Relay semi-hydrogenation of acetylene via hydrogen spillover on non-reducible Al2O3-supported single-atom catalysts Authors: Xianhui Wang, Luo Ran, Kaige Tian, Xin Chang, Guodong Sun, Chunlei Pei, Sai Chen, Zhi-Jian Zhao, Jinlong Gong Published: 2026 ABSTRACT AbstractHydrogen spillover involves the migration of hydrogen atoms from metal to another component, yet it remains challenging to achieve on non-reducible supports. This paper describes the observation of hydrogen spillover on non-reducible Al2O3-supported Pd single-atom catalysts, facilitated by phosphorus−hydroxyl (P−OH) bridge sites, for acetylene semi-hydrogenation. Through in situ spectroscopy, kinetic analysis, and computational analyses, we show that Pd single atoms promote the heterolytic activation of H2, while phosphorus doping on Al2O3 lowers energy barrier for hydrogen spillover via the formation of P−OH species. Gas-phase acetylene collides and hydrogenates with as-formed P−OH, following Eley-Rideal type mechanism, making inert Al2O3 active in hydrogenation reaction. This unconventional relay pathway reduced Pd usage by one order of magnitude and improved catalytic stability, with over 95% ethylene selectivity at 99% acetylene conversion in ethylene-rich stream at 100 °C for more than 200 hours. This work elucidates the mechanism of hydrogen spillover on non-reducible Al2O3 and relay hydrogenation on supported single-atom catalysts.Subject terms: Heterogeneous catalysis, Chemical engineering, Catalytic mechanisms Introduction Ethylene is an important building block to produce poly-ethylene1,2. However, ethylene is produced from the cracking of oil fractions with 0.6-2% acetylene impurities, which would irreversibly poison the Ziegler-Natta polymerization catalysts, lowering the quality of polyethylene3–5. The chemo-selective hydrogenation of acetylene to ethylene is one of the most effective methods to remove the trace acetylene impurities in ethylene products. Among all catalysts investigated, Pd-based catalysts have attracted industrial and academic interest6–8. However, the strong H2 activation on Pd is normally accompanied by over-hydrogenation of the substrate (C≡C and C=C bonds), leading to an undesired C−C bond product9–11. Thus, the key challenge for selective semi-hydrogenation of acetylene lies in the precise control of H2 activation and subsequent hydrogenation9,12,13. Hydrogen spillover is an important process in catalytic hydrogenation involving the migration of active hydrogen species from metal sites to support surface12,14–16. When Pd is in contact with reducible metal oxides, hydrogen atoms created by dissociative chemisorption of H2 molecules on Pd become proton−electron pairs, and are then delivered to oxide surfaces17. However, the hydrogen spillover induces the continuous reduction of reducible oxides, leading to the reconstruction of the metal−oxide interface18–21. In contrast, non-reducible metal oxides such as alumina oxide (Al2O3) play an important role in dispersing and stabilizing the metallic sites22,23, yet the possibility of hydrogen spillover on alumina remains debated16,24,25. Previous research showed that spillover on aluminum oxide is mediated by low-coordinated aluminum centers but is about ten orders of magnitude slower than on reducible oxides16. Insulators like Al2O3 lack both accessible electronic conduction bands and reducible cations capable of stabilizing mobile H+−e− pairs, making them generally considered inactive for hydrogen migration. In this work, we demonstrate the activeness of hydrogen spillover on non-reducible Al2O3-supported Pd single-atom catalysts mediated by a phosphorus−hydroxyl (P−OH) bridge site for semi-hydrogenation of acetylene. Using in situ spectroscopy, kinetic analysis, and density functional theory (DFT) calculations, we reveal that phosphorus (P) occupied penta-coordinated Al3+ ions (AlV3+) in the form of P−O−Al, which stabilizes the isolated Pd atoms for H2 dissociation and lowers the energy barrier for hydrogen spillover via the P−OH. Gas-phase acetylene collided and hydrogenated with the as-formed P−OH, following the Eley-Rideal type mechanism, making the inert Al2O3 active in the hydrogenation reaction. As a result, the Pd1/P-Al2O3 catalysts with extremely low Pd loading at 50 ppm exhibited over 99% of acetylene conversion and 95% of ethylene selectivity at 100 °C for the semi-hydrogenation of acetylene in excess ethylene during a 200-hour test. This strategy was also applicable to other metals, e.g., Ru, Cu, and Ni, supported on P/Al2O3 for acetylene hydrogenation. This study provides evidence for relay hydrogenation on non-reducible oxides supported single-atom catalysts promoted by hydrogen spillover. Results Location and coordination of P on Al2O3 The P-modified Al2O3 was prepared by a simple impregnation method. The structure of P-modified Al2O3 (denoted as xP-Al, x is the P loading, Supplementary Table 1) supports were controlled by varying the P loadings. The as-synthesized P-Al supports and γ-Al2O3 exhibited similar characteristic peaks in both peak position and intensity, as shown in X-ray Diffraction (XRD) profiles (Supplementary Fig. 1)26–28. Based on the element and valence states analysis by X-ray photoelectron spectroscopy (XPS), P atoms were enriched on the surface of Al2O3 (Supplementary Table 2) and kept in +5 valence state (Supplementary Fig. 2)29,30. Solid-state 27Al Magic-Angle-Spinning Nuclear Magnetic Resonance (27Al MAS-NMR) was performed to confirm the location of P on surface-specific Al3+ sites (Fig. 1a, Supplementary Table 3). In particular, the signal corresponding to penta-coordinated AlV3+ (chemical shift at 36.4 ppm) diminished with increasing P loading and disappeared completely in the 2.4P-Al sample with 2.4 wt% of P loading23,31–33. The consumption of AlV3+ increased linearly as a function of P content with P content up to 2.4% (Fig. 1b), while the signal for octahedral AlVI3+ at around 14 ppm progressively intensified (Supplementary Table 3). These results suggest that P is likely located at AlV3+ via oxygen bridging. Complementary, the 31P MAS-NMR spectra demonstrated a single resonance peak at −6.9 ppm chemical shift, corresponding to the P−O−Al coordination (Fig. 1c, Supplementary Table 4)34,35. However, further increasing P loading for 4.8P-Al, new resonances emerged at 44.2 ppm in 27Al MAS-NMR and at −21.7 ppm in 31P MAS-NMR spectra, which are ascribed to the formation of AlPO4, indicating the incipient aggregation of P species36,37. Evidence of location and coordination of P and Pd on Pd1/P-Al catalyst. a 27Al MAS-NMR spectra of P modified Al2O3 supports, 600 MHz at room temperature. b linear relation between P content and consumed AlV3+ specie. c 31P MAS-NMR spectra of P promoted Al2O3 supports, 600 MHz at room temperature. EDS mapping of d P and e Pd on 0.005Pd1/2.4P-Al for TEM image in Supplementary Fig. 3. f, g AC-HAADF-STEM images of 0.005Pd1/2.4P-Al, where Pd atoms were marked by red circle. h Fourier transforms (FTs) of the EXAFS spectra of 0.03Pd1/2.4P-Al at Pd K-edge after H2 reduction at 150 °C. i Proposed structure for Pd1/P-Al. Structural characteristics of single-atom Pd on P-Al2O3 support The introduction of P induced sufficient sites for Pd dispersion. Transmission electron microscopy (TEM) and aberration-corrected high-angle-annular-dark-field scanning transmission electron microscopy (AC-HAADF-STEM) images of 2.4P-Al supported Pd catalyst with 50 ppm of Pd loading (denoted as 0.005Pd1/2.4P-Al) evidenced that Pd was atomically dispersed on support, without observable Pd nanoparticles or clusters (Fig. 1d–g, Supplementary Fig. 3). Fourier-transformed (FT) k2-weighted extended X-ray absorption fine structure (EXAFS) spectra demonstrated a broad and asymmetric peak in R space, reflecting the coexistence of multiple coordination path. A multiple scattering path fitting model with Pd−O coordination at 2.01 Å and Pd−P coordination at 2.36 Å was employed to reproduce the experimental EXAFS data accurately. The coordination numbers for Pd−O and Pd−P were determined to be 2.0 and 1.2, respectively (Fig. 1h, Supplementary Fig. 4, Supplementary Table 5). The negligible contribution of the Pd−Pd scattering path further confirmed the atomic dispersion of Pd atoms. In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) using CO as the probe molecule for 0.005Pd1/2.4P-Al catalyst exhibited a single peak at 2097 cm-1 for CO adsorption on Pd single atoms, while an obvious peak caused by bridge-adsorbed CO on two adjacent Pd atoms appeared at around 1967 cm-1 for 0.005Pd/Al, demonstrating the high dispersion of Pd atoms on 0.005Pd1/2.4P-Al catalyst (Supplementary Fig. 5)38–40. Figure 1i demonstrates the proposed structure for Pd1/P-Al with atomically dispersed Pd atoms coordinated by one phosphorus and two oxygen atoms. By contrast, considerable Pd particles with sizes of 1.0 and 0.9 nm appeared on 0.005Pd/Al and 0.005Pd/1.2P-Al (Supplementary Figs. 6 and 7), indicating the important roles of P in anchoring and dispersing Pd nanoparticles. Catalytic performance in acetylene semi-hydrogenation The synthesized Pd1/P-Al catalysts were evaluated for acetylene semi-hydrogenation in a gas mixture of acetylene, hydrogen, and excess ethylene (molar ratio = 1:30:30). Over the unpromoted 0.005Pd/Al catalyst, full acetylene conversion was achieved at 100 °C, but ethylene selectivity was determined to be 14%, with ethane as the main product (Fig. 2a, see Catalytic Test in Methods for details, Supplementary Fig. 8). A progressive improvement in ethylene selectivity was observed with increasing P loading. Notably, both 0.005Pd1/2.4P-Al and 0.005Pd1/4.8P-Al (Supplementary Fig. 9) markedly suppressed ethane formation, reaching an ethylene selectivity of about 95% with >99% and 91% of acetylene conversion at 100 °C, respectively. Further increasing the reaction temperature to 150 °C, both the 0.005Pd1/2.4P-Al and 0.005Pd1/4.8P-Al catalysts achieved full acetylene conversion, while retaining high ethylene selectivity of approximately 90%. By contrast, the ethylene selectivity dropped drastically to −453% for 0.005Pd/Al. To further verify the critical role of isolated Pd single atoms in chemo-selective hydrogenation of acetylene, P/Al-supported Pd nanoparticle catalysts were fabricated by elevating the pre-reduction temperature from 150 °C to 200 °C and 300 °C (Supplementary Fig. 10a). The ethylene selectivity dropped markedly to 84% and 22% for the Pd nanoparticle counterparts pre-reduced at 200 °C and 300 °C, respectively (Supplementary Fig. 10b). Catalytic performance of P-Al supported Pd single-atom catalysts in acetylene semi-hydrogenation. a Acetylene conversion and ethylene selectivity over 0.005Pd/Al, 0.005Pd/1.2P-Al, 0.005Pd1/2.4P-Al, and 0.005Pd1/4.8P-Al. Reaction conditions: 0.6 vol% C2H2, 18 vol% H2, and 18 vol% C2H4 gas mixed with balanced N2, GHSV = 30,000 mL·gcat−1·h−1, 0.1 MPa, T = 100 °C, catalyst 100 mg. b Comparison of space-time yield (STY) normalized by metal amount of 0.005Pd1/2.4P-Al with reported catalysts at different reaction temperatures. Detailed information of each catalyst is provided in Supplementary Table 6. c Long-term stability test over 0.005Pd1/2.4P-Al at 100 °C for 200 hours. Reaction conditions: 0.6 vol% C2H2, 18 vol% H2, and 18 vol% C2H4 gas mixed with balanced N2, GHSV = 30,000 mL·gcat−1·h−1, 0.1 MPa, T = 100 °C, catalyst 100 mg. Space-time yield (STY) was measured at different reaction conditions, and the optimal 0.005Pd1/2.4P-Al catalyst exhibited the highest STY of 144.64 mol·gPd-1·h-1 at the low temperature of 100 °C. The STY normalized by active metal amount indicates that, despite the variations in metal loading (Pd, Au, Ru, Ni, and Cu), temperature, and gaseous hourly space velocity (GHSV) (Fig. 2b, Supplementary Table 6), 0.005Pd1/2.4P-Al catalysts achieved superior values by reducing Pd usage by one order of magnitude, which outperforms PdAg/Al2O3 whereas 0.03 wt % Pd is necessary for commercial (mimic) catalysts and reported catalysts that required higher temperatures for acetylene conversion. The long-term catalytic stability of the 0.005Pd1/2.4P-Al catalyst was further evaluated under co-feeding conditions with excess ethylene, a critical test scenario that closely mimics industrial operating environments and is therefore essential for assessing practical applicability. As illustrated in Fig. 2c, the catalyst demonstrated exceptional durability, maintaining stable performance over an extended period exceeding 200 hours. Throughout the test, a consistent acetylene conversion of >99% was achieved, accompanied by a high ethylene selectivity sustained above 95%. Remarkably, post-reaction characterization confirmed that Pd species remained atomically dispersed without noticeable aggregation (Supplementary Fig. 11). Meanwhile, the used 0.005Pd1/2.4P-Al catalyst exhibited identical CO-DRIFTS spectra to the fresh sample, without the appearance of bridge-adsorbed CO on the Pd nanoparticle (Supplementary Fig. 5). These results collectively highlight the structural and catalytic robustness of the catalyst under practical conditions. Proximity-dependent semi-hydrogenation rate To figure out the contribution of Pd sites and P-Al support in acetylene semi-hydrogenation, we physically mixed 0.005Pd1/2.4P-Al catalyst and support (2.4P-Al or Al2O3) (mass ratio, 1:1) with different proximity (Fig. 3a). In dual-bed mixture, 0.005Pd1/2.4P-Al and 2.4P-Al support were arranged in a dual-bed configuration, with a 2 mm inert quartz sand as the contact inhibitor. In the granule mixture, granules of 0.005Pd1/2.4P-Al and 2.4P-Al support with the size of 40–60 mesh were well mixed, narrowing the distance to approximately 250–400 μm. In powder mixture, 0.005Pd1/2.4P-Al and 2.4P-Al support powder were finely ground to get the intimate interaction between these two compositions. Proximity-dependent semi-hydrogenation rate between 0.005Pd1/2.4P-Al and 2.4P-Al. a Schematic diagram for the physical mixture of 0.005Pd1/2.4P-Al and support with different proximity. b Acetylene semi-hydrogenation over the physical mixture of 0.005Pd1/2.4P-Al and an equal amount of 2.4P-Al support with different proximity. c Acetylene semi-hydrogenation over the powder mixture of 0.005Pd1/2.4P-Al and different quality of 2.4P-Al support. The red solid lines are to guide the eyes. Reaction conditions: 0.6 vol% C2H2, 18 vol% H2, and 18 vol% C2H4 gas mixed with balanced N2, GHSV = 600,000 mL·gcat−1·h−1, 0.1 MPa, T = 100 °C, Catalyst 5 mg. The hydrogenation over the physical mixture was operated at high GHSV (600,000 mL·gcat-1·h-1) to eliminate the influence of internal and external diffusion and highlight the intrinsic promotion effect of hydrogen spillover on acetylene hydrogenation (See detailed information in Supplementary Fig. 12). The ethylene formation rate over pristine 0.005Pd1/2.4P-Al catalyst was determined to be 251.49 mol·gPd-1·h-1. Notably, the introduction of 2.4P-Al in the powder mixture significantly enhanced ethylene formation rate to 397.78 mol·gPd-1·h-1 (Fig. 3b, Supplementary Fig. 13a), which is 1.6 times to that of 0.005Pd1/2.4P-Al, although 2.4P-Al itself is inactive for acetylene hydrogenation (Supplementary Fig. 14). In sharp contrast, neither the dual-bed nor the granule mixture configuration elicited any significant improvement in ethylene formation. For the powder mixture with different amounts of 2.4P-Al support, a systematic increase in acetylene conversion was observed with increasing 2.4P-Al content (Fig. 3c, Supplementary Fig. 13b), while the ethylene formation rate was restricted to about 405.21 mol·gPd-1·h-1. By contrast, the addition of Al2O3 to the physical mixture of 0.005Pd1/2.4P-Al and Al2O3 produced virtually no enhancement in the performance (Supplementary Fig. 15). This difference indicates that P might play a significant role in communicating Pd and nominally inactive supports. The inferior hydrogenation performance over 0.005Pd/AlPO4 further highlights the crucial role of P-Al support in promoting acetylene conversion (Supplementary Fig. 16). Accelerated hydrogen spillover on Al2O3 via P−OH bridging sites To figure out the potential communication behavior between Pd and P-Al support, we performed the H2-temperature programmed reduction (H2-TPR). As shown in Supplementary Fig. 17, 2.4P-Al support exhibited two reduction peaks at 284 °C and 461 °C, ascribed to the reduction of P=O and P−O−Al, respectively. Comparatively, the temperature of reduction peaks shifted to 228 °C and 424 °C for 0.005Pd1/2.4P-Al. It seems that the introduction of Pd single atoms contributes to the reduction of support. In situ DRIFTS of H-D exchange experiments were conducted to indicate the H2 activation and spillover. As shown in Supplementary Fig. 18, H2 treatment has a negligible effect on the surface of Al2O3 and P-Al, thus indicating that the hydroxyl groups on Al2O3 and P-Al surface are saturated prior to H2 exposure. After D2 treatment on 0.005Pd/Al, negative peaks appeared at 3763 and 3663 cm-1 and positive peaks rose simultaneously at 2747 and 2702 cm-1, which means that the H atoms of the hydroxyl groups on the γ-Al2O3 are replaced by D atoms (Fig. 4a, b, Supplementary Fig. 19a). Additionally, another negative peak at 3730 cm-1 ascribed to the hydroxyl groups linked to AlV3+ was also consumed with the introduction of D2. Comparatively, in addition to the Al−OH groups at 3761 and 3663 cm-1, two new negative peaks emerged at 3790 and 3678 cm-1 with corresponding positive peaks appearing at 2786 and 2711 cm-1 for 0.005Pd1/2.4P-Al (Fig. 4c, d, Supplementary Fig. 19b). The new band at 3790 cm-1 is typically assigned to hydroxyl groups on Al2O3, and the specific band at 3678 cm-1 is ascribed to the hydroxyl groups linked to the P (P−OH). The assignments of the IR bands are summarized in Supplementary Table 7. Meanwhile, the disappearance of hydroxyl groups connected to AlV3+ (110) at 3730 cm-1 for 0.005Pd1/2.4P-Al further confirmed the occupation of AlV3+ sites by P heteroatom, which is in accordance with the 27Al MAS-NMR results. These results demonstrate that the introduction of atomically dispersed Pd facilitates the reduction of the support by hydrogen spillover mediated by the P−OH. With the appearance of hydrogen spillover, electrons can also diffuse across 2.4P-Al support, as evidenced by the reduction of WO3 in the physical mixture of 0.005Pd1/2.4P-Al and WO3 under H2-Ar flow at room temperature (Supplementary Fig. 20). Meanwhile, the reduction of P-Al supports in H2-TPR results further demonstrated that surface P dopant made the support reducible and the consequent electron transfer possible (Supplementary Fig. 17). Taking together, these observations support a coupled H+−e− migration mechanism over the phosphorus-modified Al2O3 support, in which hydrogen spillover is accompanied by simultaneous electron transfer. Hydrogen spillover and hydrogenation on P modified Al2O3 support. In situ DRIFTS of H2-D2 exchange and hydrogenation experiments over a, b 0.005Pd1/2.4P-Al and c, d 0.005Pd/Al. The a.u. stands for arbitrary units. e Proposed relay hydrogenation scheme for (1) H2 activation, (2) H spillover, and (3) hydrogenation over Pd1/P-Al catalyst. P−OH as the hydrogenation sites To monitor the role of P−OH during acetylene semi-hydrogenation, we switched the gas environment to a diluted C2H2 flow after D2 treatment. For 0.005Pd1/2.4P-Al catalyst, DRIFTS analysis revealed the predominant consumption of P−OD species, accompanied by the detection of C2H2D2 product in the outlet gas (Fig. 4a, b, Supplementary Figs. 21a and 22). However, Al−OD species on Al2O3 were inert toward acetylene hydrogenation (Fig. 4c, d, Supplementary Fig. 21b). These results suggest that the surface P−OH likely serves as active hydrogenation species. This behavior differs from the conventional Horiuti-Polanyi (H-P) type mechanism, where hydrogen and acetylene competitively adsorb on a single active site to complete the stepwise hydrogenation process of C≡C triple bond to C=C double bond41,42. The surface coverage of P−OH species significantly inhibited the adsorption of both acetylene and ethylene. In accordance with the C2H2-TPD results (Supplementary Fig. 23), the absence of multi-peak reaction intermediates in the 2800-3200 cm-1 region for the adsorption of acetylene on 0.005Pd1/2.4P-Al further indicated the inhibited adsorption of acetylene on a hydrogen-pre-covered surface (Supplementary Fig. 24). By contrast, these characteristic peaks of C2H2 adsorption are obvious on 0.005Pd/Al. These observations suggest that acetylene hydrogenation may follow an unconventional Eley-Rideal (E-R) type mechanism, where gas-phase acetylene reacts directly with pre-covered H species to form ethylene43. Kinetic studies were further performed in a differential reactor to prove the E-R reaction mechanism after eliminating the influence of internal and external diffusion (Supplementary Fig. 25). Across a broad range of partial pressures of H2 (2.4-36.5 kPa) and C2H2 (0.4-2.3 kPa), the hydrogenation systems achieved high ethylene selectivity (higher than 90%) to exclude the undesired over-hydrogenation reaction. A kinetic analysis was conducted to clarify the rate-determining steps (RDS) of acetylene hydrogenation over 0.005Pd/Al and 0.005Pd1/2.4P-Al. Since acetylene hydrogenation generally follows the H-P type mechanism, we considered a stepwise hydrogenation process in the elementary steps for both catalysts (See the Kinetic analysis section in Supplementary Information for details). The experimental reaction order with respect to C2H2 was found to be 0.96 over 0.005Pd/Al (Supplementary Fig. 26a) and 0.94 over 0.005Pd1/2.4P-Al (Supplementary Fig. 27a), respectively. Whereas the reaction order with respect to H2 was determined to be 0.91 (Supplementary Fig. 26b) and 0.47 (Supplementary Fig. 27b) in the range of 2.4-6.0 kPa for 0.005Pd/Al and 0.005Pd1/2.4P-Al, and close to 0 in the range of 12.2-36.5 kPa for 0.005Pd/Al and 18.2-36.5 kPa for 0.005Pd1/2.4P-Al. Comparatively, the reaction order of H2 fell within a narrower range of 0-0.5 over 0.005Pd1/2.4P-Al, consistent with the E-R mechanism. At low H2 partial pressures, the hydrogen species generated from H2 dissociation on Pd sites is efficiently consumed by the relay hydrogenation step. However, at high H2 partial pressure, the zero-order kinetics with respect to H2 suggested that P−OH contributed from H spillover was saturated, and thus decreased the competitive adsorption of C2H2 on the surface. Further increasing the H2 partial pressure did not enhance ethylene formation rate. Accordingly, in the proposed E-R reaction mechanism, the rate-determining step (RDS) was identified to be the first hydrogenation step (see the Kinetic analysis section in Supplementary Information for details). The apparent activation energy (Ea) was estimated by Arrhenius-type plots. The Ea was determined to be 36.1 and 53.3 kJ∙mol-1 for 0.005Pd/Al and 0.005Pd1/2.4P-Al, respectively (Supplementary Fig. 28). The higher Ea for 0.005Pd1/2.4P-Al should be ascribed to the poor capability in H2 activation (Supplementary Fig. 29). Based on these observations, a scheme for P−OH evolution and hydrogenation is proposed, that is, H species generated from the heterolytic dissociation of H2 on Pd single atom spilt to P-Al surface and reduced P=O, forming P−OH. Subsequently, the gas-phase acetylene collided with surface P−OH species, forming ethylene (Fig. 4e). By contrast, both acetylene and H2 strongly adsorbed on the Pd nanoparticle, which would result in the undesired over-hydrogenation, leading to the formation of ethane. DFT calculation We then performed DFT calculations to demonstrate the occurrence of hydrogen spillover and hydrogenation on P-Al surface. The location and atomic structure of P species on Al2O3 were confirmed by optimizing the adsorption models of P atoms or POx clusters on the surface of Al2O3. On a (3×3) Al2O3 (110) surface44, additional O atoms in POx cluster were indispensable to maintain a high valence state, i.e., up to +5, for P according to the XPS results. As shown in Fig. 5a, 41 distinct local energy minima were identified (Supplementary Figs. 30-34, Supplementary Tables 8-11). Among these configurations, the structure featuring P−O tetrahedron with one P=O and three P−O−Al bonds was found to be the most stable structure (Fig. 5b). The DFT results further revealed that P occupied the AlV3+ sites in a 1:1 stoichiometry, consistent with the observed depletion of AlV3+ signal in 27Al MAS-NMR spectra. Therefore, it is concluded that the P dopant is enriched yet atomically dispersed on the surface of Al2O3 and preferentially occupies surface AlV3+ in the form of P−O−Al. Hydrogenation mechanism over P-Al-supported Pd single-atom catalysts. a The net atomic charge of P with corresponding stability of all models have been considered. AlPO4 with +2.80 |e| net atomic charge was employed for reference (Supplementary Fig. 30). b The optimal model for PO3 site on Al2O3 (110). The orange sphere denotes P atom, while O atoms are in red and Al atoms are in light blue. c Calculated energy barriers for hydrogen spillover on Pd/Al and Pd1/P-Al. d Calculated energy barriers for acetylene semi-hydrogenation on Al2O3 and P-Al surfaces. The asterisk (*) represents the adsorbed state of the species on the catalyst surface. TS1-TS3 represent the transition states of the stepwise hydrogenation. The orange sphere denotes P atom, while O atoms are in red and Al atoms are in light blue. In acetylene and ethylene molecules, the gray spheres denote C atoms, while H atoms are in white. Regarding the crucial hydrogen spillover process, DFT calculations revealed a high energy barrier (1.45 eV) on pristine Al2O3 that strongly inhibited the hydrogen spillover. In comparison, P was found to function as an efficient hydrogen transfer medium via the formation of P−OH and significantly decreased the energy barrier (0.14 eV) for hydrogen spillover on P-Al surface, despite of the higher energy barrier for H2 heterolytic activation, as shown in Fig. 5c. This DFT result is in correspondence with the in situ H2-D2 exchange spectroscopic results for hydrogen migration (P−OH and P−OD evolutions, Fig. 4a–d). Such enhanced hydrogen spillover enabled continuous replenishment of active hydrogen species, thereby maintaining high hydrogen coverage on the catalyst surface. Furthermore, the results showed that C2H2, C2H4, and hydrogen acted as electron donors to POx sites (Supplementary Figs. 35–37), among which, hydrogen exhibited stronger adsorption capacity and preferentially covered the P=O site under all different hydrogen coverages (Supplementary Table 12). Hence, the subsequent adsorption of both acetylene and ethylene was drastically weakened. This weakened ethylene absorption capacity favors its desorption, which was beneficial for ethylene selectivity. Our DFT calculations further revealed the utilization of split hydrogen atoms on P-Al surface for hydrogenation (Fig. 5d). With the pre-covered hydrogen atoms from hydrogen spillover, the potential energy diagram of the elementary steps for the selective hydrogenation of acetylene were evaluated. Non-dissociated adsorption of acetylene was taken as the initial reaction states, as proved by the adsorption configuration and corresponding adsorption energy of acetylene and H atoms (Supplementary Fig. 38). As shown in the energy profiles of the stepwise hydrogenation of acetylene, the barrier for the first H addition transition state (TS1) was lower on P-Al surface (1.16 eV) than it over pristine Al2O3 surface (1.52 eV). This expected result demonstrated that the promoted hydrogen spillover is favorable for hydrogenation. Moreover, the insurmountable barrier (1.49 eV) for the second hydrogen addition transition state (TS2) over Al2O3 further restricted its hydrogenation capacity. By contrast, P modification significantly reduced the energy barrier for the second hydrogen addition step, aligning with the kinetic experiments, that is, the first hydrogenation step is the RDS. Comparing the desorption with the further hydrogenation barrier (1.48 eV), C2H4 preferentially desorbed from P-Al surface instead of over-hydrogenation, resulting in the superior ethylene selectivity. In comparison, Al2O3 presented a stronger tendency towards further hydrogenation, which is detrimental to ethylene selectivity. Related Me/P-Al catalysts Considering the functional role of Pd single atoms in H2 activation, we explored replacing Pd with other metals. As shown in Fig. 6, Me/2.4P-Al (Me = Ru, Cu, and Ni) exhibited superior catalytic performance compared to their counterparts Me/Al. Specially, while 0.01Ru/Al was inactive for acetylene hydrogenation, 0.01Ru/2.4P-Al achieved complete acetylene conversion and 93% of ethylene selectivity at 150 °C, this underscores the importance of the metal center in activating H2 and the role of hydrogen spillover in supplying active hydrogen species. The inferior low-temperature performance for 0.01Ru/2.4P-Al was ascribed to its low H2 activation capacity at low temperature, as evidenced by the notably small H2 desorption peak area in H2-TPD results after the pre-treatment under H2 flow at 100 °C (Supplementary Fig. 39). Acetylene semi-hydrogenation over related Me/P-Al catalysts (Me = Ru, Cu, and Ni). a Acetylene conversion and ethylene selectivity over 0.01Ru/2.4P-Al and 0.01Ru/Al at 100 °C and 150 °C. b Acetylene conversion and ethylene selectivity over 0.1Cu/2.4P-Al and 0.1Cu/Al at 250 °C, over 0.05Ni/2.4P-Al and 0.05Ni/Al at 200 °C respectively. Reaction conditions: 0.6 vol% C2H2, 18 vol% H2, and 18 vol% C2H4 gas mixed with balanced N2, GHSV = 30,000 mL·gcat-1·h-1, 0.1 MPa, Catalyst 100 mg. Discussion In summary, this work described an unconventional relay hydrogenation process, where the non-reducible Al2O3 was devoted to acetylene hydrogenation, while the Pd single atom contributed to H2 activation. The optimized 0.005Pd1/2.4P-Al catalyst, with an extremely low Pd loading of 50 ppm and 2.4 wt% of P loading, exhibited over 99% of acetylene conversion and 95% of ethylene selectivity in excess ethylene during a 200-hour test at 100 °C. We demonstrate that phosphorus dopant on Al2O3 significantly decreased the energy barrier for the hydrogen spillover from a Pd single atom to non-reducible Al2O3 via the formation of phosphorus−hydroxyl. Acetylene molecules subsequently reacted with as-formed P−OH, following an E-R-type mechanism. Besides Pd single atom, other metals, e.g., Ru, Cu, and Ni, with H2 activation ability were also applicable for the hydrogen spillover mediated hydrogenation process. Methods Preparation of catalysts Pretreatment of Al2O3 γ-Al2O3 (99.5%, Shanghai Adamas Co., Ltd.) was pre-heated at 700 °C for 2 h in air before the impregnation. Synthesis of xP-Al Phosphorus-doped Al2O3 with different phosphorus loading (1.2, 2.4, and 4.8 wt%), named as xP-Al (x = 1.2, 2.4, and 4.8), was prepared via a simple wet impregnation method. Typically, the required amount of diammonium phosphate (99.5%, Aladdin biological technology Co., Ltd.) was dissolved in deionized water. Then, 1 g of γ-Al2O3 (99.5%, Shanghai Adamas Co., Ltd.) was added to the clear solution. The mixture was stirred and kept in an ultrasonic bath for 1 h and dried at 80 °C overnight. The xP-Al support was finally obtained after being calcined at 600 °C in air for 4 h. Synthesis of 0.005Pd1/xP-Al Pd single atoms were deposited onto xP-Al support using a simple wet impregnation method. An appropriate amount of PdCl2 (99.9%, J&K China Chemical Co., Ltd.) based on 0.005 wt% of Pd loading was dissolved in dilute HCl (99%, Macklin biological technology Co., Ltd.) to form a clear solution. Then 1 g of xP-Al was added to the light brown solution. The mixture was stirred and kept in an ultrasonic bath for 1 h, and dried at 80 °C overnight. Finally, the dried sample was reduced at 150 °C under 10 vol% H2-N2 flow (50 mL min−1). Synthesis of 0.005Pd/Al The procedure for the preparation of 0.005Pd/Al was the same as 0.005Pd1/xP-Al, and the support is γ-Al2O3 instead of P-Al. Typically, an appropriate amount of PdCl2 (99.9%, J&K China Chemical Co., Ltd.) based on 0.005 wt% of Pd loading was dissolved in dilute HCl (99%, Macklin biological technology Co., Ltd.) to form a clear solution. Then 1 g of γ-Al2O3 was added to the light brown solution. The mixture was stirred and kept in an ultrasonic bath for 1 h, and dried at 80 °C overnight. Finally, the dried sample was reduced at 150 °C under 10 vol% H2-N2 flow (50 mL min-1). Synthesis of Me (Ru, Cu, and Ni)/2.4P-Al The preparation of Me (Ru, Cu, and Ni)/2.4P-Al catalysts was identical to 0.005Pd1/2.4P-Al, except for the metal species. Typically, an appropriate amount of metal precursor (Ru, Cu, or Ni nitrate, 99.5%, Aladdin biological technology Co., Ltd.) based on the metal loading was supported on P-Al via a simple wet impregnation method. The metal precursor was dissolved in deionized water, and the solution was then mixed with 1 g of 2.4P-Al, followed by stirring and kept in an ultrasonic bath for 1 h. The mixture was then dried at 80 °C in air overnight. Finally, the dried sample was reduced at 150 °C under 10 vol% H2-N2 flow (50 mL min-1). Synthesis of 0.005Pd/AlPO4 The procedure for the preparation of 0.005Pd/AlPO4 was the same as 0.005Pd1/xP-Al and 0.005Pd/Al, and the support is AlPO4 instead of P-Al or γ-Al2O3. Characterization X-ray powder diffraction (XRD) patterns were collected on a Rigaku Smartlab, operated at 60 kV and 220 mA with 2θ values from 20° to 80° at 8° min-1 using graphite filtered Cu Kα radiation (λ = 1.5406 Å). The elemental composition of the catalysts was analyzed by inductively coupled plasma optical emission spectroscopy (ICP-OES) (VISTA-MPX, Varian). Before measurements, the catalysts were dissolved in the mixed solutions of hydrofluoric acid and aqua regia. X-ray photoelectron spectroscopy (XPS) was recorded on a ThermoFisher ESCALAB-250Xi instrument equipped with an Al Kα X-ray source (hν = 1486.6 eV). The binding energies were calibrated using the C 1 s peak at 284.5 eV as a reference. Transmission electron microscopy (TEM) was performed on a JEM-2100F under a working voltage of 200 kV. The transmission electron microscope is equipped with an energy-dispersive X-ray spectroscopy (EDS) detector cooled by liquid nitrogen for elemental analysis. Aberration-corrected high-angle-annular-dark-field scanning transmission electron microscopy (AC-HAADF-STEM) measurements were carried out on a JEM ARM 200 F from JEOL operated at 200 kV. Solid-state 27Al and 31P Magic-Angle-Spinning Nuclear Magnetic Resonance (MAS-NMR) were recorded on a JEOL JNM ECZ600R spectrometer (Tokyo, Japan). The experiments were carried out at a frequency of 600 MHz and a field strength of 14.1 T with scanning of 300 times and relaxation delay of 2 s using a 4 mm diameter zirconia tube at room temperature. All samples were calcined at 600 °C in air before MAS-NMR measurements. In situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) for H2-D2 exchange and hydrogenation experiments were recorded on a Thermo Scientific Nicolet IS50 spectrometer, equipped with a Harrick Scientific DRIFT cell and a mercury-cadmium-telluride MCT detector cooled by liquid N2. The output products were analyzed on Hiden QIC-20 mass spectrometer (C2H2 and C2H2D2, m/e equals to 26 and 30, respectively). Typically, the samples to be measured were reduced at 150 °C under 10 vol% H2-N2 flow (40 mL min-1) and cooled to 30 °C. Background spectra were then collected. The spectra for H2-D2 exchange were collected after switching the gas from H2 to D2. Subsequently, the spectra for the hydrogenation experiments were collected after switching the gas from D2 to C2H2. The desired spectra for H2-D2 exchange and hydrogenation experiments were acquired after subtracting the corresponding background spectra. In situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) for C2H2 adsorption was recorded on the same apparatus. Typically, the samples to be measured were reduced at 150 °C under 10 vol% H2-N2 flow (40 mL min-1) and cooled to −30 °C to avoid the hydrogenation on the surface of catalysts. Background spectra were then collected. The spectra for C2H2 adsorption were collected after switching the gas from H2 to C2H2. The desired spectra for C2H2 adsorption were acquired after subtracting the corresponding background spectra. In situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) for CO adsorption was recorded on the same apparatus. Typically, the samples to be measured were reduced at 150 °C under 10 vol% H2-N2 flow (40 mL min-1) and cooled to −100 °C. Background spectra were then collected. The spectra for the tested samples were collected after treatment in a 10 vol% CO-He flow (40 mL min-1) and subsequent Ar purging (40 mL min-1). The desired spectra for CO adsorption were acquired after subtracting the corresponding background spectra. H2 temperature-programmed desorption (H2-TPD) was performed on the Micromeritics Auto Chem II 2920 chemisorption apparatus. Typically, 100 mg of the developed catalysts were reduced in a U-type reactor at 150 °C under 10 vol% H2-N2 flow (40 mL min-1) and cooled to −40 °C. The samples were then treated under 10 vol% H2-He (40 mL min-1) for another 1 h. Finally, the samples were heated to 800 °C at a rate of 10 °C min-1 under He flow (30 mL min-1). H2 temperature-programmed reduction (H2-TPR) was also performed on the Micromeritics Auto Chem II 2920 chemisorption apparatus. Typically, 100 mg of the catalysts were heated in U-type reactor at 300 °C for 1 h under Ar flow (40 mL min-1) and then cooled to 80 °C. Finally, the samples were heated to 800 °C at a rate of 10 °C min-1 under 10 vol% H2-Ar flow (30 mL min-1). C2H2 and C2H4 temperature-programmed desorption (C2H2, C2H4-TPD) were also performed on the Micromeritics Auto Chem II 2920 chemisorption apparatus. Typically, 100 mg of the developed catalysts were reduced in a U-type reactor at 150 °C under 10 vol% H2-N2 flow (40 mL min-1) and cooled to -40 °C. The samples were then pre-adsorbed with C2H2 or C2H4 for another 1 h. Finally, the samples were heated to 600 °C at a rate of 10 °C min-1 under the He flow (30 mL min-1) after the purification. H2-D2 exchange experiments were also performed on the Micromeritics Auto Chem II 2920 chemisorption apparatus and the output products were analyzed on Hiden QIC-20 mass spectrometer (HD, m/e equals to 3). Typically, 100 mg of the developed catalysts were reduced in a U-type reactor at 150 °C under 10 vol% H2-N2 flow (40 mL min-1) and cooled to −40 °C. The gas flow was then switched to 10 vol% D2-N2 flow (40 mL min-1) and heated to 600 °C at a rate of 10 °C min-1. The X-ray absorption spectra (XAS) data at Pd K edge was collected at BL14W1 station in Shanghai Synchrotron Radiation Facility (SSRF). The electron storage ring was operated at 3.5 GeV with a maximum current of 250 mA. The data were collected at room temperature in a fluorescence excitation mode. The tested sample was pre-reduced under flowing H2-N2 and cooled down to room temperature. After that, the reduced sample was pelletized as disks of 13 mm diameter with 1 mm thickness. The obtained EXAFS spectra were analyzed using Athena and Artemis software ver. 0.9.25 implemented in the Demeter package. Catalytic tests The acetylene semi-hydrogenation reaction was performed on a fixed-bed reactor, equipped with a quartz tubular reactor that is 8 mm in inner diameter and 26 cm in length. 100 mg of the developed catalysts (40–60 mesh), diluted with 1.4 g quartz, was packed into the tubular reactor. Prior to the hydrogenation reaction, the tested catalysts were reduced at 150 °C under 10 vol% H2-N2 flow (40 mL min-1). Afterward, a mixture of 0.6 vol% C2H2, 18 vol% H2, and 18 vol% C2H4 balanced with N2 at a GHSV of ~ 30,000 mL·gcat-1·h-1 was fed and the pressure for the reaction system was set to be 1 bar. The bed temperature was gradually increased from 50 °C to 150 °C with a gap of 10 °C, and the exhausted gas was injected into the gas chromatograph. The products were analyzed by GC 2060 (Shanghai Ruimin Instrument)) equipped with a flame ionization detector (FID) and detector columns (PORAPAK-N). Acetylene conversion and ethylene selectivity were calculated according to Eq. (1) and Eq. (2) 4,10,45: Supplementary information Supplementary information Transparent Peer Review file Source data Source Data Acknowledgements We gratefully acknowledge the technical support from the Advanced Instrumental Analysis Center, School of Chemical Engineering and Technology, Tianjin University, for their provision of high-performance characterization services. We acknowledge the Shanghai Synchrotron Radiation Facility (SSRF) for providing the experimental platform and resources essential to this study. Author contributions J.L.G. supervised the project. X.H.W. and R.L. performed the experiments, characterization, and DFT calculation. X.H.W., S.C., Z.-J.Z., and J.L.G. designed the project. K.G.T. and G.D.S. contributed to characterization of the hydrogenation catalysts. X.C. and C.L.P. contributed to useful discussion. All the authors contributed to the writing of the manuscript. Peer review Peer review information Nature Communications thanks Kohsuke Mori, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding S.C. discloses support for the research of this work from the National Key R&D Program of China [2023YFA1507800] and the National Natural Science Foundation of China [grant number 22478279]. J.L.G. discloses support for the research of this work from the National Natural Science Foundation of China [grant number 22121004]. X.H.W. discloses support for the research of this work from the National Natural Science Foundation of China [grant number 22408266]. This work is also supported by the Haihe Laboratory of Sustainable Chemical Transformations, the Program of Introducing Talents of Discipline to Universities (grant number BP0618007) and the XPLORER PRIZE. R.L., K.G.T., G.D.S., X.C., C.L.P., and Z.-J.Z. declares no relevant funding. Data availability The data generated in this study are provided in Supplementary Information/Source Data file. Data are available from the corresponding authors upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Xianhui Wang, Luo Ran. Contributor Information Sai Chen, Email: saic@tju.edu.cn. Zhi-Jian Zhao, Email: zjzhao@tju.edu.cn. Jinlong Gong, Email: jlgong@tju.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-74503-0. References Associated Data Supplementary Materials Supplementary information Transparent Peer Review file Source Data Data Availability Statement The data generated in this study are provided in Supplementary Information/Source Data file. Data are available from the corresponding authors upon request. Source data are provided with this paper.