1. Introduction
Amidoboranes, also named as amidotrihydroborates (MNH
2BH
3, MAB), constitute a constantly growing family of hydrogen-rich ammonia borane [
1,
2,
3,
4] derivatives. Amidoboranes comprise metal cations and monovalent amidoborate anions [NH
2BH
3−, AB
−]. Known for nearly 80 years, they were recently rediscovered and are now the subject of intense research [
5] (
Figure 1).
Ammonia borane, the parent compound of amidoborane salts, was synthesized and characterized for the first time by Shore and Perry in 1955 [
1]. Sixty years after its first synthesis, ammonia borane is still the subject of intensive experimental and theoretical studies [
1,
2,
3,
4,
6,
7,
8,
9], and has been the subject of several reviews [
10,
11]. Ammonia borane has one of the highest gravimetrical hydrogen capacities (19.6 wt%) among all known chemical compounds [
1]. Thus, it is a very convenient starting material for preparation of new hydrogen storage materials [
12].
The first amidoborane salt, sodium amidoborane (NaAB), was synthesized and described by Schlesinger and Burg in 1938 [
13]. For almost half a century, there was no interest in the chemistry of amidoboranes. In the late 1980s and early 1990s, several amidoborane compounds were investigated in the group of Shore, namely: lithium amidoborane (LiAB), sodium amidoborane (NaAB), potassium amidoborane (KAB), magnesium amidoborane (Mg(AB)
2) and zinc amidoborane (Zn(AB)
2) [
14,
15,
16]. In the mid-1990s, the possibility of using LiAB as reducing agent in organic chemistry was discussed by Myers et al. [
17].
A huge acceleration of research focused on metal amidoboranes—and especially their hydrogen storage properties—has been observed during the last decade. Since 2008, a number of important papers and reports were released describing amidoboranes of: lithium, LiAB [
18,
19,
20,
21]; sodium, NaAB [
18,
21,
22,
23,
24]; potassium, KAB [
25]; rubidium, RbAB [
26]; cesium, CsAB [
26]; magnesium, Mg(AB)
2 [
27,
28,
29,
30]; calcium, Ca(AB)
2 [
19,
31,
32]; strontium, Sr(AB)
2 [
33]; zinc, Zn(AB)
2 [
34]; aluminum, Al(AB)
3 [
35,
36,
37]; and yttrium, Y(AB)
3 [
38]. Several bimetallic amidoborane salts have been prepared, as well, i.e., lithium–sodium, LiNa(AB)
2 [
39]; lithium–aluminum, LiAl(AB)
4 [
35,
40]; sodium–aluminum, NaAl(AB)
4 [
41]; sodium–magnesium, NaMg(AB)
3 [
42]; potassium–magnesium, KMg(AB)
3 [
43]; rubidium–magnesium, RbMg(AB)
3 [
43]; disodium–magnesium, Na
2Mg(AB)
4 [
44]; and dipotassium–magnesium, K
2Mg(AB)
4 [
44,
45]. Amidoborane compounds containing metal centers coordinated with molecules of the solvent (which could not be later desorbed in post processing, e.g., THF complexes of europium amidoborane, Eu(AB)
2 [
15], ytterbium amidoborane, Yb(AB)
2 [
15], and calcium amidoborane, Ca(AB)
2 [
30]) have also been described. One can also find information about unsuccessful attempts of synthesis of other amidoboranes, e.g., those of titanium [
19] and iron [
46].
Apart from those, systems containing molecules of solvent trapped in the crystal structure of the compound are also known, e.g., amidoborane ammoniates of: lithium, LiAB∙NH
3 [
47]; magnesium, Mg(AB)
2∙NH
3 [
48] and Mg(AB)
2∙2NH
3 [
49]; calcium, Ca(AB)
2∙NH
3 [
50] and Ca(AB)
2∙2NH
3 [
51]; and aluminum, Al(AB)
3∙NH
3 [
35,
52]. The are also amidoborane hydrazinates of: lithium, LiAB∙N
2H
4 [
53]; calcium, Ca(AB)
2∙N
2H
4 [
54]; and complex of lithium amidoborane with ammonia borane, LiAB∙AB [
55]. Another group of amidoborane derivatives is that of compounds containing amidoborate anions with a hydrocarbon functional group, e.g., LiNMe
2BH
3 [
56,
57], LiNPr
2BH
3 [
57], LiNH
2B(C
6F
5)
3 [
58], NaNMe
2BH
3 [
59], KNMe
2BH
3 [
60], K(NMe
2BH
2NMe
2BH
3) [
61], and many others, which have been described in the review by Stennett and Harder [
62]. Complex compounds containing amidoborate anions and other hydrocarbon groups have also been synthesized [
63,
64,
65,
66,
67].
Amidoboranes were subjects of several patent applications. Among them, the most important were probably the US patents of Torgersen et al. [
68], Burrell et al. [
69], and Balema et al. [
70], which covered most of the known and probably synthesizable compositions.
There is also a number of theoretical papers on modeling of the properties and crystal structures of selected amidoboranes, e.g., electronic structure and dehydrogenation characteristics of LiAB [
71,
72,
73]; crystal structure and dehydrogenation mechanism of LiAB [
74,
75], NaAB [
74,
75], KAB [
75], LiNa(AB)
2 [
76] and Na
2Mg(AB)
4 [
77]; crystal structure, and electronic and mechanical properties of Mg(AB)
2, Ca(AB)
2 and Sr(AB)
2 [
78]; spectroscopic properties of LiAB [
79]; influence of homopolar dihydrogen bonding on hydrogen storage properties [
80]; or supramolecular interactions [
60].
In this review, we present a comprehensive summary of amidoborane salts containing only metal cations and amidoborate anions without any other constituents, substituents, solvent molecules, or additional functionalities. We compare their crystal structures, hydrogen storage properties, 11B NMR spectra, vibrational spectra, and other features. We analyze physicochemical properties of amidoborane salts as compared to those of ammonia borane to show an impact of introduction of different metal cations to the crystal structure in place of one acidic proton.
2. Crystal Structures of Metal Amidoboranes
Up till now, 19 amidoborane salts were reported (11 monometallic and eight bimetallic) [
5,
18,
19,
20,
21,
22,
23,
24,
25,
26,
27,
28,
29,
30,
31,
32,
33,
34,
35,
36,
37,
38,
39,
40,
41,
42,
43,
44,
45,
62]. Usually, amidoboranes are white or grayish solids. However, the color of the compound may strongly depend on the grain size of the material or even on the synthesis method. All amidoborane salts are highly sensitive against water and moisture, which cause hydrolysis to ammonia borane and respective metal hydroxide [
23,
81] or to boric acid [
82]. Because of high reactivity against moisture and water, amidoborane compounds are typically stored and processed under inert atmosphere, mostly common argon, nitrogen or helium. Some of the compounds seem to be unstable at room temperature, thus they should be cooled down and stored at low temperature to preserve their properties [
24].
All amidoborane salts exhibit properties typical of protonic–hydridic hydrogen storage materials because the amidoborate anions contain both the hydrogen atoms with partial positive charge (H
δ+ in NH
2 groups) and those with partial negative charge (H
δ– in BH
3 groups). Amidoborate anions differ from ammonia borane molecules by substitution of one protonic hydrogen atom in each molecule with a metal cation. The shortest M–N distances in amidoborane crystals (e.g., Li–N, 1.98 Å [
18]; Na–N, 2.35 Å [
18]; K–N, 3.10 Å [
25]; Ca–N, 2.383 Å [
19]; etc.) are obviously much longer than the H–N distances in ammonia borane molecule (N–H, 0.85 Å [
3]); the M–N bonds in the former are usually quite ionic [
18].
The parent ammonia borane crystallizes in the
I4mm space group (HT form [
3]) or in the
Pmn21 (the LT form below 224 K [
4]). Amidoborane salts are not isostructural with ammonia borane and crystallize in various lower symmetry space groups. In
Table 1 and
Table 2, we have listed crystallographic data available in the literature for a handful comparison of ammonia borane with mono- and bimetallic amidoborane salts, for which crystal structures have been solved or their diffraction patterns have been successfully indexed in a given unit cell.
Amidoboranes form crystals where amidoborate anions coordinate metal cations with either (NH
2) or/and (BH
3) groups. The bond lengths in amidoborate anions (N–H, ca. 0.9–1.1 Å; B–H, ca. 1.1–1.2 Å; and B–N, ca. 1.56–1.58 Å) are comparable with the corresponding bond lengths in ammonia borane molecules (HT form: N–H, 0.85(7) Å; and B–H, 1.11(4) Å [
3]) (
Table 2). It is worth mentioning that the B–N bond length is usually slightly shorter (by 0.012–0.067 Å) in amidoborate anions than in ammonia borane molecules (HT form: 1.597(3) Å [
3]). Volume of one formula unit of all amidoborane salts may be estimated with reasonable accuracy by simply adding the calculated volumes of amidoborate anion (ca. 60 Å
3) and the respective metal cations.
Among all known amidoborane salts, only two pairs of compounds are monometallic α-LiAB and NaAB crystallizing in
Pbca space group [
18] and bimetallic Na
2Mg(AB)
4 and K
2Mg(AB)
4 crystallizing in
I41/a space group [
44,
45].
Light alkali metal (α-LiAB, β-LiAB, NaAB, and KAB) amidoboranes crystallize in
Pbca space group [
18,
20,
21,
25]. β-LiAB [
20,
21] and KAB [
25], with Z = 16, have twice as big unit cell than α-LiAB and NaAB, both with Z = 8 [
18]. RbAB and CsAB crystallize with Z = 4 in monoclinic space group
P21/c and orthorhombic space group
Pnam, respectively [
26]. Ca(AB)
2 and Sr(AB)
2 both crystallize in
C2 space group but they are not isostructural and their atomic coordinates are quite different from each other [
19,
43]. If space group assignment is correct, these would be the only amidoboranes that lack the inversion center (polar space group) and their unit cells exhibit an uncompensated dipole moment. In fact, the heavy-atom sublattice of Ca(AB)
2 may be symmetrized to centrosymmetric
C2/m at crude threshold. LiNa(AB)
2 and NaAl(AB)
4 have the lowest (triclinic) symmetry among all amidoborane salts [
39,
41]. Na
2Mg(AB)
4 and K
2Mg(AB)
4 form isostructural crystals with the highest symmetry among amidoborane salts (
I41/a) [
44,
45]. Crystal structures of amidoborane salts listed above are presented in
Figure 2 and
Figure 3.
Crystal structure of Y(AB)
3 [
38] and NaMg(AB)
3 [
41] were not solved but their powder X-ray patterns were successfully indexed and fitted using the LeBail method. Crystal structures of other synthesized amidoborane salts have not yet been determined.
The crystal structures of metal amidoboranes are often unprecedented and show great variety (
Figure 2 and
Figure 3); this is due to a low symmetry of the amidoborate anion, and its flexibility to bind to metal cations, involving both the lone pair on N atom, and the three hydride terminal at B atom. We presented general perspective views over each crystal structure showing unit cells along selected axis perpendicular to the plane of the picture (
Figure 2). To analyze crystal packing and cation–anion connectivity, for simplicity, we have also represented the entire amidoborate anion by a single dummy atom, X, sitting in the middle of the B–N bond in an axonometric view over the unit cells (
Figure 3). Such representation leads to facile analysis of anion–cation sublattice as well as inter-ionic interactions. Moreover, the nearest anionic neighbors of a given cation may now be identified. To show interactions between metal cations and amidoborate anions, we also present fragments of the unit cells of amidoborane salts showing coordination polyhedra of metal cations (
Figure 4).
The crystal structures of α-LiAB [
18,
19] and NaAB [
18,
21] consist of double layers of cations and anions, with dihydrogen bonds linking the layers together. Interestingly, one of the anions is much closer to the cation than two others; thus, the crystal structures of α-LiAB [
18,
19] and NaAB [
18,
21] can be viewed as quasi-molecular as far as weaker stabilizing interactions are omitted. By quasi-molecular, we mean here that there is just one close cation–anion contact, the others being substantially longer. β-LiAB [
20,
21] has a similar structure but with somewhat more complex pattern of cation–anion bonding with the local electric dipoles alternating along the
c axis and the
ab diagonal. KAB [
25] repeats the double-layer motif, but the layers are now more puckered and each potassium cation is linked to three close anions thus forming a polymeric network. Both heaviest alkali metal amidoboranes, those of rubidium and cesium, do not show the presence of any low-dimensional polymeric sublattices; they are ionic compounds where anions are shared by at least three and even up to five metal cations [
26]. For RbAB and CsAB, the distances M–B and M–N (Rb-B: 3.189–5.591 Å, Rb-N: 3.076–3.143 Å [
26]; Cs-B: 3.680–3.854 Å, and Cs-N: 3.340–3.535 Å [
26]) are comparable with respective distances in borohydrides (Rb-B: 3.541 Å [
83]; and Cs-B: 3.71 Å [
83]) and amides (Rb-N: 3.140–3.355 Å [
84]; and Cs-N: 3.515 Å [
85]). The changes observed in the Li–Cs series are obviously due to an increasing ionic radius of alkali metal cations, and enhances ionicity of chemical bonding down the Group 1.
A similar trend may be observed when comparing the crystal structures of Ca(AB)
2 [
19] and Sr(AB)
2 [
43]. While the former is quasi-molecular, with two closest anions linked to Ca site in the quasi-linear fashion, the latter comprises bent Sr(AB)
2 units that are in fact interconnected in 2D sheets, with each AB
− anion shared between two neighboring Sr
2+ cations. One may expect that crystal structure of not-yet-synthesized Ba(AB)
2 would form extended polymeric network, just like RbAB or CsAB [
26].
Crystal structure of LiNa(AB)
2 [
39] contains isolated Li
2(AB)
42− dianions (linked only via weak dihydrogeen bonds into 1D chains) as well as Na
+ cations. This form clearly originates from transfer of AB
− anions from a stronger Lewis base, NaAB, to a Lewis acid, LiAB. The same situation is observed for Na
2Mg(AB)
4 [
44], which consists of isolated Mg(AB)
42− dianions and Na
+ cations. K
2Mg(AB)
4 [
45] and NaAl(AB)
4 [
41] also comprise the isolated Mg(AB)
42− and Al(AB)
4− anions, respectively, together with alkali metal cations.
Coordination spheres of metal cations in the lattice of amidoborane salts are dependent on the size of the cation and type of crystal structure. Li
+ cations are coordinated by four amidoborate anions in α-LiAB [
19], β-LiAB [
21] and LiNa(AB)
2 [
39]. In LiNa(AB)
2 [
39], Li
+ cations are coordinated by three N atoms and one hydridic H atom, each in a different top of a distorted tetrahedron, while in α-LiAB [
19] and β-LiAB [
21] Li
+ cations are surrounded by one N atom and hydridic H atoms, five and six, respectively, where one or two H atoms sit in one top of a distorted tetrahedron. Na
+ is coordinated by four AB
− anions in NaAB [
21], by five AB
− anions in LiNa(AB)
2 [
39], and by six AB
− anions in NaAl(AB)
4 [
41] and Na
2Mg(AB)
4 [
44], but only in NaAB [
21] Na
+ are coordinated by N atoms. Metal cations in KAB [
25], RbAB [
26], CsAB [
26] and Ca(AB)
2 [
19] are coordinated by five or six AB
− anions, where two of them are facing metal cations with N atmos. Sr
2+ cations are coordinated at least by two N atoms form neighbor AB
− anions [
33].
Coordination number of Sr
2+ cations cannot be derived due to lack of hydrogen positions in the solution of Sr(AB)
2 crystal structure [
33]. In bimetallic LiNa(AB)
2 [
39], NaAl(AB)
4 [
41], Na
2Mg(AB)
4 [
44] and Na
2Mg(AB)
4 [
45], tetrahedral complex anions are formed by Li
+, Al
3+ and Mg
2+ cations that are coordinated by four AB anions through their N atoms [
39,
41,
44,
45]. Respective counter ions, Na
+ [
39,
41,
44] and K
+ [
45], are coordinated only by hydridic H atoms form amidoborate groups.
3. NMR, IR and Raman Spectra of Metal Amidoboranes
Metal amidoboranes exhibit certain similarities in physicochemical properties and spectroscopic characterization (NMR, FTIR, and Raman). Several diagnostic features can be seen for all of them, which help in identification of these compounds (
Figure 5, for more detailed view of selected amidoborane salts see
S1, S6, S8, S10, S12, S13, S16, S19, S22, S23, S25, S27, S31 Supplementary Materials).
Boron-11 nuclear magnetic resonance spectroscopy (either in solution or in the solid state) is one of the most common and accurate techniques for characterization of amidoboranes.
11B NMR spectra of these salts contain a single characteristic signal at ca. −21 ppm (similar to that seen for the parent ammonia borane [
18,
23,
35]), which comes from [BH
3] groups. It can easily be distinguished from the signals originating from [BH
4] groups (ca. −45 ppm), [BH
2] groups (ca. −10 ppm), or other chemical moieties containing boron atoms in a form different to [BH
3] groups.
11B NMR measurements are usually carried out in THF-d
8 solution despite relatively low solubility of metal amidoboranes in THF.
11B NMR spectra of amidoboranes consist of a quartet with an intensity ratio of 1:3:3:1. The signal is split due to coupling of boron-11 with three hydrogen-1 nuclei. The quartet is centered at ca. −21 ppm, with J-coupling value in the range of 80–100 Hz [
16,
23,
25,
26,
31,
35,
86,
87]. On the other hand, the
11B MAS NMR measurements for the samples in the solid state give one broad peak at about −21 ppm [
18,
25,
29,
38,
39,
40,
43,
45]. In
Figure 4, we have shown a collection of the
11B NMR spectra, while, in
Table 3, we have listed
11B NMR chemical shifts and J-coupling constants for various amidoboranes present studied so far.
Two papers have reported “non-standard” NMR spectra of amidoborane salts; Shimoda et al. studied NaAB with
23Na NMR [
88] and KAB with temperature resolved
39K NMR spectroscopy [
89].
Infrared absorption and Raman scattering spectroscopy are two other common—and complementary—methods for characterization of metal amidoboranes. The IR and Raman spectra of metal amidoboranes, similar to neat ammonia borane, consist of the two main distinct groups of bands related to stretching vibrations of NH and BH bonds, which fall in the 3200–3400 cm
−1 and 2000–2400 cm
−1 range, respectively (
Table 4 and
Table 5). Other bands originate from deformations of the NH
2 group (1400–1650 cm
−1), deformations of BH
3 group (1000–1350 cm
−1), stretching vibrations of the BN bond (below 1000 cm
−1), and others. It is hard to determine precise range of wavenumbers for these bands because of their partial overlap. Unfortunately, there is no single well defined band with a fixed wavenumber that would be distinctive and common for all metal amidoboranes as the chemical vicinity of amidoborate anion is different for each of them; thus, the band positions in the vibrational spectra turn out to be sensitive markers of interatomic interactions in the solid state. The only generalization that one may make is that the upper bound of the NH stretching region in the spectra of amidoboranes (3200–3400 cm
−1) is usually blueshifted with respect to the respective band for ammonia borane (3200–3320 cm
−1) [
2,
23]. This indicates some stiffening of the N–H vibrons.
4. Thermal Decomposition of Metal Amidoboranes
Thermal decomposition of amidoborane salts is an exothermic multistep process leading to amorphic polymeric (BNH)
n–type products at a relatively low temperature (100–250 °C) or to BN at higher temperature (>350 °C). Hydrogen is the main gaseous product of thermal decomposition of amidoboranes. The first step of the thermal decomposition of amidoborane salts occurs at ca. 60–110 °C resulting in desorption of a single molecule of hydrogen from each formula unit (
Table 5). Further decomposition steps occur at temperatures higher than 100 °C. The second molecule of hydrogen per formula unit is usually desorbed below 250 °C. From the point of view of hydrogen storage for support of low temperature fuel cell systems for automotive use, only hydrogen available below 100 °C can be useful [
93]. In this context, most amidoborane salts can offer at most only 2/5 of their hydrogen content.
For comparison, neat ammonia borane undergoes three exothermic decomposition steps (70–112 °C, 135–180 °C, and >450 °C) while releasing one hydrogen molecule per formula unit in each step, which is equivalent to ca. 6.5 wt% mass loss [
2,
7,
74,
76,
94]. Neat ammonia borane decomposes via a rapid uncontrollable process while evolving hydrogen contaminated with borazine and monomeric aminoborane [
94]. The decoposition can be accelerated using an addition of specific catalyst [
77] or changing of the process conditions [
95,
96,
97].
Most amidoborane salts decompose with melting via exothermic reactions (ca. −5 kJ/mol∙H
2) which are less exothermic than decomposition of ammonia borane (−20 kJ/mol∙H
2). Thermal decomposition of amidoboranes is thermodynamically favorable because of both the enthalpic and the entropic contributions to Gibbs free energy of this reaction being negative. Surprisingly, several bimetallic amidoborane salts decompose via slightly endothermic processes, namely: Ca(AB)
2 (+3.5 kJ/mol) [
31], NaMg(AB)
3 (+3.4 kJ/mol) [
42], Na
2Mg(AB)
4 [
45] and K
2Mg(AB)
4 [
45]. This result proves that thermodynamics of the decomposition process may to some extend be tuned by designing a composition of bimetallic amidoborane salts. A rule of thumb here is that the acid–base reactions stabilize the complex (bimetallic) system [
12], e.g.,
where KAB serves as a Lewis base, and Mg(AB)
2 as a Lewis acid. This, in turn, leads to less negative (or even positive) enthalpy of thermal decomposition of the product.
Possibility of onboard regeneration of a decomposed material is very important when considering amidoborane salts as hydrogen source for automotive use. According to DOE targets, reloading of an “empty” hydrogen store should be possible under mild pressure of hydrogen (ca. 5 bars) within 5 min [
93]. The condition of thermodynamic reversibility of absorption–desorption reactions at close-to-ambient (p,T) conditions is an equivalent (except for so-called high-entropy storage systems) to the enthalpy of thermal decomposition falling around +40 kJ/mol. Clearly, since the thermal decomposition of amidoboranes is exothermic or only slightly endothermic, the regeneration process is thermodynamically unfavorable. Indeed, regeneration of decomposed amidoborane salts with gaseous hydrogen under high pressure has not yet been successfully completed, even in the case of the salts decomposing in endothermic reactions [
18,
42]. Thus, chemical pathways of regeneration of the discharged material are now under development.
There has been some discussion in the literature whether hydrogen desorbed upon heating of light alkali metal amidoboranes is pure or it contains contaminants which might be harmful for fuel cells catalysts and membranes (
Figure 6). A number of papers were published reporting alkali metal amidoboranes as sources of pure hydrogen, e.g., LiAB [
18,
19,
86], NaAB [
18,
86] and KAB [
25]. On the other hand, several other groups reported for these materials the desorption of hydrogen contaminated with ammonia: α-LiAB [
20,
25,
39,
89,
90], β-LiAB [
20] and NaAB [
23,
24,
87,
88,
98]. Some amidoborane salts (KAB [
25,
89], Mg(AB)
2 [
29], Al(AB)
3 [
35], and LiAl(AB)
4 [
35,
40]) were reported to desorb pure hydrogen and, currently, there are no contradictory reports. The remaining known amidoborane compounds (RbAB [
26], CsAB [
26], Sr(AB)
2 [
43], Y(AB)
3 [
38], LiNa(AB)
2 [
39], NaAl(AB)
4 [
41], NaMg(AB)
3 [
42], Na
2Mg(AB)
4 [
44]) decompose releasing hydrogen contaminated with ammonia, diborane or borazine, all of which are harmful for PEM and alkaline fuel cells.
Mechanism of thermal decomposition of amidoboranes is not well determined because of differences in reported observation, however, a few groups proposed possible scenarios of this process. The first step of the most often cited mechanism of hydrogen evolution is identical to that determined for ammonia borane [
99], and it relies on intermolecular recombination of protonic and hydridic hydrogen atoms to form dihydrogen molecule [
18,
23,
24].
Spielmann et al. performed a detailed study on the mechanism of dehydrogenation of Ca(AB)
2 suggesting a head-to-tail dimerization of amidoborate anions yielding [HN-BH-NH-BH
3]
2− ions [
32], which is in good agreement with mechanism presented earlier by Stowe et al. [
99]. Xiong et al., describing thermolysis of LiAB and NaAB, pointed out that the key feature enabling formation of dihydrogen is presence of both positively and negatively charged hydrogen atoms in amidoborate anions [
18]. Local recombination of H
δ+ and H
δ− atoms does not involve mass transport through different phases [
18]. In the next paper, Xiong et al. reported formation of amorphous BN along with reformation of NaH upon heating NaAB to 200 °C (see
S4 in Supplementary Materials) [
22]. Luedke et al. proposed metal ion assisted hydride transfer involving the scission of M–N and B–H bonds resulting in the formation of MH and dimerization of two amidoborate anions upon thermolysis of amidoborane salts [
86]. MH further reacts with an imide group of the dimer to release hydrogen molecule. Next, hydrogen molecule is formed in recombination of H
δ+ and H
δ− atoms [
86]. Fijalkowski et al. [
23] further proposed an intermolecular transformation of two amidoborate anions forming hypothetical product containing BH
3NHNaBH
3− anions and Na[NH
3]
+ cations, which could easily evolve weakly coordinated ammonia [
23,
24]. Later, this hypothesis was substantiated by Fijalkowski et al. [
87]. The intermolecular step responsible for both ammonia evolution and initiation of the polymerization of the solid residue was finally suggested to correspond to trimerisation of amidoborate anions with the concomitant formation of BH
3NH
2BH
2NH
2BH
3− anions and ammonia molecules [
87]. The latter hypothesis finds strong support from the X-ray, Raman and FTIR studies [
87]. One can find signals from LiBH
3NH
2BH
2NH
2BH
3 and NaBH
3NH
2BH
2NH
2BH
3 in the X-ray powder patterns of, e.g., α-LiAB [
18], β-LiAB [
20], NaAB [
21], LiNa(AB)
2 [
39] and LiAB∙AB [
55], synthesized in recent years. LiBH
3NH
2BH
2NH
2BH
3 and NaBH
3NH
2BH
2NH
2BH
3 were earlier observed by Evans [
100] and Ryan [
21] but at that time their role as intermediates of the thermal decomposition of LiAB and NaAB had not yet been understood. Selected mechanism discussed above are presented in
Figure 7 [
19,
23,
32,
86,
87].
In
Table 6, we summarize experimentally obtained data on thermal decomposition of amidoborane salts. We would like to point out that, for most of the presented compounds, mass loss is not connected exclusively with evolved hydrogen but also other compounds (e.g., ammonia, borazine, and diborane) that are found to be undesired hydrogen contaminants.