Reduction behavior of MOFs in hydrogen spilloverIn order to deeply understand the hydrogen spillover behavior of varied MOFs including reducible and nonreducible ones, we first examined the thermal stability of frequently-used MOFs with different reduction potential energy28 (Supplementary Fig. 1) (Zn-based ZIF-8, Zr-based UiO-66, Co-based ZIF-67, Fe-based MIL-101 and Cu-based MOF-2, Supplementary Figs. 2–6) and their composite with immobilized Pt nanoparticles (denoted as MOFs/Pt, Supplementary Figs. 7–11) in flowing H2 gas by thermogravimetric analysis (TGA).As shown by the TGA curves (Fig. 1a–c), MOFs/Pt with high reduction potential energy (Cu-based, Fe-based, and Co-based ones) are more readily decomposed than the pristine MOFs. Using Cu-based MOFs as an example, the temperature at which the weight fraction w decreases to 65 wt% (Tw65) for Cu-MOF-2 and Cu-MOF-2/Pt in H2 flows is 386 °C and 323 °C, respectively. The evident difference of ΔTw65 is caused by distinct decomposition mechanisms. In the Cu-MOF-2/Pt system, hydrogen atoms are produced by the dissociation of gaseous H2 molecules on the embedded Pt surface, which subsequently migrate from Pt to Cu-MOF-2 via hydrogen spillover. As the split hydrogen atoms have higher reactivity than H2, hydrogenolysis of Cu-MOF-2 happens more vigorously, where Cu-O coordination bonds are cleaved and Cu2+ species are reduced at lower temperatures. The reduction of Cu2+ ions is proved by X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS) (Supplementary Figs. 12 and 13). A similar phenomenon is observed in the Fe-based and Co-based systems (Supplementary Figs. 14 and 15).Fig. 1: Thermal stability of several typical MOFs and MOFs/Pt in flowing H2 gas.TGA curve and ΔTw65 of MOFs and MOFs/Pt: a Cu-MOF-2 system. b Fe-MIL-101 system. c Co-ZIF-67 system. d Zr-UiO-66 system. e Zn-ZIF-8 system.Conversely, Fig. 1d, e indicates that such hydrogen spillover-promoted decomposition does not take effect in the Zr-UiO-66/Pt and Zn-ZIF-8/Pt systems. This is likely a consequence of their much lower reduction potential energy, meaning that the migrated hydrogen atoms cannot reduce Zr4+ and Zn2+ ions, and thus the frameworks of Zr-UiO-66/Pt and Zn-ZIF-8/Pt remain stable against hydrogenolysis (Supplementary Figs. 16–19).To sum up, analogously to all the reported results25,27,29, the inherent crystal structure of the reducible MOFs is destroyed by activated hydrogen, which greatly reduces their practical value as the catalytic supports. Therefore, we focus the study on the hydrogen spillover effect of more meaningful and stable nonreducible MOFs. Zn-based ZIF-8 is chosen as the platform material considering its high thermal stability (Supplementary Figs. 20–24), functional group flexibility (Supplementary Fig. 25), and extremely low reduction potential energy (Supplementary Fig. 1).Characterization of sandwich model catalystThe sandwich nanostructured MOFs@Pt@MOFs catalyst19 with the outer MOFs-shell of varied thickness and functional groups serves as a perfect model for exploring hydrogen spillover, ensuring the identical diffusion pathways of hydrogen atoms starting from Pt to the different external surfaces on MOFs. Accordingly, highly symmetrical and uniform Zn-ZIF-8@Pt@Zn-ZIF-8 nanocubes were prepared by electrostatic adsorption of pre-synthesized 3 nm Pt nanoparticles onto 100 nm Zn-ZIF-8 nanocube cores30 (Supplementary Fig. 26) followed by liquid-phase epitaxial growth of another Zn-ZIF-8 shells (Supplementary Fig. 27). Note that Zn-ZIF-8 is composed of Zn2+ and 2-methylimidazole. Characterizations of the morphology and structure clearly validate the successful synthesis of sandwich Zn-ZIF-8@Pt@Zn-ZIF-8 nanocubes (Supplementary Figs. 28–32). In addition, XPS depth profiles show that the original Zn-ZIF-8@Pt@Zn-ZIF-8 possesses only C, N, O, and Zn elements whereas the Pt element is discerned after argon plasma etching, suggesting that Pt nanoparticles are completely encapsulated by the outer Zn-ZIF-8 shell (Supplementary Fig. 33).The thickness of the outer Zn-ZIF-8 layer is then adjusted from 15 to 50 nm, in order to make it as a ruler for precisely measuring the spillover distance of hydrogen atoms (Supplementary Fig. 34). Note that the sample is synthesized in water, so the water molecules are unavoidably adsorbed in Zn-ZIFs pores if no special activation treatment is performed (denoted as Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O)). In order to further obtain Zn-ZIFs@Pt@Zn-ZIFs homologs, solvent-assisted ligand exchange/reduction strategy31,32 was adopted to replace 2-methyl group in Zn-ZIF-8 with different functional groups including CHO, OH, NO2 and NH2 (Supplementary Fig. 35), denoted as Zn-ZIFs (CHO), Zn-ZIFs (OH), Zn-ZIFs (NO2) and Zn-ZIFs (NH2), respectively. As manifested in Fig. 2a, b, the morphology and crystal structure of homologs remain unchanged after the ligand exchange/reduction process. Figure 2c presents the attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectra of Zn-ZIF-8@Pt@Zn-ZIF-8 before and after ligand exchange/reduction. The incorporations of new functional groups are confirmed by appearance of νC=O at 1700 cm−1and βC=O at 531 cm−1 (CHO), δC-O at 1459 cm−1 and νO-H at round 3310 cm−1 (OH), νas(N-O) at 1549 cm−1 and νs(N-O) at 1359 cm−1(NO2), νC-N at 1178 cm−1 and νN-H at the range of 3100–3500 cm−1 (NH2), respectively. The H-nuclear magnetic resonance (1H NMR) spectra quantitatively reveal that the ligand replacement ratio of all homologs exceeds 85% (Fig. 2d).Fig. 2: Solvent-assisted ligand exchange/reduction process for Zn-ZIF-8@Pt@Zn-ZIF-8.a TEM images of Zn-ZIFs@Pt@Zn-ZIFs homologs (scale bar: 50 nm). b XRD patterns. c ATR-FTIR spectra and d 1H NMR spectra of Zn-ZIFs@Pt@Zn-ZIFs homologs after ligand exchange.Evaluation of hydrogen spillover efficiencyZn-ZIF-8 is characteristic of a very small aperture window (0.34 nm) (Supplementary Fig. 36), allowing selective diffusion of H2 (0.29 nm) rather than large-sized molecules like cyclooctene (0.6 nm) (Supplementary Fig. 37). The adsorption experiments prove its size-sieving effect (Supplementary Fig. 38), guaranteeing that the reaction between hydrogen atom and organic reactant occurs only on the external surface of the MOFs matrix (Supplementary Figs. 39 and 40). Therefore, catalytic hydrogenation of cyclooctene is selected to evaluate the hydrogen spillover behavior of Zn-ZIFs@Pt@Zn-ZIFs homologs with different functional groups and shell thicknesses (Fig. 3a, Supplementary Figs. 34 and 41–44).Fig. 3: Spillover hydrogenation of cyclooctene.a Hydrogenation pathway of cyclooctene on Zn-ZIFs@Pt@Zn-ZIFs. b Catalytic conversion of cyclooctene by Zn-ZIFs@Pt@Zn-ZIFs homologs. c Catalytic stability of Zn-ZIFs@Pt@Zn-ZIFs homologs. d Linear relationship between spillover intensity and spillover distance of Zn-ZIFs@Pt@Zn-ZIFs homologs (80-min reaction). e Plot of lnk as a function of (1/T) for various catalysts. All reactions were performed using catalysts with the same amount of Pt NPs. All the reactions are carried out under a shell thickness of 15 nm, temperature of 80 °C, and time of 120 min if not specially mentioned. (error bar: standard deviation).First, we explore the effect of functional groups on spillover hydrogenation while controlling the shell thickness of all the samples to be 15 nm. As shown in Supplementary Table S1 and Fig. 3b, Zn-ZIF-8@Pt@Zn-ZIF-8 with water removal exhibits very low catalytic activity towards cyclooctene hydrogenation even under the elevated reaction conditions (gray curve in Fig. 3b, Supplementary Fig. 45), demonstrating that the hydrogen spillover is negligible in pure Zn-ZIF-8 below 100 °C. Our result is consistent with earlier reports33,34 that Pt@ZIF-8 is inactive for the hydrogenation of organic molecules. In sharp contrast, Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) without water removal shows unexpectedly high catalytic activity towards cyclooctene hydrogenation (red curve in Fig. 3b). These results clearly indicate that water plays a crucial role in the migration of activated hydrogen. For other homologs with water removal, under the same condition, the hydrogenation can take place on Zn-ZIFs@Pt@Zn-ZIFs (OH), Zn-ZIFs@Pt@Zn-ZIFs (CHO) and Zn-ZIFs@Pt@Zn-ZIFs (NH2) (Fig. 3b), manifesting the occurrence of hydrogen spillover from Pt nanoparticles to MOFs surface via these H acceptors. Among these homologs, the highest activity of Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) is attributed to the high mobility of molecular water. Furthermore, the recovered Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) catalyst of stable structure (Supplementary Fig. 46) is rapidly deactivated due to the loss of water, whereas the other homologs maintain high catalytic activity, validating the major role of water in hydrogen migration (Fig. 3c and Supplementary Table S2).Next, we change the shell thicknesses of the above four homologs including Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O), Zn-ZIFs@Pt@Zn-ZIFs (OH), Zn-ZIFs@Pt@Zn-ZIFs (CHO) and Zn-ZIFs@Pt@Zn-ZIFs (NH2). Here, cyclooctene hydrogenation conversion/100 is defined as the spillover intensity. The spillover intensity generally increases with thinner shell thickness, higher reaction temperature, and longer reaction time (Supplementary Fig. 47 and Supplementary Tables S3 and S4). At the specific temperature and time, the spillover intensity fits well to an equation: y = ax + b, where y refers to the spillover intensity, a represents the spillover decay factor, x stands for the shell thickness, and b is the maximum spillover intensity close to 1 (Fig. 3d and Supplementary Fig. 48). Noteworthily, Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) exhibits the smallest spillover decay factor of 0.01997 at 80 °C and 80-min reaction, and its hydrogen spillover even exceeds 50 nm (red line in Fig. 3d), representing an amazing distance that has been never found on nonreducible oxides. In addition, the sandwich catalysts with large-sized Pt NPs (ca. 5 nm) exhibit nearly the same spillover decay factor as the above-used one but lower hydrogenation activity, meaning that particle size only affects hydrogen activation independent of spillover intensity (Supplementary Figs. 49–51). Furthermore, the catalytic kinetics is studied. As shown in Supplementary Figs. 52–55, all four homolog catalysts display the zero-reaction order characteristic in the initial period, enabling us to quantitively extract the activation energy (Ea) using the Arrhenius equation: k = Ae−Ea/RT. Accordingly, Ea is determined to be 37.37 kJ/mol for Zn-ZIFs@Pt@Zn-ZIFs (H2O), 39.98 kJ/mol for Zn-ZIFs@Pt@Zn-ZIFs (OH), 51.14 kJ/mol for Zn-ZIFs@Pt@Zn-ZIFs (CHO) and 55.81 kJ/mol for Zn-ZIFs@Pt@Zn-ZIFs (NH2) (Fig. 3e and Supplementary Fig. 56). Such low Ea suggests the kinetically favorable pathway of Zn-ZIFs@Pt@Zn-ZIFs in the catalytic hydrogenation through hydrogen spillover at low temperatures.In order to elucidate the process of hydrogen spillover, isotope deuterium labeling experiments were adopted to track the trajectories of water and hydrogen by detecting the probe molecule (cyclooctene) mass spectrometry (Supplementary Fig. 57). As shown in Supplementary Fig. 58, the hydrogenation and deuteration of cyclooctene are hardly detected in Zn-ZIF-8(D2O)-D2 system, indicating that H-D exchange on probe cyclooctene is difficult to occur under the mild condition. At the same time, complete spillover hydrogenation and spillover deuteration can be clearly observed in Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O)-H2 system and Zn-ZIF-8@Pt@Zn-ZIF-8 (D2O)-D2 system, respectively. Interestingly, whether in the Zn-ZIF-8@Pt@Zn-ZIF-8 (D2O)-H2 system or in the Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O)-D2 system, cyclooctene is hydrogenated to cyclooctane, [D1]-cyclooctane and [D2]-cyclooctane. The above analysis demonstrates that H2 splitting occurs on Pt NPs, and the activated hydrogen atoms diffuse across the MOF structure by the water-assist path accompanied by exchange with water. In addition, as for the Zn-ZIFs@Pt@Zn-ZIFs (CHO)-D2 system, the product of [D2]-cyclooctane clarifies the migration of D atoms across MOFs containing CHO functional groups.Visual evidence of hydrogen spilloverThe color change of WO3 is a classic visual method for assessing the hydrogen spillover effect35. Upon mixing WO3 with varied Zn-ZIFs@Pt@Zn-ZIFs homologs, color-change experiments further confirm the distance and functional group-dependent hydrogen spillover in Zn-ZIFs (Supplementary Figs. 59 and 60).Exploration of hydrogen spillover mechanismTo decipher the spillover mechanism in Zn-ZIFs, we selected two representative catalysts, H2O-mediated Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) and functional group-mediated Zn-ZIFs@Pt@Zn-ZIFs (CHO). The coordination nature of metal atoms was investigated by in situ X-ray absorption spectroscopy (XAS)-XRD combined spectroscopy. Clearly, both Zn K-edge X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) spectra of Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) and Zn-ZIFs@Pt@Zn-ZIFs (CHO) barely change under hot hydrogen atmosphere (Fig. 4a, b, Supplementary Figs. 61–66 and Supplementary Table S5), indicating that hydrogen spillover in Zn-ZIFs does not change either the oxidation state or coordination environment of Zn atoms. In addition, the corresponding XRD patterns show that both samples maintain high structural stability during the whole H2-heat process (Fig. 4c and Supplementary Fig. 66).Fig. 4: Effect of metal nodes on hydrogen spillover: in situ XAS-XRD characterization of Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) and Zn-ZIFs@Pt@Zn-ZIFs (CHO).a Zn K-edge XANES and b Zn K-edge EXAFS spectra in R-space collected on as-prepared Zn-ZIFs@Pt@Zn-ZIFs under air at room temperature or hydrogen at 80 °C. c, XRD patterns of as-prepared Zn-ZIFs@Pt@Zn-ZIFs under air at room temperature or hydrogen at 80 °C.To further explore the H-binding sites in nonreducible MOFs, we employed in situ XPS technology to detect the valence of surface elements (Supplementary Figs. 67–70). The Zn 2p and N 1 s peaks of as-prepared Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) are located at 1021/1044 eV and 389.8 eV, respectively, and these peaks remain almost constant under hydrogen atmosphere. Note that the slight shift of these peaks is likely caused by the introduction of H (or electron) in the system that affects the Zn-N bond. As for Zn-ZIFs@Pt@Zn-ZIFs (CHO), its Zn 2p peaks also remain unchanged during the entire in situ test, whereas the corresponding O 1s peaks clearly shift to higher energy (Supplementary Fig. 70), indicating that the O species of the aldehyde group is the binding site of the H atom. Altogether, Zn species retain their original valence state after contact with H2 even at elevated temperatures, and hydrogen spillover occurs on these nonreducible MOFs and does not involve the metal redox process. Therefore, we suppose that the oxygen site of the aldehyde group acts as a binding site for the hydrogen atom, which is then transferred to the oxygen site of the next aldehyde group. For the Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) system, the embedded water acts as the functional group to promote hydrogen spillover. As proved by Supplementary Fig. 71, the water-containing catalyst delivers high catalytic activity for cyclooctene hydrogenation, whereas the water-free catalyst is inactive.Theoretical calculation of hydrogen migrationTo probe the mechanism of hydrogen adsorption and migration on Zn-ZIFs supports, first-principles atomistic simulation calculations were carried out. As shown in Supplementary Fig. 72, 1.61 eV is required to transfer a dissociated hydrogen atom among the carbon sites of Zn-ZIF-8 (C1 → C2), which is energetically improbable at ambient temperature25. Furthermore, this calculated migration energy of the dissociated hydrogen atom on Zn-ZIF-8 is only slightly influenced by the presence of water in the pores. Regardless of whether hydrogen atoms migrate in the form of H3O· or in the water environment (VASPsol), the energy barriers are both about 1.5 eV (Supplementary Figs. 73 and 74). Hence, we deduce that electrons migrate along the framework of Zn-ZIF-8 simultaneously with water-assisted proton hopping, which is affected by the number of water molecules. Along the water molecular chain, the energy barrier of proton migration is only 0.2 eV (Fig. 5a); while this value reaches 0.66 eV when involved with two water molecules (i.e., H5O2+) (Supplementary Fig. 75). This calculation explains the reduced catalytic activity of the recovered sample (Fig. 3c).Fig. 5: Mechanism of hydrogen migration on Zn-ZIF-8 (H2O) and Zn-ZIFs (CHO). The method used is the nudged elastic band (NEB).a Hydrogen spillover (activation energy Eact) on Zn-ZIF-8 (H2O) (water-assisted proton hopping of left model): Eact = 0.2 eV. b Hydrogen spillover on aldehyde-Zn-ZIFs (steps O1-TS1-2-O2 of left model): Eact = 0.47 eV.In regard to aldehyde-Zn-ZIFs, dissociative hydrogen is found to adsorb on the O site (Supplementary Fig. 70). The transfer of hydrogen species between neighboring oxygen sites has an activation energy barrier of 0.47 eV (Fig. 5b). In comparison, for nitro-ZIFs, although H also migrates through the O··H··O pathway, its migration barrier is as high as 0.86 eV (Supplementary Fig. 76), which explains its low hydrogenation activity (Fig. 3b).Catalytic application of controlled spillover hydrogenationFinally, the hydrogen spillover effect in MOFs is applied to regulate practical catalytic reactions. Partial hydrogenation products of halogen-substituted N-heteroarenes are known to be core structural motifs in both fine and bulk chemicals, while the strong adsorption between substrate and metal catalyst often leads to over-hydrogenation and dehalogenation36. We choose 5-chloroquinoline as a substrate based on the fact that its large size prevents diffusion through the pores, so the hydrogenation is completely derived from spillover hydrogen (Fig. 6a). Impressively, the Zn-ZIFs@Pt@Zn-ZIFs (-CHO and H2O) catalysts not only deliver a high catalytic activity similar to the commercial Pt/C, reducible TiO2-supported Pt and nonreducible Al2O3-supported Pt catalysts, but also endow an unprecedentedly high selectivity of >99% to the primary product of 5-chloro-1,2,3,4-tetrahydroquinoline under the same reaction condition (Fig. 6b, Supplementary Figs. 77–79 and Supplementary Table S6). Noteworthily, this result is comparable to the best-reported heterogeneous catalysts (Supplementary Table S7). In addition, with the increase of shell thickness, the conversion rate of 5-chloroquine gradually decreases while the selectivity remains stable (Fig. 6c and Supplementary Figs. 80–83), which is consistent with the concept of spillover distance decay factor (Fig. 3d).Fig. 6: Catalytic performance of Zn-ZIFs@Pt@Zn-ZIFs catalysts.a Scheme of spillover hydrogenation of 5-chloroquinolines. b Conversion ratio and selectivity of 5-chloroquinolines catalyzed by various catalysts. c Conversion ratio and selectivity of 5-chloroquinolines catalyzed by Zn-ZIF-8@Pt@Zn-ZIF-8 (H2O) with different shell thickness. Note: the reactions are carried out under the time of 240 min in group b and 200 min in group c. All reactions were performed using catalysts with the same amount of Pt NPs. (error bar: standard deviation).
