Thank you for visiting Nature.com. The version of browser you are using has limited CSS support. For best results, we recommend that you use a newer version of your browser (or disable Compatibility Mode in Internet Explorer). In the meantime, to ensure ongoing support, we are showing the site without styling or JavaScript.
       The development of hydrogen technologies lies at the heart of the green economy. As a precondition for realizing hydrogen storage, active and stable catalysts for the hydrogenation (de)hydrogenation reaction are required. Until now, this area has been dominated by the use of expensive precious metals. Here, we propose a novel low-cost cobalt-based catalyst (Co-SAs/NPs@NC) in which highly distributed single-metal sites are synergistically coupled with fine nanoparticles to achieve efficient formic acid dehydrogenation. Using the best material of atomically dispersed CoN2C2 units and encapsulated nanoparticles of 7-8 nm size, using propylene carbonate as solvent, an excellent gas production of 1403.8 ml g-1 h-1 was obtained, and there was no loss after 5 cycles. activity, which is 15 times better than commercial Pd/C. In situ assay experiments show that, compared to related single metal atom and nanoparticle catalysts, Co-SAs/NPs@NC enhances the adsorption and activation of the key monodentate intermediate HCOO*, thereby promoting subsequent CH bond cleavage. Theoretical calculations show that integration of cobalt nanoparticles promotes the conversion of the d-band center of a single Co atom into an active site, thereby enhancing the coupling between the carbonyl O of the HCOO* intermediate and the Co center, thereby lowering the energy barrier.
       Hydrogen is considered an important energy carrier for the current global energy transition and can be a key driver of achieving carbon neutrality1. Due to its physical properties such as flammability and low density, safe and efficient storage and transport of hydrogen are key issues in realizing the hydrogen economy2,3,4. Liquid organic hydrogen carriers (LOHCs), which store and release hydrogen through chemical reactions, have been proposed as a solution. Compared to molecular hydrogen, such substances (methanol, toluene, dibenzyltoluene, etc.) are easy and convenient to handle5,6,7. Among various traditional LOHCs, formic acid (FA) has relatively low toxicity (LD50: 1.8 g/kg) and H2 capacity of 53 g/L or 4.4 wt%. Notably, FA is the only LOHC that can store and release hydrogen under mild conditions in the presence of suitable catalysts, thus not requiring large external energy inputs1,8,9. In fact, many noble metal catalysts have been developed for the dehydrogenation of formic acid, for example, palladium-based catalysts are 50-200 times more active than inexpensive metal catalysts10,11,12. However, if you take into account the cost of active metals, for example, palladium is more than 1000 times more expensive.
       Cobalt, The search for highly active and stable heterogeneous base metal catalysts continues to attract the interest of many researchers in academia and industry13,14,15.
       Although inexpensive catalysts based on Mo and Co, as well as nanocatalysts made from noble/base metal alloys,14,16 have been developed for FA dehydrogenation, their gradual deactivation during the reaction is inevitable due to the occupation of active sites of metals, CO2, and H2O by protons. or formate anions (HCOO-), FA contamination, particle aggregation and possible CO poisoning17,18. We and others recently demonstrated that single-atom catalysts (SACs) with highly dispersed CoIINx sites as active sites improve the reactivity and acid resistance of formic acid dehydrogenation compared to nanoparticles17,19,20,21,22,23,24. In these Co-NC materials, N atoms serve as major sites to promote FA deprotonation while enhancing structural stability through coordination with the central Co atom, while Co atoms provide H adsorption sites and promote CH22 scission, 25,26. Unfortunately, the activity and stability of these catalysts are still far from modern homogeneous and heterogeneous noble metal catalysts (Fig. 1) 13 .
       Excess energy from renewable sources such as solar or wind can be produced by electrolysis of water. The hydrogen produced can be stored using LOHC, a liquid whose hydrogenation and dehydrogenation are reversible. In the dehydrogenation step, the only product is hydrogen, and the carrier liquid is returned to its original state and hydrogenated again. Hydrogen could eventually be used in gas stations, batteries, industrial buildings, and more.
       Recently, it was reported that the intrinsic activity of specific SACs can be enhanced in the presence of different metal atoms or additional metal sites provided by nanoparticles (NPs) or nanoclusters (NCs)27,28. This opens up possibilities for further adsorption and activation of the substrate, as well as for modulation of the geometry and electronic structure of monatomic sites. Therefore, substrate adsorption/activation can be optimized, providing better overall catalytic efficiency29,30. This gives us the idea of ​​creating appropriate catalytic materials with hybrid active sites. Although improved SACs have shown great potential in a wide range of catalytic applications, their role in hydrogen storage is, to our knowledge, unclear. In this regard, we report a versatile and robust strategy for the synthesis of cobalt-based hybrid catalysts (Co-SAs/NPs@NCs) consisting of defined nanoparticles and individual metal centers. The optimized Co-SAs/NPs@NC exhibit excellent formic acid dehydrogenation performance, which is better than non-noble nanostructured catalysts (such as CoNx, single cobalt atoms, cobalt@NC and γ-Mo2N) and even noble metal catalysts. In-situ characterization and DFT calculations of active catalysts show that individual metal sites serve as active sites, and the nanoparticles of the present invention enhance the d-band center of Co atoms, promote adsorption and activation of HCOO*, thereby lowering the energy barrier of the reaction. .
       Zeolite imidazolate frameworks (ZIFs) are well-defined three-dimensional precursors that provide catalysts for nitrogen-doped carbonaceous materials (metal-NC catalysts) to support various types of metals37,38. Therefore, Co(NO3)2 and Zn(NO3)2 combine with 2-methylimidazole in methanol to form the corresponding metal complexes in solution. After centrifugation and drying, CoZn-ZIF was pyrolyzed at different temperatures (750–950 °C) in an atmosphere of 6% H2 and 94% Ar. As shown in the figure below, the resulting materials have different active site characteristics and are named Co-SAs/NPs@NC-950, Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750 (Figure 2a). ) . Specific experimental observations of some key steps in the synthesis process are detailed in Figures 1 and 2. C1-C3. Variable temperature powder X-ray diffraction (VTXRD) was performed to monitor the evolution of the catalyst. After the pyrolysis temperature reaches 650 °C, the XRD pattern changes significantly due to the collapse of the ordered crystal structure of ZIF (Fig. S4) 39 . As the temperature further increases, two broad peaks appear in the XRD patterns of Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750 at 20–30° and 40–50°, representing the peak of amorphous carbon (Fig. C5). 40. It is worth noting that only three characteristic peaks were observed at 44.2°, 51.5° and 75.8°, belonging to metallic cobalt (JCPDS #15-0806), and 26.2°, belonging to graphitic carbon (JCPDS # 41-1487). The X-ray spectrum of Co-SAs/NPs@NC-950 shows the presence of graphite-like encapsulated cobalt nanoparticles on the catalyst41,42,43,44. The Raman spectrum shows that Co-SAs/NPs@NC-950 appears to have stronger and narrower D and G peaks than the other samples, indicating a higher degree of graphitization ( Figure S6 ). In addition, Co-SAs/NPs@NC-950 exhibits higher Brunner-Emmett-Taylor (BET) surface area and pore volume (1261 m2 g-1 and 0.37 cm3 g-1) than other samples and most ZIFs are NC derivatives. materials (Figure S7 and Table S1). Atomic absorption spectroscopy (AAS) shows that the cobalt content of Co-SAs/NPs@NC-950, Co-SAs/NPs@NC-850 and Co-SAs/NPs@ is 2.69 wt%, 2.74 % wt. and 2.73% wt. NC-750 respectively (Table S2). The Zn content of Co-SAs/NPs@NC-950, Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750 gradually increases, which is attributed to the increased reduction and volatilization of Zn units. Increase in pyrolysis temperature (Zn, boiling point = 907 °C) 45.46. Elemental analysis (EA) showed that the percentage of N decreases with increasing pyrolysis temperature, and the high O content may be due to the adsorption of molecular O2 from exposure to air. (Table S3). At a certain cobalt content, nanoparticles and isolated coatoms coexist, resulting in a significant increase in catalyst activity, as discussed below.
       Schematic diagram of the synthesis of Co-SA/NPs@NC-T, where T is the pyrolysis temperature (°C). b TEM image. c Image of Co-SAs/NPs@NC-950 AC-HAADF-STEM. Individual Co atoms are marked with red circles. d EDS Template Co-SA/NPs@NC-950.
       Notably, transmission electron microscopy (TEM) demonstrated the presence of various cobalt nanoparticles (NPs) with an average size of 7.5 ± 1.7 nm only in Co-SA/NPs@NC-950 ( Figures 2 b and S8). These nanoparticles are encapsulated with graphite-like carbon doped with nitrogen. The lattice fringe spacing of 0.361 and 0.201 nm corresponds to graphitic carbon (002) and metallic Co (111) particles, respectively. In addition, high-angle aberration-corrected annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) revealed that Co NPs in Co-SAs/NPs@NC-950 were surrounded by abundant atomic cobalt (Fig. 2c). However, only atomically dispersed cobalt atoms were observed on the support of the other two samples (Fig. S9). Energy dispersive spectroscopy (EDS) HAADF-STEM image shows uniform distribution of C, N, Co and segregated Co NPs in Co-SAs/NPs@NC-950 (Fig. 2d). All these results show that atomically dispersed Co centers and nanoparticles encapsulated in N-doped graphite-like carbon are successfully attached to NC substrates in Co-SAs/NPs@NC-950, while only isolated metal centers.
       The valence state and chemical composition of the obtained materials were studied by X-ray photoelectron spectroscopy (XPS). The XPS spectra of the three catalysts showed the presence of elements Co, N, C and O, but Zn was present only in Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750 (Fig. 2). ). C10). As pyrolysis temperature increases, total nitrogen content decreases as nitrogen species become unstable and decompose into NH3 and NOx gases at higher temperatures (Table S4) 47 . Thus, the total carbon content gradually increased from Co-SAs/NPs@NC-750 to Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-950 (Figures S11 and S12). The sample pyrolyzed at higher temperature has a lower proportion of nitrogen atoms, which means that the amount of NC carriers in Co-SAs/NPs@NC-950 should be less than that in other samples. This leads to stronger sintering of cobalt particles. The O 1s spectrum shows two peaks C=O (531.6 eV) and C–O (533.5 eV), respectively (Figure S13) 48 . As shown in Figure 2a, the N 1s spectrum can be resolved into four characteristic peaks of pyridine nitrogen N (398.4 eV), pyrrole N (401.1 eV), graphite N (402.3 eV) and Co-N (399.2 eV). Co-N bonds are present in all three samples, indicating that some N atoms are coordinated to monometallic sites, but the characteristics differ significantly49. Application of higher pyrolysis temperature can significantly reduce the content of Co-N species from 43.7% in Co-SA/NPs@NC-750 to 27.0% in Co-SAs/NPs@NC-850 and Co 17.6%@ NC-950. in -CA/NPs, which corresponds to an increase in the C content (Fig. 3a), indicating that their Co-N coordination number may change and be partially replaced by C50 atoms. The Zn 2p spectrum shows that this element exists predominantly in the form Zn2+. (Figure S14) 51. The spectrum of Co 2p exhibits two prominent peaks at 780.8 and 796.1 eV, which are attributed to Co 2p3/2 and Co 2p1/2, respectively (Figure 3b). Compared with Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750, the Co-N peak in Co-SAs/NPs@NC-950 is shifted to the positive side, indicating that one Co atom to the surface -SAs/NPs@NC-950 has a higher degree of electron depletion, resulting in a higher oxidation state. It is worth noting that only Co-SAs/NPs@NC-950 showed a weak peak of zero-valent cobalt (Co0) at 778.5 eV, which proves the existence of nanoparticles resulting from the aggregation of cobalt SA at high temperatures.
       a N 1s and b Co 2p spectra of Co-SA/NPs@NC-T. c XANES and d FT-EXAFS spectra of Co-K-edge of Co-SAs/NPs@NC-950, Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750. e WT-EXAFS contour plots of Co-SAs/NPs@NC-950, Co-SAs/NPs@NC-850, and Co-SAs/NPs@NC-750. f FT-EXAFS fitting curve for Co-SA/NPs@NC-950.
       Time-locked X-ray absorption spectroscopy (XAS) was then used to analyze the electronic structure and coordination environment of the Co species in the prepared sample. Cobalt valence states in Co-SAs/NPs@NC-950, Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750 Edge structure revealed by normalized near-field X-ray absorption of Co-K edge ( XANES) spectrum. As shown in Figure 3c, the absorption near the edge of the three samples is located between the Co and CoO foils, indicating that the valence state of Co species ranges from 0 to +253. In addition, a transition to lower energy was observed from Co-SAs/NPs@NC-950 to Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750, indicating that Co-SAs/ NPs@NC-750 has a lower oxidation state. reverse order. According to the linear combination fitting results, the Co valence state of Co-SAs/NPs@NC-950 is estimated to be +0.642, which is lower than the Co valence state of Co-SAs/NPs@NC-850 (+1.376). Co-SA/NP @NC-750 (+1.402). These results indicate that the average oxidation state of cobalt particles in Co-SAs/NPs@NC-950 is significantly reduced, which is consistent with the XRD and HADF-STEM results and can be explained by the coexistence of cobalt nanoparticles and single cobalt. . Co atoms 41. Fourier transform X-ray absorption fine structure (FT-EXAFS) spectrum of the Co K-edge shows that the main peak at 1.32 Å belongs to the Co-N/Co-C shell, while the scattering path of metallic Co -Co is at 2.18 only in Co-SAs Å found in /NPs@NC-950 (Fig. 3d). Moreover, the wavelet transform (WT) contour plot shows the maximum intensity at 6.7 Å-1 attributed to Co-N/Co-C, while only Co-SAs/NPs@NC-950 shows the maximum intensity attributed to 8.8. Another intensity maximum is at Å−1 to the Co–Co bond (Fig. 3e). In addition, EXAFS analysis performed by the lessor showed that at pyrolysis temperatures of 750, 850 and 950 °C, the Co-N coordination numbers were 3.8, 3.2 and 2.3, respectively, and the Co-C coordination numbers were 0. 0.9 and 1.8 (Fig. 3f, S15 and Table S1). More specifically, the latest results can be attributed to the presence of atomically dispersed CoN2C2 units and nanoparticles in Co-SAs/NPs@NC-950. In contrast, in Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750, only CoN3C and CoN4 units are present. It is obvious that with increasing pyrolysis temperature, N atoms in the CoN4 unit are gradually replaced by C atoms, and cobalt CA aggregates to form nanoparticles.
       Previously studied reaction conditions were used to study the effect of preparation conditions on the properties of various materials (Fig. S16)17,49. As shown in Figure 4 a, the activity of Co-SAs/NPs@NC-950 is significantly higher than that of Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750. Notably, all three prepared Co samples showed superior performance compared to standard commercial precious metal catalysts (Pd/C and Pt/C). In addition, the Zn-ZIF-8 and Zn-NC samples were inactive towards formic acid dehydrogenation, indicating that the Zn particles are not active sites, but their effect on the activity is negligible. In addition, the activity of Co-SAs/NPs@NC-850 and Co-SAs/NPs@NC-750 underwent secondary pyrolysis at 950°C for 1 hour, but was lower than that of Co-SAs/NPs@NC-750 . @NC-950 (Fig. S17). Structural characterization of these materials showed the presence of Co nanoparticles in the re-pyrolyzed samples, but the low specific surface area and absence of graphite-like carbon resulted in lower activity compared to Co-SAs/NPs@NC-950 (Figure S18–S20). The activity of samples with different amounts of Co precursor was also compared, with the highest activity shown at 3.5 mol addition (Table S6 and Figure S21). It is obvious that the formation of various metal centers is influenced by the hydrogen content in the pyrolysis atmosphere and the pyrolysis time. Therefore, other Co-SAs/NPs@NC-950 materials were evaluated for formic acid dehydrogenation activity. All materials showed moderate to very good performance; however, none of them were better than Co-SAs/NPs@NC-950 (Figures S22 and S23). The structural characterization of the material showed that with increasing pyrolysis time, the content of monoatomic Co-N positions gradually decreases due to the aggregation of metal atoms into nanoparticles, which explains the difference in activity between samples with a pyrolysis time of 100-2000. difference. 0.5 h, 1 h, and 2 h (Figures S24–S28 and Table S7).
       Graph of gas volume versus time obtained during dehydrogenation of fuel assemblies using various catalysts. Reaction conditions: FA (10 mmol, 377 μl), catalyst (30 mg), PC (6 ml), Tback: 110 °C, Tactical: 98 °C, 4 parts b Co-SAs/NPs@NC-950 ( 30 mg), various solvents. c Comparison of gas evolution rates of heterogeneous catalysts in organic solvents at 85–110 °C. d Co-SA/NPs@NC-950 recycling experiment. Reaction conditions: FA (10 mmol, 377 µl), Co-SAs/NPs@NC-950 (30 mg), solvent (6 ml), Tback: 110 °C, Tactical: 98 °C, each reaction cycle lasts one hour . Error bars represent standard deviations calculated from three active tests.
       In general, the efficiency of FA dehydrogenation catalysts is highly dependent on the reaction conditions, especially the solvent used8,49. When using water as solvent, Co-SAs/NPs@NC-950 showed the highest initial reaction rate, but deactivation occurred, possibly due to protons or H2O18 occupying the active sites. Testing of the catalyst in organic solvents such as 1,4-dioxane (DXA), n-butyl acetate (BAC), toluene (PhMe), triglyme and cyclohexanone (CYC) also did not show any improvement, and in propylene carbonate (PC) ) (Fig. 4b and Table S8). Likewise, additives such as triethylamine (NEt3) or sodium formate (HCCONa) had no further positive effect on catalyst performance (Figure S29). Under optimal reaction conditions, the gas yield reached 1403.8 mL g−1 h−1 (Fig. S30), which was significantly higher than all previously reported Co catalysts (including SAC17, 23, 24). In various experiments, excluding reactions in water and with formate additives, selectivity up to 99.96% for dehydrogenation and dehydration was obtained (Table S9). The calculated activation energy is 88.4 kJ/mol, which is comparable to the activation energy of noble metal catalysts (Figure S31 and Table S10).
       In addition, we compared a number of other heterogeneous catalysts for formic acid dehydrogenation under similar conditions (Fig. 4c, tables S11 and S12). As shown in Figure 3c, the gas production rate of Co-SAs/NPs@NC-950 exceeds that of most known heterogeneous base metal catalysts and is 15 and 15 times higher than that of commercial 5% Pd/C and 5% Pd/C, respectively. once. % Pt/C catalyst.
       An important feature of any practical application of (de)hydrogenation catalysts is their stability. Therefore, a series of recycling experiments using Co-SAs/NPs@NC-950 were carried out. As shown in Figure 4 d, the initial activity and selectivity of the material remained unchanged over five consecutive runs (see also Table S13). Long-term tests were conducted and gas production increased linearly over 72 hours (Figure S32). The cobalt content of the used Co-SA/NPs@NC-950 was 2.5 wt%, which was very close to that of the fresh catalyst, indicating that there was no obvious leaching of cobalt (Table S14). No obvious color change or aggregation of metal particles was observed before and after the reaction ( Figure S33 ). AC-HAADF-STEM and EDS of materials applied in long-term experiments showed retention and uniform dispersion of atomic dispersion sites and no significant structural changes (Figures S34 and S35). The characteristic peaks of Co0 and Co-N still exist in XPS, proving the coexistence of Co NPs and individual metal sites, which also confirms the stability of the Co-SAs/NPs@NC-950 catalyst (Figure S36).
       To identify the most active sites responsible for formic acid dehydrogenation, selected materials with only one metal center (CoN2C2) or Co NP were prepared based on previous studies17. The order of formic acid dehydrogenation activity observed under the same conditions is Co-SAs/NPs@NC-950 > Co SA > Co NP (Table S15), indicating that atomically dispersed CoN2C2 sites are more active than NPs . The reaction kinetics show that hydrogen evolution follows first-order reaction kinetics, but the slopes of several curves at different cobalt contents are not the same, indicating that the kinetics depends not only on formic acid, but also on the active site (Fig. 2). C37). Further kinetic studies showed that, given the absence of cobalt metal peaks in X-ray diffraction analysis, the kinetic order of the reaction in terms of cobalt content was found to be 1.02 at lower levels (less than 2.5%), indicating an almost uniform distribution of monoatomic cobalt centers. main. active site (figs. S38 and S39). When the content of Co particles reaches 2.7%, r suddenly increases, indicating that the nanoparticles interact well with individual atoms to obtain higher activity. As the content of Co particles further increases, the curve becomes nonlinear, which is associated with an increase in the number of nanoparticles and a decrease in monatomic positions. Thus, the improved LC dehydrogenation performance of Co-SA/NPs@NC-950 results from the cooperative behavior of individual metal sites and nanoparticles.
       An in-depth study was carried out using in situ diffuse reflectance Fourier transform (in situ DRIFT) to identify reaction intermediates in the process. After heating the samples to different reaction temperatures after adding formic acid, two sets of frequencies were observed (Fig. 5a). Three characteristic peaks of HCOOH* appear at 1089, 1217 and 1790 cm-1, which are attributed to the out-of-plane CH π (CH) stretching vibration, CO ν (CO) stretching vibration and C=O ν (C=O) stretching vibration, 54, 55 respectively . Another set of peaks at 1363 and 1592 cm-1 corresponds to the symmetric OCO vibration νs(OCO) and the asymmetric OCO stretching vibration νas(OCO)33.56 HCOO*, respectively. As the reaction progresses, the relative peaks of HCOOH* and HCOO* gradually weaken. Generally speaking, the decomposition of formic acid involves three main steps: (I) adsorption of formic acid at active sites, (II) removal of H through the formate or carboxylate pathway, and (III) combination of two adsorbed H to produce hydrogen. HCOO* and COOH* are key intermediates in determining the formate or carboxylate pathways, respectively57. In our catalytic system, only the characteristic HCOO* peak appeared, indicating that the decomposition of formic acid occurs only through the formic acid pathway58. Similar observations were made at lower temperatures of 78 °C and 88 °C (fig. S40).
       In situ DRIFT spectra of HCOOH dehydrogenation on a Co-SAs/NPs@NC-950 and b Co SAs. The legend indicates on-site reaction times. c Dynamics of gas volume obtained using different isotope labeling reagents. d Kinetic isotope effect data.
       Similar in situ DRIFT experiments were carried out on the related materials Co NP and Co SA to study the synergistic effect in Co-SA/NPs@NC-950 ( Figures 5 b and S41). Both materials show similar trends, but the characteristic peaks of HCOOH* and HCOO* are slightly shifted, indicating that the introduction of Co NPs changes the electronic structure of the monoatomic center. A characteristic νas(OCO) peak appears in Co-SAs/NPs@NC-950 and Co SA but not in Co NPs, further indicating that the intermediate formed upon addition of formic acid is monodentate formic acid perpendicular to the plane salt surface. and is adsorbed onto SA as the active site 59 . It is worth noting that a significant increase in the vibrations of the characteristic peaks π(CH) and ν(C = O) was observed, which apparently led to distortion of HCOOH* and facilitated the reaction. As a result, the characteristic peaks of HCOOH* and HCOO* in Co-SAs/NPs@NC almost disappeared after 2 min of reaction, which is faster than that of monometallic (6 min) and nanoparticle-based catalysts (12 min). . All these results confirm that nanoparticle doping enhances the adsorption and activation of intermediates, thereby accelerating the reactions proposed above.
       To further analyze the reaction pathway and determine the rate determining step (RDS), the KIE effect was carried out in the presence of Co-SAs/NPs@NC-950. Here, different formic acid isotopes such as HCOOH, HCOOD, DCOOH and DCOOD are used for KIE studies. As shown in Figure 5c, the dehydrogenation rate decreases in the following order: HCOOH > HCOOD > DCOOH > DCOOD. In addition, the calculated values ​​of KHCOOH/KHCOOD, KHCOOH/KDCOOH, KHCOOD/KDCOOD and KDCOOH/KDCOOD were 1.14, 1.71, 2.16 and 1.44, respectively (Fig. 5d). Thus, CH bond cleavage in HCOO* exhibits kH/kD values ​​>1.5, indicating a major kinetic effect60,61, and appears to represent the RDS of HCOOH dehydrogenation on Co-SAs/NPs@NC-950.
       In addition, DFT calculations were performed to understand the effect of doped nanoparticles on the intrinsic activity of Co-SA. The Co-SAs/NPs@NC and Co-SA models were constructed based on the experiments shown and previous works (Figs. 6a and S42)52,62. After geometric optimization, small Co6 nanoparticles (CoN2C2) coexisting with monoatomic units were identified, and the Co-C and Co-N bond lengths in Co-SA/NPs@NC were determined to be 1.87 Å and 1.90 Å, respectively. , which is consistent with the XAFS results. The calculated partial density of states (PDOS) shows that single Co metal atom and nanoparticle composite (Co-SAs/NPs@NC) exhibit higher hybridization near the Fermi level compared to CoN2C2, resulting in HCOOH. The decomposed electron transfer is more efficient ( Figures 6b and S43). The corresponding d-band centers of Co-SAs/NPs@NC and Co-SA were calculated to be -0.67 eV and -0.80 eV, respectively, among which the increase of Co-SAs/NPs@NC was 0.13 eV, which contributed that after the introduction of NP, the adsorption of HCOO* particles by the adapted electronic structure of CoN2C2 occurs. The difference in charge density shows a large electron cloud around the CoN2C2 block and the nanoparticle, indicating a strong interaction between them due to electron exchange. Combined with Bader charge analysis, it was found that atomically dispersed Co lost 1.064e in Co-SA/NPs@NC and 0.796e in Co SA ( Figure S44 ). These results indicate that the integration of nanoparticles leads to electron depletion of Co sites, resulting in an increase in Co valence, which is consistent with the XPS results (Fig. 6c). The Co-O interaction characteristics of HCOO adsorption onto Co-SAs/NPs@NC and Co SA were analyzed by calculating the crystalline orbital Hamiltonian group (COHP)63. As shown in Figure 6 d, negative and positive values ​​of -COHP correspond to the antibonding state and binding state, respectively. The bond strength of Co-O adsorbed by HCOO (Co-carbonyl O HCOO*) was assessed by integrating the -COHP values, which were 3.51 and 3.38 for Co-SAs/NPs@NC and Co-SA, respectively. HCOOH adsorption also showed similar results: the increase in the integral value of -COHP after nanoparticle doping indicated an increase in Co-O bonding, thereby promoting the activation of HCOO and HCOOH ( Figure S45 ).
       Lattice structure of Co-SA/NPs@NC-950. b PDOS Co-SA/NP@NC-950 and Co SA. c 3D isosurface of the difference in charge densities of HCOOH adsorption on Co-SA/NPs@NC-950 and Co-SA. (d) pCOHP of Co-O bonds adsorbed by HCOO on Co-SA/NPs@NC-950 (left) and Co-SA (right). e Reaction pathway of HCOOH dehydrogenation on Co-SA/NPs@NC-950 and Co-SA.
       To better understand the superior dehydrogenation performance of Co-SA/NPs@NC, the reaction path and energy were established. Specifically, FA dehydrogenation involves five steps, including the conversion of HCOOH to HCOOH*, HCOOH* to HCOO* + H*, HCOO* + H* to 2H* + CO2*, 2H* + CO2* to 2H* + CO2, and 2H* in H2 (Fig. 6e). The adsorption energy of formic acid molecules on the catalyst surface through carboxylic oxygen is lower than through hydroxyl oxygen (Figures S46 and S47). Subsequently, due to the lower energy, the adsorbate preferentially undergoes OH bond cleavage to form HCOO* rather than CH bond cleavage to form COOH*. At the same time, HCOO* uses monodentate adsorption, which promotes the breaking of bonds and the formation of CO2 and H2. These results are consistent with the presence of a νas(OCO) peak in in situ DRIFT, further indicating that FA degradation occurs through the formate pathway in our study. It is important to note that according to KIE measurements, CH dissociation has a much higher reaction energy barrier than other reaction steps and represents an RDS. The energy barrier of the optimal Co-SAs/NPs@NC catalyst system is 0.86 eV lower than that of Co-SA (1.2 eV), which significantly improves the overall dehydrogenation efficiency. Notably, the presence of nanoparticles regulates the electronic structure of the atomically dispersed coactive sites, which further enhances the adsorption and activation of intermediates, thereby lowering the reaction barrier and promoting hydrogen production.
       In summary, we demonstrate for the first time that the catalytic performance of hydrogen production catalysts can be significantly improved by using materials with highly distributed monometallic centers and small nanoparticles. This concept has been validated by the synthesis of cobalt-based single-atom metal catalysts modified with nanoparticles (Co-SAs/NPs@NC), as well as related materials with only single-metal centers (CoN2C2) or Co NPs. All materials were prepared by a simple one-step pyrolysis method. Structural analysis shows that the best catalyst (Co-SAs/NPs@NC-950) consists of atomically dispersed CoN2C2 units and small nanoparticles (7-8 nm) doped with nitrogen and graphite-like carbon. It has excellent gas productivity up to 1403.8 ml g-1 h-1 (H2:CO2 = 1.01:1), H2 and CO selectivity of 99.96% and can maintain constant activity for several days. The activity of this catalyst exceeds the activity of certain Co SA and Pd/C catalysts by 4 and 15 times, respectively. In situ DRIFT experiments show that compared to Co-SA, Co-SAs/NPs@NC-950 exhibits stronger monodentate adsorption of HCOO*, which is important for the formate pathway, and dopant nanoparticles can promote HCOO* activation and C–H acceleration . The bond cleavage was identified as RDS. Theoretical calculations show that Co NP doping increases the d-band center of single Co atoms by 0.13 eV through interaction, enhancing the adsorption of HCOOH* and HCOO* intermediates, thereby reducing the reaction barrier from 1.20 eV for Co SA to 0 .86 eV. He is responsible for outstanding performance.
       More broadly, this research provides ideas for the design of new single-atom metal catalysts and advances understanding of how to improve catalytic performance through the synergistic effect of metal centers of different sizes. We believe that this approach can be easily extended to many other catalytic systems.
       Co(NO3)2 6H2O (AR, 99%), Zn(NO3)2 6H2O (AR, 99%), 2-methylimidazole (98%), methanol (99.5%), propylene carbonate (PC, 99% ) ethanol (AR, 99.7%) was purchased from McLean, China. Formic acid (HCOOH, 98%) was purchased from Rhawn, China. All reagents were used directly without additional purification, and ultrapure water was prepared using an ultrapure purification system. Pt/C (5% mass loading) and Pd/C (5% mass loading) were purchased from Sigma-Aldrich.
       The synthesis of CoZn-ZIF nanocrystals was carried out based on previous methods with some modifications23,64. First, 30 mmol Zn(NO3)2·6H2O (8.925 g) and 3.5 mmol Co(NO3)2·6H2O (1.014 g) were mixed and dissolved in 300 ml of methanol. Then, 120 mmol of 2-methylimidazole (9.853 g) was dissolved in 100 ml of methanol and added to the above solution. The mixture was stirred at room temperature for 24 hours. Finally, the product was separated by centrifugation at 6429 g for 10 min and washed thoroughly with methanol three times. The resulting powder was dried in vacuum at 60°C overnight before use.
       To synthesize Co-SAs/NPs@NC-950, dry CoZn-ZIF powder was pyrolyzed at 950 °C for 1 h in a gas flow of 6% H2 + 94% Ar, with a heating rate of 5 °C/min. The sample was then cooled to room temperature to obtain Co-SA/NPs@NC-950. For Co-SAs/NPs@NC-850 or Co-SAs/NPs@NC-750, the pyrolysis temperature was varied to 850 and 750 °C, respectively. Prepared samples can be used without further processing, such as acid etching.
       TEM (transmission electron microscopy) measurements were performed on a Thermo Fisher Titan Themis 60-300 “cube” microscope equipped with an aberration corrector for imaging and a 300 kV probe shaping lens. HAADF-STEM experiments were carried out using FEI Titan G2 and FEI Titan Themis Z microscopes equipped with probes and image correctors, as well as DF4 four-segment detectors. EDS elemental mapping images were also obtained on an FEI Titan Themis Z microscope. XPS analysis was performed on an X-ray photoelectron spectrometer (Thermo Fisher model ESCALAB 250Xi). XANES and EXAFS spectra of Co K-edge were collected using an XAFS-500 table (China Spectral Instruments Co., Ltd.). Co content was determined by atomic absorption spectroscopy (AAS) (PinAAcle900T). X-ray diffraction (XRD) spectra were recorded on an X-ray diffractometer (Bruker, Bruker D8 Advance, Germany). Nitrogen adsorption isotherms were obtained using a physical adsorption apparatus (Micromeritics, ASAP2020, USA).
       The dehydrogenation reaction was carried out in an argon atmosphere with air removed according to the standard Schlenk method. The reaction vessel was evacuated and refilled with argon 6 times. Turn on the condenser water supply and add catalyst (30 mg) and solvent (6 ml). Heat the container to the desired temperature using a thermostat and allow it to equilibrate for 30 minutes. Formic acid (10 mmol, 377 μL) was then added to the reaction vessel under argon. Turn the three-way burette valve to depressurize the reactor, close it again, and begin measuring the volume of gas produced using a manual burette (Figure S16). After the time required for the reaction to complete, a gas sample was collected for GC analysis using a gas-tight syringe purged with argon.
       In situ drift experiments were performed on a Fourier transform infrared (FTIR) spectrometer (Thermo Fisher Scientific, Nicolet iS50) equipped with a mercury cadmium telluride (MCT) detector. The catalyst powder was placed in a reaction cell (Harrick Scientific Products, Praying Mantis). After treating the catalyst with a stream of Ar (50 ml/min) at room temperature, the sample was heated to a given temperature, then bubbled with Ar (50 ml/min) in a HCOOH solution and poured into the in-situ reaction cell. for reaction. Model heterogeneous catalytic processes. Infrared spectra were recorded at intervals ranging from 3.0 seconds to 1 hour.
       HCOOH, DCOOH, HCOOD and DCOOD are used as substrates in propylene carbonate. The remaining conditions correspond to the HCOOH dehydrogenation procedure.
       First principles calculations were performed using the density functional theory framework within the Vienna Ab initio modeling package (VASP 5.4.4) 65,66. A superunit cell with a graphene surface (5 × 5) with a transverse dimension of approximately 12.5 Å was used as a substrate for CoN2C2 and CoN2C2-Co6. A vacuum distance of more than 15 Å was added to avoid interaction between adjacent substrate layers. The interaction between ions and electrons is described by the projected amplified wave (PAW) method65,67. The Perdue-Burke-Ernzerhoff (PBE) generalized gradient approximation (GGA) function proposed by Grimm with van der Waals correction68,69 was used. The convergence criteria for total energy and force are 10−6 eV/atom and 0.01 eV/Å. The energy cutoff was set at 600 eV using a Monkhorst-Pack 2 × 2 × 1 K-point grid. The pseudopotential used in this model is constructed from the electronic configuration into the C 2s22p2 state, N 2s22p3 state, Co 3d74s2 state, H 1 s1 state, and O 2s22p4 state. The adsorption energy and electron density difference are calculated by subtracting the energy of the gas phase and surface species from the energy of the adsorbed system according to adsorption or interface models70,71,72,73,74. The Gibbs free energy correction is used to convert DFT energy into Gibbs free energy and takes into account vibrational contributions to entropy and zero point energy75. The ascending image-nudging elastic band (CI-NEB) method was used to search for the transition state of the reaction76.
       All data obtained and analyzed during this study are included in the article and supplementary materials or are available from the corresponding author on reasonable request. Source data is provided for this article.
       All code used in the simulations accompanying this article is available from the corresponding authors upon request.
       Dutta, I. et al. Formic acid supports low-carbon economy. adverb. Energy materials. 12, 2103799 (2022).
       Wei, D., Sang, R., Sponholz, P., Junge, H. and Beller, M. Reversible hydrogenation of carbon dioxide to formic acid in the presence of lysine using Mn-claw complexes. Nat. Energy 7, 438–447 (2022).
       Wei, D. et al. Towards a hydrogen economy: development of heterogeneous catalysts for hydrogen storage and release chemistry. ACS Energy Letters. 7, 3734–3752 (2022).
       Modisha P.M., Ouma S.N.M., Garijirai R., Wasserscheid P. and Bessarabov D. Prospects for hydrogen storage using liquid organic hydrogen carriers. Energy Fuels 33, 2778–2796 (2019).
       Niermann, M., Timmerberg, S., Drunert, S. and Kaltschmitt, M. Liquid organic hydrogen carriers and alternatives for the international transport of renewable hydrogen. update. support. energy. Open 135, 110171 (2021).
       Preister P, Papp K and Wasserscheid P. Liquid organic hydrogen carriers (LOHC): towards a hydrogen-free hydrogen economy. application. Chemical. resource. 50, 74–85 (2017).
       Chen, Z. et al. Development of reliable palladium catalysts for formic acid dehydrogenation. AKS catalogue. 13, 4835–4841 (2023).
       Sun, Q., Wang, N., Xu, Q. and Yu, J. Nanopore-supported metal nanocatalysts for efficient hydrogen production from liquid-phase hydrogen storage chemicals. adverb. Matt. 32, 2001818 (2020).
       Seraj, JJA, et al. An efficient catalyst for the dehydrogenation of pure formic acid. Nat. communicate. 7, 11308 (2016).
       Kar S., Rauch M., Leithus G., Ben-David Y. and Milstein D. Efficient dehydrogenation of pure formic acid without additives. Nat. Gatar. 4, 193–201 (2021).
       Li, S. et al. Simple and effective principles for the rational design of heterogeneous formic acid dehydrogenation catalysts. adverb. Matt. 31, 1806781 (2019).
       Liu, M. et al. Heterogeneous catalysis of hydrogen storage technology using formic acid-based carbon dioxide. adverb. Energy materials. 12, 2200817 (2022).