*2.2. Hydrogen Storage Applications*

Ammonia borane (AB) is a white crystalline solid that was first prepared by Shore and Parry in 1955 [23]. Over the past decade, this compound has attracted considerable attention as portable hydrogen storage materials, according to its high gravimetric hydrogen contents (*ca.* 20% by weight) [24–29]. A very pertinent review dedicated to this compound and related derivatives as dihydrogen sources was proposed by Staubitz *et al.* in 2010 [29].

In the pristine state, AB is almost stable under inert conditions up to about 100 °C and decomposes within the range 100–200 °C through a two-step exothermic process where two equivalent H2, as well as undesired by-products, such as borazine B3N3H6 and NH3, are evolved [24,25]. This decomposition suffers from three important problems: (1) the process is exothermic, which means that the storage reversibility is thermodynamically impossible in acceptable operating conditions; (2) the dehydrogenation temperature is too high for the portable/mobile application prospects; (3) the emission of undesired by-products is detrimental, as they are incompatible with the use of proton exchange membrane fuel cell (PEMFC) [27].

A promising solution seems to be the decrease of the particle size at the nanoscale (<10 nm) via confinement of the borane in a porous compound (*i.e.*, scaffold) [30]. As an illustration, Gutowska *et al*. showed that AB confined in the mesoporosity of silica SBA (Santa Barbara Amorphous)-15 has improved dehydrogenation behavior in comparison to the pristine hydride, with an onset at 70 °C and the liberation of borazine-free H2 [31]. The destabilization of AB is generally explained by two phenomena. The first one is the nanosizing of the hydride particle. At the nanoscale, both kinetics and thermodynamics might be altered by both size and interface effects. In fact, the surface energy value could be different as a result of the interactions between the active confined material and the scaffold. The second phenomenon is associated with H<sup>į</sup><sup>+</sup> ···Hįí surface interactions, with Hįí of the BH3 moiety of AB and H<sup>į</sup><sup>+</sup> belonging to surface/terminal hydroxyl groups (OH) generally found on carbonaceous or oxide nano-scaffolds. Such acid-base interactions enhance H2 release, but usually lead to an unstable material at room conditions [25].

An improved strategy we recently demonstrated is to use nano-scaffolds free of reactive surface groups [22]. For that purpose, we used the B1.0N0.9-NPs annealed at 2000 °C, which we labeled B1.0N0.9-NP2000. As measured by energy dispersive X-ray spectrometry (EDX), boron, nitrogen and oxygen contents are 43.55, 55.7 and 0.75 wt%, respectively. Unfortunately, they exhibit a Brunauer-Emmett-Teller (BET)-specific surface area of 21.8 m2 ·g<sup>í</sup><sup>1</sup> , which is low to achieve the nanoconfinement of AB. Therefore, we applied a ball-milling process of this sample to tentatively increase the specific surface area, leading to the sample labeled B1.0N0.9-NP2000BM. In comparison to B1.0N0.9-NP2000, the sample B1.0N0.9-NP2000BM shows a considerably increased BET-specific surface area with 200.5 m2 ·g<sup>í</sup><sup>1</sup> and a total pore volume of 0.424 cm3 ·g<sup>í</sup><sup>1</sup> as measured by the Barrett-Joyner-Halenda (BJH) analysis. As a result of the ball-milling, the HRTEM images (Figure 6a,b) of the sample showed that cleavage of the walls occurred through the basal planes. In addition, Figure 6c show that the stacking sequence can in some cases be disordered similarly to those of *t*-BN after ball-milling.

As a result of the BET and TEM investigations, we successfully demonstrated that the walls of the hollow-cored BN-NPs could be opened to provide porosity after ball-milling.

**Figure 6.** HRTEM images of the sample B1.0N0.9-NP2000BM evidencing in (**a**) and (**b**), a cleavage of the walls in the area delimited by the white arrows, and in (**c**), a disordering of the stacking sequence.

Hydrogen storage materials can be confined within porous scaffolds by melt infiltration (if the active hydrogen storage material melts and do not decompose) or solution infiltration. In our procedure, a solution of AB in tetrahydrofuran (THF) was infiltrated into the framework of the sample B1.0N0.9-NP2000BM according to an optimized procedure described elsewhere [25]. A nanocomposite labeled AB@B1.0N0.9-NP2000BM was formed. It was stored at 3–4 °C. The successful impregnation of AB in B1.0N0.9-NP2000BM was followed by N2 adsorption/desorption analysis of the nanocomposite. A BET-specific surface area of 6.7 m2 ·g<sup>í</sup><sup>1</sup> and a total pore volume of 0.023 cm3 ·g<sup>í</sup><sup>1</sup> are measured, which demonstrates that AB was inserted into the hollow core and blocked the pores of the nano-scaffolds. More interesting is that the decomposition of AB is down to 81 °C (compared to 110 °C for the pristine AB in our conditions) and that a major evolution of H2 is identified by MS. In our experimental conditions, the only by-product was identified to be NH3 above 80 °C. At 80 °C, a weight loss of 1.7 wt% was measured, which means an effective gravimetric hydrogen storage capacity of 3.4 wt% by considering a weight ratio equal to 1:1 in AB@B1.0N0.9-NP2000BM.

Our results confirmed the remarkable benefit of hollow-cored BN-NPs on the dehydrogenation behavior of AB. The performance is comparable to the dehydrogenation results of AB confined into a magnesium metal organic framework (MOF) [32,33] or nickel MOF [34], whereas only nanoconfinement

is considered here. Most interesting, the MS results suggest that there is no detectable trace of borazine as a gaseous by-product. Another important observation standing from the thermogravimetric analysis coupled mass spectrometry (TGA-MS) result is that AB@B1.0N0.9-NP2000BM is stable at room conditions. Accordingly, the stability of AB@B1.0N0.9-NP2000BM at <40 °C is clearly attributed to the absence of surface H<sup>į</sup><sup>+</sup> , and the improvement of the dehydrogenation properties of AB in AB@B1.0N0.9-NP2000BM can be exclusively ascribed to the effect of nanoconfinement.

#### **3. Experimental Section**

The synthesis of borazine was carried out in an argon atmosphere, using argon/vacuum lines and Schlenk-type flasks. Argon (>99.995%) was purified by passing through successive columns of phosphorus pentoxide (Sigma-Aldrich, Saint Quentin, France), sicapent (Millipore S.A.S, Molsheim, France) and copper oxide-based catalysts (Sigma-Aldrich, Saint Quentin, France). The Schlenk flasks were dried at 120 °C overnight before pumping under vacuum and before filling with argon for the synthesis. Sodium borohydride (NaBH4, 98.5%, powder from Sigma-Aldrich, Saint Quentin, France), ammonium sulfate ((NH4)2SO4, 99.0% from Sigma-Aldrich (Saint Quentin, France) and tetraethylene glycol dimethyl ether(CH3O(CH2CH2O)4CH3, 99.0%, from Sigma-Aldrich (Saint Quentin, France) were used as-received. It should be mentioned that ammonium sulfate was dried at 120 °C inside an oven for three days, then put under vacuum during cooling for 1 h. Manipulation of the chemical products was made inside an argon-filled glove box (Jacomex BS521; Dagneux, France) dried with phosphorus pentoxide.

Borazine Synthesis: The operating procedure, adapted from the literature [18], was previously reported by our group [19]. FTIR (Caesium Iodide (CsI) windows/cm<sup>í</sup><sup>1</sup> ): (N–H) = 3451 medium; (B–H) = 2509 medium; (B–N) = 1454 small; (B–N–B) = 897 medium. <sup>1</sup> H NMR (300 MHz/CDCl3/ppm): = 3.30–5.35 (quadruplet, 3H, B*H*), 5.35–6.05 (triplet, 3H, N*H*). 11B NMR (96.29 MHz/C6D6/ppm): = 30.1 (br).

Nanoparticle Preparation: The experimental set-up is composed of a nebulized spray generator (RBI, Meylan, France), in which the spray is generated by a piezoelectric device (barium titanate). Frequency (800 kHz) and power (100 W) alimentations are adjusted to obtain the aerosol. The aerosol temperature is first held at 15 °C by a regulated water circulation to avoid borazine evaporation and/or condensation. The piezoelectric device generates an ultrasound beam, which is directed to the liquidgas interface; a fountain formed at the surface followed by the generation of the spray, resulting from vibrations at the liquid surface and cavitations at the gas-liquid interface.

The borazine was directly introduced in the aerosol generating chamber under nitrogen, then aerosolized and carried to the pyrolysis furnace with a 0.5 mL·min<sup>í</sup><sup>1</sup> nitrogen flow rate. The thermal decomposition of borazine was performed in a hot alumina tube containing an isothermal zone of 0.1 m in length. The fast heating rate implies gaseous species generation leading to powder formation by a chemical vapor condensation route. The particles were finally trapped into two collectors placed before the vacuum pump and containing filter-barriers made of microporous alumina (pore size of 1 m). Yield was estimated to be 0.22 g·min<sup>í</sup><sup>1</sup> . After synthesis, the particles are stored inside an argon-filled glove-box. In a typical experiment, 27 mL (21.9 g) of borazine is used to produce 6.5 g of B1.0N0.9-NPs. However, the exact yield is difficult to estimate, because of the design of the spray-pyrolysis system. A non-negligible/considerable quantity of powders, deposited in the furnace tube, cannot be recovered. To study the evolution of their crystallization degree, 2 g of the B1.0N0.9-NPs are placed into boron nitride boats and then introduced in a graphite furnace (Gero 5 Model HTK 8). The furnace chamber is subsequently suctioned with a pump charged with nitrogen before heating. A cycle of ramping at 10 °C·min<sup>í</sup><sup>1</sup> is used to heat the sample to the desired temperature (in the range 1400–2000 °C) with a holding time of 1 h, before cooling down to RT at 10 °C·min<sup>í</sup><sup>1</sup> . Chemical analysis found (wt%): B, 50.0; N, 49.4; O, 0.6. The milling process of B1.0N0.9-NP2000 is performed under inert condition (argon) with a planetary ball-miller (Retsch PM100; Haan, Germany). The described process has been optimized (in terms of mass, ratio balls/BN, time, rotation) to our conditions. Typically, degassed B1.0N0.9-NP2000 (at 150 °C under dynamic vacuum for 24 h) is introduced into a stainless steel reactor (25 mL). Balls in stainless steel are added (weight ratio balls: B1.0N0.9-NP2000 of 20). The milling process is performed at 600 rpm for 1 h. The as-obtained B1.0N0.9-NP2000BM is finally sieved.

The infiltration of ammonia borane is performed as follows: the host material B1.0N0.9-NP2000BM (100 mg) is degassed at 150 °C under dynamic vacuum for 24 h in a Schlenk tube and then cooled to 0 °C. In an argon-filled glove box, a concentrated solution of ammonia borane (100 mg, 97%; Sigma Aldrich, Saint Quentin, France) is prepared using 0.5 mL of anhydrous THF (Sigma Aldrich, Saint Quentin, France). The ammonia borane solution is injected into the Schlenk tube containing B1.0N0.9-NP2000BM kept under static vacuum and at 0 °C. By capillary action, the ammonia borane solution fills the channels of the host rapidly, which is evidenced by vigorous effervescence. When the effervescence stops, the sample is put under ultrasonic treatment for 20 min at 0 °C. Finally, the as-obtained sample AB@B1.0N0.9-NP2000BM (weight ratio B1.0N0.9-NP2000BM:AB of 1) is dried under dynamic vacuum for 48 h at 0 °C. The composite samples obtained are denoted AB@B1.0N0.9-NP2000BM. Samples are transferred in an argon-filled vial and then stored in a fridge at 3–4 °C.

Characterizations: The B1.0N0.9-NPs and annealed samples are first mounted on carbon film-covered stainless pads for scanning electron microscopy (SEM, Hitachi S4800, Tokyo, Japan) including Energy Dispersive X-ray Spectroscopy (EDX, EDAX/TSL Genesis 4000, Tokyo, Japan). Due to the insulating properties of BN, the samples are sputtered with 10 nm of a Pd/Au mixture to prevent charging during SEM observations. In parallel, the same samples are ultrasonicated in ethanol, and the resulting solution is afterwards deposited on a collection of hollow carbon-film-covered copper grids for transmission electron microscopy (TEM, TOPCON 002B working at 200 kV, Tokyo, Japan) observation. Samples were characterized using a Philips PW 3040/60 X'Pert PRO X-ray diffraction system (Eindhoven, The Netherland). Powder samples are prepared by placing ~100 mg on the XRD sample holder (PVC), and the sintered pieces were put down on the XRD sample holder for data collection. Cu KĮ (Ȝ = 1.54 Å) radiation with a Ni filter was used with a working voltage and a current of 40 kV and 30 mA, respectively. Scans were continuous from 2ș = 10°–90° with a time per step of 0.85 s in increments of 0.017°. Peak positions and relative intensities were characterized by comparison with JCPDS (Joint Committee on Powder Diffraction Standards) files of the standard material (JCPDS card No 34-0421). Debye-Scherrer line broadening was used to calculate the average crystallite sizes from each XRD pattern. The transmission electron microscopy (TEM) studies of B1.0N0.9-NP2000BM samples were carried out with a JEOL (Tokyo, Japan) GmbH 2010F transmission electron microscope (Cs = 1 mm) operating at 200 kV. The characterization of the samples was performed by N2 adsorption/desorption (Sorptomatic 1990 Series, Thermo Fisher Scientific Inc, Waltham, MA, USA). Thermogravimetric analysis (TGA) measurements (repeated at least three times to ensure the reproducibility of the results) were performed with a Mettler Toledo TGA/SDTA 851e (Schwerzenbach, Switzerland) under the following conditions: sample mass 9–10 mg, aluminum crucible of 100 L with a pinhole, heating rate of 5 °C·min<sup>1</sup> , temperature range 25–200 °C and atmosphere of N2 (60 mL·min<sup>1</sup> ). The purity of H2 was analyzed by mass spectrometry (Canon Anelva Corporation MQA200TS, Tokyo, Japan) coupled to the TGA apparatus.

#### **4. Conclusions**

This article reviews our recent advancements in the synthesis and energy application of nanostructured boron nitride. Such materials exhibit chemical and physical properties that are significantly different from those of bulk and microsized materials. In our approach, we discussed our recent strategy based on borazine, which has been used as a vapor phase pyrolysis precursor for the synthesis and fabrication of BN nanoparticles. In particular, we demonstrated the possibility of tailoring the nanostructure of these nanoparticles by a further annealing process at high temperature (2000 °C), leading to hollow-cored BN-NPs that are strongly facetted, forming polygonal particles with an interlayer spacing of 3.34 Å. The ball-milling of these nanostructures allowed developing the specific surface area of the material while hollow-cored BN-NPs porous shell structures were obtained. They show a BET-specific surface area of 200.5 m2 ·g<sup>í</sup><sup>1</sup> , a total pore volume of 0.287 cm3 ·g<sup>í</sup><sup>1</sup> and a narrow pore size distribution centered at 3.5 nm. They were used as nano-scaffolds of ammonia borane in order to improve its dehydrogenation properties to form a nanocomposite able to liberate pure H2 in the temperature range 40–80 °C in our conditions. The only trace of by-product being detected at >80 °C is ammonia. Considering the regenerability of ammonia borane [35], our results suggest that our composite material is a safe and practical hydrogen storage material. This improvement is exclusively ascribed to the nanoconfinement effect.
