Solid-Phase Hydrogen Storage Based on NH3BH3-SiO2 Nanocomposite for Thermolysis

Current H2-proton exchange membrane fuel cell systems available for commercial applications employ heavy and high-risk physical hydrogen storage containers. However, these compressed or liquefied H2-containing cylinders are only suitable for ground-based electric vehicles, because although highly purified H2 can be stored in a cylinder, it is not compatible with unmanned aerial vehicles (UAVs), which require a lighter and more stable energy source. Here, we introduce a chemical hydrogen storage composite, composed of ammonia borane (AB) as a hydrogen source and various heterogeneous catalysts, to elevate the thermal dehydrogenation rate. Nanoscale SiO2 catalysts with a cotton structure dramatically increase the hydrogen evolution rate on demand, while simultaneously lowering the startup temperature for AB thermolysis. Results show that the dehydrogenation reaction of AB with a cotton-structured SiO2 nanocatalyst composite occurs below 90°C, the reaction time is less than a minute, and the hydrogen generation yield is over 12 wt%, with an activation energy of 63.9 kJ·mol-1.


Introduction
Hydrogen-an abundant element in the universe-is a theoretically eco-friendly fuel and a clean energy source for fuel cells, as it produces no environmentally hazardous exhaust. Recent interest in hydrogen fuel has centered on how to safely store hydrogen with enhanced energy density. Essentially, there are two major hydrogen storage methods available: physical and chemical storage, where the latter includes sorbents, metal hydrides, and chemical hydrides [1]. Most commercially available hydrogen fuel cell-driven automobiles utilize physical storage, which is more reliable than chemical storage because pure hydrogen molecules are highly compressed, or even liquefied, to be stored in a physical storage tank. However, continuously raised safety concerns and the heavy weights of hydrogen fuel containers, which are absolutely incompatible with aerial vehicles, limit the employment of physical storage for some ground-based transportation vehicles.
Chemical hydrides store hydrogen fuel in various hydride forms such as sodium borohydride, alane, and ammonia borane (AB). These hydride compounds can be decomposed thermally or hydrolytically to generate H 2 gas. Although hydrolytic dehydrogenation of chemical hydrides occurs at temperatures lower than 80°C, the requirement of expensive catalysts for hydrolysis, the production of ammonia as a byproduct, and the low H 2 yield are practically incompatible with aerial vehicles [2][3][4][5][6][7][8]. Thermolysis of the chemical hydride does not require a noble metal catalyst and is relatively free of ammonia poisoning [9][10][11]. The H 2 yield through thermolysis of AB, which is a frequently employed hydride compound belonging to the chemical hydride group, can be increased up to 13 wt% [9]. In addition, a mixed approach has been introduced, involving a combination of both thermolysis and hydrolysis in series for the dehydrogenation of the AB-TiO 2 composite to enhance the hydrogen yield to the theoretical maximum [12]. Nevertheless, a high temperature requirement for the thermal dehydrogenation reaction and ammonia poisoning through trace amounts of water content in a reactor or reactants remain challenging.
In accordance with recent demands for unmanned aerial vehicles (UAVs) for aerial surveys, weather observations, radio communication relays, and military purposes [13], here, we propose AB as a component for a disposable hydrogen fuel composite, which can be used for proton exchange membrane fuel cells (PEMFC). As an alternative energy storage device, a PEMFC powered by hydrogen fuel can immediately take advantage of the high efficiency of PEMFC (~50%) and high specific energy density of H 2 gas (~39,500 Wh·kg -1 ). Indeed, a mechanically low-noise H 2 -PEMFC system will perform better for a long-distance and low-altitude flight with UAVs than secondary lithium batteries, which are presently widely utilized, but commonly suffer from low energy density (~250 Wh·kg -1 ) [14,15]. Safe hydrogen storage and effective hydrogen generation on demand are of considerable importance in H 2 -PEMFC power supply systems. Although some research on hydrogen storage/generation methods for PEMFC-based UAVs has been conducted [16,17], a further enhanced energy density and highly efficient chemical hydride composite to reduce the weights of UAVs and effectively optimize PEMFC-based power pack systems are still required, especially for longer duration flights with UAVs [18][19][20].
AB contains 19.6 wt% of high hydrogen content per unit of mass. However, the thermal dehydrogenation temperature should be increased up to hundreds of degrees Celsius to obtain the theoretical maximum H 2 yield. Finding an appropriate catalyst to decrease the working temperature of the AB decomposition reaction with increased H 2 yield is of considerable importance. In this work, we present a characterization of a specifically designed chemical hydride composite, composed of AB and SiO 2 nanopowders (NP). Other researchers have reported on the availability of silicon-based catalysts, including quartz wool and mesoporous silica, in the thermal decomposition of AB. However, the actual working mechanism is not yet clearly understood [21][22][23][24][25]. Furthermore, the characteristics of the AB-SiO 2 NP composite are compared with those of other AB composites, containing various catalysts such as zeolites, metal, and metal oxide NP and boric acid (BA) as an additive. The optimization of the weight ratio between the AB and the catalysts is described, as well as the screening of the composites for fast H 2 evolution rate and low dehydrogenation temperature, to meet the demands of PEMFC-driven UAVs. Fourier transform infrared (FTIR) spectroscopy (ALPHA series, Bruker Corp.) is utilized to verify the hydrogen evolution of AB by thermolysis.

Catalyst Preparation and Reactor Design.
Commercial zeolite ammonium ZSM-5 powder (CBV2314) is available from Zeolyst International (PA, USA). Laboratory preparation of the ZSM-5 in an attempt to remove tetrapropylammonium hydroxide (TPAOH) is as follows: Sodium aluminate dispersed in a basic aqueous medium is mixed in a flask with a basic Ludox® AS-40 colloidal silica suspension (40 wt% in H 2 O) and stirred at room temperature for 3 h. Then, the suspended solution (the weight percentage of Si to Al is 67) is autoclaved at 190°C and 1.013 MPa for 12 h, and finally, the hydrothermally synthesized white powder-dispersed solution is filtrated and dried at 110°C overnight to produce the sodium-substituted TPAOHfree ZSM-5 catalyst (Na-ZSM-5). Na-ZSM-5 can be further stirred in a flask with ammonium nitrate at 80°C for 4 h. After filtration and drying at 110°C, the white product is calcined in an electric furnace at 600°C for 4 h to obtain hydrogen-substituted ZSM-5 (H-ZSM-5). The other materials and reagents are of analytical grade and purchased from Sigma-Aldrich (Milwaukee, USA), unless stated otherwise.
The laboratory-made hydrogen evolution reactor is composed of three thermocouples and a pressure gauge (Figures 1(a) and 1(b)). Thermocouple 1 directly measures the inside temperature of the vial that actually contains a fuel composite of AB and various catalysts and is feedbackcontrolled for precise control of the working temperature of the dehydrogenation reaction. Thermocouples 2 and 3 determine the temperatures of the inside and outside of the reactor, respectively, and the temperature difference between them gives the approximate energy loss of the reactor. The pressure gauge determines the amount of hydrogen gas generated by the heat-liberating AB dehydrogenation reaction.

Preparation of Fuel
Composites. The weight of a fuel composite is 0.15 g. An AB : SiO 2 NP = 6 : 4 composite contains nine portions of AB (0.09 g) and six portions of SiO 2 NP (0.06 g). If the ratio were 8 : 2, then the composite would contain 0.12 g AB and 0.03 g SiO 2 NP. All reagent components are stored in a desiccator before use, to remove traces of water. After mixing both components gently in a 2 mL volume vial, the fuel composite is carefully placed in the reactor, followed by argon gas being blown into the reactor for 1 min. Then, the heating block increases the reactor temperature at a rate of 2°C min -1 .

Thermolysis of Ammonia
Borane. The theoretical maximum H 2 yield from AB is 19.6 wt%: one-third of the hydrogen is generated at around 120°C, another third is dehydrogenated at around 160°C, and the final third requires over 500°C [26][27][28] (see Supplementary Material Fig. S1). Thermal dehydrogenation of AB can avoid the ammonia poisoning commonly observed in catalytic hydrolysis of AB, which generates one unit of ammonia for every three units of generated hydrogen. Even a level of ammonia of a few ppm can poison a PEMFC, depending on the exposure time to ammonia [29]. Although sodium borohydride (d ≈ 1 074 g cm -3 ) does not contain any amine or ammonia substituents and may provide a good alterative hydrogen fuel source without ammonia poisoning, the lighter AB (d ≈ 0 78 g cm -3 ) is a better energy source for aerial vehicles.

Results and Discussion
Various catalysts with an additive such as SiO 2 NP, aluminosilicate zeolites (CBV2314, H-ZSM-5, and Na-ZSM-5), Ni NP, metal oxides (ZrO 2 , TiO 2 , and Al 2 O 3 ), and BA are compared in terms of the H 2 yield, reaction time, and onset reaction temperature for the thermal dehydrogenation of AB. The highly exothermic dehydrogenation reaction of AB Journal of Nanomaterials yields a sharp increase in temperature with the release of H 2 gas ( Figure 2). As an additive, BA, which has achieved the best known thermal dehydrogenation performance for AB, produces a H 2 yield of 9.56 wt%. This means that a 2.15 mole of H 2 gas is generated for one mole of AB, based on the ideal gas equation of state, considering that the optimized weight ratio between AB and BA for the hydrogen evolution reactor employed in this work, shown in Figure 1, is 7 : 3. The weight ratio optimization of the AB-BA composite is described in the Supplementary Material, Fig. S2. The mole-equivalent H 2 yield increases to 2.20 with the van der Waals equation, i.e., P + aN 2 /V 2 V − bN = NRT, where van der Waal constants a and b are 0.2453 bar·L 2 ·mol -2 and 0.02651 L·mol -1 , respectively. The higher the content ratio of the additives or catalysts in the hydrogen fuel composites, the faster the hydrogen evolution rate, and usually the lower the onset temperature of the dehydrogenation reactions. However, the H 2 yield as a weight percentage, defined by the weight ratio of hydrogen to the total weight of a composite, decreases as the amount of additive or catalyst in the composite increases. The onset temperature of the AB-BA composite is between 94 and 100°C, and the reaction time duration is less than 3 min. The AB-SiO 2 NP composite exhibits a H 2 yield of 9.26 wt% and a reaction time duration of 2 min, which are comparable with those of the AB-BA composite. While the temperature curve of the AB-BA composite exhibits a small shoulder peak directly before the abrupt temperature increase, the AB-SiO 2 NP composite    Figure 2: Thermolytic dehydrogenation of AB in the presence of various additives or catalysts. The set temperature is 100°C, and the heating rate is 2°C·min -1 . Dehydrogenation of AB is a strong exothermic reaction, and an increase in temperature is observed with H 2 gas evolution.
3 Journal of Nanomaterials exhibits a single sharp peak. The other catalysts exhibit relatively large and wide shoulder peaks. Table 1 summarizes thermolytic parameters of the catalytic dehydrogenation of AB with various additives and catalysts. Only the thermolytic parameters of SiO 2 NP are comparable with those of BA, and the CBV2314 catalyst actually acts as an anticatalyst in AB thermolysis. The H 2 evolution curves of the other metal or metal oxide NP, including Ni, ZrO 2 , anatase TiO 2 , and Al 2 O 3 are also provided (see Supplementary Material Fig. S3), but these exhibit poorer H 2 yields than that of SiO 2 NP, because the metal and metal oxides are significantly heavier and probably less porous than SiO 2 NP, as shown in Figure 3 and Supplementary Material Fig. S4.
In general, heat conduction mechanisms in powders and flake media differ from each other [30,31]. Heat transfer in a heat conductor is similar to electron transfer in an electric semiconductor. Electron movement in randomly oriented bulk NP becomes faster through directional electron transfer [32,33]. Similarly, a mechanical arrangement of AB-based fuel composite can allow a faster heat transfer rate in the in-plane direction of the composite. Effective heat transfer during the exothermic dehydrogenation of AB thermolysis enables enhanced hydrogen evolution compared with a simple powder-based AB composite. Hence, it can be assumed that a unidirectional compression force provides the AB-SiO 2 NP composite with a favorable orientation for better heat conduction in the fuel composite. Supplementary Material Fig. S5 illustrates how to prepare directionally favorable AB-SiO 2 NP composite. After gently mixing both components in a vial, the composite is carefully pressed from the top until only half of the original volume is left.
Even though the compositional optimization of the AB-SiO 2 NP composite shows that the 8 : 2 composite appears to provide the best H 2 yield with a minimized time requirement for a given amount of AB composite (Table 1), a greater H 2 yield can be obtained at a different optimized condition with the unidirectionally pressed AB-SiO 2 NP composite ( Figure 4). Note that hydrogen gas is liberated even before the sharp temperature peaks are observed. A physically favorable arrangement of the composite increases the H 2 yield by over 20% and decreases the reaction time by approximately 50%. Table 2 summarizes the thermolytic parameters of various AB-SiO 2 NP composites.
The AB thermolysis reaction requires heat to initiate the dehydrogenation reaction, and the set temperature of the reactor strongly affects the H 2 yield, as shown in Figure 5. No reaction occurs at a set temperature of 50°C, meaning that the AB-SiO 2 NP composite is stable under normal atmospheric conditions. As the set temperature increases from 80 to 110°C, the reaction time and H 2 yield increase from 77 to 140 s and 10.37 to 11.67 wt%, respectively. Because a typical PEMFC system functions correctly at a working temperature of around 85°C, it would be reasonable to set the reactor temperature to approximately the same as the operation temperature of the PEMFC. Otherwise, additional equipment for insulation is required to thermally separate the PEMFC system from the hydrogen fuel supplier. The activation energy (E a ) of the SiO 2 NPcatalyzed dehydrogenation reaction of AB is 63.9 kJ·mol -1 (Supplementary Material Fig. S6). This is just one-third of the typical activation barrier of 180 kJ·mol -1 to the AB decomposition reaction [34]. Figure 6 shows the FTIR spectra of the AB-SiO 2 NP composite. Boron, nitrogen, and hydrogen atoms in a tetrahedral backbone with covalent single bonds lead to three characteristic FTIR bands for AB (Supplementary Material Fig. S7): the peaks observed in a range of 648-1822 cm -1 are assigned to B-N stretching vibrations, peaks between 2260 and 2820 cm -1 are assigned to B-H stretching vibrations, and N-H stretching modes are observed in a range of 3167-3712 cm -1 [21,35,36]. The evolution of hydrogen gas by heating the composite increases the bond order along the principal axis of AB. This means that the π-bond character becomes involved, in addition to the σ-bond character, and stretching frequencies are upshifted upon the bond order increase. More importantly, the B-H and N-H stretching intensities diminish when a large quantity of hydrogen gas is generated. As shown in Figure 6, both the B-H and N-H stretching intensities are considerably reduced after the thermal decomposition of AB, and the same result is observed with additive BA (Supplementary Material Fig. S7).

Conclusions
In this work, AB, a stable chemical hydride at room temperature and pressure, was thermally decomposed to produce hydrogen gas at a temperature of approximately 90°C, in the presence of SiO 2 NP. We designed our own hydrogen evolution reactor and hydrogen fuel composite to directly compare the results for the AB-SiO 2 NP composite with that of AB-BA, which has achieved the best thermolytic performance in other studies with a composite ratio of AB : BA of 8 : 2. The H 2 yield of the AB-SiO 2 NP composite (AB : SiO 2 NP = 7 : 3 in wt%) was 9.26 wt%, and the yield of the mechanically pressed one (8.5 : 1.5 composite in wt%) was 12 wt%, with a reaction time of less than a minute, which is comparable to that of the AB-BA composite. In general, chemical hydride-based hydrogen fuel composites for H 2 -PEMFC for aerial vehicles should satisfy the following criteria: It should be sufficiently stable not to be dehydrogenated at a conventional atmospheric condition (<50°C), have a high capability for hydrogen storage to save  Journal of Nanomaterials weight, achieve a fast hydrogen evolution rate on demand at a temperature of less than 90°C, and have an in-built hydrogen fuel filtration system for the sake of ammoniasusceptible PEMFC (we did not seriously consider the byproduct issue in the present work) [1]. The hydrogen evolution curves of the AB-SiO 2 NP composites showed that none of the compositional ratios investigated in this work thermally decomposed at a temperature below 50°C. Once the thermal dehydrogenation of AB occurred, the reaction ended within a couple of minutes. More precise optimization of the weight ratio between the two components, con-cerning not only compositional optimization but also the physical shape, density, and molecular orientation, could additionally enhance the hydrogen storage capability and       Journal of Nanomaterials eventually reduce the total weight of a power generation package for UAVs. Given a thermally shielded fuel cartridge, the AB-SiO 2 NP composite could lead to the commercialization of disposable hydrogen fuel for UAVs.

Data Availability
The data used to support the findings of this study are included within the article. Additional information could be obtained from the supplementary materials.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper.

Authors' Contributions
The first and second authors contributed equally to this work.

Acknowledgments
This study was supported by the Agency for Defense Development (ADD) of Korea (Grant number 17-114-407-038).

Supplementary Materials
Fig. S1: TGA of pure AB. Thermal behavior of silica gel and SiO2 NP is also shown for comparison. Notable two peaks shown at 120 OC and 160 OC stems from formation of polyamidoboranes and/or polyimidoboranes by the AB decomposition. Note that a slight mass loss of silica gel and SiO2 NP is due to evaporation of the surface-adsorbed water vapor and unspecified polymeric contaminants. Fig. S2: the weight ratio optimization of the AB-BA composites. The onset temperature and the H2 yield of each composite are as follows, respectively: 94 OC and 8.21 wt% for AB BA = 6 4 composites; 98 OC and 9.56 wt% for AB BA = 7 3 composites; 100 OC and 5.55 wt% for AB BA = 8 2 composites. Fig. S3: hydrogen evolution curves of AB composites containing various metal or metal oxide NP such as Ni (black), ZrO2 (red), anatase TiO2 (blue), and Al2O3 (green). H2 yield of each composite is 4.27, 7.44, 3.74, and 3.97 wt% in order. Note that AB content in each composite is the same as 60 wt%. Fig. S4: photo images (×1200) of various metal or metal oxide NP catalysts including Ni (d ≈ 8 91 g cm -3 ), ZrO2 (d ≈ 5 68 g cm -3 ), anatase TiO2 (d ≈ 3 8 g cm -3 ), and Al2O3 (d ≈ 3 99 g cm -3 ). Fig. S5: unidirectional compression of AB-SiO2 NP composite to reduce the volume by half. Fig. S6: Arrhenius plot based on the hydrogen evolution curves at 4 different temperatures shown in Figure 5. Note that the obtained E a from the slope measures 63.9 kJ·mol -1 . Fig. S7: (a) FTIR spectra of AB flake and AB powder. Both spectra commonly show 3 characteristic bands, i.e., BN stretching, BH stretching, and NH stretching bands. (b) FTIR spectra of AB and BA composites before and after the thermal dehydrogenation reaction. Note that the BH and NH stretching peaks are simultaneously reduced after the dehydrogenation reaction. (Supplementary Materials)