H8=XC3H8⋅CC3H8gPtwhere \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${X}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}XC3H8 is the C3H8 conversion, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}CC3H8 is the C3H8 gas flow rate, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{g}}_{\mathrm{Pt}}$$\end{document}gPt is the mass of Pt. The turnover frequency, TOF (s−1), was calculated using the equation,4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${TOF}=\frac{{{r}}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}\cdot {M}}{{D}}$$\end{document}TOF=rC3H8⋅MDwhere M is the relative atomic mass of Pt, and D is the Pt dispersion. PC3H8 and PO2 are the initial partial pressures of propane and oxygen, and α and β are the partial reaction orders. To obtain the partial reaction order α, the PC3H8 was changed between 0.16 and 0.22 kPa, and PO2 was maintained at 1 kPa. Similarly, for partial reaction order β, the PO2 was adjusted between 0.8 and 1.4 kPa, and PC3H8 was fixed at 0.2 kPa. Theoretical calculation methods All DFT calculations were performed using the Vienna Ab initio Simulation Package (VASP). A generalized gradient approximation (GGA)45 based on revised Perdew-Burke-Ernzerhof (RPBE) exchange-correlation functional46 was used. The projector augmented wave (PAW) method with a 450 eV cut-off energy was applied to describe the core-electron potential47,48. Structure optimizations were finished until the electronic energy and force of the structure converged to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-5}$$\end{document}10−5 eV and 0.03 eV/Å, respectively. The Pt (211) and Pt (221) slabs were represented by using (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2\times 4$$\end{document}2×4) supercells of 16 layers. The bottom five slabs were fixed to simulate the bulk structure. The Pt(100) and Pt(111) slabs were modeled using 4 × 4 supercells with six atomic layers, and the bottom two layers were fixed to simulate the bulk structure. A vacuum zone of 15 Å was set. The truncated octahedral Pt38 particle was constructed and placed within a \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$25$$\end{document}25 Å × 25 Å × 25 Å cubic periodic box. Monkhorst-Pack \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k$$\end{document}k points mesh49 of 3 × 3 × 1 was used for the slabs and 1 × 1 × 1 for the particle. The adsorption energy of atomic oxygen was calculated as:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{\mathrm{O}}={E}_{\mathrm{surface}+{\mathrm{O}}^{*}}-{E}_{\mathrm{surface}}-\frac{1}{2}{E}_{{\mathrm{O}}_{2}}$$\end{document}EadsO=Esurface+O*−Esurface−12EO2where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{E}}_{\mathrm{surface}+{\mathrm{O}}^{*}}$$\end{document}Esurface+O* is the total energy of the Pt catalysts with adsorbed oxygen, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{E}}_{\mathrm{surface}}$$\end{document}Esurface and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{E}}_{{\mathrm{O}}_{2}}$$\end{document}EO2 are the energies of bare Pt catalysts and gas phase O2, respectively. For propane adsorption, Grimme D3 level dispersion50 was included to describe the van der Waals (vdW) interaction. And the adsorption energy of propane was calculated as:6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{ads}}^{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}={E}_{\mathrm{surface}+{\mathrm{C}}_{3}{\mathrm{H}}_{8}^{*}}-{E}_{\mathrm{surface}}-{E}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EadsC3 H8=Esurface+C3H8*−Esurface−EC3H8where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{surface}+{\mathrm{C}}_{3}{\mathrm{H}}_{8}^{*}}$$\end{document}Esurface+C3H8* is the total energy of the Pt catalysts with adsorbed propane, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{\mathrm{surface}}$$\end{document}Esurface and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{{\mathrm{C}}_{3}{\mathrm{H}}_{8}}$$\end{document}EC3H8 are the energies of reference Pt catalysts and gas phase C3H8, respectively. For propane adsorption on O-precovered Pt surfaces, the O-precovered system was used as the reference state. To assess the role of dispersion, propane adsorption energies calculated without the D3 correction were additionally summarized and compared with the RPBE + D3 results in Supplementary Table 7. The comparison shows that the adsorption energies of molecular propane on Pt are strongly influenced by dispersion. To model the O precovered Pt slabs and Pt38, several O arrangements were considered based on previous literature51,52, and the most stable configuration was adopted for propane adsorption calculations. Transition states were determined using the climbing-image nudged elastic band (CI-NEB) method53,54 and the dimer method55. Transition states were confirmed by the presence of a single imagary frequency. The force convergence criterion for transition-state optimization was set to 0.05 eV/Å. Vibrational frequencies of the transition state were calculated using the finite differences approach with a displacement size of 0.02 Å. All transition states have been verified by ensuring that the transition state has only one significant imaginary vibrational frequency. Supplementary information Supplementary Information Transparent Peer Review file Supplementary Information Transparent Peer Review file Source data Source Data Source Data