Boron Hydrogen Compounds: Hydrogen Storage and Battery Applications

About 25 years ago, Bogdanovic and Schwickardi (B. Bogdanovic, M. Schwickardi: J. Alloys Compd. 1–9, 253 (1997) discovered the catalyzed release of hydrogen from NaAlH4. This discovery stimulated a vast research effort on light hydrides as hydrogen storage materials, in particular boron hydrogen compounds. Mg(BH4)2, with a hydrogen content of 14.9 wt %, has been extensively studied, and recent results shed new light on intermediate species formed during dehydrogenation. The chemistry of B3H8−, which is an important intermediate between BH4− and B12H122−, is presented in detail. The discovery of high ionic conductivity in the high-temperature phases of LiBH4 and Na2B12H12 opened a new research direction. The high chemical and electrochemical stability of closo-hydroborates has stimulated new research for their applications in batteries. Very recently, an all-solid-state 4 V Na battery prototype using a Na4(CB11H12)2(B12H12) solid electrolyte has been demonstrated. In this review, we present the current knowledge of possible reaction pathways involved in the successive hydrogen release reactions from BH4− to B12H122−, and a discussion of relevant necessary properties for high-ionic-conduction materials.


Introduction
Boron hydrogen compounds have been intensively studied for almost a century since the pioneering studies of A. Stock [1]. Boron hydrogen compounds are also energetic materials and were considered as rocket or jet fuels [2]; however, the toxicity of boranes has prevented their extended use. Currently, nontoxic compounds such as ammonia-borane are also studied as hypergolic propellants [3,4]. Recently, many different applications of boron hydrogen compounds have emerged [5]. In particular, compounds derived from closo-hydroborates such as B 12 H 12 2− have found many new applications, including new all-solid-state batteries, medical applications, and as catalysts [6][7][8][9][10][11]. Since the discovery of catalyzed hydrogen release in NaAlH 4 by Bogdanovic and Schwickardi [12], light boron and aluminum hydrides were intensively studied and reviewed as potential hydrogen storage materials [13][14][15][16][17][18][19][20][21][22]. The dehydrogenation reactions of metal borohydrides ultimately lead to hydrogen, metal and boron, or metal borides. In this reaction process, intermediate species are formed, particularly compounds with closo-hydroborate anion B 12 H 12 2− [23,24]. B 12 H 12 2− is particularly stable and can thus also act as a detrimental thermodynamic sink for further dehydrogenation reactions. The properties of closo-hydroborates and related anions were addressed in several recent publications [6,[25][26][27][28]. New research on the thermal properties of closo-hydroborate salts revealed a high-temperature phase transition in Na 2 B 12 H 12 leading to a superionic phase [29]. Thus, the controlled dehydrogenation of a borohydride salt can be used to safely prepare new closoand nidohydroborate salts for potential battery applications [30] without using toxic boranes such as B 10 H 14 , which were used for the synthesis of this large boron species [31].
In this review, we first describe experimental results on hydrogen storage in Mg(BH 4 ) 2 , which has a large hydrogen content of 14.9 wt %. Hydrogen storage in other borohydrides, such as LiBH 4 , was recently reviewed [32]. Recent results on potential dehydrogenation In this review, we first describe experimental results on hydrogen storage in Mg(BH4)2, which has a large hydrogen content of 14.9 wt %. Hydrogen storage in other borohydrides, such as LiBH4, was recently reviewed [32]. Recent results on potential dehydrogenation intermediates for Mg(BH4)2 provide new insights on the potential reaction intermediates and are reported here. In this context, we then present recent results based on DFT calculations to explore possible reaction paths for successive dehydrogenation reactions starting from BH4 − . These paths are described in more detail in the following section, which discusses the formation and reactions of B3H8 − , as this ion is considered to be one of the reaction intermediates during the dehydrogenation of borohydride compounds. The high-temperature dehydrogenation of B3H8 − leads to the formation of closohydroborate anions B10H10 2− and B12H12 2− , which form excellent solid ionic conductors for new all-solid-state batteries [30]. The properties of these ionic conductors are presented in the last section.

Magnesium Borohydride
Among the many compounds considered for hydrogen storage, Mg(BH4)2 is particularly interesting and has been studied by many authors. The earlier studies on Mg(BH4)2 were reviewed in detail in 2016 [22]. Mg(BH4)2 has a hydrogen content of 14.9 mass % [22,33]. This compound can be prepared in different crystalline modifications, and high pressure-phase transitions were also observed [33]. Porous γ-Mg(BH4)2 can also adsorb 0.8 H2 at low temperatures and 100 bar to achieve a total hydrogen mass content of 17.4% [33]. High-pressure phase δ-Mg(BH4)2 has a very high volumetric hydrogen content of 147 g H2/L. Mg(BH4)2 can also form amorphous solids. Overall dehydrogenation reaction is, in fact, a multistep reaction (see Figure 1) with various reaction intermediates, such as Mg(B3H8)2, MgH2, and MgB12H12, which were proposed both experimentally and theoretically [22,[34][35][36]. MgB2 is the decomposition product obtained after heating to 500 °C [37]. Boron-rich MgB7 films are obtained by heating volatile Mg(B3H8)2 solvates with dimethyl ether and diethyl ether [38].  [22,[34][35][36][37][38][39][40][41][42][43][44] MgB 2 can be rehydrogenated, although under drastic conditions (950 bar H 2 at 400 • C) [40]. The rehydrogenation of MgB 2 can be accelerated with THF, MgH 2 , and Mg [41]. Mechanically milled mixtures of MgB 2 , THF, and 40 mol % Mg could thus ab-Molecules 2021, 26, 7425 3 of 14 sorb 6 wt % of H 2 at 300 • C under 700 bar of H 2 , which is less drastic than that without THF. Recently, rehydrogenation at room temperature with mechanical activation by ball milling was reported [42]. These rehydrogenation reactions of MgB 2 demonstrate the principle that hydrogen storage in Mg(BH 4 ) 2 is indeed reversible. A recent combined experimental and theoretical study concluded that the initial stages of rehydrogenation are associated with the formation of σ bonds of hydrogen with boron on the reactive edges of the MgB 2 solid [43].  Figure 1) show that the reaction kinetics can be influenced by various parameters, which also include the initial crystalline modification of Mg(BH 4 ) 2 .
The overall enthalpy of reaction for the dehydrogenation of Mg(BH 4 ) 2 (∆ f H • = −208 kJ/mol) to form MgB 2 (∆ f H • = −91.96 kJ/mol) and hydrogen can be calculated [45][46][47] to be equal to +116 kJ/mol, i.e., less than 30 kJ/mol per hydrogen molecule released, which is, in principle, in the correct range for a hydrogen storage material [13]. The first step of a dehydrogenation reaction of BH 4 − is likely to be the breaking of a B-H bond. Isotope exchange reactions of Mg(BH 4 ) 2 with D 2 allow for producing a complete exchange to form Mg(BD 4 ) 2 , and the corresponding activation energy was estimated to be about 51 kJ/mol [48]. For Ca(BH 4 ) 2 , the corresponding activation energy was found to be 82 and 98.5 kJ/mol for the reverse reaction, confirming that breaking a bond with hydrogen or deuterium is the rate-limiting step [49]. Theoretical calculations of potential defects in Mg(BH 4 ) 2 suggest that, in the initial phase of the dehydrogenation, a H − ion is formed that can diffuse in the lattice [50]. On the other hand, gas diffusion in the solid is also a contribution to exchange kinetics, as was shown by isotope exchange reactions with the highly porous modification of γ-Mg(BH 4 ) 2 with a high surface area compared to a ball-milled sample with a strongly reduced surface aera [51].
The reaction kinetics of hydrogen release in Mg(BH 4 ) 2 can be significantly enhanced by various additives, such as TiCl 3 [52] or NbF 5 and TiO 2 [53]. Lewis bases in the form of solvates of Mg(BH 4 ) 2 can also accelerate the hydrogen release [54]. As shown in Figure 1, the THF solvate releases H 2 gas below 200 • C to form Mg (B 10 H 10 12 2− was also observed, but with THF and dimethyl ether, B 12 H 12 2− remained a minor reaction product. The physical properties of Mg(BH 4 ) 2 .3THF were recently investigated in detail [55]. In this compound, Mg 2+ is coordinated to 2 BH 4 − ions and 3 THF molecules. The orientational mobilities of the BH 4 − ions are not particularly sensitive to the presence of THF. The authors concluded that "the presence of THF also disrupts the stability of the crystalline phase leading to enhanced kinetics for the dehydrogenations". Recently, Tran et al. [56] [38,59]. Kim et al. [38] reported the formation of boron-rich MgB 7 films upon heating under vacuum above 425 • C due to some evaporation of Mg under these conditions. Thermogravimetry (TG) experiments [59] revealed a 30 wt % mass loss setting in above ca 80 • C corresponding to the evolution of B 2 H 6 , B 5 H 9 and H 2 . The residual solid after heating to 200 • C was a mixture of mainly Mg(BH 4 ) 2 , Mg(B 10 H 10 ), and Mg (B 12 H 12 ), and the combined evolution of H 2 , B 2 H 6 , and B 5 H 9 was confirmed by mass spectrometry [60]. The addition of activated (ball-milled) MgH 2 to Mg(B 3 H 8 ) 2 results in a strong reduction in borane evolution and up to 88% conversion back to Mg(BH 4 ) 2 at 100 • C. The presence of activated MgH 2 thus substantially decreases the formation of (closo-hydro)borates and provides the necessary hydrogen for the conversion of B 3 H 8 − back into BH 4 − . These experiments suggest that, while Lewis acids may favor the dehydrogenation reactions of Mg(BH 4 ) 2 , they do not necessarily catalyze the rehydrogenation reactions, as transition metal halides do not appear to affect the rehydrogenation of MgB 2 [40,61]. THF and other Lewis bases appear to accelerate both the dehydrogenation and rehydrogenation reactions of Mg(BH 4 ) 2 , and encourage more studies to even further improve the kinetics.

DFT Calculations
The results presented above for Mg(BH 4 ) 2 suggest the formation of various intermediate species such as B 2 5 anions, and some other species were reviewed [62].
In order to assess the driving forces for different reactions, thermodynamic information can be very useful, but experimental data are very scarce. For alkali borohydrides, thermodynamical data are available [47], but only few other experimental data are available. Using the experimental values of the formation enthalpy of Mg(BH 4 ) 2 [45] and La(BH 4 ) 3 [63], the formation enthalpy of other M(BH 4 ) 2 and M(BH 4 ) 3 compounds were estimated, assuming that the lattice enthalpy of bromides and borohydrides with the same metal ion were identical within about 15 kJ/mol [46]. The experimental formation enthalpy of NH 4 B 3 H 8 (−530 ± 33 kJ/mol) [64], (NH 4 ) 2 B 10 H 10 (−359.2 ± 10 kJ/mol) [65], and of guanidinium and other nitrogen-based closo-borates was reported [66]. Recently, new heat capacity measurements for Na, K, Rb, Cs, Mg, Ca borohydrides were reported [67]. The knowledge of all thermodynamic properties in principle allows for quantitatively describing the phase diagram of a system, which was performed using available data for the Mg-B-H system [68].
In the absence of experimental data, theoretical data are obtained. It is quite challenging to obtain accurate results of formation enthalpies using DFT. Nguyen et al. [69] calculated for the formation enthalpy of (NH 4 ) 2 B 10 H 10 with the G3 method the value of −184 kJ/mol, which is quite different from the experimental value of −359.2 kJ/mol. For α-Mg(BH 4 ) 2 , formation enthalpy values ranging from −67 to −277 kJ/mol were reported in the literature [68], while the experimental value was −208 kJ/mol [45]. Zhang et al. [23] computed relative formation energies of potential solid intermediates formed during the dehydrogenation of Mg(BH 4 ) 2 , in combination with a Monte Carlo-based structure prediction method. They predicted a potential Mg 3 (B 3 3− ion, while Mg(B 3 H 8 ) 2 was found to be very high in relative energy and thereby unlikely to be formed.
The principal difficulty for estimating the formation enthalpy of crystalline solids is the evaluation of lattice energy, as different approaches (volume-based, Kaputinski equation etc.) lead to different values. Further, lattice energies can only be computed for crystalline materials, preferentially on the basis of experimental structure data, but experiments showed that a significant fraction of the reaction intermediates remain amorphous, complicating things even further.
DFT calculations in the gas phase are quite reliable, and allow for obtaining good structural data and vibrational frequencies, in particular when anharmonicity is included. Several studies report the formation enthalpy of borohydride ions in the gas phase [69][70][71][72]  Anharmonic DFT calculations allow for obtaining improved agreement with experimental vibrational spectra, from which heat capacity data were calculated [73]. Figure 2 compares experimental [74] and DFT calculated [69][70][71][72] formation enthalpy data for neutral and anionic boron hydrogen species. Figure 2 shows that the calculated formation enthalpy for a given species (e.g., B 3 H 8 − ) can differ by about 100 kJ/mole for different sources. These values are derived, for instance, from isodesmic reactions with known formation heat [69], thus generating a potential propagation of errors if the initial formation enthalpy values are different. We outline all reported values to highlight the limitations of the accuracy of these data. crystalline materials, preferentially on the basis of experimental structure data, but experiments showed that a significant fraction of the reaction intermediates remain amorphous, complicating things even further.
DFT calculations in the gas phase are quite reliable, and allow for obtaining good structural data and vibrational frequencies, in particular when anharmonicity is included. Several studies report the formation enthalpy of borohydride ions in the gas phase [69][70][71][72] Anharmonic DFT calculations allow for obtaining improved agreement with experimental vibrational spectra, from which heat capacity data were calculated [73]. Figure 2 compares experimental [74] and DFT calculated [69][70][71][72] formation enthalpy data for neutral and anionic boron hydrogen species. Figure 2 shows that the calculated formation enthalpy for a given species (e.g., B3H8 − ) can differ by about 100 kJ/mole for different sources. These values are derived, for instance, from isodesmic reactions with known formation heat [69], thus generating a potential propagation of errors if the initial formation enthalpy values are different. We outline all reported values to highlight the limitations of the accuracy of these data.   [70], who showed that a sequence of BH3 additions followed by H2 elimination from B2H6 to B6H10 is overall exothermic, but with two less stable reaction intermediates (B3H9 and B4H8) that can act as barrier steps for the kinetics. Figure 2 shows that anionic species with 9-12 boron   Figure 2 shows that the experimental formation enthalpies of neutral species are all positive [74], with values ranging from 36 kJ/mol (for B 2 H 6 ) to 210 kJ/mol for (B 2 H 4 Figure 2 shows that anionic species with 9-12 boron atoms are the most stable, which indicates that there is a thermodynamic driving force towards these anions. The most stable species in this figure is the closo B 12 H 12 2− ion, and its stability is related to its 3-dimensional aromaticity [6]. The formation enthalpy of B 12 H 12 2− in the gas phase was estimated to be between −325.5 and −428.6 kJ/mol according to different theoretical studies [72,75,76]. One key intermediate in the overall dehydrogenation reactions of BH 4 − appears to be ion B 3 H 8 − , which is discussed in the next section.

Formation and Reactions of B 3 H 8 −
As mentioned above, the formation of Mg(B 3 H 8 ) 2 was observed during the decomposition of Mg(BH 4 ) 2 under dynamic vacuum [54,77], and Y(B 3 H 8 ) 3 was obtained after heating Y(BH 4 ) 3 under hydrogen pressure of 1-10 bar [78]. There are several reports in the literature on the synthesis of B 3 3 . The efficient synthesis of alkali metal octahydrotriborates (M = Na, K, Rb, Cs) from the reaction of MBH 4 with 2 equivalents of dimethyl sulfide borane was reported [82]. The formation of ion B 2 H 7 − was observed by NMR for the reaction of LiBH 4 with THF.BH 3 in THF [83], and during the solvothermal reaction of BH 4 − with CH 2 Cl 2 at 70 • C [84]. The reaction of BD 4 − requires higher temperatures (90 • C) [84], which suggests that the rate-determining reaction step is associated with the breaking of a boron-hydrogen (deuterium) bond, which could be the formation of a reactive Lewis adduct of BH 3 (B 10 H 10 ) (B 12 H 12 ), as demonstrated by Gigante et al. [86]. This synthesis starts with the conversion of NaBH 4 into (Et 4 N)BH 4 , which reacts solvothermally with CH 2 Cl 2 to form (Et 4 N)B 3 H 8 . (Et 4 N)B 3 H 8 is then heated in toluene to 185 • C to form a mixture of (Et 4 N) 2 B 10 H 10 and (Et 4 N) 2 B 12 H 12 , which can then either be separated by fractional crystallization or directly converted with sodium tetraphenylborate into ionic conductor Na 4 (B 10 H 10 )(B 12 H 12 ).

Closoborates and Related Species as Solid Ionic Conductors
Solid ionic conductors for lithium or sodium batteries allow for avoiding the use of a flammable organic electrolyte and are thus expected to considerably improve the safety of batteries. A good solid electrolyte must fulfill several empirical criteria, according to [87]: -"open structure" with a low coordination number of the mobile ion; - The presence of structural phase transitions at low pressure. In the case of AgI, the ambient pressure wurtzite structure (space group P6 3 mc) transforms at 3 kbar and 315 K into a NaCl structure (space group Fm-3m), thus going from a rather covalent network with coordination number 4 to a rather ionic structure with coordination number 6. The associated charge fluctuations between ions can potentially be coupled to vibrational motions and thus dynamically favor ionic conduction.
For practical applications, the conductivity of the material should be higher than 1 mS/cm. Further, the material should have high chemical and thermal stability, and a high electrochemical stability window. Additionally, it must be electronically insulating to avoid battery self-discharge or shortage. Further, the electrolyte should be deformable in order to accommodate the volume changes of anode and cathode materials upon lithium or sodium insertion and removal. This can thus limit the formation of fractures that reduce the performance of the battery. Lastly, the material should not be toxic and be cheap enough for the considered applications.
The discovery of superionic conductivity in the high-temperature phases of LiBH 4 [88] and Na 2 B 12 H 12 [29] has stimulated new research for similar compounds with high ionic conductivity at lower temperatures. These compounds include closo-hydroborates, nidohydroborates (B 11 [89] reported LD 50 values for Na 2 B 10 H 10 and Na 2 B 12 H 12 administered orally to rats to be around or higher than 7.5 g/kg of body weight for both compounds. The crystal chemistry of inorganic hydroborates except BH 4 − was recently presented in detail [90], while the crystal chemistry of salts with BH 4 − was addressed in an earlier review [18]: "All nonmolecular hydroborate crystal structures can be derived by simple deformation of the close-packed anionic lattices, i.e., cubic close packing (ccp) and hexagonal close packing (hcp), or bodycentered cubic (bcc), by filling tetrahedral or octahedral sites" [90]. This observation can be illustrated considering group-subgroup relationships of encountered crystal structures, as illustrated in Figure 3 for some relevant compounds [90][91][92][93][94][95][96][97][98][99][100][101][102]. Crystal packing is governed by large anions, leaving in some space groups empty cationic sites, which, of course, favor ionic conduction. For instance, β-Na 2 B 12 H 12 crystallizes in the Pm-3n space group with a statistical population of 6 sites occupied by 4 Na + ions.
Perturbations of the anionic sublattice further allow for stabilizing the conductive phase at lower temperatures. This was first demonstrated for solid solutions of LiBH4 with LiBr and LiI [103]. Phase stability and ionic conductivity in mixed LiBH4-LiX (X = Cl, Br) was recently studied in detail [104]. Perturbations of the structure by ball milling or partial substitution was demonstrated for Na2B12H12 with a partial introduction of iodine ions in the closo-hydroborate [105]. In a further step, solid solutions of closo-hydroborate and closo-carbahydroborates, and solid solutions of nido-hydroborates with closo-hydroborates were studied [106][107][108][109][110][111][112]. Representative examples of mixed borate ionic conductors are shown in Table 1.  . Group-subgroup relationships between space groups (in Herrmann-Mauguin notation) of closo-hydro borates and some closo-halogeno borates. t, "translationengleich" subgroups; k, "klassengleich" subgroups. The mechanism of ionic conduction in these compounds is related to the dynamical properties of the borohydride or carbohydride ions. These properties can be addressed using NMR [113] and neutron techniques [114], in conjunction with temperature-dependent conductivity and X-ray diffraction, and are supported by theoretical calculations [76,77,88]. A detailed study of ionic conductor Na 4 (B 12 H 12 )(B 10 H 10 ) [115] with all these techniques revealed 3 different regimes with increasing temperature. Below −50 • C, conductivity remains very low. Above this temperature, an apparent activation energy of 0.6 eV was found, related to significant couplings of anionic and cationic motions. Above 70 • C, activation energy decreases to 0.37 eV, as thermal energy leads to noncorrelated ionic motions.
One important aspect of solid ionic conductors is their electrochemical stability, which is a critical limit for a reversible battery application. Asakura et al. [116] developed a linear sweep voltammetry method to reliably measure the electrochemical stability of borohydride-based solid electrolytes. The measured oxidative stability of LiBH 4 of 2.0 V vs. Li + /Li was significantly smaller than that in initial reports claiming a stability of up to 5 V [117]. For Na 4 (B 12 H 12 )(B 10 H 10 ), two oxidation onsets at 3.02 and 3.22 V vs. Na + /Na were tentatively assigned to the onset of decomposition of the less stable [B 10 H 10 ] 2− and more stable [B 12 H 12 ] 2− ions, respectively [116]. Closo-carborane ions are even more stable, as for Na 4 (CB 11 H 12 ) 2 (B 12 H 12 ), where a large anodic current was observed above 4 V vs. Na + /Na, together with a small onset at 2.93 V. For Li 2 (CB 9 H 10 )(CB 11 H 12 ), the onset of decomposition was observed at 2.86 V vs. Li + /Li [116]. Nido-borates are electrochemically less stable. The oxidative stability limit for Na 5 (B 11 H 14 )(B 12 H 12 ) 2 was 2.6 V vs. Na + /Na, and for LiB 11 H 14 , 2.6 V vs. Li + /Li [107].
These developments have also led to several all-solid-state battery prototypes based on these mixed borate ionic conductors. Duchêne et al. [118] presented a 3 V sodium battery using Na 4 (B 12 H 12 )(B 10 H 10 ), and Murgia et al. [119] showed Na stripping/plating over >500 h in a Na cell with Na 4 (CB 11 H 12 ) 2 (B 12 H 12 ). Recently, Asakura et al. [120] demonstrated a 4 V sodium battery with the same solid-state conductor, Na 4 (CB 11 H 12 ) 2 (B 12 H 12 ). These results show that closo-hydroborates and their derivatives are very promising materials for chemically and electrochemically stable all-solid-state ionic conductors.

Conclusions
In the last 20 years, many studies on borohydride species have considerably increased our knowledge on the properties of these materials. For hydrogen storage applications, the kinetics and reversibility of the dehydrogenation reactions remain a major challenge for practical applications. The chemistry of borohydrides from BH 4 − to B 12 H 12 2− in the gas phase and in solution has been theoretically and experimentally addressed; however, in solids, these studies are very challenging, as structural data of potential reaction intermediates such as Mg(B 3 H 8 ) 2 are elusive, and not all intermediates can be observed. If the reaction intermediates are amorphous, X-ray diffraction cannot be used, and theoretical approaches can lead to many different potential structures. The presence of additional hydrides or of Lewis bases such as THF, as shown for the reactions of KB 3 H 8 and Mg(B 3 H 8 ) 2 , strongly modifies the reaction products upon heating. We are thus still very far from a full microscopic understanding of these hydrogenation-dehydrogenation reactions and in the search for optimal catalysts for these processes.
For hydrogen storage, B 3 H 8 − is an interesting species that can be rehydrogenated back to BH 4 − . Even though only 25% of the hydrogen is available for this reversible hydrogen storage, the temperatures (less than 200 • C) and kinetics of these reactions approach practical conditions. The closo-hydroborate ions that are formed and identified as intermediates of dehydrogenation reactions have found new and very promising applications as solid-state ionic conductors, as they present many very favorable properties for this use. The recent demonstration of a 4 V all-solid-state battery using solid sodium electrolyte Na 4 (CB 11 H 12 ) 2 (B 12 H 12 ) [120] highlights this potential. Whether compounds such as Mg(B 10 H 10 ), which can be obtained starting from Mg(BH 4 ) 2 .2THF, are applicable for new Mg-based batteries remains to be demonstrated. In the preparation of these closo-hydroborates and their derivatives, starting from BH 4 − instead of neutral boranes, has the great advantage to reduce the toxicity of the reactants. B 2 H 6 , B 5 H 9 and B 10 H 14 are highly toxic and thereby not really suitable for industrial production processes of closo-hydroborates at a higher scale. Thus, boron-hydrogen compounds have a future for new green energy applications.

Conflicts of Interest:
The author declares no conflict of interest.