Introduction

Solid-state hydrogen storage has been widely considered as the safest and most effective way to achieve the widespread use of H2 as a prominent energy carrier in the future1. Despite decades of extensive efforts, it is still a big challenge to find a suitable H-rich material that features high safety, low cost and fast kinetics for H2 absorption and desorption under moderate conditions2,3,4. Metal-N-H systems have been extensively investigated as potential materials for solid-state hydrogen storage since the discovery of the hydrogen storage performance of the Li-N-H system, which was found to reversibly store 11.5 wt. % hydrogen5,6. Inspiringly, it was verified that partial substitution of Mg for Li in the Li-N-H composite to form the Li-Mg-N-H system could significantly decrease the desorption enthalpy change7,8,9. Among such composites, the Mg(NH2)2-2LiH system is one of the most promising hydrogen storage materials due to its favorable thermodynamics (39 kJ mol−1 H2), which is lowered by 35% compared with the LiNH2-LiH system, as well as its relatively high H2 content and good reversibility8,9,10,11. According to the H2 sorption process as described in Equation (1), 5.6 wt. % H2 could be reversibly stored in this system at an operating temperature of 90°C under 1 bar of equilibrium hydrogen pressure, based on the thermodynamic calculation12,13, which could satisfy the practical requirements of proton exchange membrane (PEM) fuel cells.

Nevertheless, due to both the sluggish interface reaction between Mg(NH2)2 and LiH in the preliminary step and the following mass transport across the thus-formed imide layer, which results in a rather high activation energy barrier (Ea: ~102–120 kJ mol−1), a temperature higher than 200°C is required to achieve a reasonable H2 sorption rate13,14,15. In addition, another big challenge for practical application of the Li-Mg-N-H system is the undesirable release of ammonia resulting from the insufficient reaction of Mg(NH2)2 with LiH, which not only significantly degrades the H2 cycling capacity due to the loss of elemental N, but also would damage the active catalyst in the PEM fuel cells16,17. Previous experimental results have clearly established that, in terms of mitigating the above-mentioned problems, intimate contact between Mg(NH2)2 and LiH through the reduction of their particle sizes is an effective way of improving the hydrogen storage properties of the Li-Mg-N-H system11,18. It is well-known, however, that particles with small sizes are intrinsically unstable and prone to grow into larger crystallites19. Thus, during long-term cyclic heat treatment for H2 release and uptake, phase separation and particle growth are inevitable, which would significantly degrade the sorption kinetics19. Therefore, an efficient host, which is both capable of uniformly confining the nanostructured morphology of the thus-synthesized complex hydrides during repeated heating and is chemically inert, is essential for preserving the nanosize effects on the active substrates inside.

To this end, nanoconfinement of complex hydrides in porous scaffolds, which can serve as containers, has emerged as an intriguing strategy to synthesize nanostructured materials and therefore improve their hydrogen storage properties19,20,21,22,23,24,25. The porous templates can effectively, on the one hand, facilitate the formation of nanoparticles and, on the other hand, maintain the shape and size of the active materials during cycles of hydrogen sorption19. In general, the preparation process involves melt infiltration or solution infiltration, which requires good wettability of the porous scaffolds by the precursors. Moreover, in order to achieve a relatively high loading level with good distribution, many repetitions of the infiltration steps and the following drying are required26. This makes the nanoconfinement time-consuming and tedious, while not allowing control over the homogeneous distribution of confined composites26,27. More importantly, the Li-Mg-N-H system is composed of multiphase components and all of them have high chemical activity, low solubility, and/or high melting points, which therefore hinders the direct synthesis of a nanosized Li-Mg-N-H composite by the conventional nanoconfinement technique.

Herein, we report a novel in-situ reaction strategy based on an electrospinning technique, which is inexpensive, simple and versatile, for the preparation of nanowoven materials, consisting of long entangled nanofibers of different materials, to fabricate nanowires (NWs) of Li-Mg-N-H composite uniformly coated with hierarchical porous carbon. As illustrated in Fig. 1, the overall synthetic procedure for the carbon-coated Li-Mg-N-H NWs originates from electrospinning coupled with the in-situ fabrication of the porous carbon coating and the consecutive transformation of the precursors to form the Li-Mg-N-H composite. Due to the special physicochemical properties of the components in the Li-Mg-N-H system, as mentioned above, our synthetic procedure is based on the in-situ reaction between MgCl2 and LiN3 as the electrospinning precursors. Upon heat treatment concurrent with carbonization to produce the uniform carbon coating, LiN3 firstly decomposes into Li3N with vigorous release of nitrogen, which then reacts with MgCl2 to produce Mg3N2 via a metathesis reaction (2). With further increase of temperature, the interaction between the thus-formed Mg3N2 and the residual Li3N results in the formation of disordered LiMgN-type structures (Fig. S1 and S2 in the Supporting Information)28. Afterwards, annealing under H2 atmosphere induces the hydrogenation according to Equation (3)29, which, as a result, forms the composite of Mg(NH2)2 and LiH, with LiCl as the by-product. The formation of Mg(NH2)2, LiH and LiCl, as detected by X-ray diffraction of the bulk precursors after annealing (Fig. S1), could verify the feasibility of this procedure.

Figure 1
figure 1

Schematic diagram of the in-situ technique to fabricate the carbon-coated Li-Mg-N-H NWs via electrospinning using a single-needle nozzle (top) and typical SEM images of the resulting products at different stages during preparation (bottom).

Full size image

Results

Scanning electron microscope (SEM) images of the as-electrospun NWs are shown in Fig. 1. When a high voltage is applied between the outer metallic capillary, containing the mixed precursor solution of LiN3, NH4MgCl3 and poly(vinyl alcohol) (PVA) and the collector, the liquid jets are stretched by electrostatic forces to generate textile NWs with a uniform coating of PVA. From the SEM images, the formation of a highly interconnected network of the as-electrospun NWs with a diameter of ~250 nm is clearly observed. Subsequently, the as-collected NWs were dewatered at ~85°C and the temperature was then programmed to rise to 250°C under dynamic vacuum, meanwhile realizing the transformation of NH4MgCl3 into MgCl2, accompanied by the release of the thus-formed ammonia and hydrogen chloride. Simultaneously, the formation of Li3N due to the decomposition of LiN3 was verified by the color change (Fig. S3) to the characteristic color of Li3N, i.e., ruby red from white, across this heating step. On elevating the calcination temperature to 550°C under N2 atmosphere, the PVA substrates were then transformed into pyrolysis carbon upon thermal decomposition, which homogeneously puts “carbonaceous clothes” on the active composite inside, accompanied by the metathesis reaction between MgCl2 and Li3N. As shown in Fig. 1, the carbonized NWs perfectly maintain the structural integrity of the as-electrospun polymers and, compared to polymeric NWs with smooth surfaces, exhibit rough surfaces with small relics and mesoporous structure on the outside wall. This is supposed to stem from the explosive release of N2 during the decomposition of LiN3, which could push away the surrounding PVA substrate from the point where LiN3 exists, leading to the formation of numerous pores after the carbonization of relatively flexible polymers into relatively stiff carbon sheets at high temperature27. Magnified images of Li-Mg-N-H NWs (Fig. 2a) can further demonstrate the presence of open pores on the external surface, with pore diameters in the range of 2–40 nm. It also was found that the diameter of the as-electrospun PVA NWs was significantly decreased after carbonization, due to the decomposition of the polymer during calcination30. In the high-resolution transmission electron microscope (HRTEM) image (Fig. 2b), not only the wirelike nanostructure, but also the porous interior with mesopores and macropores can be clearly observed. This verifies the homogenous distribution of pores, which can effectively facilitate the H2 transportation across the carbon matrix, owing to the explosive decomposition of LiN3 inside the carbonized wires, indirectly confirming the uniform dispersion of Li-Mg-N-H composite. The HRTEM image (inset of Fig. 2b) also reveals the presence of small mesopores (~2.5 nm) inside the disordered carbon sheets. The N2 adsorption-desorption isotherms of the as-prepared NWs exhibit typical type-IV hysteresis, with an obvious increase at a high relative pressure (P/Po = 0.80–0.99), indicative of the presence of both macropores and mesopores, due to the compression of PVA by the strong power of the N2 stream and the subsequent solidification from carbonization, which agrees well with the TEM results. In addition, the isotherms show increases in adsorption at a low relative pressure, suggesting the presence of a large number of micropores in the NWs after calcination. These pores may be derived from the free volume and microporosity in the carbon NWs that is attributed to the disordered packing of the turbostratic carbon sheets and clusters as a result of the carbonization process of the nanostructured polymers31,32,33. Taking advantage of the one-dimensional (1D) and hierarchical porous architecture, the Li2Mg(NH)2@C possesses a relatively large Brunauer-Emmett-Teller (BET) surface area of around 438 m2 g−1 and pore volume of 0.86 cm3 g−1 according to the N2 sorption isotherms (Fig. S4). Moreover, the pore size distribution further demonstrates the formation of mesopores in the as-prepared NWs. Accordingly, on account of the special physico-chemical properties of every individual component in the as-electrospun matrices, the hierarchical porous carbon NWs can be obtained via a simple single-nozzle electrospinning. It is worth noting that the energy dispersive spectroscopy (EDS) elemental map (Fig. 2d) of C coincides well with the structure of the NWs, while the maps of Cl, Mg and N agree well with the C map, which directly demonstrates the homogeneous dispersion of the precursors inside the carbon NWs.

Figure 2
figure 2

(a) High-magnification SEM image of a single Li-Mg-N-H NW; (b) TEM image of Li-Mg-N-H NWs (inset: high-magnification TEM image (scale bar 5 nm), in which the marked area clearly shows the small and large mesopores coated with carbon). TEM image (c) and the corresponding elemental mapping (d) of the as-prepared Li-Mg-N-H NWs with nanoconfinement of the precursors. (e) High-resolution XRD spectra of the Li-Mg-N-H NWs upon heating to various temperatures.

Full size image

After the carbothermal treatment and annealing in the H2 atmosphere, high-resolution X-ray diffraction (HRXRD) verified the formation of LiCl, as shown in Fig. 2e, while other products are undiscernible due to a lack of long-range order. The characteristic peaks assigned to the N-H bonds of Mg(NH2)2 can be clearly detected, however, from the Fourier transform infrared (FTIR) spectra (Fig. S5), indicating the presence of Mg(NH2)2 embedded in the porous NWs. Moreover, the in-situ HRXRD confirms the generation of Li2Mg(NH)2 with increasing temperature, providing direct evidence of the successful synthesis of the porous carbon-coated Li2Mg(NH)2 NWs. In addition, the surrounding carbon matrix appears to have an amorphous structure, acting as a strong buffer to accommodate the sintering of nanostructured Li2Mg(NH)2 during the cycles of H2 sorption. As deduced from the Debye-Scherrer equation, the average crystallite size of the carbon-coated Li2Mg(NH)2 is only ~4 nm. Clearly, the electrospinning will, on the one hand, eliminate the tedious process of infiltration and drying and, at the same time, is suitable for good encapsulation and distribution of the precursors inside the as-electrospun NWs34, while the blocking of pores and/or agglomeration of active composite on the outside surface of the templates is usually observed during nanoconfinement, which significantly limits complete pore filling and good dispersion of complex hydrides inside the scaffolds26. Finally, the in-situ metathesis reaction between MgCl2 and Li3N stemming from the well-dispersed precursors inside the thus-formed carbon NWs, followed by the hydrogenation process, leads to the formation of the nanosized Li-Mg-N-H composite with homogeneous distribution. Therefore, as a result of the explosive release of N2 from the as-electrospun wires upon carbothermal treatment, the obtained carbon-coated Li-Mg-N-H-based NWs possess a well-designed 1D nanostructure and a three-dimensional (3D) interconnected texture full of micropores, mesopores and macropores, which can effectively hinder the growth and agglomeration of the confined nanoparticles during heat treatment and promote H2 access through the porous carbonaceous framework to interact with the active substrates. This is the first report, to the best of our knowledge, on the in-situ synthesis of a carbon-coated Li-Mg-N-H-based system with 3D porous nanostructures.

To compare the dehydrogenation behavior of the Li-Mg-N-H-based NWs with that of a ball-milled composite of Mg(NH2)/2LiH, mass spectrometry (MS), thermogravimetry (TG) and volumetric temperature-programmed desorption (TPD) characterizations were conducted, as shown in Fig. 3. The bulk Li-Mg-N-H system after ball milling starts to dehydrogenate above 140°C, with dehydrogenation peaking at ~199°C, along with the simultaneous emission of ammonia, which is generally consistent with reports in the literature35,36,37,38,39,40,41. In comparison, the carbon-coated Li-Mg-N-H NWs exhibited onset and peak temperatures of H2 desorption that were downshifted to ~78°C and 113°C, respectively, which are 62°C and 86°C lower, respectively, than for their ball-milled counterpart. It is also noteworthy that the generation of the toxic by-product, i.e., NH3, was significantly depressed in the thermal dehydrogenation of the Li-Mg-N-H NWs, which is favorable for both reversibility and onboard practical applications as a hydrogen-storage medium. The quantitative H2 desorption measurement by the TG method confirmed the dramatically reduced operating temperature of the carbon-coated Li-Mg-N-H NWs, which corresponds well with the MS spectra. A weight loss of 5.5 wt. % H2 could be achieved in the temperature range between 78°C and 155°C. From the in-situ HRXRD results (Fig. 2e), the formation of Li2Mg(NH)2 upon dehydrogenation is confirmed to start at a temperature of ~110°C, which is consistent with the TG-MS results.

Figure 3
figure 3

(a) Mass spectra (top) and thermogravimetry curves (bottom) of the carbon-coated Li-Mg-N-H NWs and bulk Mg(NH2)2/2LiH composite after complete hydrogenation. Hydrogen desorption (b) and absorption (c) curves of the carbon-coated Li-Mg-N-H NWs at different temperatures, with the ball-milled Mg(NH2)2/2LiH composite at 135°C included for comparison. Carbon was not considered as an active component for the hydrogen storage measurements.

Full size image

The isothermal H2 desorption and absorption properties (Fig. 3b, c) were further characterized by volumetric measurements to clarify the effects of both the reduction in the particle size and the carbonaceous coating with highly porous nanostructure on the modification of the Li-Mg-N-H system. The ball-milled composite shows a H2 capacity of approximately 0.5 wt. % and 0.65 wt. % for desorption and absorption, respectively, within 60 min, even at the high temperature of 135°C. On the contrary, the carbon-coated Li-Mg-N-H NWs display a fast hydrogenation at 135°C, with a capacity reaching up to 5.5 wt. % and a subsequent hydrogen release as high as ~5.2 wt. % can be implemented over the period of 60 min. In particular, upon decreasing the operating temperature to 105°C, the H2 sorption from the Li-Mg-N-H NWs can still proceed and a complete cycle of H2 uptake and release (~5.5 wt. %) can be achieved with an extended holding time (Fig. S6). In order to quantitatively estimate the enormously enhanced hydriding/dehydriding kinetics for the as-prepared carbon-coated Li-Mg-N-H NWs, the apparent activation energies (Ea) during the H2 absorption/desorption were determined by combining the Johnson-Mehl-Avrami equation with the Arrhenius equation after fitting the experimental results at various temperatures. Based on the slope of the linear plots of ln (k) (k, rate constant) versus 1/T (T, absolute temperature), the activation energy for hydrogenation and dehydrogenation of the porous carbon-coated Li-Mg-N-H NWs was verified to be around 23.9 and 24.5 kJ mol−1 H2, respectively, very much lower than for the bulk counterpart (Fig. S7). This directly confirms the significant enhancement of kinetics for the carbon-coated Li-Mg-N-H NWs, due to the reduction of particle size and therefore diffusion distances.

Another important issue with Li-Mg-N-H composite for hydrogen storage application is the significant degradation of H2 capacity upon cycling, owing to the release of ammonia and the particle growth and phase separation during thermal treatment. The cycling behaviour of the carbon-coated Li-Mg-N-H NWs, in comparison with their bulk counterpart, is shown in Fig. 4 and Fig. S8, using isothermal dehydrogenation and hydrogenation. The H2 capacity, as cycling of the bulk Li-Mg-N-H composite proceeds, is severely degraded from 0.7 wt. % to 0.35 wt. % after only 5 cycles of sorption, which is equal to a capacity retention lower than 50%, even with a dwell time of 400 min for one cycle at 130°C. In contrast, in the case of the carbon-coated Li-Mg-N-H NWs cycled under the same conditions, but over a significantly shorter time, the H2 capacity still reaches 5.3 wt. % and the capacity retention is up to 94.6% after 20 consecutive cycles of hydrogenation and dehydrogenation. The characteristic peaks of Mg(NH2)2 and Li2Mg(NH)2 can be clearly observed, even through 20 cycles of sorption (Fig. S9) and the porous structure of the Li2Mg(NH)2@C NWs is well maintained after cyclic heat treatment (Fig. S10). Furthermore, the curves of absorption and desorption during the cyclic H2 sorption are almost identical to each other (Fig. S11), indicating the greatly enhanced hydrogen storage performance, i.e., capacity and kinetics were well-preserved, which is attributed to a strong tolerance of the porous carbonaceous coating towards particle growth and sintering.

Figure 4
figure 4

(a) Reversible H2 sorption performance of the carbon-coated Li-Mg-N-H NWs at 130°C and (b) H2 capacity dependence of the carbon-coated Li-Mg-N-H NWs (green) and their ball-milled counterpart (black) on cycle number.

Full size image

Discussion

It is well known that the high activation energy barriers for both the hydrogenation and the dehydrogenation of the Li-Mg-N-H composite results from the sluggish interface reaction due to the slow diffusion of atoms (e.g., Li+, Mg2+ and H+ ions) across amide-imide and imide-hydride phase boundaries and the mass transport along the as-formed layer of imides18,42. The reduction of particle size into nanometer range can significantly decrease the diffusion pathways across amides/imides and hydrides, increase their surface area and develop closer proximity between Mg(NH2)2 and LiH, which would lead to the significantly improved kinetics and thermodynamics towards superior hydrogen storage performance. In the case of the ball-milled Li-Mg-N-H composite, the sizes of most of the particles are larger than 1 μm (Fig. S12) and therefore, the surface-to-volume ratio is small, which endows it with high activation energy. By contrast, it is notable that the in-situ fabrication of the carbon-coated Li2Mg(NH)2 via electrospinning followed by calcination and annealing can, on the one hand, realize the direct synthesis of nanosized Li2Mg(NH)2 (~4 nm), significantly improving its hydrogen storage performance (Fig. 3), while, on the other hand, the space-confinement function of the porous carbon coating can well preserve the nanostructural features down to the nanometer scale during consecutive cycles of heating, which therefore gives the Li2Mg(NH)2 stable reversibility without apparent degradation, even up to 20 cycles of de-/re-hydrogenation (Fig. 4).

In summary, we have successfully prepared hierarchically porous Li2Mg(NH)2@C NWs via a simple and scalable single-nozzle electrospinning technique with a subsequent carbothermal reaction, which realizes the in-situ formation of a porous carbon coating serving as template, with the Li-Mg-N-H composite produced via the subsequent reaction under physical restriction. Our results obviously verified the significantly improved hydrogen storage performance, including complete reversibility at a temperature as low as 105°C and tremendous reduction of the hydriding/dehydriding temperature of the Li2Mg(NH)2, with the reduction of particle size down to the nanometer scale. Induced by the space-confinement of the porous carbon coating, the Li2Mg(NH)2@C NWs present very good cycling stability, close to the theoretical value over 20 cycles of de-/re-hydrogenation at 130°C. To the best of our knowledge, no such highly stable cycling at relatively low temperature has been reported for Li-Mg-N-H based materials. However, it should be noted that, due to the extra weight of the porous carbon NWs and the by-product of LiCl, the actual reversible H2 capacity in the whole system is limited. Therefore, further works in the search of an effective solvent to remove the inactive LiCl and/or a new synthetic strategy to achieve a high loading percent of precursors are still required to achieve a high hydrogen storage capacity in the whole system. Nonetheless, these results demonstrate the potential of this strategy in further improving the hydrogen storage performance of complex hydrides and designing the nanowire-like structure with a wide range of materials for developing high-performance energy storage applications.

Methods

Synthesis of hierarchical porous Li2Mg(NH)2@C nanowires

Poly(vinyl alcohol) (PVA, 0.5 g, Alfa Aesar, Mv = 88000, CAS: 9002-89-5) was mixed with deionized water (5 mL) and stirred in a water bath at 90°C for 8 h to make the PVA solution. After cooling down to room temperature, LiN3 solution (0.5 mL, 20 wt. %) and NH4MgCl3 (0.05 g) in water were then added and vigorously stirred for 6 h to make a homogeneous spinning dope. The resultant precursor solution was poured into a syringe with a 18-gauge blunt tip needle. The flow rate of the solution was approximately 500 μL h−1 and was controlled by a syringe pump (Longer, TJP-3A, China). A grounded stainless steel plate was horizontally placed 10 cm from the needle to collect the as-electrospun NWs. A high voltage of 15 kV was applied by a high-voltage power supply. The as-collected electrospun nanowires were firstly dewatered at a temperature of 85°C under dynamic vacuum for 15 h and then calcined at 550°C for 3 h under dynamic N2 atmosphere to obtain the porous carbon NWs containing the precursors (heating rate 1°C min−1). Finally, the thus-formed carbon NWs were annealed under 150 bar H2 at 300°C for 12 h for a complete hydrogenation, as schematically illustrated in Fig. 1 and then further annealed under vacuum for dehydrogenation at 150°C, which leads to the formation of the hierarchically porous Li2Mg(NH)2@C NWs. According to elemental analysis, the resulting nanowires contained 9.86 mass% Li, 5.7 mass% Mg, 6.23 mass% N, 2.8 mass% O, 16.7 mass% Cl and 58.5 mass% C, which gives the mass concentration of Li2Mg(NH)2 of ~15.2 mass% in the as-prepared carbon NWs.

Preparation of the ball-milled composite

The ball-milling of Mg(NH2)2 and LiH with a molar ratio of 1:2 was conducted via a planetary QM-1SP2 for 10 h. The ball-to-powder ratio was 30:1, with a milling speed of 500 rpm. The milling procedure was carried out by alternating between 30 min of milling and 6 min of rest. In order to prevent contamination by air, all handling and manipulation of the materials were performed in an argon-filled glove box with a recirculation system to keep H2O and O2 levels below 1 ppm.

Materials characterization

Mass spectrometry (MS; Hidden HPR 20) was conducted from room temperature, using a heating rate of 2°C min−1 under dynamic nitrogen with a purge rate of 80 ml min−1. Differential scanning calorimetry (DSC) measurements were performed with a TAQ 2000 DSC under N2 with a gas flow of 40 mL min−1 at a heating rate of 2°C min−1. Nitrogen absorption/desorption isotherms (Brunauer-Emmett-Teller (BET) technique) at the temperature of liquid nitrogen via a Quantachrome NOVA 4200e instrument were collected to characterize the pore structure of the samples. The pore volumes and pore size distributions were obtained by using the Barrett-Joyner-Halenda (BJH) model from the adsorption branches of isotherms. The phase composition of the powders was analyzed by X-ray diffraction (XRD, D8 Advance, Bruker AXS) with Cu Kα radiation. Amorphous tape was used to prevent any possible reactions between the sample and air during the XRD measurement. High-resolution X-ray diffraction data were collected by a Mythen-II detector at the Powder Diffraction Beamline, Australian Synchrotron. The sample was loaded into a pre-dried 0.7 mm boron-silica glass capillary tube and the capillary tube was then sealed with vacuum grease in a glove box filled with argon. In order to identify the phase transformations during dehydrogenation, time-resolved in-situ measurements were conducted using a Cyberstar hot-air blower to heat the capillary from room temperature to the desired temperature, at a constant heating rate of 2°C min−1. Fourier transform infrared (FTIR, Magna-IR 550 II, Nicolet) analysis was conducted to determine the chemical bonding. During the FTIR measurements (KBr pellets), samples were loaded into a closed tube with KBr for measurement in an argon-filled glove box. The morphology of the samples was evaluated using a field emission scanning electron microscope (FE-SEM, JEOL 7500FA, Tokyo, Japan) and a transmission electron microscope (TEM, JEOL 2011 F, Tokyo, Japan). Elemental analysis was performed with an Elemen Tar Vario EL3 Elemental Analyser.

The hydrogen storage properties of the thus-synthesized Li2Mg(NH)2@C NWs were investigated on a Sievert's apparatus, denoted as a gas reaction controller (GRC, Advanced Materials Corp., USA). The apparatus was carefully calibrated from the H2 sorption of a LaNi5 reference sample with an accuracy of ±1%, and, typically, a ~800 mg sample was loaded into a stainless-steel autoclave for hydrogenation and dehydrogenation measurements. The H2 absorption kinetics measurements were performed at various temperatures with an initial pressure of 35 atm and the desorption properties were detected at various temperatures under a hydrogen pressure below 0.02 atm. The pressure-concentration isotherm (PCI) measurements were performed at the desired temperatures and the equilibrium time for each point was 600 s. For comparison purposes, the carbon scaffold masses were excluded when determining the amount of hydrogen released from the relevant composites containing carbon scaffolds.