Ammonia Borane: An Extensively Studied, Though Not Yet Implemented, Hydrogen Carrier

Ammonia borane H3N−BH3 (AB) was re-discovered, in the 2000s, to play an important role in the developing hydrogen economy, but it has seemingly failed; at best it has lagged behind. The present review aims at analyzing, in the context of more than 300 articles, the reasons why AB gives a sense that it has failed as an anodic fuel, a liquid-state hydrogen carrier and a solid hydrogen carrier. The key issues AB faces and the key challenges ahead it has to address (i.e., those hindering its technological deployment) have been identified and itemized. The reality is that preventable errors have been made. First, some critical issues have been underestimated and thereby understudied, whereas others have been disproportionally considered. Second, the potential of AB has been overestimated, and there has been an undoubted lack of realistic and practical vision of it. Third, the competition in the field is severe, with more promising and cheaper hydrides in front of AB. Fourth, AB has been confined to lab benches, and consequently its technological readiness level has remained low. This is discussed in detail herein.


Introduction
Recent years have witnessed a renewed interest in hydrogen, which is considered to be a high-potential energy carrier of the near-future energy mix. New momentum has been created at the international level [1] as a result of two decades of intense research and development at both academic and industrial scales. Further efforts are, however, needed to knock down the remaining technological and scientific barriers related to the whole chain of the hydrogen economy, viz. production [2], distribution [3], storage [4,5] and end-use [6,7].
The storage issue is particularly problematic. Also the conventional physical methods (i.e., compression up to 700-900 bars and liquefaction at −253 • C), new solutions involving materials have been developed [4,5]. With porous materials, cryo-adsorption (physisorption) of molecular hydrogen has shown to be effective in allowing high gravimetric hydrogen storage capacities; for instance, higher than 7 wt% H 2 at −196 • C [8]. As with all of the other materials, hydrogen storage is chemical and thus involves elemental hydrogen in chemical binding (chemisorption) [9]. There are alkali/alkaline-earth hydrides like magnesium hydride MgH 2 , intermetallic compounds like lanthanum-nickel alloy LaNi 5 , complex hydrides like sodium alanate NaAlH 4 and chemical hydrides like lithium borohydride LiBH 4 [10][11][12][13][14]. There are also BH and BNH materials [15,16], represented by sodium borohydride NaBH 4 [17], ammonia borane H 3 N−BH 3 [18] and hydrazine borane H 4 N 2 −BH 3 [19], for example. Sodium borohydride has been much investigated from the late 1990s, primarily as liquid-state hydrogen carrier, for which the main challenge has been dehydrogenation by catalyzed hydrolysis [20]. After more than 20 years of research, now is the time to scale it up [21]. Ammonia borane emerged in the mid-2000s, firstly as an alternative to sodium borohydride as a liquid-state hydrogen carrier [22], and secondly as Ammonia borane is an old compound. Its history commences before the 2000s, in the first half of the twentieth century [27]; this is addressed in Section 2. Ammonia borane has been the object of a new interest dating back to the mid-2000s. The interest stretched beyond the hydrogen storage application. Ammonia borane has been also considered, though briefly, as a potential liquid fuel for direct liquid-fed fuel cells. This aspect is covered in Section 3. Ammonia borane is a potential liquidstate hydrogen carrier as well as a solid-state hydrogen storage material. Research dedicated to the former use is discussed in Section 4, and that of the latter application is addressed in Section 5. These four sections are put into perspective in Section 6 so as to stress on the main challenges that remain, assess the technological readiness level of the aforementioned applications and give an objective analysis of opportunities for the industrial development of ammonia borane as a hydrogen carrier.

Ammonia Borane, an Old Compound
Ammonia borane is the result of works carried out by two American research groups between the 1930s and the 1950s. The first research group, led by Schlesinger, long sought to synthesize it [28][29][30][31][32]. All of their attempts were thwarted because of unsuitable experimental conditions in terms of reactants, stoichiometry, temperature and/or solvent [27]. The successful synthesis of ammonia borane, as a crystalline compound, was reported by the second research group, represented by Shore and Parry [33][34][35]. Ammonia borane was prepared by a metathesis reaction of lithium borohydride with an ammonium salt in diethyl ether, followed by hydrogen evolution by proton-hydride combination, as shown below: Ammonia borane is an old compound. Its history commences before the 2000s, in the first half of the twentieth century [27]; this is addressed in Section 2. Ammonia borane has been the object of a new interest dating back to the mid-2000s. The interest stretched beyond the hydrogen storage application. Ammonia borane has been also considered, though briefly, as a potential liquid fuel for direct liquid-fed fuel cells. This aspect is covered in Section 3. Ammonia borane is a potential liquid-state hydrogen carrier as well as a solid-state hydrogen storage material. Research dedicated to the former use is discussed in Section 4, and that of the latter application is addressed in Section 5. These four sections are put into perspective in Section 6 so as to stress on the main challenges that remain, assess the technological readiness level of the aforementioned applications and give an objective analysis of opportunities for the industrial development of ammonia borane as a hydrogen carrier.

Ammonia Borane, an Old Compound
Ammonia borane is the result of works carried out by two American research groups between the 1930s and the 1950s. The first research group, led by Schlesinger, long sought to synthesize it [28][29][30][31][32]. All of their attempts were thwarted because of unsuitable experimental conditions in terms of reactants, stoichiometry, temperature and/or solvent [27]. The successful synthesis of ammonia borane, as a crystalline compound, was reported by the second research group, represented by Shore and Parry [33][34][35]. Ammonia borane was prepared by a metathesis reaction of lithium borohydride with an ammonium salt in diethyl ether, followed by hydrogen evolution by proton-hydride combination, as shown below: LiBH 4  This is one of the safest paths allowing the synthesis of pure ammonia borane, which is thus being currently used (while consuming different alkali borohydrides like NaBH 4 and ammonium salts like (NH 4 ) 2 SO 4 ) [36][37][38][39][40][41][42].
As a new compound, ammonia borane was then much studied. For example, the distance of the N−B bond of the gas-state molecule was calculated to be 1.66 Å [43]. The formation enthalpy and formation entropy of gas-state ammonia borane were calculated as −83.3 kJ mol −1 and 243.9 J mol −1 K −1 , comparing favorably to the values of the isoelectronic ethane (−83.8 kJ mol −1 and 229.6 J mol −1 K −1 ) [44]. The Mulliken charge of H δ+ and that of H δ− of ammonia borane were calculated as +0.21 and −0.195, respectively [45]. An electron transfer from the H 3 N group to the BH 3 one was evidenced [46], in good agreement with the formation of a dative bond between the two groups. The ammonia borane molecule is polarized, and a dipole moment of about 5.6 D was calculated, for instance [47]. As discussed at length in references [18,48,49], ammonia borane as a molecule was studied extensively between the 1960s and the 1980s, and ammonia borane as a material has been well studied within the past there decades. For example, the crystal structure of the material, though first determined in the 1950s [33,50,51], was scrutinized more in detail (at different temperatures and under pressure) in recent years [52][53][54].
It is now known that, at atmospheric pressure, ammonia borane has a low-temperature phase (orthorhombic, space group Pmn2 1 ) and a high-temperature phase (tetragonal I4mm), the transition taking place at −48 • C. The recent interest on ammonia borane as a material has been explained by its promises as a solid-state hydrogen carrier [55]. This is discussed in the next sections.
However, the capacity of ammonia borane to generate hydrogen (i.e., its role as a hydrogen carrier) has long been recognized. Between the 1960s and the early 1980s, it was patented as a solid propellant hydrogen generator [56][57][58][59]. Recent works confirmed the properties of ammonia borane as a propellant and an energetic material [60][61][62][63][64][65], but efforts in this direction have been limited. Today, ammonia borane is first and foremost considered to be a hydrogen carrier, with the released hydrogen being used for electricity generation by conversion in a fuel cell.

Aqueous Ammonia Borane as Anodic Fuel
Ammonia borane is a water-soluble reductant that can be electro-oxidized in alkaline conditions at metal (e.g., gold) electrodes [66], the reaction being often reported as However, in the electro-oxidation conditions (<30 • C and <1 mol L −1 ), the anhydrous borate anion BO 2 − cannot form, since the thermodynamically stable form is the dihydrated one, namely the tetrahydroxyborate anion B(OH) 4 − [67]. The following reaction is thus more accurate: Each of the hydridic hydrogens H δ− of the BH 3 group of ammonia borane are electro-oxidized into a proton while two electrons per hydrogen are released. These electrons can then be involved in an electro-reduction reaction; that of oxygen O 2 , for example: These reactions form the basis of the direct ammonia borane fuel cell (DABFC; Figure 2), for which the overall reaction is: Energies 2020, 13, 3071

of 45
Energies 2020, 13, x FOR PEER REVIEW 4 of 45 standard hydrogen electrode, specific energy of 9300 Wh kg −1 and pure compound capacity of 5700 Ah kg −1 [70]. In operation, the borohydride cell should be slightly better performing.

Figure 2.
Schematic representation of (a) the direct ammonia borane fuel cell and (b) the direct borohydride fuel. At the right is given the standard potential versus the standard hydrogen electrode, the specific energy and the pure compound capacity. The anode (in yellow) and the cathode (in blue) are separated by a mmebrane (in grey).
The first DABFC was reported by Yao et al. [71]. They isolated an efficient silver anode electrocatalyst (among other metals like platinum and gold), and manganese oxide MnO2 as a cathode electrocatalyst showing tolerance to ammonia borane. The fuel cell showed, at ambient temperature, an energy density of about 2000 mAh g −1 at 1 mA cm −2 and 0.9 V. Shortly afterward, Zhang et al. [68,69] reported their DABFC made of a carbon-supported platinum (30-46.6 wt%) electrodes, showing a power density of 110 mW cm −2 at 185 mA cm −2 , 0.6 V and 45 °C. These studies opened the way for developing efficient anode electrocatalysts, which were especially inactive towards the unwanted heterogeneous hydrolysis of ammonia borane: H3N−BH3 + 3H2O → NH3 + B(OH)3 + 3H2 (6) H3N−BH3 + 4H2O → NH4 + + B(OH)4 − + 3H2 Zhang et al. [72] worked on electro-oxidation on a gold electrode in the presence of thiourea. Direct electro-oxidation was found to occur, but the number of electrons exchanged was determined as being two, out of the theoretical maximum of six, because of the occurrence of hydrolysis. Belén Molina Concha et al. [73] confirmed, by on-line differential electrochemical mass spectrometry and in situ Fourier transform infrared spectroscopy, the occurrence of heterogeneous hydrolysis, especially at low potential. Hydrogen was detected, and the number of electrons exchanged was found to be about three. At higher potential, above 0.6 V, the number of electrons was about six because of preponderant electro-oxidation. The authors stressed the key role of the intermediate species BH3OH − : Schematic representation of (a) the direct ammonia borane fuel cell and (b) the direct borohydride fuel. At the right is given the standard potential versus the standard hydrogen electrode, the specific energy and the pure compound capacity. The anode (in yellow) and the cathode (in blue) are separated by a mmebrane (in grey).
Theoretically, the fuel cell features a standard potential of +1.62 V versus the standard hydrogen electrode, a specific energy of 8400 Wh kg −1 and a pure compound capacity of 5200 Ah kg −1 [68,69]. For comparison, the direct borohydride fuel cell ( Figure 2) has the following properties: theoretically, eight electrons released per borohydride anion BH 4 − , a standard potential of +1.64 V versus the standard hydrogen electrode, specific energy of 9300 Wh kg −1 and pure compound capacity of 5700 Ah kg −1 [70]. In operation, the borohydride cell should be slightly better performing. The first DABFC was reported by Yao et al. [71]. They isolated an efficient silver anode electrocatalyst (among other metals like platinum and gold), and manganese oxide MnO 2 as a cathode electrocatalyst showing tolerance to ammonia borane. The fuel cell showed, at ambient temperature, an energy density of about 2000 mAh g −1 at 1 mA cm −2 and 0.9 V. Shortly afterward, Zhang et al. [68,69] reported their DABFC made of a carbon-supported platinum (30-46.6 wt%) electrodes, showing a power density of 110 mW cm −2 at 185 mA cm −2 , 0.6 V and 45 • C. These studies opened the way for developing efficient anode electrocatalysts, which were especially inactive towards the unwanted heterogeneous hydrolysis of ammonia borane: Zhang et al. [72] worked on electro-oxidation on a gold electrode in the presence of thiourea. Direct electro-oxidation was found to occur, but the number of electrons exchanged was determined as being two, out of the theoretical maximum of six, because of the occurrence of hydrolysis. Belén Molina Concha et al. [73] confirmed, by on-line differential electrochemical mass spectrometry and in situ Fourier transform infrared spectroscopy, the occurrence of heterogeneous hydrolysis, especially at low potential. Hydrogen was detected, and the number of electrons exchanged was found to be about three. At higher potential, above 0.6 V, the number of electrons was about six because of preponderant electro-oxidation. The authors stressed the key role of the intermediate species BH 3 OH − : Nagle et al. [74,75] focused on gold in nanoporous morphology. The electrode was more active than bulk gold, with an oxidation current of 13.1 mA cm −2 versus 2.65 mA cm −2 . Barsuk et al. [76] also showed the beneficial effect of the nanoporous morphology, in their case for a silver anode electrocatalyst. To further enhance the electrocatalytic activity of gold, Karabiberoglu et al. [77] alloyed it with silver. Zhang et al. [78] focused on core-shell nanoparticles with an amorphous core of iron and a shell of platinum. They were much more active than commercial platinum (with about 2.5 and 0.75 mA µg −1 respectively at −0.7 V). Kiran et al. [79] developed a titanium carbide anode electrocatalyst. A fuel cell made of it, and of carbon-supported platinum as the cathode, generated a maximum power density of 85 mW cm −2 at 105 mA cm −2 and 80 • C. Olu et al. [80] explored the potential of a carbon-supported palladium anode electro-catalyst. A peak power density of 181 mW cm −2 was measured at 25 • C, with a similar performance being found for a comparable platinum electro-catalyst. Above all, this work showed that the performance depends on the anode's texture. Zadick et al. [81] alloyed palladium with nickel, such as Ni 3 Pd nanoparticles, supported onto a carbon support. Alternatively, silver and cobalt were used instead of palladium. The noble-free nickel-cobalt nanoparticles exhibited the best results, being both active and durable in the ammonia borane electro-oxidation conditions. Wang et al. [82] confirmed the potential of noble metal-free nickel-based electro-catalysts, especially that of carbon fiber-supported nanowires of nickel-copper selenide (better performing than the iron and cobalt counterparts). Similar conclusions were obtained by Wu et al. [83]. They successfully developed carbon fiber supported porous nickel-copper oxide nanowire arrays that showed a low onset potential with −0.316 V and high faradic efficiency (>98%), for example.
The electro-oxidation of ammonia borane has been little studied in comparison with the electro-oxidation of the borohydride anion of sodium borohydride. The fuel cells based on these two reactions are, however, comparable, in particular as regards the issues they both suffer from. The first main issue is, as briefly reported above, the occurrence of metal electrode-catalyzed heterogeneous hydrolysis that competes with the electro-oxidation reaction and lowers the number of electrons exchanged, and thus the faradic capacity of ammonia borane. The second main issue is about crossover of ammonia borane or other intermediate species through the ion exchange membrane of the fuel cell [72]: the ammonium ion NH 4 + (Equation (5)) is likely to crossover a cation exchange membrane, and the key intermediate species BH 3 OH − is prone to crossover an anion exchange membrane. Crossover is harmful for the performance of the cathode electro-catalyst. As far as is known, there is no study reporting a membrane specifically developed for a direct ammonia borane fuel cell. On a final matter, ammonia borane necessarily costs more than sodium borohydride, as the former is synthesized from the latter [18]. The electro-oxidation of ammonia borane is very likely facing another issue, which has been neglected so far. In aqueous solution, ammonia borane generates ammonium ions in equilibrium with ammonia, and these species (a weak Brønsted acid and a Lewis base, respectively) may have a negative effect on the fuel cell by deteriorating the electrode, the electro-catalyst and/or the membrane. The features of ammonia borane as an anodic fuel are less attractive than those of sodium borohydride. This can explain why, in comparison, the electro-oxidation of ammonia borane has been little studied. Few fuel cells were constructed and tested, but these efforts remained at the lab scale and, in any case, the technology appears to have been abandoned now.
The electro-oxidation of ammonia borane remains of scientific interest. There is much to be learned from it, with regard to the intermediate species BH 3 OH − (Equation (8)), for example. This species is believed to play an important role in the electro-oxidation of the borohydride anion as well [84], Energies 2020, 13, 3071 6 of 45 and the electro-oxidation of ammonia borane might allow better understanding of the electro-oxidation of the borohydride anion.

In Water
From a technological point of view, the hydrolysis of ammonia borane (Equations (6) and (7)) does not present any advantage over the hydrolysis of sodium borohydride. This is discussed hereafter. In all aspects, the latter reaction is superior (Figure 3). Evidence of this comes from the fact that the hydrolysis of ammonia borane has not been scaled up yet: no demonstrator, nor prototype and device, has been reported, and the effective gravimetric hydrogen storage capacity of the couple ammonia borane-water under operation is unknown [21].
Improved catalytic performance will take place certainly and soon, achieved by means of further changes and developments. The number of potential heterogenous catalysts is illimited. This may be illustrated as follows. Starting with twelve transition metals (i.e., Fe, Co, Ni, Cu, Zn, Ag, Pd, Rh, Ru, Ir, Pt, Au) and considering bi-and tri-atomic systems with these metals, as well as other elements like B and P, there would be more than 2900 combinations ( Figure 4). Then, supporting these combinations onto various supports (e.g., silica or carbon in various forms, nitrides, graphene and modified graphene, metal organic frameworks which exceed 10,000 candidates and so forth), for example onto ten supports for example, there would be more than 40,000 potential catalysts. The possibilities seem to be almost limitless.
Improved catalytic performance will take place certainly and soon, achieved by means of further changes and developments. The number of potential heterogenous catalysts is illimited. This may be illustrated as follows. Starting with twelve transition metals (i.e., Fe, Co, Ni, Cu, Zn, Ag, Pd, Rh, Ru, Ir, Pt, Au) and considering bi-and tri-atomic systems with these metals, as well as other elements like B and P, there would be more than 2900 combinations ( Figure 4). Then, supporting these combinations onto various supports (e.g., silica or carbon in various forms, nitrides, graphene and modified graphene, metal organic frameworks which exceed 10,000 candidates and so forth), for example onto ten supports for example, there would be more than 40,000 potential catalysts. The possibilities seem to be almost limitless. Water is, first and foremost, a reactant. Half of the released H2 is due to it, as one hydridic hydrogen H δ− of the BH3 group of ammonia borane reacts with one protic hydrogen H δ+ of water: H3N−B(H2)⋅⋅⋅H δ− ⋅⋅⋅H δ+ ⋅⋅⋅OH → H3N−B(H2)−OH + H2 (11) No side-reaction exists, which makes the hydrogen generation so simple. Water is almost always used in excess. This makes sense when a catalyst is under development. Accordingly, water is also a solvent. However, this makes no sense when the application is foreseen because of a negative impact on the storage capacities ( Figure 5). When water is used in stoichiometric conditions (Equations (6) and (7)), the theoretical gravimetric hydrogen storage capacities of the couple ammonia borane-water is 7.1 and 5.8 wt% H2, respectively. The solubility limit of ammonia borane in water is 11.4 mol L −1 [128], which is similar to a molar ratio H2O/H3N−BH3 of 4.9 and a theoretical gravimetric hydrogen storage capacity of about 5 wt% H2. At lower concentrations of ammonia borane, e.g., 1 mol L −1 (and even lower), as in those applied for investigating a wide variety of heterogeneous catalysts, the theoretical gravimetric hydrogen storage capacity is lower than 0.6 wt% H2, thus precluding any technological perspective. A very limited number of studies have explored the use of ammonia borane in a solid state (as powder [129] or in the form of pellets [130]) combined with the addition of water in near-stoichiometric conditions. In a recent work, an all-in-one formulation consisting of nickel nanoparticles intimately combined to AB nanoparticles was reacted with a limited amount of water [129]. An effective gravimetric hydrogen storage capacity of 4.8 wt% H2 was determined at 43.3 °C, whereas the conversion was lower than 100% (i.e., 78.3%). Further improvements in the composite formulation should allow the achievement of higher capacities, the target being 6.8 wt% H2. However, the hydrolysis reaction goes beyond the catalytic accelerator. According to Equation (6), water as a reactant, and ammonia and boric acid as by-products have to be taken into account. According to Equation (7), where there is a slight excess of water in comparison to the reaction shown by Equation (6), the by-product is ammonium tetrahydroxyborate. Each of these reactant and by-products are important, at least as important as the catalyst, but little attention has been paid to them.
Water is, first and foremost, a reactant. Half of the released H 2 is due to it, as one hydridic hydrogen H δ− of the BH 3 group of ammonia borane reacts with one protic hydrogen H δ+ of water: No side-reaction exists, which makes the hydrogen generation so simple. Water is almost always used in excess. This makes sense when a catalyst is under development. Accordingly, water is also a solvent. However, this makes no sense when the application is foreseen because of a negative impact on the storage capacities ( Figure 5). When water is used in stoichiometric conditions (Equations (6) and (7)), the theoretical gravimetric hydrogen storage capacities of the couple ammonia borane-water is 7.1 and 5.8 wt% H 2 , respectively. The solubility limit of ammonia borane in water is 11.4 mol L −1 [128], which is similar to a molar ratio H 2 O/H 3 N−BH 3 of 4.9 and a theoretical gravimetric hydrogen storage capacity of about 5 wt% H 2 . At lower concentrations of ammonia borane, e.g., 1 mol L −1 (and even lower), as in those applied for investigating a wide variety of heterogeneous catalysts, the theoretical gravimetric hydrogen storage capacity is lower than 0.6 wt% H 2 , thus precluding any technological perspective. A very limited number of studies have explored the use of ammonia borane in a solid state (as powder [129] or in the form of pellets [130]) combined with the addition of water in near-stoichiometric conditions. In a recent work, an all-in-one formulation consisting of nickel nanoparticles intimately combined to AB nanoparticles was reacted with a limited amount of water [129]. An effective gravimetric hydrogen storage capacity of 4.8 wt% H 2 was determined at 43.3 • C, whereas the conversion was lower than 100% (i.e., 78.3%). Further improvements in the composite formulation should allow the achievement of higher capacities, the target being 6.8 wt% H 2 .  (6) and (7)) are shown. The capacities calculated for an ammonia borane concentration of 11.4 mol L −1 (solubility limit of ammonia borane in water) and a concentration of 1 mol L −1 are also shown (5 and 0.6 wt% H2 respectively).
The hydrolysis of ammonia borane is often compared to that of sodium borohydride to underline the similarities between these two reactions. There is, however, a substantial difference. Ammonia is generated when ammonia borane is used. Ammonia is highly soluble in water and reacts to form its acidic form (pKa = 9.25): This does not hinder the loss of some ammonia in the gas phase, which is also exacerbated by the exothermic nature (−156 kJ mol −1 ) of the hydrolysis reaction [131]. This is well known, especially since the work published by Ramachandran and Gagare in 2007 [37]. The point is that ammonia is thermodynamically stable in hydrolytic conditions (enthalpy of dehydrogenation of 92.4 kJ mol −1 ); in other words, it cannot be dehydrogenated at a temperature below 600 °C [132]. Hence, the release of some ammonia together with hydrogen is unavoidable. The issue can be addressed by using a trap consisting of an aqueous solution of sulfuric acid [133] or of copper sulfate [128]. This is an efficient approach for a laboratory set-up. It is, however, questionable whether such a trap is implementable for a scaled-up device. The fact is that the ammonia problem is being grossly underrated whilst ammonia is detrimental to low-temperature polymer electrolyte membrane fuel cells (because of the poisoning of platinum anode and reaction with acidic membrane) [134].
As mentioned above, the hydrolysis reaction is spontaneous, and thus exothermic. This has negative consequences. The exothermicity of the reaction means that some water is vaporized and expelled out together with the generated hydrogen. Borates are very likely to be carried out by the steam. No report has been found on this, but such an issue was reported for the exothermic hydrolysis of sodium borohydride [135]. In such conditions, hydrogen is not pure, water condensates on any cold surface and the carried borates precipitate in the piping, resulting in accumulation and the clogging of the system. Such problems were encountered in operation conditions for a commercial device based on the hydrolysis of sodium borohydride [136].
The exothermicity of the hydrolysis reaction also means that there is a transition from a reactive B−H bond in H3N−BH3 to a stable B−O bond, e.g., in B(OH)3 (Equation (6)). Borates are indeed a stable species, the B−O bond being relevantly compared to the C−O bond of carbon dioxide [137]. They are hardly transformed, that is, reduced, back into the starting hydride, whether it be sodium borohydride [138] or ammonia borane, but it should be noted that the direct or indirect regeneration of ammonia borane from the borates forming upon hydrolysis has not been investigated yet. This  (6) and (7)) are shown. The capacities calculated for an ammonia borane concentration of 11.4 mol L −1 (solubility limit of ammonia borane in water) and a concentration of 1 mol L −1 are also shown (5 and 0.6 wt% H 2 respectively).
The hydrolysis of ammonia borane is often compared to that of sodium borohydride to underline the similarities between these two reactions. There is, however, a substantial difference. Ammonia is generated when ammonia borane is used. Ammonia is highly soluble in water and reacts to form its acidic form (pK a = 9.25): This does not hinder the loss of some ammonia in the gas phase, which is also exacerbated by the exothermic nature (−156 kJ mol −1 ) of the hydrolysis reaction [131]. This is well known, especially since the work published by Ramachandran and Gagare in 2007 [37]. The point is that ammonia is thermodynamically stable in hydrolytic conditions (enthalpy of dehydrogenation of 92.4 kJ mol −1 ); in other words, it cannot be dehydrogenated at a temperature below 600 • C [132]. Hence, the release of some ammonia together with hydrogen is unavoidable. The issue can be addressed by using a trap consisting of an aqueous solution of sulfuric acid [133] or of copper sulfate [128]. This is an efficient approach for a laboratory set-up. It is, however, questionable whether such a trap is implementable for a scaled-up device. The fact is that the ammonia problem is being grossly underrated whilst ammonia is detrimental to low-temperature polymer electrolyte membrane fuel cells (because of the poisoning of platinum anode and reaction with acidic membrane) [134].
As mentioned above, the hydrolysis reaction is spontaneous, and thus exothermic. This has negative consequences. The exothermicity of the reaction means that some water is vaporized and expelled out together with the generated hydrogen. Borates are very likely to be carried out by the steam. No report has been found on this, but such an issue was reported for the exothermic hydrolysis of sodium borohydride [135]. In such conditions, hydrogen is not pure, water condensates on any cold surface and the carried borates precipitate in the piping, resulting in accumulation and the clogging of the system. Such problems were encountered in operation conditions for a commercial device based on the hydrolysis of sodium borohydride [136].
The exothermicity of the hydrolysis reaction also means that there is a transition from a reactive B−H bond in H 3 N−BH 3 to a stable B−O bond, e.g., in B(OH) 3 (Equation (6)). Borates are indeed a stable species, the B−O bond being relevantly compared to the C−O bond of carbon dioxide [137]. They are hardly transformed, that is, reduced, back into the starting hydride, whether it be sodium borohydride [138] or ammonia borane, but it should be noted that the direct or indirect regeneration of Energies 2020, 13, 3071 9 of 45 ammonia borane from the borates forming upon hydrolysis has not been investigated yet. This might be explained by the fact that more than one borate forms upon the hydrolysis of ammonia borane.
As shown by Equations (6) and (7), two borates are likely to form, i.e., boric acid B(OH) 3 and the tetrahydroxyborate anion B(OH) 4 − , especially for diluted aqueous solutions [133,139]. In addition to these species, other unidentified borates have been suggested [22,131]. The reaction of B(OH) 3 and B(OH) 4 − to form equilibria towards borates is in fact well documented [140][141][142]. Rachiero et al. [143] suggested the formation of polyborates: For concentrated aqueous solutions (10.8 mol L −1 ), a tetraborate like (NH 4 ) 2 B 4 O 5 (OH) 4 ·1.4H 2 O was reported by Ramachandran and Gagare [37]. Recently, for a comparable concentrated solution (10 mol L −1 ), Valero-Pedraza et al. [144] observed the precipitation of (NH 4 The formation of this species was supposed to be as follows: According to this path, there are two free ammonia molecules. If lost, they could not be integrated into the recycling cycle of the borates, which would run counter to a few green chemistry principles [145]. The tetraborate (NH 4 ) 2 B 4 O 5 (OH) 4 ·2H 2 O is an analog of borax Na 2 B 4 O 5 (OH) 4 ·2H 2 O, and borax is one of the reactants used for the industrial production of sodium borohydride via the so-called Brown-Schlesinger process. These findings suggest that, with sustained efforts, one may gain a better understanding of these borate species while working on their imperative recyclability. In any case, the recyclability requires a large amount of energy, and the cost of the process will be dependent on the energy cost [146], suggesting the crucial role of green electricity for efficient cost reduction [147].

In Methanol
Methanol can be used instead of water as a source of protic hydrogens H δ+ that can react with the hydridic hydrogens H δ− of ammonia borane. The first report about the methanolysis of ammonia borane dates back to 2007, and it is insightful on several levels. Ramachandran and Gagare [37], after comparing a series of aliphatic alcohols, found that methanol provided the best results. The couple ammonia borane-methanol, in the presence of 1 mol% ruthenium chloride, was able to generate 3 equivalents of hydrogen in less than 1 min at ambient conditions. However, a 15% excess of methanol was found to be necessary to obtain the three equivalents of hydrogen (Equation (15)). A precipitating "crude borate" was isolated and identified as being an ammonium tetramethoxyborate salt such as [NH 4 B(OCH 3 ) 4 ] 5 ·2CH 3 OH: It is not clearly specified whether the methanolic solution, in addition to the precipitating "crude borate", was analyzed or not. It is also not clearly specified whether the purity of the released hydrogen was analyzed or not. Interestingly, the recyclability of the ammonium tetramethoxyborate salt was investigated. It was transformed back into ammonia borane (yield 81% and purity >98%), by using lithium alanate LiAlH 4 as a reducing agent and ammonium chloride NH 4 Cl as a source of ammonia: The reaction was performed in a sealed reactor, with tetrahydrofuran as a solvent, for 8 h at ambient conditions. Hence, the cycle was closed. From this work, four interesting aspects stand out: an excess of methanol is required to obtain a conversion of 100%, one of the by-products is a polyborate salt, the release (or not) of ammonia is unclear, and the polyborate salt can be effectively recycled back into ammonia borane.
While the aforementioned report has the advantage of putting several aspects of the methanolysis reaction forward, none of these aspects has been the subject of a systematic and deep investigation. As for the hydrolysis reaction, efforts have been focused on heterogeneous catalysts. Listing the catalysts reported since 2007 goes beyond the scope of the present article. The reader may refer to a recent review article [148]. A variety of nanostructured transition metal catalysts (also called nanocatalysts) have been developed, with a special focus on copper [149][150][151][152], cobalt [153] and cobalt-palladium combinations [154,155], nickel [153,156,157], palladium [158][159][160], rhodium [161][162][163][164] and ruthenium [165][166][167].
The reaction shown by Equation (16) was proposed to explain the formation of the ammonium tetramethoxyborate salt [NH 4 B(OCH 3 ) 4 ] 5 ·2CH 3 OH, forming from an aqueous solution of ammonia borane at 2 mol L −1 [37]. For more diluted solutions, the following equation is generally reported: When the literature is reviewed, it stands out that this reaction is rather unknown. The amount of methanol has not been optimized in order to optimize the effective gravimetric hydrogen storage capacity of the couple ammonia borane-methanol (versus the theoretical value of 3.8 wt% H 2 for the stoichiometry in Equation (17) (Figure 6). The release of some ammonia, together with hydrogen, has not been checked and measured. The nature of the borate by-products, depending on the amount of methanol used and/or the drying process, has not been determined yet, except for the aforementioned tetramethoxyborate salt. The knowledge generated by the hydrolysis reaction suggests that the effective capacities will be lower than the theoretical value; ammonia could be found in the hydrogen stream and more than one borate may form. These have to be seen as outstanding issues.
Energies 2020, 13, x FOR PEER REVIEW 10 of 45 While the aforementioned report has the advantage of putting several aspects of the methanolysis reaction forward, none of these aspects has been the subject of a systematic and deep investigation. As for the hydrolysis reaction, efforts have been focused on heterogeneous catalysts. Listing the catalysts reported since 2007 goes beyond the scope of the present article. The reader may refer to a recent review article [148]. A variety of nanostructured transition metal catalysts (also called nanocatalysts) have been developed, with a special focus on copper [149][150][151][152], cobalt [153] and cobalt-palladium combinations [154,155], nickel [153,156,157], palladium [158][159][160], rhodium [161][162][163][164] and ruthenium [165][166][167].
The reaction shown by Equation (16) was proposed to explain the formation of the ammonium tetramethoxyborate salt [NH4B(OCH3)4]5⋅2CH3OH, forming from an aqueous solution of ammonia borane at 2 mol L −1 [37]. For more diluted solutions, the following equation is generally reported: When the literature is reviewed, it stands out that this reaction is rather unknown. The amount of methanol has not been optimized in order to optimize the effective gravimetric hydrogen storage capacity of the couple ammonia borane-methanol (versus the theoretical value of 3.8 wt% H2 for the stoichiometry in Equation (17) ( Figure 6). The release of some ammonia, together with hydrogen, has not been checked and measured. The nature of the borate by-products, depending on the amount of methanol used and/or the drying process, has not been determined yet, except for the aforementioned tetramethoxyborate salt. The knowledge generated by the hydrolysis reaction suggests that the effective capacities will be lower than the theoretical value; ammonia could be found in the hydrogen stream and more than one borate may form. These have to be seen as outstanding issues. The successful regeneration of ammonia borane using the ammonium tetramethoxyborate salt [NH4B(OCH3)4]5⋅2CH3OH as reactant [37] has great appeal. The hydrogen cycle with the couple ammonia borane-methanol can be closed. More broadly, ammonia borane can be directly synthesized, namely without forming sodium borohydride in the first step, from another possible methanolysis product, which is trimethoxyborate B(OCH3)3. Indeed, Ramachandran et al. [168] developed a one-pot synthesis of ammonia borane, taking place in tetrahydrofuran at 0 °C: These achievements are arguments in favor of the methanolysis reaction, for which the hydrogen cycle could be closed (Figure 7). The successful regeneration of ammonia borane using the ammonium tetramethoxyborate salt [NH 4 B(OCH 3 ) 4 ] 5 ·2CH 3 OH as reactant [37] has great appeal. The hydrogen cycle with the couple ammonia borane-methanol can be closed. More broadly, ammonia borane can be directly synthesized, namely without forming sodium borohydride in the first step, from another possible methanolysis product, which is trimethoxyborate B(OCH 3 ) 3 . Indeed, Ramachandran et al. [168] developed a one-pot synthesis of ammonia borane, taking place in tetrahydrofuran at 0 • C: These achievements are arguments in favor of the methanolysis reaction, for which the hydrogen cycle could be closed (Figure 7).
Energies 2020, 13, x FOR PEER REVIEW 11 of 45 Figure 7. A way to close the hydrogen cycle with the couple ammonia borane-methanol, provided that a single by-product forms (Equation (17)).
The decomposition path of ammonia borane is generally described in a fairly straightforward manner. The first step is the induction [174]. Ammonia borane, which becomes more mobile under heating [175], dimerizes into an ionic dimer, the diammoniate of diborane: The diammoniate of diborane is the reactive species, the one decomposing. It is more reactive than ammonia borane. The decomposition is then stepwise.
From the diammoniate of diborane, which reacts with an additional ammonia borane molecule, the first equivalent of hydrogen is generated. This can be illustrated as follows: It is important to point out that the reaction is generally reported in the open literature as follows [176]: Ammonia borane transforms into polyaminoborane (non-crystalline) and hydrogen by dehydrocoupling. However, gaseous by-products like aminoborane H2N=BH2, diborane B2H6, ammonia NH3 and/or borazine B3N3H6 are also generated ( Figure 8) [177]. This first decomposition step is exothermic, with an enthalpy of e.g., −21 kJ mol −1 [178].

Figure 7.
A way to close the hydrogen cycle with the couple ammonia borane-methanol, provided that a single by-product forms (Equation (17)).
The decomposition path of ammonia borane is generally described in a fairly straightforward manner. The first step is the induction [174]. Ammonia borane, which becomes more mobile under heating [175], dimerizes into an ionic dimer, the diammoniate of diborane: The diammoniate of diborane is the reactive species, the one decomposing. It is more reactive than ammonia borane. The decomposition is then stepwise.
From the diammoniate of diborane, which reacts with an additional ammonia borane molecule, the first equivalent of hydrogen is generated. This can be illustrated as follows: It is important to point out that the reaction is generally reported in the open literature as follows [176]: Ammonia borane transforms into polyaminoborane (non-crystalline) and hydrogen by dehydrocoupling. However, gaseous by-products like aminoborane H 2 N=BH 2 , diborane B 2 H 6 , ammonia NH 3 and/or borazine B 3 N 3 H 6 are also generated ( Figure 8) [177]. This first decomposition step is exothermic, with an enthalpy of e.g., −21 kJ mol −1 [178]. The second step of the ammonia borane decomposition, taking place above 120-130 °C [172], generates borazine as the main product [179]. The second equivalent of hydrogen is also generated: According to this reaction, polyiminoborane [−BH=NH−]n forms. The aforementioned two steps take place at temperatures lower than 200 °C.
The third equivalent of hydrogen is generated above 200 °C. The total release of it requires high temperatures, well beyond 500 °C, and this results boron nitride BN forming (semi-crystalline above 1170 °C) [180]: Boron nitride is a highly stable ceramic, and its rehydrogenation is thermodynamically impossible in affordable conditions (in terms of temperature and pressure of H2). The dehydrogenation of ammonia borane thus has to be restricted to about two equivalents of hydrogen to avoid the formation of boron nitride.
The decomposition is in fact more complex than the simplistic chain-growth polymerization path (Equations (21) and (22)) based on heteropolar dihydrogen N−H δ+ ⋅⋅⋅H δ− −B interactions and the head-to-tail dehydrocoupling of ammonia borane ( Figure 9a). Basically, this path does not rationalize the release of the unwanted gaseous products like diborane and the formation of a polymeric residue with tri-and tetra-coordinated boron atoms, thus being more complex than the aforementioned linear polymers [181]. Wolstenholme et al. [182] and Roy et al. [183] showed the existence of counterintuitive homopolar dihydrogen B−H δ− ⋅⋅⋅H δ− −B interactions (Figure 9b). Petit and Demirci [184] reported homopolar dihydrogen N−H δ+ ⋅⋅⋅H δ+ −N interactions noticeably contributing to hydrogen release. Al-Kukhun et al. [177] showed that the BH3 group of ammonia borane makes a superior contribution to the release of hydrogen than the NH3 group, and that the heteropolar and homopolar interactions have comparable contributions. Otherwise, Petit et al. [185] analyzed the polymeric products forming upon the evolution of one equivalent of hydrogen from two different ammonia boranes, and found that all of the molecules do not simultaneously decompose and polymerize ( Figure 10). The polymeric products were found to be composed of various molecular structures, suggesting a mixture of polyaminoborane, polyiminoborane and polyborazylene, in good agreement with Summerscales and Gordon's observations [186]. In other words, various reaction pathways, which are intra-and inter-molecular, take place during the two-step decomposition of ammonia borane, up to about 200 °C [187]. It results in a polymeric residue generally denoted as BNHx (with x < 2) or BNH<2.
In summary, ammonia borane in a pristine state is not appropriate for hydrogen storage, and this is for four main reasons. Ammonia borane decomposes more than it dehydrogenates. The process The second step of the ammonia borane decomposition, taking place above 120-130 • C [172], generates borazine as the main product [179]. The second equivalent of hydrogen is also generated: According to this reaction, polyiminoborane [−BH=NH−] n forms. The aforementioned two steps take place at temperatures lower than 200 • C.
The third equivalent of hydrogen is generated above 200 • C. The total release of it requires high temperatures, well beyond 500 • C, and this results boron nitride BN forming (semi-crystalline above 1170 • C) [180]: Boron nitride is a highly stable ceramic, and its rehydrogenation is thermodynamically impossible in affordable conditions (in terms of temperature and pressure of H 2 ). The dehydrogenation of ammonia borane thus has to be restricted to about two equivalents of hydrogen to avoid the formation of boron nitride.
The decomposition is in fact more complex than the simplistic chain-growth polymerization path (Equations (21) and (22)) based on heteropolar dihydrogen N−H δ+ ···H δ− −B interactions and the head-to-tail dehydrocoupling of ammonia borane ( Figure 9a). Basically, this path does not rationalize the release of the unwanted gaseous products like diborane and the formation of a polymeric residue with tri-and tetra-coordinated boron atoms, thus being more complex than the aforementioned linear polymers [181]. Wolstenholme et al. [182] and Roy et al. [183] showed the existence of counterintuitive homopolar dihydrogen B−H δ− ···H δ− −B interactions (Figure 9b). Petit and Demirci [184] reported homopolar dihydrogen N−H δ+ ···H δ+ −N interactions noticeably contributing to hydrogen release. Al-Kukhun et al. [177] showed that the BH 3 group of ammonia borane makes a superior contribution to the release of hydrogen than the NH 3 group, and that the heteropolar and homopolar interactions have comparable contributions. Otherwise, Petit et al. [185] analyzed the polymeric products forming upon the evolution of one equivalent of hydrogen from two different ammonia boranes, and found that all of the molecules do not simultaneously decompose and polymerize ( Figure 10). The polymeric products were found to be composed of various molecular structures, suggesting a mixture of polyaminoborane, polyiminoborane and polyborazylene, in good agreement with Summerscales and Gordon's observations [186]. In other words, various reaction pathways, which are intra-and inter-molecular, take place during the two-step decomposition of ammonia borane, up to about 200 • C [187]. It results in a polymeric residue generally denoted as BNH x (with x < 2) or BNH <2 . at low temperature and under hydrogen pressure. This thus suggests the requirement of chemical rehydrogenation. The polymeric residue forming upon decomposition is of complex composition and not well identified, which makes the recycling approach tough.
It is in that context that strategies for destabilizing ammonia borane and making it suitable for hydrogen storage were introduced. There are four. They all aim at dehydrogenating ammonia borane rapidly, at temperatures below 100 °C and with the formation of a single polymeric product.   It is in that context that strategies for destabilizing ammonia borane and making it suitable for hydrogen storage were introduced. There are four. They all aim at dehydrogenating ammonia borane rapidly, at temperatures below 100 °C and with the formation of a single polymeric product.   In summary, ammonia borane in a pristine state is not appropriate for hydrogen storage, and this is for four main reasons. Ammonia borane decomposes more than it dehydrogenates. The process takes place slowly at temperatures over 70 • C and more rapidly above 120 • C. The decomposition (and the dehydrogenation) is exothermic. Accordingly, the rehydrogenation of the dehydrogenated form of ammonia borane requires energy to offset the heat lost (due to the exothermic breaking of B−H and N−H bonds). This suggests difficult, from a thermodynamic point of view, rehydrogenation at low temperature and under hydrogen pressure. This thus suggests the requirement of chemical rehydrogenation. The polymeric residue forming upon decomposition is of complex composition and not well identified, which makes the recycling approach tough.
It is in that context that strategies for destabilizing ammonia borane and making it suitable for hydrogen storage were introduced. There are four. They all aim at dehydrogenating ammonia borane rapidly, at temperatures below 100 • C and with the formation of a single polymeric product.

Ammonia Borane in an Aprotic Solvent
With protic solvents, hydrolysis and alcoholysis take place, as discussed in the previous sections. With aprotic solvents, ammonia borane can be isolated as a molecule and the dihydrogen N−H δ+ ···H δ− −B interactions are disrupted. This makes dehydrogenation easier.
Ammonia borane is soluble in ionic liquids. Bluhm et al. [171] investigated an equal weight mixture of ammonia borane and 1-butyl-3-methylimidazolium chloride (oversaturated solution), the solubility of the borane being 47.5 g in 100 g of solvent [188]. The mixture consisted of ammonia borane in both solubilized and solid states. Heated at 85 • C, the mixture released hydrogen immediately, showing no induction period (versus an induction time of about 3 h for the pristine borane). It released 1 mol H 2 per mol H 3 N−BH 3 in 67 min and 2.2 mol H 2 in 330 min [189]. Nakagawa et al. [188] reported that more than 2 mol H 2 per mol H 3 N−BH 3 , with limited impurity emissions, can be released between 80 and 130 • C, without any reaction between ammonia borane and 1-butyl-3-methylimidazolium chloride. Consistent results were reported elsewhere [190]. The ionic liquid was suggested to act as a promoter favoring the formation of the diammoniate of diborane. Upon dehydrogenation, neutral polyaminoborane and then unsaturated cross-linked polyborazylene were reported to form. Elsewhere, a complex solid solution was instead suggested [188]. Several other ionic liquids were reported in reference [189] and elsewhere [191][192][193], including 1-butyl-2,3-dimethylimidazolium chloride, 1-ethyl-2,3-dimethylimidazolium ethyl-sulfate, 1-butyl-3-methylimidazolium tetra-fluoroborate, 1-butyl-3-methylimidazolium triflate and 1-butyl-3-methylimidazolium hexafluorophosphate, among others. Beyond the works cited above, where the last dates back to 2015, the use of ionic liquids has not been further studied. One plausible explanation is the difficult extraction of the dehydrogenation products because of the high boiling point of the ionic liquids.
Ammonia borane is soluble in organic solvents. In triglyme, its solubility is 34.4 g in 100 g of solvent. Kostka et al. [194] studied solutions at various concentrations, from 0.1 to 6 mol L −1 . Improved dehydrogenation properties were found, in comparison with the pristine ammonia borane. For instance, the solution at 6 mol L −1 generated 1 mol H 2 per mol H 3 N−BH 3 in less than 1 h at 70 • C. There was no induction period, also for other solvents like glyme, diglyme and tetraglyme. Dehydrocoupling of ammonia borane in triglyme follows a complex path where, in addition to borazine, several cyclic products form (Figure 11), especially for high concentrations of ammonia borane. The cyclic products were suggested to be intermediates before the formation of borazine and then polyborazylene [195].
Ammonia borane, solubilized in an organic solvent, is able to generate pure hydrogen with good kinetics, provided that a (homogenous) catalyst is used. Denney et al. [196] reported an iridium pincer complex, denoted (POCOP)Ir(H) 2 with POCOP as [η 3 -1,3-(OP t Bu 2 ) 2 C 6 H 3 ]. The complex (1 mol%) was able to catalyze the dehydrocoupling of ammonia borane (0.5 mol L −1 in tetrahydrofuran) towards the release of one equivalent of hydrogen in less than 5 min at room temperature: The oligomeric product precipitated. Another iridium complex, [Ir(H) 2 (dppm) 2 ]OTf, was reported elsewhere [197]. Keaton et al. [198] investigated a N-heterocyclic carbene nickel complex for dehydrogenating ammonia borane (25 wt%) in diglyme at 60 • C. A small amount of the complex (1 mol%) was able to catalyze the rapid dehydrogenation of ammonia borane to a high extent (ca. 2.8 mol H 2 per mol H 3 N−BH 3 ) in less than 4 h. Conley and Williams [199] developed a Shvo's cyclopentadienone-ligated ruthenium complex that catalyzed the generation of two equivalents of hydrogen and the formation of borazine from ammonia borane in a mixture of diglyme and benzene, at 70 • C: Afterwards, iron [200,201], ruthenium [202][203][204] and palladium complexes [205,206] were reported as possible efficient homogeneous catalysts for dehydrogenating ammonia borane in solution. Details about all of the homogeneous catalysts reported above are available in several review papers authored by active researchers in the field [207][208][209]. Recently, Hasenbeck et al. [210] reported a metal-free organocatalyst, the 6-tert-butyl-2-thiopyriperformed. At a concentration of 1 mol%, it was able to dehydrogenate ammonia borane at reflux in tetrahydrofuran in 2 h (Equation (25)). A turnover frequency of 88 h −1 was achieved.
The dehydrocoupling of ammonia borane in an organic solvent and in the presence of a homogenous catalyst offers some advantages, such as a simple dehydrogenation pathway resulting in a single product like borazine or polyborazylene, the release of pure hydrogen (apart from the solvent), faster hydrogen release kinetics in comparison to the uncatalyzed reaction and temperatures of reaction lower than 100 • C (e.g., at ambient conditions). However, there are also drawbacks, which are significant barriers to the technological implementation of this destabilization strategy, namely: the vapor pressure of the organic solvents that inevitably pollutes the hydrogen stream; the homogenous nature of the catalyst that makes its extraction from the reaction slurry tough; the limited cyclability of the catalyst; and the low solubility of ammonia borane in the mentioned solvents, which negatively impact the effective gravimetric hydrogen storage capacities (Figure 12). Using a liquid hydrogen fuel resembling gasoline has appeal, and it is regrettable that this liquid ammonia borane approach has not been scaled up yet. This would have allowed a better understanding of the advantages and drawbacks of it. for B-(cyclodiborazanyl)aminoborohydride, CTB for cyclotriborazane, µADB for µ-aminodiborane and BZ for borazine. Adapted from ref. [194].
Ammonia borane, solubilized in an organic solvent, is able to generate pure hydrogen with good kinetics, provided that a (homogenous) catalyst is used. Denney et al. [196] reported an iridium pincer complex, denoted (POCOP)Ir(H)2 with POCOP as [η 3 -1,3-(OP t Bu2)2C6H3]. The complex (1 mol%) was able to catalyze the dehydrocoupling of ammonia borane (0.5 mol L −1 in tetrahydrofuran) towards the release of one equivalent of hydrogen in less than 5 min at room temperature: The oligomeric product precipitated. Another iridium complex, [Ir(H)2(dppm)2]OTf, was reported elsewhere [197]. Keaton et al. [198] investigated a N-heterocyclic carbene nickel complex for dehydrogenating ammonia borane (25 wt%) in diglyme at 60 °C. A small amount of the complex (1 mol%) was able to catalyze the rapid dehydrogenation of ammonia borane to a high extent (ca. 2.8 mol H2 per mol H3N−BH3) in less than 4 h. Conley and Williams [199] developed a Shvo's cyclopentadienone-ligated ruthenium complex that catalyzed the generation of two equivalents of hydrogen and the formation of borazine from ammonia borane in a mixture of diglyme and benzene, at 70 °C: Afterwards, iron [200,201], ruthenium [202][203][204] and palladium complexes [205,206] were reported as possible efficient homogeneous catalysts for dehydrogenating ammonia borane in solution. Details about all of the homogeneous catalysts reported above are available in several review papers authored by active researchers in the field [207][208][209]. Recently, Hasenbeck et al. [210] reported a metal-free organocatalyst, the 6-tert-butyl-2-thiopyriperformed. At a concentration of 1 mol%, it was able to dehydrogenate ammonia borane at reflux in tetrahydrofuran in 2 h (Equation (25)). A turnover frequency of 88 h −1 was achieved.
The dehydrocoupling of ammonia borane in an organic solvent and in the presence of a homogenous catalyst offers some advantages, such as a simple dehydrogenation pathway resulting in a single product like borazine or polyborazylene, the release of pure hydrogen (apart from the With DADB for diammoniate of diborane, CDB for cyclodiborazane, BCDB for B-(cyclodiborazanyl)aminoborohydride, CTB for cyclotriborazane, µADB for µ-aminodiborane and BZ for borazine. Adapted from ref. [194]. limited cyclability of the catalyst; and the low solubility of ammonia borane in the mentioned solvents, which negatively impact the effective gravimetric hydrogen storage capacities (Figure 12). Using a liquid hydrogen fuel resembling gasoline has appeal, and it is regrettable that this liquid ammonia borane approach has not been scaled up yet. This would have allowed a better understanding of the advantages and drawbacks of it.

Chemical Doping of Solid Ammonia Borane
By adding a chemical dopant, solid mixtures made of ammonia borane as the main component can be obtained. The first mixture was proposed by Benedetto et al. [211]. They showed that the addition of 1-2 mol% of platinum, in the form of hexachloroplatinic salt, resulted in improved

Chemical Doping of Solid Ammonia Borane
By adding a chemical dopant, solid mixtures made of ammonia borane as the main component can be obtained. The first mixture was proposed by Benedetto et al. [211]. They showed that the addition of 1-2 mol% of platinum, in the form of hexachloroplatinic salt, resulted in improved dehydrogenation properties, with a significantly reduced induction period and faster kinetics. Upon reaction, reduced platinum was recovered, but no analyses were performed with regard to the purity of the released hydrogen and the nature of the polymeric residue. The positive effect of a dopant on the induction period of ammonia borane was later confirmed by Heldebrant et al. [175]. The period was measured to be as long as about 6 h for pristine ammonia borane at 80 • C, whereas the addition of ammonium chloride (5 wt%) resulted in the absence of an induction period. A series of chlorides were afterwards considered to be possible dopants. He et al. [212] tested nickel chloride NiCl 2 and cobalt chloride CoCl 2 (for both 2 mol%). Ammonia borane was destabilized, liberating hydrogen (and no borazine) at 59 • C and upwards. Benzouaa et al. [213] worked with cobalt chloride CoCl 2 , iron chloride FeCl 3 and aluminum chloride AlCl 3 . They all destabilized ammonia borane, which featured decreased onset temperatures of decomposition (52, 65 and 85 • C respectively, versus 99 • C for the pristine ammonia borane, the last temperature being typical of an ammonia borane sample with a purity higher than 97% and of an experiment performed at a heating rate of 5 • C min −1 ). Reduced induction periods and much less borazine were noticed when, for example, cobalt chloride was used. Better results were reported for copper chloride [214,215]. Despite improved decomposition properties, doping with metal chlorides does not totally prevent the formation of borazine, and the polymeric residue formed is of complex composition [213][214][215][216]. The metal cation Cu 2+ of copper chloride plays a Lewis acid role, destabilizing the B−N bond of ammonia borane through the formation of a reactive intermediate, such as Cu· · · NH 2 -BH 2 , then facilitating the head-to-tail dehydrocoupling of ammonia borane. Comparable results were reported for magnesium and calcium chlorides [217]. In the presence of the former, ammonia borane dehydrogenated purely, starting from 40 • C, whereas traces of ammonia were detected when ammonia borane was doped with the latter halide. The release of ammonia was also reported for ammonia borane doped with potassium bromide [218]. The additive was found to favor the formation of diammoniate of diborane as well as the formation of ammonium ions that catalyzed the dehydrogenation of ammonia borane. Overall, the addition of metal halide has positive effects, but the evolution of gaseous by-products is not totally suppressed. It is also unclear what the nature of the polymeric residue and the post-thermolysis state of the chemical additive are.
Metal hydrides are also possible dopants. Note that the mixtures made of one mole of ammonia borane and one mole of another hydride, which are actually composites, are out of the scope of the present work. In such composites, the addition of the hydride cannot be seen as a simple destabilization of ammonia borane. The composites have to be considered to be new materials.
Kang et al. [219] demonstrated the positive effect of magnesium hydride MgH 2 as a chemical additive. The milled mixture, consisting of 1 mol H 3 N−BH 3 and 0.5 mol MgH 2 , showed an apparent activation energy of 66 kJ mol −1 for the release of one equivalent of hydrogen (versus 183 kJ mol −1 for pristine ammonia borane). Up to 200 • C, it released around 1.9 equivalents of hydrogen: A fraction of magnesium hydride was suggested to participate in the dehydrogenation process via the formation of Mg-B-N-H intermediate species [220]. Tang et al. [221] used magnesium-nickel hydride Mg 2 NiH 4 (1 mol) to destabilize ammonia borane (4 mol). Pure hydrogen (two equivalents) was released, according to a two-step process occurring at 50 and 110 • C (one equivalent for each step). Improved dehydrogenation properties were also reported for other double cation hydrides, such as NaMgH 3 , KMgH 3 and RbMgH 3 [222]. Choi et al. [223] compared magnesium hydride MgH 2 , titanium hydride TiH 2 , zirconium hydride ZrH 2 and calcium hydride CaH 2 . Doping 2 mol H 3 N−BH 3 with 1 mol MgH 2 and 0.1 mol TiH was found to be efficient in lowering the dehydrogenation temperature of ammonia borane. Kobayashi et al. [224] showed that lithium alanate Li 3 AlH 6 improves the dehydrogenation kinetics of ammonia borane (for a 1:3 mixture) via the formation of an intermediate containing lithium amidoborane LiNH 2 BH 3 . Wan et al. [225] destabilized ammonia borane (3 mol) by using aluminum hydride AlH 3 (1 mol). The release of more than 12 wt% of pure hydrogen was measured up to 250 • C. Ammonia borane was destabilized by Coulombic attraction between H δ+ of NH 3 of ammonia borane and H δ− of AlH 3 . With more reactive metal hydrides like sodium hydride NaH, chemical doping is not relevant because a derivative of ammonia borane forms. This is discussed hereafter, in Section 5.5, which is dedicated to amidoboranes.
Other chemical dopants have been reported so far. Neiner et al. [226,227] used pre-milled hexagonal boron nitride. The dopant showed a slight beneficial effect by favoring the disruption of the dihydrogen N−H δ+ ···H δ− −B interactions in ammonia borane, but it also had a negative impact, by promoting the formation of borazine by a surface-enhanced mechanism. Pre-milling boron nitride was important to obtain nanocrystallites of about 10 nm, thereby developing a higher specific surface area of 251 m 2 g −1 than the bulk counterpart (12 m 2 g −1 ). The chemical dopant destabilization strategy is based on grain-to-grain contact between the dopant and ammonia borane, and nanosizing the dopant allows increased contacts. In this way, elsewhere, silicon [228], silica [229] and nickel [230] were used in the form of nanosized dopants.
Some of the studied dopants contain oxygen and protic hydrogen. This is a paradox, because it is like protonolysis (comparable to hydrolysis or methanolysis), and stable B−O bonds are likely to form. A first example of an oxygen-containing dopant is boric acid B(OH) 3 . Hwang and Varma [231] demonstrated its positive effect on the thermolysis of ammonia borane. For example, more than two equivalents of hydrogen were released at 85 • C, and only trace amount of ammonia (<30 ppm) was detected. The involvement of the tetrahydroxyborate ion B(OH) 4 − and water was noticed.
However, surprisingly, the polymeric residue was suggested to be a polyborazylene-like species, with no mention of likely B−O bonds. Ergüven et al. [232] explained the role of boric acid as that of a Lewis acid (versus a Lewis base for ammonia borane). Shin et al. [233] destabilized ammonia borane with mucic acid, a dicarboxylic acid with the formula C 6 H 10 O 8 . The 8:2 mixture produced pure H 2 (10.8 wt%) at 80 • C in 1 min. This performance is attractive. Similar attractive results were reported by Kim et al. [234], who used d-mannitol C 6 H 14 O 6 , an alcohol with six OH groups, as a dopant. In both works, the polymeric residue was made of B−O bonds. This raises the question: is such an approach more attractive than hydrolysis, which allows the generation of a lot hydrogen at 20 • C, and with faster kinetics? There are also protic hydrogen-free oxygen-containing dopants. Kim et al. [235,236] investigated polyetheral additives like tetraethylene glycol dimethyl ether C 10 H 22 O 5 , which enhanced the rate of hydrogen generation from ammonia borane under heating (e.g., 125 • C). It, however, did not hinder the formation of unwanted gaseous products like borazine. The great thing about Kim et al.'s works is that they pushed the logic to the extreme [237]. They constructed a hydrogen generator based on ammonia borane and the aforementioned additive, with the generator integrating a 200 We polymer electrolyte membrane fuel cell ( Figure 13). Under operation, a hydrogen generation rate as high as 3.3 L min −1 was reached, being enough to operate the fuel cell for 25 min with 58 g of ammonia borane. A purifying system (i.e., acidic active carbons trapping borazine) allowed the feeding of the fuel cell with pure hydrogen. This demonstrator was the first one ever reported with ammonia borane as a hydrogen carrier. ammonia borane and the aforementioned additive, with the generator integrating a 200 We polymer electrolyte membrane fuel cell ( Figure 13). Under operation, a hydrogen generation rate as high as 3.3 L min −1 was reached, being enough to operate the fuel cell for 25 min with 58 g of ammonia borane. A purifying system (i.e., acidic active carbons trapping borazine) allowed the feeding of the fuel cell with pure hydrogen. This demonstrator was the first one ever reported with ammonia borane as a hydrogen carrier. The addition of a chemical dopant results in the destabilization of ammonia borane, which releases hydrogen at temperatures below 100 °C and with faster kinetics. The destabilization is even, in some cases, too effective, leading to the ammonia borane being unstable at ambient temperature. The released hydrogen is purer than the hydrogen generated by pristine ammonia borane, but the chemical dopant does not seem to be efficient enough to suppress the formation of any gaseous byproduct. The purity of the released hydrogen is, however, not clearly specified, and is globally controversial. The main limitation is inherent to the destabilization strategy itself, which is based on grain-to-grain contacts. The most efficient destabilization could be expected from the smallest grains, namely, at the nanoscale. Another critical issue with the doping strategy concerns the residue recovered upon dehydrogenation. Globally, very little is known about the state of the chemical dopant, and the nature of the polymeric residue is rather complex, though it is claimed to be polyborazylene-like in some instances. This also leads to substantial uncertainties in how to separate the dopant from the polymeric residue and then recycle all of the products. Recycling is a critical issue. This is discussed in Section 5.6 for the polymeric residue, but it is important to notice here that the presence of B−O bonds in the polymeric residue makes the recycling more complicated. If this destabilization strategy should be further investigated, because it is the best for obtaining highly effective gravimetric hydrogen storage capacities (Figure 14), efforts should be focused on nanoengineering. The addition of a chemical dopant results in the destabilization of ammonia borane, which releases hydrogen at temperatures below 100 • C and with faster kinetics. The destabilization is even, in some cases, too effective, leading to the ammonia borane being unstable at ambient temperature. The released hydrogen is purer than the hydrogen generated by pristine ammonia borane, but the chemical dopant does not seem to be efficient enough to suppress the formation of any gaseous by-product. The purity of the released hydrogen is, however, not clearly specified, and is globally controversial. The main limitation is inherent to the destabilization strategy itself, which is based on grain-to-grain contacts. The most efficient destabilization could be expected from the smallest grains, namely, at the nanoscale. Another critical issue with the doping strategy concerns the residue recovered upon dehydrogenation. Globally, very little is known about the state of the chemical dopant, and the nature of the polymeric residue is rather complex, though it is claimed to be polyborazylene-like in some instances. This also leads to substantial uncertainties in how to separate the dopant from the polymeric residue and then recycle all of the products. Recycling is a critical issue. This is discussed in Section 5.6 for the polymeric residue, but it is important to notice here that the presence of B−O bonds in the polymeric residue makes the recycling more complicated. If this destabilization strategy should be further investigated, because it is the best for obtaining highly effective gravimetric hydrogen storage capacities (Figure 14), efforts should be focused on nano-engineering.

Nanoconfinement of Ammonia Borane
By infiltrating ammonia borane into the porosity of a scaffold, it is nanosized, and nanosized ammonia borane (also called ammonia borane nanophase) is able to show modified decomposition Figure 14. Gravimetric hydrogen storage capacity (GHSC) of ammonia borane (100 to 50% weight) mixed with a chemical dopant (0 to 50% in weight). With respect to the bars with no border, the capacity is based on 100% on the hydrogens of ammonia borane. For the bars with the white borders, it is assumed that 75% of the hydrogen of ammonia borane can be released. The capacities (in wt% H 2 ) are all given.

Nanoconfinement of Ammonia Borane
By infiltrating ammonia borane into the porosity of a scaffold, it is nanosized, and nanosized ammonia borane (also called ammonia borane nanophase) is able to show modified decomposition properties, with dehydrogenation being greatly favored. This is the nanoconfinement strategy [238][239][240][241].
The first work dealing with the nanoconfinement of ammonia borane used silica SBA-15 (specific surface area of 900 m 2 g −1 ) as a scaffold. Gutowska et al. [23] infiltrated by capillary action the mesoporosity of the silica SBA-15 using a saturated methanolic solution of ammonia borane. In doing so, AB@SBA-15, consisting in 50 wt% of ammonia borane and showing a specific surface area below 50 m 2 g −1 , was produced. AB@SBA-15 showed attractive dehydrogenation properties by releasing hydrogen from 50 • C. Moreover, the enthalpy of hydrogen release was measured as −1 kJ mol −1 , suggesting an almost neutral process. Comparable improved dehydrogenation properties were reported for mesoporous silica MCM-41 [242][243][244], silica hollow nanospheres [245] and silica aerogel [246]. Note that the infiltration of AB is feasible with another suitable solvent, i.e., tetrahydrofuran [247], but the most important aspect is to avoid the presence of ammonia borane outside the porosity, because non-confined ammonia borane (also called an excess of ammonia borane) forms aggregates outside the scaffold porosity and behaves like pristine ammonia borane [248,249]. With silica as the scaffold, the purity of the released H 2 is debated. It was found to be pure [250] or not (due to the formation of borazine and ammonia) [23,242].
There are three effects to explain the better dehydrogenation properties of the ammonia borane nanophase ( Figure 15). First, the reduction of the size of the ammonia borane particles results in more defect sites (i.e., active sites initiating the dehydrogenation process) [243,250] and in reduced diffusion distances (facilitating the release of hydrogen) [251]. Second, the hydrogens H δ+ of the surface SiO−H groups catalytically activate ammonia borane through dihydrogen O−H δ+ ···H δ− −B interactions [252], resulting in an acid-base reaction between SiO−H and BH 3 and the formation of surface SiO−B groups [245]: Porous carbon is also a potential scaffold of ammonia borane. Improved dehydrogenation properties for ammonia borane were reported for carbon cryogels [254,255], activated carbon [256,257], mesoporous carbon CMK-3 [258], microporous carbon [259] and N-containing carbon nanotubes [260]. The positive effects described for the silica scaffolds are valid for the carbon scaffolds ( Figure 15). First, Tang et al. [261] clearly showed that nanoparticles with a diameter of 4 nm were able to dehydrogenate starting from 50 °C, whereas larger nanoparticles (diameter of 10 nm) dehydrogenated from 67 °C. So et al. [262] showed that the dehydrogenation of ammonia borane Third, the ammonia borane nanophase is amorphous and, accordingly, it has an altered dihydrogen N−H δ+ ···H δ− −B network [253], and more mobile and active molecules [174]. Porous carbon is also a potential scaffold of ammonia borane. Improved dehydrogenation properties for ammonia borane were reported for carbon cryogels [254,255], activated carbon [256,257], mesoporous carbon CMK-3 [258], microporous carbon [259] and N-containing carbon nanotubes [260]. The positive effects described for the silica scaffolds are valid for the carbon scaffolds ( Figure 15). First, Tang et al. [261] clearly showed that nanoparticles with a diameter of 4 nm were able to dehydrogenate starting from 50 • C, whereas larger nanoparticles (diameter of 10 nm) dehydrogenated from 67 • C. So et al. [262] showed that the dehydrogenation of ammonia borane could be tuned by controlling the scaffold pore size in such a way that the narrower the pore size is, the lower the peak temperature of the H 2 release. In other words, microporosity (a pore size of <2 nm) should be favored [263]. Second, the surface COO−H (and/or CO−H) groups provide acidic hydrogens H δ+ that destabilize the ammonia borane molecules by acid-base reactions, even at temperatures as low as 3-4 • C [256], which indicates that this effect overpasses the nanosizing effect. This is interpreted as catalytic activation (i.e., acid catalysis) [264]. This was evidenced by the detection of CO−B and COO−B bonds in the materials upon dehydrogenation [265,266]. The catalytic activation, however, does not hinder the formation and release of ammonia [267].
Metal organic frameworks (MOFs) have also been designated as potential scaffolds of ammonia borane. The first ever reported for that purpose was water-free JUC-32-Y, made of unsaturated Y 3+ metal sites and 1,3,5-benzenetricarboxylate rigid linkers [268]. Ammonia borane was infiltrated by capillary action and the as-obtained nanophase released hydrogen from 50 • C, with the peak temperature of the dehydrogenation being 84 • C. Onset temperatures of dehydrogenation lower than 70 • C were reported for a series of MOFs, such as Mg-MOF-74 [269], Zn-MOF-74 [270], MIL-101 [271] and MOF-5 [272]. This was explained by the nanosizing effect [273]. Better yet, Chung et al. [274] showed a linear relationship between the temperature of hydrogen release and the reciprocal of the particle size of the ammonia borane nanophase (Figure 16). On the other hand, Jeong et al. [272] stressed on the importance of using MOFs with small pores. Geometric hindrance caused by the small pores was suggested to prevent the formation of borazine. In fact, it is believed that a large surface area and porosity are essential to avoid the formation of borazine [275]. Another effect, which is catalytic, was suggested [276][277][278][279]. The unsaturated metals of the MOFs are believed to act as Lewis acid sites, favoring the destabilization of ammonia borane through M α+ ···NH 3 −BH 3 interactions. For example, the formation of Fe−NH 2 groups was identified for ammonia borane confined into Fe-MIL-5 upon dehydrogenation [270]. With saturated metal sites, ammonia was produced because of a less efficient catalytic effect [271,274,275]. Another catalytic effect (Figure 15), due to the O-functional groups (belonging to the linkers) of the MOFs, was suggested to take place together with the previous catalytic effect. These groups interact with the BH 3 group of ammonia borane, thereby destabilizing the B−H and B−N bonds. Srinivas et al. [270,277] gave evidence of this, via the formation of B−O bonds upon dehydrogenation.
Another way to obtain a nanophase of ammonia borane is to blend it with a polymer. Poly(methyl acrylate) is a possible scaffold [280]. The as-obtained nanophase released hydrogen and some ammonia from 70 • C. Comparable behaviors were reported for polyacrylamide and polyvinylpyrrolidone [281,282]. With hyper-crosslinked porous polymers, ammonia borane is infiltrated by the capillary effect. For instance, Tang et al. [283] used poly(styrene-co-divinylbenzene) with a specific surface area of 757 m 2 g −1 and a pore volume of 0.88 cm 3 g −1 . The ammonia borane nanophase was in the form of 4-nm nanoparticles, and released pure hydrogen from 50 • C. A last synthetic approach is nanostructuration by electrospinning, where nanofibers of a polymer encapsulating ammonia borane are elaborated. Useful polymers for that purpose are polystyrene [284], polyvinylpyrrolidone [285], poly(methyl methacrylate) [286], polyethylene oxide [287] and polyacrylamide [288]. In each case, embedded ammonia borane nanophase showed improved dehydrogenation properties in comparison with the bulk counterpart, but the release of some ammonia, diborane and/or borazine could not be avoided. The first, obvious, effect explaining the better dehydrogenation properties is nanosizing (Figure 15). The other effect is related to the occurrence of C=O···BH 3 interactions between the polymer and the  Another way to obtain a nanophase of ammonia borane is to blend it with a polymer. Poly(methyl acrylate) is a possible scaffold [280]. The as-obtained nanophase released hydrogen and some ammonia from 70 °C. Comparable behaviors were reported for polyacrylamide and polyvinylpyrrolidone [281,282]. With hyper-crosslinked porous polymers, ammonia borane is infiltrated by the capillary effect. For instance, Tang et al. [283] used poly(styrene-co-divinylbenzene) with a specific surface area of 757 m 2 g −1 and a pore volume of 0.88 cm 3 g −1 . The ammonia borane nanophase was in the form of 4-nm nanoparticles, and released pure hydrogen from 50 °C. A last synthetic approach is nanostructuration by electrospinning, where nanofibers of a polymer encapsulating ammonia borane are elaborated. Useful polymers for that purpose are polystyrene [284], polyvinylpyrrolidone [285], poly(methyl methacrylate) [286], polyethylene oxide [287] and polyacrylamide [288]. In each case, embedded ammonia borane nanophase showed improved dehydrogenation properties in comparison with the bulk counterpart, but the release of some ammonia, diborane and/or borazine could not be avoided. The first, obvious, effect explaining the better dehydrogenation properties is nanosizing (Figure 15). The other effect is related to the occurrence of C=O⋅⋅⋅BH3 interactions between the polymer and the ammonia borane. These interactions weaken the B−H, N−H and B−N bonds of ammonia borane, thereby favoring the release of hydrogen but also driving the release of ammonia because of weakened B−N bonds [289,290].
There are few other examples of scaffolds. Boron nitride, in the form of porous hollow nanopolyhedra (a mean diameter of 70 nm, wall thickness of 5 nm and total pore volume of 0.424 cm 3 g −1 ), was found to effectively nanosize and destabilize ammonia borane from 40 °C [291]. The destabilization was due to a pure nanoconfinement effect, because the surface of the boron nitride scaffold was totally hydrogen-free. The release of some NH3 was, however, detected. Another example is manganese oxide. With porous hollow cubes (diameter of 400 nm, wall thickness of 40 nm and porous volume of 1.05 cm 3 g −1 ) [292] and porous hollow spheres (diameter of 1-2 µm nm, wall thickness of 80 nm and porous volume of 0.47 cm 3 g −1 ) [293], the ammonia borane nanophase released pure hydrogen from 60 °C due to the nanosizing effect.
The nanosizing effect has clearly shown to be successful. Pure hydrogen can be generated at very low temperatures, though the presence of gaseous by-products is still possible. However, overly low temperatures, like the ambient temperature, should be avoided for safety and technical reasons. Given the variety of the scaffolds reported so far, there is room for finding the optimal conditions for There are few other examples of scaffolds. Boron nitride, in the form of porous hollow nanopolyhedra (a mean diameter of 70 nm, wall thickness of 5 nm and total pore volume of 0.424 cm 3 g −1 ), was found to effectively nanosize and destabilize ammonia borane from 40 • C [291]. The destabilization was due to a pure nanoconfinement effect, because the surface of the boron nitride scaffold was totally hydrogen-free. The release of some NH 3 was, however, detected. Another example is manganese oxide. With porous hollow cubes (diameter of 400 nm, wall thickness of 40 nm and porous volume of 1.05 cm 3 g −1 ) [292] and porous hollow spheres (diameter of 1-2 µm nm, wall thickness of 80 nm and porous volume of 0.47 cm 3 g −1 ) [293], the ammonia borane nanophase released pure hydrogen from 60 • C due to the nanosizing effect.
The nanosizing effect has clearly shown to be successful. Pure hydrogen can be generated at very low temperatures, though the presence of gaseous by-products is still possible. However, overly low temperatures, like the ambient temperature, should be avoided for safety and technical reasons. Given the variety of the scaffolds reported so far, there is room for finding the optimal conditions for the release of pure hydrogen. In fact, the nanoconfinement strategy faces two salient issues. Ironically, the first issue is the scaffold. It has a weight and, because of it, there is a mass penalty, resulting in a substantial drop of the theoretical gravimetric hydrogen storage capacity of the pair scaffold-ammonia borane ( Figure 17). The second issue is related to the formation of highly stable B−O bonds upon the dehydrogenation of ammonia borane infiltrated into most of the scaffolds reported above. This is an added complication for the tricky recycling of the polymeric residue (cf. Section 5.6). Here, also, the lack of effort towards scaling up is to be regretted. the first issue is the scaffold. It has a weight and, because of it, there is a mass penalty, resulting in a substantial drop of the theoretical gravimetric hydrogen storage capacity of the pair scaffoldammonia borane (Figure 17). The second issue is related to the formation of highly stable B−O bonds upon the dehydrogenation of ammonia borane infiltrated into most of the scaffolds reported above. This is an added complication for the tricky recycling of the polymeric residue (cf. Section 5.6). Here, also, the lack of effort towards scaling up is to be regretted. Figure 17. Gravimetric hydrogen storage capacity (GHSC) of ammonia borane (20 to 50% in weight) confined into a scaffold (80 to 50% in weight). For comparison, pure ammonia borane (100% in weight) is also shown. With respect to the bars with no border, the capacity is based on 100% on the hydrogens of ammonia borane. For the bars with the white borders, it is assumed that 75% of the hydrogen of ammonia borane can be released. The capacities (in wt% H2) are all given.

Alkali and Alkaline-Earth Derivatives of Ammonia Borane
The last, but not the least, of the destabilization strategies is to make ammonia borane evolve towards a derivative. In a simple way, one of the protic hydrogens H δ+ of ammonia borane is substituted by a cation, like an alkali metal, leading to a new compound; an amidoborane. The first amidoborane, sodium amidoborane NaNH2BH3, was discovered in the 1930s by Schlesinger et al. [29,31]. A few others were discovered in the 1990s. Shore and co-workers studied some amidoboranes, including sodium, lithium and potassium salts [294][295][296]. Myers et al. [297] synthesized the lithium counterpart, LiNH2BH3. In recent history, relating to ammonia borane, the first derivative to emerge was an alkaline earth compound, namely calcium amidoborane Ca(NH2BH3)2 [298]. Since then, a number of amidoboranes have been discovered. It goes beyond the scope of the present work to survey all of the mono-and multi-metallic amidoboranes reported so far. All have been discussed at length in a series of reviews [299][300][301][302][303], and hereafter only the monometallic alkali and alkaline-earth amidoboranes are tackled.
Lithium amidoborane LiNH2BH3 is easily prepared by, for example, ball-milling an equimolar mixture of ammonia borane and lithium hydride LiH [304]: It is worth briefly mentioning that there are alternative routes for preparing alkali amidoboranes, as surveyed in details in refs. [299][300][301][302][303]. Lithium amidoborane is a crystalline solid, having two allotropes, the α and the β phases; both crystallize in the orthorhombic space group Pbca, but the unit cell of the β phase is twice that of the α phase [305]. Lithium amidoborane has a high gravimetric hydrogen density, with 13.7 wt% H2, and this makes it attractive in the field. Under heating, it dehydrogenates according to a nearly thermally neutral event (−3 kJ mol −1 ), and at temperatures below 100 °C [306]. The purity of the released hydrogen is not clear. It was claimed to be either pure Figure 17. Gravimetric hydrogen storage capacity (GHSC) of ammonia borane (20 to 50% in weight) confined into a scaffold (80 to 50% in weight). For comparison, pure ammonia borane (100% in weight) is also shown. With respect to the bars with no border, the capacity is based on 100% on the hydrogens of ammonia borane. For the bars with the white borders, it is assumed that 75% of the hydrogen of ammonia borane can be released. The capacities (in wt% H 2 ) are all given.

Alkali and Alkaline-Earth Derivatives of Ammonia Borane
The last, but not the least, of the destabilization strategies is to make ammonia borane evolve towards a derivative. In a simple way, one of the protic hydrogens H δ+ of ammonia borane is substituted by a cation, like an alkali metal, leading to a new compound; an amidoborane. The first amidoborane, sodium amidoborane NaNH 2 BH 3 , was discovered in the 1930s by Schlesinger et al. [29,31]. A few others were discovered in the 1990s. Shore and co-workers studied some amidoboranes, including sodium, lithium and potassium salts [294][295][296]. Myers et al. [297] synthesized the lithium counterpart, LiNH 2 BH 3 . In recent history, relating to ammonia borane, the first derivative to emerge was an alkaline earth compound, namely calcium amidoborane Ca(NH 2 BH 3 ) 2 [298]. Since then, a number of amidoboranes have been discovered. It goes beyond the scope of the present work to survey all of the mono-and multi-metallic amidoboranes reported so far. All have been discussed at length in a series of reviews [299][300][301][302][303], and hereafter only the monometallic alkali and alkaline-earth amidoboranes are tackled.
Lithium amidoborane LiNH 2 BH 3 is easily prepared by, for example, ball-milling an equimolar mixture of ammonia borane and lithium hydride LiH [304]: It is worth briefly mentioning that there are alternative routes for preparing alkali amidoboranes, as surveyed in details in refs. [299][300][301][302][303]. Lithium amidoborane is a crystalline solid, having two allotropes, the α and the β phases; both crystallize in the orthorhombic space group Pbca, but the unit cell of the β phase is twice that of the α phase [305]. Lithium amidoborane has a high gravimetric hydrogen density, with 13.7 wt% H 2 , and this makes it attractive in the field. Under heating, it dehydrogenates according to a nearly thermally neutral event (−3 kJ mol −1 ), and at temperatures below 100 • C [306]. The purity of the released hydrogen is not clear. It was claimed to be either pure [307] or polluted with some ammonia [308]. With sodium amidoborane NaNH 2 BH 3 (orthorhombic, with a Pbca space group), the high weight of sodium results in a lower gravimetric hydrogen density, with 9.6 wt% H 2 . It can be prepared by gentle ball-milling [309], or even by a simple mixing in an inert atmosphere [310], because sodium hydride is more reactive than lithium hydride [311]. Under heating, at 80 • C, it was found to release 0.5 equivalents of hydrogen in about 1.5 min, whereas a time of about 50 min was required for the lithium compound [307]. Traces of ammonia were, however, detected [312].
The synthesis of potassium amidoborane requires safer vessel (a stainless steel autoclave), and a solvent like tetrahydrofuran, because potassium hydride, a base stronger than the aforementioned metal hydrides, reacts violently with ammonia borane [313]. Potassium amidoborane crystallizes in the orthorhombic space group Pbca. It carries 7.3 wt% of hydrogen. It is less stable than the parent ammonia borane, releasing 1.5 equivalents of pure hydrogen up to 100 • C. Note that it cannot be compared to the previous alkali amidoboranes because of discrepancies in the conditions of the dehydrogenation experiments. Though less attractive in terms of gravimetric hydrogen density, rubidium amidoborane RbNH 2 BH 3 (4.4 wt% H 2 ) and cesium amidoborane CsNH 2 BH 3 (3.1 wt% H 2 ) were synthesized by reacting the metal with ammonia borane in anhydrous tetrahydrofuran; for example [314]: Both are crystalline [315]. The former crystallizes in a monoclinic system, with the space group P2 1 /c. The latter crystallizes in a primitive orthorhombic structure and the space group Pnam. They decompose over the temperature range of 55-100 • C while liberating hydrogen and ammonia. In sum, the most attractive amidoboranes for solid-state hydrogen storage are the lithium and sodium compounds ( Figure 18); they carry more hydrogen and are able to release quite pure hydrogen below 100 • C [302]. [307] or polluted with some ammonia [308]. With sodium amidoborane NaNH2BH3 (orthorhombic, with a Pbca space group), the high weight of sodium results in a lower gravimetric hydrogen density, with 9.6 wt% H2. It can be prepared by gentle ball-milling [309], or even by a simple mixing in an inert atmosphere [310], because sodium hydride is more reactive than lithium hydride [311]. Under heating, at 80 °C, it was found to release 0.5 equivalents of hydrogen in about 1.5 min, whereas a time of about 50 min was required for the lithium compound [307]. Traces of ammonia were, however, detected [312]. The synthesis of potassium amidoborane requires safer vessel (a stainless steel autoclave), and a solvent like tetrahydrofuran, because potassium hydride, a base stronger than the aforementioned metal hydrides, reacts violently with ammonia borane [313]. Potassium amidoborane crystallizes in the orthorhombic space group Pbca. It carries 7.3 wt% of hydrogen. It is less stable than the parent ammonia borane, releasing 1.5 equivalents of pure hydrogen up to 100 °C. Note that it cannot be compared to the previous alkali amidoboranes because of discrepancies in the conditions of the dehydrogenation experiments. Though less attractive in terms of gravimetric hydrogen density, rubidium amidoborane RbNH2BH3 (4.4 wt% H2) and cesium amidoborane CsNH2BH3 (3.1 wt% H2) were synthesized by reacting the metal with ammonia borane in anhydrous tetrahydrofuran; for example [314]: Both are crystalline [315]. The former crystallizes in a monoclinic system, with the space group P21/c. The latter crystallizes in a primitive orthorhombic structure and the space group Pnam. They decompose over the temperature range of 55-100 °C while liberating hydrogen and ammonia. In sum, the most attractive amidoboranes for solid-state hydrogen storage are the lithium and sodium compounds ( Figure 18); they carry more hydrogen and are able to release quite pure hydrogen below 100 °C [302]. The better dehydrogenation properties of the alkali amidoboranes (versus ammonia borane) can be explained in two ways. First, the compound becomes ionic, and substantial changes occur in chemical bonding by substituting one protic hydrogen H δ+ with a metal cation M + [316]: the M−N The better dehydrogenation properties of the alkali amidoboranes (versus ammonia borane) can be explained in two ways. First, the compound becomes ionic, and substantial changes occur in chemical bonding by substituting one protic hydrogen H δ+ with a metal cation M + [316]: the M−N bond is longer than the H−N bond in ammonia borane, and the B−N bond is slightly shorter, with, for example, 1.53 Å for sodium amidoborane and 1.6 Å for ammonia borane. This has implications ( Figure 19). The B−N bond is then stronger, the NH 2 group is stronger in terms of Lewis basicity and the hydridic hydrogen H δ− , becoming more basic, is more reactive than its counterpart in ammonia borane [317]. Second, the intermolecular dihydrogen N−H δ+ ···H δ− −B interactions are weakened with the presence Energies 2020, 13, 3071 24 of 45 of the metal cation [313]; that is, the H δ+ ···H δ− distances are longer. A distance of 2.25 Å was suggested for lithium amidoborane, for example [317]. This is longer than the distance 2.02 Å in ammonia borane, but still shorter than the sum of the van der Waals radii for two H atoms (2.4 Å). This is an evidence of weaker dihydrogen N−H δ+ ···H δ− −B interactions [314]. The consequence of these two effects is a lower stability of the amidoboranes in comparison with ammonia borane. If this destabilization strategy is compared to the ones discussed above, it is the simplest; there is no solvent, no catalyst, no chemical dopant and no scaffold. The amidoborane only needs to be heated up so that it dehydrogenates below 100 °C. Against this background, one can only regret that there has been no scaling up. Data collected in more realistic conditions, from a technological point of view, would have allowed a more technical discussion.

The Critical Issue of Regeneration
In comparison with the efforts made to dehydrogenate ammonia borane, there has been very little effort made to hydrogenate the polymeric residues, those recovered upon thermolytic dehydrogenation [186,324].
Hausdorf et al. [325] used a polymeric residue denoted BNH<2, which was prepared from ammonia borane heated at 90 °C for 12 h and then 180 °C for 2 h. The polymeric residue was first digested in hydrochloric acid (Figure 20), the product ammonium chloride NH4Cl was recovered to produce ammonia, the product boron trichloride BCl3 was hydrodechlorinated to form diborane B2H6 and finally the as-obtained ammonia and diborane were reacted to form ammonia borane. Instead of hydrochloric acid, the superacid systems AlBr3/HBr/CS2 and AlCl3/HCl/toluene were also used, with the regeneration leading to ammonia borane, with a yield of 60% [326]. An alternative path was suggested by Davis et al. [327]. Using polyborazylene as a model polymeric residue, a digestionreduction path using a thiol and a tin hydride was suggested. The B−N bonds of the polymeric residue were broken by the action of the thiol, and the as-formed intermediates were reduced in the presence of an excess tin hydride, allowing the formation of ammonia borane with a yield of 67%. These two regeneration schemes were later combined by Tan et al. [328]. They proposed a derived method based on the use of the superacid system AlCl3/HCl/CS2 for the digestion step and a tin hydride for the reduction step. An improved yield of 89% was calculated. With alkaline earth metal cations, the positive charge +2 is counterbalanced by two anions [NH 2 BH 3 ] − . Because of the acute toxicity of beryllium, beryllium amidoborane Be(NH 2 BH 3 ) 2 was investigated virtually, by computational calculations [318]. This is unfortunate, as beryllium amidoborane would have a theoretical gravimetric hydrogen density of 14.7 wt% H 2 . The synthesis of magnesium amidoborane Mg(NH 2 BH 3 ) 2 was awkward. Both magnesium hydride and magnesium, as reactants, seemed to be inert towards ammonia borane, which was explained for the former reactant by a weaker Lewis basicity [319]. Luo et al. [320] found the solution to such inertness. To obtain magnesium amidoborane (12 wt% H 2 ), the mixture of magnesium hydride (or magnesium) and ammonia borane was aged under an argon atmosphere: The reaction required 45 days with magnesium hydride (6 months with magnesium) at ambient temperature. The as-obtained amidoborane (monoclinic, space group C2) is not more attractive than ammonia borane for solid-state hydrogen storage. It dehydrogenates between 100 and 300 • C, through a complex multistep process, resulting in the release of 4.2 equivalents of hydrogen, the formation of polyborazylene-like residues and the presence of a small amount of magnesium hydride. The calcium derivative, Ca(NH 2 BH 3 ) 2 , which requires an intense milling to form and isolate a pure phase, is quite challenging. The reactants, calcium hydride CaH 2 and ammonia borane, are still present as minor phases [321]. Calcium amidoborane crystallizes in a monoclinic C2 unit cell, and the B−N bond (1.546 Å) is shorter than that of ammonia borane, suggesting more Lewis basic hydridic hydrogens H δ− . Calcium amidoborane (10.1 wt% H 2 ) is less stable, as it releases hydrogen from 80 • C. About four equivalents of hydrogen are formed over the temperature range 80-250 • C. Strontium amidoborane Sr(NH 2 BH 3 ) 2 (6.9 wt% H 2 ) was reported to form by making pre-milled strontium hydride SrH 2 and two equivalents of pre-milled ammonia borane react, the process being performed at 45 • C for 2 h [322]. The final product contained 55 wt% of crystalline strontium amidoborane (isostructural to calcium amidoborane), 34 wt% of an unidentified phase, 7 wt% of strontium hydride and 4 wt% of ammonia borane. As the last representative, amorphous barium amidoborane Ba(NH 2 BH 3 ) was obtained by the reaction of metallic barium with ammonia borane in tetrahydrofuran at −10 • C [323].
Its dehydrogenation properties are comparable to those of calcium amidoborane. It was found to release hydrogen (5 wt% out of a theoretical capacity of 5.1 wt%) from 81 to 163 • C. To sum up, alkaline earth amidoboranes are less attractive than their alkali counterparts because they are more difficult to synthesize, they have lower gravimetric hydrogen densities ( Figure 18) and they are more thermally stable.
If this destabilization strategy is compared to the ones discussed above, it is the simplest; there is no solvent, no catalyst, no chemical dopant and no scaffold. The amidoborane only needs to be heated up so that it dehydrogenates below 100 • C. Against this background, one can only regret that there has been no scaling up. Data collected in more realistic conditions, from a technological point of view, would have allowed a more technical discussion.

The Critical Issue of Regeneration
In comparison with the efforts made to dehydrogenate ammonia borane, there has been very little effort made to hydrogenate the polymeric residues, those recovered upon thermolytic dehydrogenation [186,324].
Hausdorf et al. [325] used a polymeric residue denoted BNH <2 , which was prepared from ammonia borane heated at 90 • C for 12 h and then 180 • C for 2 h. The polymeric residue was first digested in hydrochloric acid (Figure 20), the product ammonium chloride NH 4 Cl was recovered to produce ammonia, the product boron trichloride BCl 3 was hydrodechlorinated to form diborane B 2 H 6 and finally the as-obtained ammonia and diborane were reacted to form ammonia borane. Instead of hydrochloric acid, the superacid systems AlBr 3 /HBr/CS 2 and AlCl 3 /HCl/toluene were also used, with the regeneration leading to ammonia borane, with a yield of 60% [326]. An alternative path was suggested by Davis et al. [327]. Using polyborazylene as a model polymeric residue, a digestion-reduction path using a thiol and a tin hydride was suggested. The B−N bonds of the polymeric residue were broken by the action of the thiol, and the as-formed intermediates were reduced in the presence of an excess tin hydride, allowing the formation of ammonia borane with a yield of 67%. These two regeneration schemes were later combined by Tan et al. [328]. They proposed a derived method based on the use of the superacid system AlCl 3 /HCl/CS 2 for the digestion step and a tin hydride for the reduction step. An improved yield of 89% was calculated.
Energies 2020, 13, x FOR PEER REVIEW 26 of 45 Figure 20. Regeneration of ammonia borane by acidic digestion of the solid residue BNH<2, followed by hydrodechlorination of boron trichloride and decomposition of ammonium chloride into ammonia, and finally a Lewis acid-base reaction between diborane and ammonia. Adapted from ref. [325].
Another regeneration strategy was proposed by Sutton et al. [329]. Polyborazylene was used as model polymeric residue and, through a one-pot process (Figure 21), ammonia borane was regenerated with a yield higher than 90%. This is a simple process, where the polymer is suspended in liquid ammonia at −77 °C and then reduced by hydrazine at 40 °C for 24 h. According to Hua and Ahluwalia [330], who analyzed these processes in terms of energy consumption, efficiency and life cycle greenhouse gas emissions, none are energy efficient. Figure 20. Regeneration of ammonia borane by acidic digestion of the solid residue BNH <2 , followed by hydrodechlorination of boron trichloride and decomposition of ammonium chloride into ammonia, and finally a Lewis acid-base reaction between diborane and ammonia. Adapted from ref. [325].
Another regeneration strategy was proposed by Sutton et al. [329]. Polyborazylene was used as model polymeric residue and, through a one-pot process (Figure 21), ammonia borane was regenerated with a yield higher than 90%. This is a simple process, where the polymer is suspended in liquid ammonia at −77 • C and then reduced by hydrazine at 40 • C for 24 h. According to Hua and Ahluwalia [330], who analyzed these processes in terms of energy consumption, efficiency and life cycle greenhouse gas emissions, none are energy efficient.
Another regeneration strategy was proposed by Sutton et al. [329]. Polyborazylene was used as model polymeric residue and, through a one-pot process (Figure 21), ammonia borane was regenerated with a yield higher than 90%. This is a simple process, where the polymer is suspended in liquid ammonia at −77 °C and then reduced by hydrazine at 40 °C for 24 h. According to Hua and Ahluwalia [330], who analyzed these processes in terms of energy consumption, efficiency and life cycle greenhouse gas emissions, none are energy efficient. As for ammonia borane, the dehydrogenation of the amidoboranes is exothermic, and direct rehydrogenation is thermodynamically impossible at affordable conditions. Attempts up to 120 bars of hydrogen and 300 °C were found to be ineffective [309,317,319]. The only feasible approach is chemical reduction, as discussed above. Tang et al. [331] successfully applied the process using hydrazine as a reducing agent and ammonia as a solvent ( Figure 21) to a polymeric residue recovered upon the dehydrogenation of lithium amidoborane. The starting ammonia borane was regenerated with a yield of 63%. This approach, like the ones discussed above, cannot be applied for the reduction of B−O bonds into B−H bonds.
It is worth mentioning another approach that could be considered for any polymeric residue, whether they contain B−O bonds or not. Tang et al. [332] suggested a digestion of the polymeric residue into methanol, resulting in the methanolysis of the polymeric residue and formation of trimethoxyborate B(OCH3)3. Then, the authors applied the ammonia borane one-pot synthesis procedure developed by Ramachandran et al. [168]. The trimethoxyborate was reduced by lithium alanate in the presence of ammonium chloride acting as ammonia source (Equation (18) and Figure   Figure 21. Regeneration of ammonia borane by reduction of polyborazylene (PB) by hydrazine in ammonia. Adapted from ref. [329].
As for ammonia borane, the dehydrogenation of the amidoboranes is exothermic, and direct rehydrogenation is thermodynamically impossible at affordable conditions. Attempts up to 120 bars of hydrogen and 300 • C were found to be ineffective [309,317,319]. The only feasible approach is chemical reduction, as discussed above. Tang et al. [331] successfully applied the process using hydrazine as a reducing agent and ammonia as a solvent ( Figure 21) to a polymeric residue recovered upon the dehydrogenation of lithium amidoborane. The starting ammonia borane was regenerated with a yield of 63%. This approach, like the ones discussed above, cannot be applied for the reduction of B−O bonds into B−H bonds.
It is worth mentioning another approach that could be considered for any polymeric residue, whether they contain B−O bonds or not. Tang et al. [332] suggested a digestion of the polymeric residue into methanol, resulting in the methanolysis of the polymeric residue and formation of trimethoxyborate B(OCH 3 ) 3 . Then, the authors applied the ammonia borane one-pot synthesis procedure developed by Ramachandran et al. [168]. The trimethoxyborate was reduced by lithium alanate in the presence of ammonium chloride acting as ammonia source (Equation (18) and Figure 7). The as-obtained ammonia borane can then be reacted with lithium hydride to form lithium amidoborane, for example (Equation (28)). The energy efficiency of this process was calculated to be about 46%, and the authors concluded that the process is not effective enough.

Conclusions and Outlook
The main feature of ammonia borane is that it is versatile. Owing to the hydridic hydrogens H δ− it is made of, it can be used as an anodic fuel, a hydrolytic hydrogen carrier and a solid-state hydrogen storage material. Re-emerging in the mid-2000s, it demonstrated its potential for each of these applications, raising wide interest, as well as questions. The title of an article authored by Marder in 2007 is explicit. The author wondered whether we will soon be fueling automobiles with ammonia borane [333]. At that time, ammonia borane seemed to be full of promise-the automobile application, which is the most challenging, was targeted-but it was as if ammonia borane offered no guarantee. After 15 years of research, we may be pessimistic on the technological future of ammonia borane. The fact is, ammonia borane has barely been scaled up, which is a failure in terms of technological implementation.
The reality is that preventable errors have been made with ammonia borane. One of them has been mentioned just above. The technological readiness level of ammonia borane has remained low ( Figure 22), mainly at the laboratory and milligram scales. There should have been more actions and efforts towards scaling up, i.e., towards the construction and operation of demonstrators and prototypes. This would have helped researchers gain more insight into the problems facing its applications as an anodic fuel and hydrogen carrier. It is obvious that the technological future of ammonia borane, if any, will be promoted through scaling up. The ammonia borane nanophase, on the one hand, and lithium or sodium amidoborane, on the other, are mature enough to be investigated at the scale of a few dozen grams. Inspiration in this regard should be taken from the demonstrator based on 58 g of glycol-doped ammonia borane providing enough hydrogen to power a 200-We fuel cell [237]. Inspiration should be also taken from the few prototypes and devices using sodium borohydride as a hydrolytic hydrogen carrier [21]. It is likely that scaling up of ammonia borane went beyond the aforementioned demonstrator. Unfortunately, there has been very little communication. For instance, the company Cella Energy Ltd. (insolvent in 2018-2019 [334] and now bankrupt) developed plastic pellets consisting of ammonia borane enclosed in a polymeric shell [284], and briefly communicated about a successful unmanned aerial vehicle test flight [335], the vehicle being powered by a power system based on the ammonia borane-based pellets.
of technological implementation.
The reality is that preventable errors have been made with ammonia borane. One of them has been mentioned just above. The technological readiness level of ammonia borane has remained low ( Figure 22), mainly at the laboratory and milligram scales. There should have been more actions and efforts towards scaling up, i.e., towards the construction and operation of demonstrators and prototypes. This would have helped researchers gain more insight into the problems facing its applications as an anodic fuel and hydrogen carrier. It is obvious that the technological future of ammonia borane, if any, will be promoted through scaling up. The ammonia borane nanophase, on the one hand, and lithium or sodium amidoborane, on the other, are mature enough to be investigated at the scale of a few dozen grams. Inspiration in this regard should be taken from the demonstrator based on 58 g of glycol-doped ammonia borane providing enough hydrogen to power a 200-We fuel cell [237]. Inspiration should be also taken from the few prototypes and devices using sodium borohydride as a hydrolytic hydrogen carrier [21]. It is likely that scaling up of ammonia borane went beyond the aforementioned demonstrator. Unfortunately, there has been very little communication. For instance, the company Cella Energy Ltd. (insolvent in 2018-2019 [334] and now bankrupt) developed plastic pellets consisting of ammonia borane enclosed in a polymeric shell [284], and briefly communicated about a successful unmanned aerial vehicle test flight [335], the vehicle being powered by a power system based on the ammonia borane-based pellets. Figure 22. Estimation of the technological readiness level of ammonia borane as an anodic fuel, a liquid-state hydrogen carrier and a solid hydrogen carrier. The prototypes reported in ref. [237] and in ref. [335] have been taken into account to take the chemical doping and confinement strategies to a higher level than the other approaches. Figure 22. Estimation of the technological readiness level of ammonia borane as an anodic fuel, a liquid-state hydrogen carrier and a solid hydrogen carrier. The prototypes reported in ref. [237] and in ref. [335] have been taken into account to take the chemical doping and confinement strategies to a higher level than the other approaches.
Another preventable error has been to underestimate critical issues. The example of ammonia borane hydrolysis is characteristic and indicative. This reaction (Equations (6) and (7)) can be divided into five aspects (the optimization of the couple ammonia borane-water, catalyst and catalysis, purity of hydrogen due to the release of ammonia, nature of the borates and regeneration of ammonia borane) that are equally important. However, one of them (catalyst and catalysis) has been disproportionately scrutinized, and the others have regrettably been understudied. Consequently, it is difficult at present to argue the hydrolysis abilities in terms of effective gravimetric hydrogen storage capacities, purity of the released hydrogen, number and nature of the by-products and regeneration of ammonia borane. The same error, viz. underestimating critical issues, can be observed in the electro-oxidation, methanolysis and thermolysis of ammonia borane. An example for each reaction is as follows. In electro-oxidation, the impact of ammonia on, for example, the electrocatalyst and the membrane has not been investigated thoroughly. In methanolysis, it is not clear whether there is one borate forming upon reaction or if there are several in equilibrium. In thermolysis, the effective gravimetric hydrogen storage capacities have not been optimized and properly determined yet. Those all affect the technological maturity of ammonia borane.
We may well have overestimated the potential of ammonia borane. For instance, a mere 15 years ago, we were wondering whether ammonia borane would be a fuel for automobiles [333]. Today, it is obvious that ammonia borane is far from having demonstrated such potential. Three concrete examples illustrate the error that has been made. First, the gravimetric hydrogen density 19.5 wt% H 2 of ammonia borane was put forward. This value is far from the net, i.e., real and useable, gravimetric hydrogen storage capacities; in such a case, the storage system as a whole is taken into account; that is, the system weight (accounting for 50 wt% in a first approximation) is considered for the calculations of the net capacities ( Figure 23). The weight of the system is then seen as a 'dead mass', negatively impacting the storage capacities. For instance, in the hydrolysis of ammonia borane, the highest net capacity would be 3.5 wt% H 2 . For ammonia borane doped with 20 wt% of a chemical additive, the net capacity could be 5.9 wt% H 2 . In the thermolysis of nanoconfined ammonia borane, the highest net capacity would be 3.7 wt% H 2 . With lithium amidoborane dehydrogenating at an extent of 75%, the net capacity would reach 5.2 wt% H 2 . It is worth mentioning that an ultimate net capacity of 6.5 wt% H 2 is expected for light-duty vehicles [336]. Second, it is not clear whether the hydrogen release kinetics (in terms of rate and/or variability) are suitable for automotive applications. This question is also relevant for other applications (mobile, portable and stationary). There are some answers in hydrolysis and very few in electro-oxidation and thermolysis. Third, ammonia borane was, unintentionally and/or unmindfully, addressed as gasoline. Consider the analogy. A lot of the efforts have focused on improving and optimizing the conversion of both fuels, whereas very little consideration was given to the by-products (borates and polymeric residues, versus carbon dioxide). The analogy breaks down at this point. Ammonia borane cannot be compared with gasoline. The chemistry behind it, the issues and the challenges ahead and the absolute necessity to close the hydrogen cycle, among others, make ammonia borane radically and conceptually different from gasoline. It is hoped that ammonia borane will, in the near-future, contribute to powering electrical devices. It is unlikely, however, that it supersedes gasoline.
Energies 2020, 13, x FOR PEER REVIEW 29 of 45 Figure 23. Net (i.e., useable) gravimetric hydrogen storage capacity (GHSC, in wt% H2) for a storage system taken as a whole, in the case of (a) ammonia borane as liquid-state hydrogen carrier in water (in stoichiometric conditions such as in Equation (6)), (b) ammonia borane as liquid-state hydrogen carrier in methanol (in stoichiometric conditions such as in Equation (17)), (c) ammonia borane in saturation in the BmimCl ionic liquid (assuming the release of 75% of the borane's hydrogen), (d) 80 wt% of ammonia borane doped with 20 wt% of a chemical additive (assuming the release of 75% of ammonia borane's hydrogen), (e) ammonia borane confined into a scaffold (50-50 wt%; assuming the release of 75% of the ammonia borane's hydrogen) and (f) lithium amidoborane as derivative of ammonia borane (assuming the release of 75% of the amidoborane's hydrogen). The weight percentages of the lost hydrogen atoms, of the other atoms (e.g., B, N, C and O) and of the storage system (set at 50 wt% of the total weight) are shown.
As a final point, the competition is high. Ammonia borane has, in front of it, promising and cheaper hydrides. The cost issue has not been tackled above, appearing inappropriate because of discrepancies in many aspects, but it can be briefly discussed for comparing one material with another. For instance, ammonia borane is commonly synthesized from sodium borohydride. It is thus more costly. As long as ammonia borane is not regenerated from its own by-products, it will remain more costly than sodium borohydride, for which regeneration paths exist. It is therefore obvious that, Figure 23. Net (i.e., useable) gravimetric hydrogen storage capacity (GHSC, in wt% H 2 ) for a storage system taken as a whole, in the case of (a) ammonia borane as liquid-state hydrogen carrier in water (in stoichiometric conditions such as in Equation (6)), (b) ammonia borane as liquid-state hydrogen carrier in methanol (in stoichiometric conditions such as in Equation (17)), (c) ammonia borane in saturation in the BmimCl ionic liquid (assuming the release of 75% of the borane's hydrogen), (d) 80 wt% of ammonia borane doped with 20 wt% of a chemical additive (assuming the release of 75% of ammonia borane's hydrogen), (e) ammonia borane confined into a scaffold (50-50 wt%; assuming the release of 75% of the ammonia borane's hydrogen) and (f) lithium amidoborane as derivative of ammonia borane (assuming the release of 75% of the amidoborane's hydrogen). The weight percentages of the lost hydrogen atoms, of the other atoms (e.g., B, N, C and O) and of the storage system (set at 50 wt% of the total weight) are shown.
As a final point, the competition is high. Ammonia borane has, in front of it, promising and cheaper hydrides. The cost issue has not been tackled above, appearing inappropriate because of discrepancies in many aspects, but it can be briefly discussed for comparing one material with another. For instance, ammonia borane is commonly synthesized from sodium borohydride. It is thus more costly. As long as ammonia borane is not regenerated from its own by-products, it will remain more costly than sodium borohydride, for which regeneration paths exist. It is therefore obvious that, for the same level of performance, the sodium borohydride-based technology will be cheaper. The problem is that ammonia borane performs less well than sodium borohydride for the electro-oxidation, hydrolysis and methanolysis reactions. For example, the specific energy and pure compound capacity of sodium borohydride are higher. The hydrogen released by hydrolysis and methanolysis of sodium borohydride is purer, and free of ammonia. With respect to ammonia borane as a solid-state hydrogen carrier, the situation is different. It is certainly the best material in many aspects, such as the temperature of dehydrogenation and the amount of hydrogen that can be released. However, there is one aspect that is detrimental to ammonia borane. The dehydrogenation is exothermic, and thus the storage of hydrogen is irreversible. In other words, the polymeric residue stemming from ammonia borane cannot be hydrogenated under the pressure of hydrogen, which is a limitation. Accordingly, the question may legitimately be asked whether the main critical impediment to ammonia borane development is not the fact that it stores hydrogen irreversibly.
Ammonia borane has failed to play an important role in the deploying hydrogen economy. More optimistically, it has lagged behind. The absolute requirement to regenerate it, from its oxidation, hydrolysis, methanolysis and thermolysis products, has been and remains its Achilles heel. That is where more efforts should be expended, in order to go through significant achievements and breakthroughs. Without an industry dedicated to ammonia borane regeneration, ammonia borane will have no large-scale commercial future. Concomitantly, more efforts should be made in scaling up, in constructing and operating demonstrators and prototypes. The technological readiness level should then rise. Ammonia borane should get outside the laboratory hood and be studied in near-real and real conditions.