Nonstoichiometric Yttrium Hydride–Promoted Reversible Hydrogen Storage in a Liquid Organic Hydrogen Carrier

Open AccessCCS ChemistryRESEARCH ARTICLE1 Mar 2021Nonstoichiometric Yttrium Hydride–Promoted Reversible Hydrogen Storage in a Liquid Organic Hydrogen Carrier Yong Wu, Yanru Guo, Hongen Yu, Xiaojing Jiang, Yuxuan Zhang, Yue Qi, Kai Fu, Lei Xie, Guoling Li, Jie Zheng and Xingguo Li Yong Wu Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author , Yanru Guo Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author , Hongen Yu Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author , Xiaojing Jiang Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author , Yuxuan Zhang Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author , Yue Qi Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author , Kai Fu Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author , Lei Xie Sunan Institute for Molecular Engineering, Peking University, Nanjing 215500 Google Scholar More articles by this author , Guoling Li College of Materials Science and Engineering, Qingdao University, Qingdao 266071. Google Scholar More articles by this author , Jie Zheng *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author and Xingguo Li *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] Beijing National Laboratory for Molecular Science (BNLMS), College of Chemistry and Molecular Engineering, Peking University, Beijing 100871 Google Scholar More articles by this author https://doi.org/10.31635/ccschem.020.202000255 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail N-Ethylcarbazole (NEC) is one of the most promising liquid organic hydrogen carriers (LOHCs), but its application is limited by sluggish kinetics due to lack of high-efficiency, low-cost catalysts. This work reports a cobalt (Co)-based catalyst promoted by nonstoichiometric yttrium hydride (YH3−x) to achieve high-efficiency, reversible hydrogen storage in NEC, with >5.5 wt % reversible hydrogen storage capacity could be achieved below 473 K, and with good kinetics. The YH3−x-promoted Co-based catalyst is the first non-noble metal catalyst with high activity for NEC hydrogenation and 12H-NEC dehydrogenation reactions. A mechanistic study suggests that YH3−x facilitates the reversible hydrogen transfer both in the hydrogenation and the dehydrogenation reactions. The nonstoichiometric YH3−x contained both lattice H and H vacancies with tunable H chemical potential serve as the H donor and H acceptor for reversible hydrogen transfer. Our results support the practical application of LOHCs and inspire new approaches for the utilization of conventional metal hydrides to promote versatile H transfer reactions. Download figure Download PowerPoint Introduction Hydrogen storage and transport are critical enabling technologies that support the widespread application of hydrogen energy. Liquid organic hydrogen carriers (LOHCs) have attracted enormous academic and industrial interest for their promising applications in large scale, long-distance H2 transportation and onboard hydrogen storage due to their high capacity, good reversibility, low cost, highly safe, easy thermal management, and good compatibility with existing gasoline infrastructure.1–7 LOHCs are typically represented in their H lean form. Recently, Benzyltoluene, dibenzyltoluene, biphenyl, diphenylmethane, 2-methylindole, 2-picoline, acridine, and N-ethylcarbazole (NEC) are widely being studied.8–14 Among these LOHCs, NEC and its fully hydrogenated form (dodecahydro-N-ethylcarbazole, 12H-NEC) is one of the most promising LOHC couples due to its high capacity (5.8 wt % and 55 g H2 L−1), moderate operation temperature (dehydrogenation temperature below 473 K), low volatility, and low toxicity.15–20 Almost all the LOHCs, including NEC, suffer from sluggish hydrogenation and dehydrogenation kinetics. Currently, there remains a lack of high-efficiency, low-cost catalysts to achieve efficient hydrogenation and dehydrogenation, which is the major obstacle for the practical application of LOHCs. Moreover, typically, different catalysts are required for NEC hydrogenation and 12H-NEC dehydrogenation. Ruthenium (Ru)-based catalysts show the best hydrogenation performance, while palladium (Pd)-based catalysts exhibit the best dehydrogenation performance.21,22 The requirement of two different catalysts significantly increases the complexity and cost of the system. Undoubtedly, a single non-noble metal catalyst with both hydrogenation and dehydrogenation catalytic performance (two-direction catalyst) is highly desirable for LOHC-based hydrogen storage systems. However, two-direction catalysts for hydrogen storage in NEC are extremely rare. To the best of our knowledge, Pd2[email protected] was the only reported efficient two-directional catalyst for NEC, which was also based on noble metals.23 There are several non-noble metal, two-direction catalysts for quinoline and its derivatives, including Cu/TiO2,24 single-atom Co catalysts on porous N-doped carbon,25 Ni31Si12/Ni2[email protected]2,26 [email protected],27 and transition metal/carbon dyads.28 However, the gravimetric hydrogen storage capacities of quinoline and its derivatives are too low (< 3 wt %) for practical hydrogen storage applications. Recently, we discovered that rare-earth hydrides could promote the NEC hydrogenation activity significantly on transition metal catalysts via mediating the hydrogen transfer. Highly efficient NEC hydrogenation catalysts Ru/YH3 and Ni/Al2O3-YH3 were developed based on this strategy.29,30 However, these YH3-promoted catalysts were inefficient for the dehydrogenation of 12H-NEC. For the practical hydrogen storage and transport application, hydrogenation can be carried out ex-situ, for instance, in a hydrogenation plant, while dehydrogenation has to be achieved in situ to fulfill the hydrogen supply demand. Therefore, fast dehydrogenation of 12H-NEC is more critical. In this work, we report a highly efficient two-direction catalyst for reversible hydrogen storage in NEC/12H-NEC, which is composed of alumina-supported cobalt boride particles blended with nonstoichiometric yttrium trihydride (denoted as Co-B/Al2O3-YH3−x). Both Co-B/Al2O3 and YH3−x are two inactive components, which are mixed by a simple hand milling to promote significant performance, thereby, exhibiting activity comparable to Ru-based catalysts used in the hydrogenation of NEC, and Pd-based catalysts used in the dehydrogenation of 12H-NEC. Reversible hydrogen storage capacity over 5.5 wt % is achieved below 473 K. We demonstrate that YH3−x mediates the hydrogen transfer in both hydrogenation and dehydrogenation reactions in which the lattice H and H vacancies with variable H chemical potential play a key role. The results enabled efficient reversible hydrogen storage of LOHC using low-cost noble-metal-free catalysts, and also, inspired a new application of metal hydrides in catalyzing more general hydrogen transfer reactions. Experimental Method Sample preparation and characterization YH3−x powder was prepared by hydrogenation of metal Y granules (Trillion Metals Corporation, >99.9%) at 623 K, 4 MPa H2 for 6 h in a Sievert-type reactor and subsequently pulverized by ball milling (Fritsch Pulverisette 5) at 0.4 MPa H2, 250 rpm. The composition was estimated to be YH2.75 based on the hydrogen absorption capacity. YD3−x was prepared by the same method using D2. We prepared the Al2O3-supported cobalt boride (Co-B/Al2O3) catalyst using 100 mg of cetyltrimethylammonium bromide and 47 mg CoCl2·6H2O dissolved in 2.5 mL deionized water with ultrasonication. Then 1.5 mL NaBH4 aqueous solution (0.176 mol L−1) was added until no bubbles were observed, followed by the addition of 220 mg γ-Al2O3 (Beijing Langely Co., Ltd, 99.95%). Subsequently, the mixtures were ultrasonicated for 40 min. The Co-B/Al2O3 was separated by centrifugation (10000 rpm for 10 min) and dried in vacuum at 333 K overnight. Cobalt boride nanoparticles were prepared by the same method but without the addition of Al2O3. The Co/Al2O3 catalyst was obtained by the impregnation reduction method using 94 mg CoCl2·6H2O, 440 mg γ-Al2O3, and 20 mL deionized water. The parameters of the reduction process were programmed, as follows: heating to 393 K (3 K min−1), maintaining at 393 K for 2 h, then heating to 973 K (5 K min−1), and maintaining at 973 K for 2 h (Ar = 70 sccm and H2 = 30 sccm). The Co-B/Al2O3-YH3−x catalyst was obtained by milling 250 mg Co-B/Al2O3 and 250 mg YH3−x in an agate mortar for 5 min in an Ar-filled glove box. The Co-B/Al2O3-YD3−x catalyst was prepared by the same method using YD3−x, and the Co/Al2O3-YH3−x catalyst was prepared by the same method using Co/Al2O3. The prepared catalysts were characterized by powder X-ray diffraction (XRD; PANalytical X’Pert3 Powder, Cu Kα), inductively coupled plasma optical emission spectroscopy (ICP-OES, Prodigy 7, Leeman), X-ray photoelectron spectroscopy (XPS; Axis Ultra, Kratos Analytical Ltd.), field emission high-resolution transmission electron microscope (HRTEM; JEM-2100 F), scanning electron microscopy (SEM; Hitachi S4800), and temperature-programmed desorption (TPD) in Ar flow at temperature ramping rate of 5 K min−1 (Quantachrome Autosorb iQ). Hydrogen storage performance measurement The hydrogen storage performance was measured in an SLM microform high-pressure autoclave (150 mL, Beijing Century Senlang Co., Ltd.) with a high-precision pressure gauge (accuracy 0.01 MPa, L61 K, Hong Kong Huibang Technology Co., Ltd) and a gas flowmeter (50 mL min−1, D07, Beijing Seven Star Flow Co., Ltd.). The setup is illustrated schematically in Supporting Information Figure S1. NEC (1.00 g, Beijing InnoChem Science & Technology Co., Ltd, > 99.9%) and the catalyst (100 mg) were loaded into the autoclave. Neither solvent nor stirring bar was used. The hydrogenation rate was measured using the Sievert method for solid hydrogen storage materials, that is, by monitoring the pressure change in the system, which had been previously calibrated for moles of H2. The pressure drop was ~ 0.6 MPa when the hydrogenation was carried out at 453 K, 10 MPa initial H2 pressure. The dehydrogenation was measured at 0.1 MPa H2 pressure and the released H2 was measured by a gas flowmeter (Beijing Sevenstar D07). The LOHCs were analyzed after hydrogenation and dehydrogenation processes by 1H nuclear magnetic resonance (1H NMR, Bruker-400 M, ARX400) and Fourier-transform infrared spectroscopy (FT-IR, Bruker Tensor 27). Results and Discussion Structure characterization of the catalyst The structural characterization of the individual components in the Co-B/Al2O3-YH3−x catalyst is shown in Supporting Information Figures S2 and S3. The Co-B sample is composed of amorphous spherical particles, 10–20 nm in diameter, which aggregates slightly. The molar ratio of B/Co was ∼ 0.6, as obtained by ICP-AES. The γ-Al2O3 composed of weakly crystallized nanoparticles of ∼ 20 nm in size. YH3−x composed of highly crystallized submicron particles with an irregular shape. Despite the nonstoichiometry, YH3−x showed the typical hexagonal structure of YH3 (ICSD 98-015-4809)31. The Co-B/Al2O3-YH3−x catalyst was a simple physical mixture of the three components. The XRD pattern (Figure 1a) was dominated by the strong diffraction peaks of YH3-x and the weak peaks of Al2O3. In the TEM image (Figure 1d) and the corresponding elemental mapping (see Supporting Information Figure S4b), the features of the three individual components could be easily identified. The HRTEM images (Figure 1d) clearly showed the lattice fringes of crystalline YH3-x and confirmed the amorphous nature of the Co-B particles. Considering the preparation method and the raw materials used, it was reasonable to obtain a simple physical mixture. In the TPD measurement, pure YH3−x and Co-B/Al2O3-YH3−x showed almost identical H2 release profiles (see Supporting Information Figure S5), which again indicated weak interaction between Co-B/Al2O3 and YH3−x. Figure 1 | Structural characterization of the Co-B/Al2O3-YH3−x catalyst: (a) XRD patterns, (b and c) XPS spectra, and (d) HRTEM images. The XRD pattern of the used catalyst is also shown in (a). Download figure Download PowerPoint We employed XPS to study the oxidation states of the elements in the Co-B/Al2O3-YH3−x catalyst (Figures 1b and 1c). The peaks in the Co 2p3/2 core spectrum were apparent at 778.1 and 781.4 eV, assigned to Co(0) and Co(II), respectively. In the B 1s spectrum, the peaks at 192.0 eV and 188.1 eV are corresponding to the oxidized B and B(0) in cobalt boride.32 The B(0) peak at 188.1 eV shifted positively by 0.9 eV, compared with that in elemental B (187.1 eV), suggesting an electron transfer from boron to the vacant d-orbital of cobalt in cobalt boride.32 These XPS features are in good agreement with previous reports on cobalt borides.32–36 Catalytic performance for hydrogen storage in NEC/12H-NEC With the Co-B/Al2O3-YH3−x, NEC can absorb hydrogen at 453 K, 10 MPa H2 and the absorbed hydrogen can be reversibly desorbed at 473 K, 0.1 MPa H2 (Figure 2a). Over 5.5 wt % hydrogen can be absorbed in 2 h and reversibly desorbed in 7 h (Figure 2b).1H NMR analysis suggests that the NEC/12H-NEC interconversion is higher than 94% (see Supporting Information Figure S6).37 After three cycles of hydrogenation and dehydrogenation, there was no significant loss of catalytic activity (Figure 2c). After the hydrogenation and dehydrogenation cycles, the catalyst showed no apparent change, in terms of phase structure and morphology, as shown by the XRD patterns (Figure 1a) and the HRTEM image (see Supporting Information Figure S7), In particular, YH3−x kept its hexagonal YH3 structure after both hydrogenation and dehydrogenation. Therefore, the contribution of YH3−x to the measured hydrogen storage capacity is negligible. There were no detectible gaseous impurities in the released gas, as shown in Supporting Information Figure S8. An 8 W fuel cell could be powered by a small Co-B/Al2O3-YH3−x-catalyzed LOHC hydrogen storage system, as demonstrated in Supporting Information Figure S9. Figure 2 | (a) The conditions for the reversible hydrogen storage in NEC/12H-NEC; (b) The hydrogenation and dehydrogenation kinetics on different catalysts; (c) Reversible hydrogen storage in NEC/12H-NEC on the Co-B/Al2O3-YH3−x catalyst. (Hy: hydrogenation; De: Dehydrogenation). Download figure Download PowerPoint Currently, Ru/Al2O3 and Pd/Al2O3 are the best catalysts for NEC hydrogenation and 12H-NEC dehydrogenation, respectively.15,22 Remarkably, the noble-metal-free Co-B/Al2O3-YH3−x catalyst showed a similar hydrogenation and dehydrogenation efficiency when compared with that of the two milestone catalysts (Figure 2b). To the best of our knowledge, Pd2[email protected] reported by Forberg et al.23 is the only catalyst active for both NEC hydrogenation and 12H-NEC dehydrogenation, which exhibited similar catalytic performance, but their system required much more sophisticated preparation procedure. Indeed, the comparative study displayed in Supporting Information Table S1 shows that almost all the catalysts reported in the literature are noble-metal-based and are only active for either hydrogenation or dehydrogenation reactions.15,23,30,37–42 Favorably, our Co-B/Al2O3-YH3−x catalyst could achieve high catalytic performance for both NEC hydrogenation and 12H-NEC dehydrogenation processes without the use of expensive noble metals. Also, the performances of different catalysts were compared, as shown in Figure 2b, to provide a better understanding of the role of each component of the Co-B/Al2O3-YH3−x catalyst. We observed that Co-B/Al2O3 only showed poor activity for the hydrogenation and dehydrogenation reactions, while YH3−x was completely inactive. Besides, simply blending Co-B with YH3−x also leads to similar activity for hydrogenation and dehydrogenation, but the activity decayed very rapidly due to the agglomeration of Co-B nanoparticles (see Supporting Information Figure S10). Therefore, the catalytic activity originated from the synergism of Co-B and YH3−x, while the Al2O3 component enhanced cyclic stability. The effect of temperature and hydrogenation pressure on the catalytic reactions was investigated. The results demonstrated that the hydrogen uptake rates increased with both the hydrogen pressure (3–10 MPa H2, see Supporting Information Figure S11) and temperature (413–453 K, see Supporting Information Figure S12a), whereas the hydrogen release rates also increase with the increase of temperature (463–483 K), see Supporting Information Figure S12b). The apparent activation energies for the hydrogenation and dehydrogenation reactions obtained from Arrhenius plots were 149 ± 1 kJ mol−1 and 199 ± 7 kJ mol−1, respectively (see Supporting Information Figures S12c and d). Although the kinetics were enhanced, the activation energy value was much higher than that of Ru- and Pd-based catalysts (typically 30–70 kJ mol−1),15,20,43 suggesting a new reaction mechanism of our fabricated catalyst different from conventional surface catalysis. YH3−x-mediated H transfer mechanisms Then we demonstrated that the remarkable catalytic promotion effect of YH3−x is originated from new H transfer mechanisms mediated by YH3−x. We used the isotope labeling method to study the H transfer mechanisms (Figures 3a–3c and Supporting Information Figure S13). The hydrogenation and the dehydrogenation reactions were carried out on the deuterated catalyst Co-B/Al2O3-YD3−x, and the isotope distribution was analyzed after the reaction. The solid catalyst is heated in Ar flow to 1000 K, and the released gas is detected by a mass spectrometer (MS). The gas-phase was measured using a residual gas analyzer, and the liquid phase was analyzed by FT-IR. Figure 3 | (a–c) Analysis of the D content in the liquid, gas, and solid phases after hydrogenation of NEC and dehydrogenation of 12H-NEC on Co-B/Al2O3-YH3−x (De: 473 K, vacuum, 30 min): (a) FT-IR spectra of the hydrogenation products; (b and c) composition of the gas phase and the gas generated from heating the solid catalyst analyzed by mass spectroscopy; (d and e) The kinetic isotope effect (KIE) of NEC hydrogenation (453 K, 3 MPa H2/D2) and 12H-NEC/12D-NEC dehydrogenation (473 K, 0.1 MPa H2/D2) on different catalysts: (d) Co-B/Al2O3-YH3−x, (e) Ru/Al2O3 for hydrogenation and Pd/Al2O3 for dehydrogenation. (12D-NEC is obtained by deuteration of NEC on Ru/Al2O3 catalyst.) Download figure Download PowerPoint We first analyzed the isotope distribution after Co-B/Al2O3-YD3−x catalyzed hydrogenation. Partial deuteration was observed in the liquid product , as evidenced by the C-D stretching peak at 2160 cm−1 in the FT-IR spectrum (Figure 3a). MS analysis suggested that a large amount of H was found in the solid catalyst after hydrogenation, while the gas phase remained H2 (Figures 3b and 3c). This notable redistribution indicated that the lattice D in YD3−x could be transferred to NEC to form 12(H/D)-NEC, and the D vacancies could be replenished by H2. The hydrogen transfer mechanism is schematically illustrated in Figure 4a. Co-B is the other essential component in the composite catalyst. Here, the role of the Co-B was to activate the NEC molecules. Figure 4 | (a) The YH3−x-mediated H transfer mechanisms in NEC hydrogenation and 12H-NEC dehydrogenation reactions of the Co-B/Al2O3-YH3−x catalyst; (b and c) The PCI curve (left) and the corresponding H chemical potential (μH) diagram (right) for the H transfer processes in (b) NEC hydrogenation and (c) 12H-NEC dehydrogenation. Spontaneous H transfer requires a downward change of μH. The filled H bonding sites or bands are represented in green. Download figure Download PowerPoint We then analyzed the isotope distribution after dehydrogenation. H2, HD, and D2 were detected in the generated gas, and the solid catalyst also contained both H and D (Figures 3b and 3c). For comparison, direct H-D exchange of YD3−x with 0.1 MPa H2 at the same temperature results in much less D release. As shown in Supporting Information Figure S14, the gas phase only contained a low fraction of HD but no D2 after 2 h. Therefore, the presence of 12H-NEC facilitated the release of the lattice D from YD3−x. This is quite surprising as the release of the lattice H was only affected by the H2 partial pressure in general circumstances. A reasonable explanation is that H in 12H-NEC could be transferred into the vacancies of the YD3 lattice, which facilitated the release of the lattice D, as schematically illustrated in Figure 4c. Similarly, here, the 12H-NEC molecules needed to be adsorbed and activated by the Co-B catalyst. We further investigated the kinetic isotope effect (KIE) in hydrogenation and dehydrogenation of NEC. For the Co-B/Al2O3-YH3−x catalyst, hydrogenation using H2 is much faster than deuteration using D2, and dehydrogenation of 12H-NEC is much faster than that of 12D-NEC (Figure 3d and Supporting Information Figure S15). The ratios of the initial reaction rates rH/rD for H/D uptake and release were was 2.4 and 6.9, respectively. For comparison, there was negligible KIE observed for the hydrogenation catalyst Ru/Al2O3 and dehydrogenation catalyst Pd/Al2O3 (Figure 3e). In fact, conventional surface hydrogenation and dehydrogenation catalysts typically show no KIE.44,45 Thus, the large KIE of the Co-B/Al2O3-YH3−x catalyst suggested that H transfer played a vital role in both the hydrogenation and dehydrogenation of NEC. The H transfer mechanisms shown in Figure 4a involve H diffusion in the YH3−x lattice, which could well explain the large KIE by the different diffusion rate of H and D.46 Synergism between YH3−x and the transition metal catalyst Efficient hydrogenation and dehydrogenation require both effective activation of the LOHC molecules and fast H transfer. On conventional surface catalysts, the two processes mentioned above have to be spatially coupled: H atoms or empty H bonding sites on the catalyst surface have to be around the activated LOHC molecule. Each NEC/12H-NEC molecule requires up to 12 H atoms. As a result, H transfer is rather inefficient due to insufficient H atoms or empty H bonding sites around the activated NEC/12H-NEC molecules. Indeed, many non-noble metal catalysts are active only for hydrogenation and dehydrogenation of smaller molecules, while a few of them could be used for hydrogenation of NEC and dehydrogenation of 12H-NEC.47–49 Nonstoichiometric YH3−x is a large reservoir for both atomic H and H-bonding sites with fast hydrogen absorption and desorption kinetics, which provides an alternative H transfer pathway, as suggested by the isotope labeling results (Figure 3). On the other hand, YH3−x barely showed any activity for both hydrogenation and dehydrogenation, which indicated that YH3−x was unable to activate the NEC/12H-NEC molecules. Therefore, the Co-B/YH3−x synergism is essential for efficient hydrogenation and dehydrogenation (Figure 2b): Co-B for activation of the NEC/12H-NEC molecules and YH3−x for fast H transfer. Moreover, H transfer between the activated NEC/12H-NEC molecules requires H migration across the Co-B/YH3−x interface. In this work, the interface formed randomly during the mixing process. Although quite an excellent catalytic performance effect was achieved, the randomly formed Co-B/YH3−x interface was gradually detached after 5 hydrogenation and dehydrogenation cycles, which is one major cause of the performance decay. Further study on more accurate control of the Co-B/YH3−x interface during preparation is required to improve the H transfer efficiency. To further verify this synergetic mechanism, we studied the promotion effect of YH3−x to other Al2O3-supported transition metal (TM) catalysts. As shown in Figure 5, despite the different catalytic performance, adding YH3−x by simple hand milling leads to considerable performance enhancement for all the TM/Al2O3 catalysts investigated, which suggests that the TM/YH3−x synergism is quite general for NEC hydrogenation and 12H-NEC dehydrogenation. In this work, the TM/YH3−x synergism shares some similarity with the hydride-promoted ammonia synthesis.50–54 According to the mechanistic study, N2 was dissociated on TM or TM nitride, and the atomic N was transferred to the hydride to form imide/amide intermediates, which subsequently, yielded NH3 and regenerated the hydride upon reaction with H2. Here, the TM/hydride synergism also played a key role: TM is used to activate N2, and the hydride served as an H donor for the activated N species. Figure 5 | Catalytic performance of different catalysts in reversible hydrogen storage in NEC. The H2 absorbed/released in 2 h is used as a figure of merit. The hydrogenation conditions on Ni/Al2O3 and Ni/Al2O3-YH3 are 423 K, 3 MPa H2.29 The results of dehydrogenation on Ru/Al2O3 and Ru/Al2O3-YH3 are not measured. Download figure Download PowerPoint As shown in Figure 5, the catalytic performance also depends strongly on TM. Systematic optimization of TM is beyond the scope of the current study. Co-B/Al2O3-YH3−x exhibited the highest two-direction catalytic performance among all the noble-metal-free TM/Al2O3 catalysts. However, Co/Al2O3-YH3−x prepared by a thermal reduction in H2 only showed very low activity (see Supporting Information Figure S16). The XPS spectra (Figures 1b and 1c) suggest electron transfer between Co and B, which might lead to a more effective activation effect for both NEC and 12H-NEC on Co-B. On the other hand, Ni-B/Al2O3-YH3−x was more active for 12H-NEC dehydrogenation but less active for NEC hydrogenation compared to Ni/Al2O3-YH3−x (see Supporting Information Figure S17). Therefore, to validate the exact role of B incorporation in the TM, further investigation is required. Importance of the nonstoichiometry of YH3−x The mechanisms shown in Figure 4a require spontaneous H transfer from YH3−x to NEC in hydrogenation and from 12H-NEC to YH3−x in dehydrogenation. The H transfer between the two condensed phases does not involve H2, so the reverse H transfer direction could not be explained by the different H2 pressure in hydrogenation and dehydrogenation. The temperature difference is also too small to reverse the H transfer direction. In this section, we show that the nonstoichiometry of YH3−x is essential for the two-direction H transfer, which is seemingly contradictory to thermodynamics. In the ideal YH3 structure, H fills all the tetrahedral and octahedral interstices of the hexagonally closest packed Y lattice. The nonstoichiometric YH3−x has unoccupied octahedral interstices. The H occupancy is determined by the H2 partial pressure, which is described quantitatively by the pressure-composition isotherm (PCI).55 Yttrium could form two hydrides, YH2 and YH3. The equilibrium H2 pressure for the YH2–YH3 transition was 186 Pa at 473 K. Therefore, the hexagonal YH3 structure in YH3−x is kept in both hydrogenation (10 MPa H2) and dehydrogenation (0.1 MPa H2) reactions, which is in agreement with the XRD results (Figure 1a). However, the H concentration and H chemical potential of YH3−x were different in hydrogenation and dehydrogenation processes. We obtained a PCI curve, which provides a quantitative description of gradual filling of the H bonding sites in a metal hydride (MH) with the decreasing bonding strength when the H2 pressure increases gradually. Therefore, the PCI curve could also be regarded as the integrated density of states (DOS) of the H bonding sites in MHs. The equilibrium