A Review of the MSCA ITN ECOSTORE—Novel Complex Metal Hydrides for Efficient and Compact Storage of Renewable Energy as Hydrogen and Electricity

Hydrogen as an energy carrier is very versatile in energy storage applications. Developments in novel, sustainable technologies towards a CO2-free society are needed and the exploration of all-solid-state batteries (ASSBs) as well as solid-state hydrogen storage applications based on metal hydrides can provide solutions for such technologies. However, there are still many technical challenges for both hydrogen storage material and ASSBs related to designing low-cost materials with low-environmental impact. The current materials considered for all-solid-state batteries should have high conductivities for Na+, Mg2+ and Ca2+, while Al3+-based compounds are often marginalised due to the lack of suitable electrode and electrolyte materials. In hydrogen storage materials, the sluggish kinetic behaviour of solid-state hydride materials is one of the key constraints that limit their practical uses. Therefore, it is necessary to overcome the kinetic issues of hydride materials before discussing and considering them on the system level. This review summarizes the achievements of the Marie Skłodowska-Curie Actions (MSCA) innovative training network (ITN) ECOSTORE, the aim of which was the investigation of different aspects of (complex) metal hydride materials. Advances in battery and hydrogen storage materials for the efficient and compact storage of renewable energy production are discussed.


Introduction
The development of materials for energy storage applications is one of the great challenges of our generation. To reduce the carbon footprint and to extend the use of renewable energy sources, it is necessary to store energy for times of great demand. Over recent years, energy production from renewables has increased worldwide. This results in a demand for sustainable energy storage systems for intermittent sources.
Metal hydrides, or in general, the energy carrier hydrogen, are ideal for sustainable storage systems and current products already show their potential in practical applications. Implementations can be found in the form of metal hydride as chemical energy storage in the case of power and heat distribution of the Henn-Na hotel in Japan (H2one from Toshiba), as well as in the 700-bar hydrogen storage tanks of several electric vehicles, such as the Toyota Mirai, Honda Clarity or Hyundai Nexo [1]. An electrochemical storage character is applied already in nickel-metal hydride batteries (Ni-MH), with over 1 billion cells sold in 2015 (vs. 6 billion lithium (Li) ion battery cells) [2]. Furthermore, a recent study suggests implementing metal hydrides as a thermal battery for concentrated solar power plants [3].
Nonetheless, not only efficient and compact storage of renewable energy are needed for the future. As the energy consumption as well as performance of portable electronic devices, such as laptops, cameras and smartphones have tremendously increased over the past decades, high-energydensity power supply systems are also required. In this context, a vast amount of research is still being conducted to improve electrochemical systems such as the Li-ion battery. Further efforts concentrate on so-called "post Li-batteries" which describe materials for all-solid-state Li-ion batteries, or even batteries with alternative working ions including sodium (Na + ), magnesium (Mg 2+ ) and calcium (Ca 2+ ), while aluminium (Al 3+ )-based compounds are not considered due to the unavailability of suitable electrode and electrolyte materials [4][5][6].
The aforementioned materials are considered as cheaper and more abundant. With respect to multivalent ions, higher energy densities are potentially achievable compared to current state-of-theart technology. Nevertheless, transport properties of larger ions are challenging and directly correlated to the underlying crystal structure. Understanding the associated structural as well as dynamic changes during ion insertion and ion extraction is crucial for the improvement of the electrode materials for "post Li-batteries".
This review focusses on eight major points of interest, starting from describing the use of metal and complex hydrides for secondary batteries including Li-ion, post Li and all-solid-state batteries based on Li. Half and full-cell performance are introduced at the end of the Chapter 2.1. Chapter 2.2. turns the focus on the electrochemical performance of closo-borates, an interesting class of complex hydrides in Na-based batteries. Moving from batteries to hydrogen storage in metals, Chapter 3.1 examines the latest reports on pure metal hydrides while concentrating on one of the most promising metals, magnesium. In Chapter 3.2, the chemically bounded storage of hydrogen is discussed with an emphasis on amide-based systems. Chapter 3.3 draws attention to the eutectic metal borohydride systems that show interesting behaviour while Chapter 3.4 presents the kinetic tailoring of LiBH4 with effective additives. In Chapter 3.5, the authors consider the role of nanoconfinement in enhancing hydrogen storage capacities, while in Chapter 3.6, the synthesis and performance of rare earth borohydrides (monometallic, bimetallic and composites) are presented. The table of content presented hereafter gives further details on the sub-chapters and each chapter ends with a short conclusion on the key achievements on the progress of hydrogen used as a versatile energy storage material.

Metal Hydrides for Li-ion and Post Li Batteries
Secondary lithium-ion batteries (LiBs), as a feasible solution for energy storage of intermittent energy production, are currently one of the most effective battery systems [7]. With high performance, LiB systems exhibit an impressive energy density up to 210 Wh kg −1 (650 Wh L −1 ) compared to other battery technologies and take a vital role in markets as the most attractive electrical energy storage system [8]. Since their commercialisation by Sony in 1991 [9], significate progress has been achieved in increasing the performance of LiBs by using new cathodes. The latter were first based on lithium iron phosphate (LiFePO4), while nanostructuration and nanoconfinement also played a role in increasing performance. Nowadays, nickel-rich layered cathode materials are a focus for automotive applications [10][11][12][13][14][15]. For the anode, a material with higher capacity is still under investigation as a replacement for commercial graphite whose capacity only reaches 375 mAh g −1 (840 Ah L −1 ). Metal hydrides have recently been proposed as a potential candidate for anode replacement in LiBs due to their high intrinsic specific energy capacity; i.e., up to 2038 mAh g −1 for MgH2 and 1074 mhA g −1 for TiH2 [16,17]. Through a conversion reaction with lithium, metal hydrides are evidenced to possess remarkable properties compared to other conversion materials, such as low reaction potential, small hysteresis and high theoretical capacity.
Safety is another important request for LiBs. The concept of all-solid-state lithium-ion-batteries (ASS-LiBs) has been developed to solve vital drawbacks of conventional LiBs with liquid electrolyte. These challenges include electrolyte leakage, the narrow range of temperature operation, causing the evaporation of the solvent at higher temperatures, decomposition of the liquid electrolyte at the electrodes surface to form a solid electrolyte interphase (SEI), and expansion and explosion of the solvent upon heating. The solid electrolyte reduces the associated explosion risks and it facilitates the manufacture process, allowing smaller cells with a higher specific energy capacity and preserving contact between components [18,19]. The main challenge is to find a lithium ionic conductor with sufficient ionic conductivity (at least 10 −4 S cm −1 under working conditions), low electronic conductivity (<10 −10 S cm −1 ), and good electrochemical stability. It should also be compatible with lithium and electrode materials and have low weight density. Complex metal hydrides, such as LiBH4, LiNH2 and their combinations with other compounds, e.g., LiX (where X is a halide), phosphorous compounds, sulphurous compounds, etc., meet most of the requirements listed above, while the sensitivity to air is non-negligible and requires a compromise between performance and safety in future applications.
Starting with the possibilities of metal hydrides as battery materials, the following chapters will explore the potential of hydrides in a battery cell not only as electrodes but also as electrolytes.

Metal Hydrides as Negative Electrode Material with Liquid Electrolyte
In 2008, Oumellal et al. [16] proposed, for the first time, the use of metal hydrides as a negative electrode for Li-ion batteries. While the state-of-the-art technology in commercial Li-ion batteries is based on an intercalation/deintercalation reaction in the graphite electrode, metal hydrides undergo a conversion reaction with the lithium ion, following the general Equation (1): During the lithiation reaction, the metal hydride undergoes a reduction to its metal and lithium hydride is formed. The lithiation can be completed in one or more steps, depending on the nature of the hydride. This reaction is favorable for all the metal hydrides with Gibbs free energy of formation less negative than that of LiH. The main advantage in using metal hydrides as electrode material is their high gravimetric capacity compared to graphite. Per hydrogen atom exchanged during the reaction, one Li is stored within the electrode, meaning that the minimum quantity stored is one Li per formula unit in the case of monohydrides. On the other hand, only one Li atom is intercalated per six C atoms in a commercial graphite electrode. Therefore, as shown in Figure 1, several metal hydrides exhibit a theoretical capacity that is much higher than graphite. Furthermore, the thermodynamic potential of metal hydrides is in the range of 0.1-1 V vs. Li + /Li, allowing a high energy density.  [20] and shows the most important values for this chapter.
With a theoretical gravimetric capacity of 2038 mAh g −1 , magnesium hydride (MgH2) was the first hydride investigated as negative electrode and even though a high experimental capacity during lithiation was reached, the MgH2 electrode exhibited a poor reversibility already within the first discharge/charge cycle [17,[21][22][23]. This drawback was shown for several other metal hydrides [21][22][23][24][25] and despite the strategies suggested in the literature, poor reversibility still hinders practical application of metal hydrides as electrode materials. Several reports [18,20,21,[25][26][27] have tried to elucidate this poor reversibility, highlighting the fact that during lithiation/delithiation, the electrode undergoes severe volume changes which might lead to cracks and a loss of contact between particles. Moreover, the formation of LiH could decrease the electronic conductivity, hindering the extraction of Li from the electrode [27]. Finally, the conversion reaction is a solid-solid reaction occurring at room temperature (RT) involving complex reactions such as hydride decomposition, phase nucleation and growth, and the diffusion of atoms within the electrode. Therefore, mass transport issues could affect the reversibility as well.
With the aim of better understanding the issues related to the poor reversibility of metal hydrides, Berti et al. [28] used MgH2 thin films deposited on Cu current collectors as a model system to investigate the mechanism occurring during lithiation and delithiation, focusing on the effect of this mechanism on electrode performance. Similarly to powder-electrodes, MgH2 thin films can be fully lithiated during the first discharge but show poor reversibility during delithiation. Thanks to transmission electron microscope (TEM) microscopy, the structure and phase distribution within the electrode during discharge/charge were investigated and analysed. As displayed in Figure 2, the focused ion beam (FIB) cross sections of the film grow in thickness because of the conversion of MgH2 to Mg and LiH. During delithiation, they shrink due to partial reformation of MgH2. However, no voids or cracks are noticed along the cross section, implying that no severe structural damage or loss in contact occurred within the first cycle, which contradicts the above-mentioned initial hypothesis. Volume changes are not the main issue of poor reversibility.
Berti et al. (authors mentioned in bold belong to the MSCA ITN ECOSTORE) performed additional electrochemical impedance spectroscopy (EIS) at different steps and showed that the resistance of the lithiated electrode is indeed lower than the pristine, thus indicating that the electrons are not blocked by the presence of LiH. Their result is supported by the fact that in the TEM images ( Figure 2), the lithiated zones are always intimate mixtures of Mg within the LiH matrix; thus, the formation of the metal already provides a path for the electrons. After eliminating mechanical and conductivity issues, the authors suggested that the main cause for the poor reversibility is probably the mass transport within the electrode. Since Li can easily diffuse inside the electrode converting all MgH2, it is likely that atomic hydrogen transport is the main cause. Indeed, the diffusion coefficient of hydrogen in MgH2 is known to be low at RT, as described by Huen et al. and others [27,29]; hence, it is assumed that the reformation of MgH2 begins on the surface of the particles and proceeds through the center, which could explain its partial reformation. Consequently, in order to enhance the reversibility, high mobility and short diffusion paths for hydrogen are desirable. The mobility of species can be enhanced mainly in two ways, either by working at a higher temperature or by providing preferential pathways (i.e., adding a phase with a higher diffusion coefficient), whereas short diffusion pathways could also be provided by the nanostructuration of the electrode.
With the purpose of improving the electrochemical properties of MgH2, Huang et al. [30] studied the effect of adding TiH2 to MgH2, preparing a 0.7MgH2 + 0.3TiH2 composite by means of reactive ball milling. Indeed, TiH2 is known for its good electrical conductivity [31] and hydrogen diffusion coefficient [32], which might improve the overall electrode reaction. The electrode prepared by Huang et al. showed improved electrochemical properties compared to MgH2- [16] and TiH2- [23] based electrodes in terms of reversible capacity and polarization, highlighting a cooperative effect between these two hydrides in a conversion reaction. During lithiation, both hydrides were reduced to Mg and Ti, forming LiH, and after de-lithiation, the partial reformation of both hydrides was observed. It was demonstrated by the authors that the presence of TiH2 enhances the kinetics of the conversion of MgH2, whereas MgH2 enables the partial reformation of TiH2. Expanding the work of Huang, Berti et al. [33] prepared different mixtures of yMgH2 + (1 − y)TiH2 (with molar ratio 0.2 ≤ y ≤ 0.8) by means of ball milling to investigate the ratio effect on the conversion reaction.
In agreement with Huang et al. [30] Berti et al. [33] also found the partial reformation of both MgH2 and TiH2 in a MgH2-rich composite electrode. However, Berti et al. showed that the reformation of these hydrides does not always appear in all the mixtures, but strongly depends on the molar ratio. Indeed, for TiH2-rich composite (y = 0.2) neither MgH2 nor TiH2 were observed during the first de-lithiation of the electrode. For the equimolar composite (y = 0.5), only the partial reformation of TiH2 was clearly observed, and finally, in MgH2-rich electrode (y = 0.8) both hydrides were reformed.
Additional studies by Berti et al. [34] showed that the highest reversibility (75%) is obtained for 0.7MgH2 + 0.3TiH2. As better shown in Figure 3a, the TiH2 phase exhibits a higher reversibility than MgH2 in this composite, with 90% vs. 40%, respectively, within the first cycle. Further investigation of the reversibility of the TiH2 phase in this composite showed that when the conversion reaction is limited to this hydride (between 0.1 and 0.45 V, as shown in Figure 3b, in order to avoid both the contribution of MgH2 and Mg-Li alloy reactions), a high initial reversibility and good capacity retention in the subsequent cycles were achieved. This improvement of the performance for the TiH2 phase was attributed to the intimate mixtures and short diffusion pathways provided by the mechanochemical synthesis, which also offers a high density of Ti/LiH interfaces. This latter hypothesis is also supported by the work of Oumellal et al. [35], in which intimate mixtures of LiH + M (where M = Mg and Ti) were prepared by mechanochemical synthesis. Interestingly, when a mixture of Ti + 2LiH prepared by ball milling is de-lithiated, the reformation of TiH2 is partially observed (around 40%). This fully supports the hypothesis that good M/LiH interfaces and short diffusion paths are needed in order to extract lithium from the electrode and reform the hydride. Moreover, these latter conclusions clearly show that the preparation of the electrode is key to improving the overall conversion reaction. A direct comparison between the 0.7MgH2 + 0.3TiH2 electrodes of Berti et al. [34] and Huang et al. [30], cycled without the contribution of the alloying reaction, shows differences in the contribution of the hydrides. Berti achieved a reversible capacity for MgH2 of 0.56Li (i.e., 40%) and 0.54Li for TiH2 (90%), whereas Huang obtained 1.1Li (78%) and 0.4Li (67%), respectively. These results might be due to the different electrode preparations in these works. Indeed, the composites were prepared with different mechanochemical procedures, also adding different additives to the composites. Moreover, the electrodes were cycled at different Crates, and as clearly shown by Huang, a faster discharge/charge rate leads to a decrease in the reversible capacity and to an increase in the polarization, highlighting the importance of an appropriate C-rate on the overall performance. The fact that preparation plays a key role was also shown by Oumellal et al. [16], who cycled two MgH2 electrodes prepared in different ways under the same conditions (C/20 between 3 and 0.15 V). The electrode prepared with MgH2, ball milled for a longer period, exhibited a better performance, achieving a reversibility of 1.5Li, with an increase of 20% compared to the electrode ball milled for a shorter time (1.1Li). In light of these facts, it is clear that not only the composition of the electrode is relevant for the sake of the performance, but also its preparation and working conditions must be taken into account with the purpose of further improving the reversibility and cycle-life of metal hydride-based electrodes. Apart from mechanochemical synthesis, the cycling performance of metal hydride electrodes has shown to be improved by nanoconfinement using carbonaceous scaffolds [36][37][38][39]. Prior to application in Li-ion batteries, nanoconfinement was shown to modify both the thermodynamic and kinetic properties of metal hydrides for hydrogen storage (Chapter 3.5. and Refs. [40,41]). Therefore, several research groups have implemented this bottom-up approach. Oumellal et al. [36] reported that MgH2 infiltrated in high surface area graphite can achieve a high reversible capacity of 500 mAh g −1 . Similarly, Huen et al. [37] showed that it is possible to enhance the reversibility of sodium alanate (NaAlH4) by nanoconfinement from 30% up to 70%. Using operando X-ray powder diffraction (XPD) to investigate the conversion reactions of NaAlH4 during discharge, it was found that the formation of intermediate LiNa2AlH6 is dominant in the ball-milled sample. However, in the nanoconfined sample, the formation of another intermediate Na3AlH6 phase occurs and competes with the formation of LiNa2AlH6. In general, the confinement within a scaffold improves the reactivity of metal hydrides not only by shortening the diffusion path, but also by altering the conversion mechanism.
Additionally, carbonaceous scaffolds are similar to graphite and allow intercalation of Li ions [37,[42][43][44]. Since the reaction potential of some metal hydrides and carbon are in the same range, it is difficult to separate and avoid the intercalation of Li into carbon. Therefore, in order to further enhance the performance of nanoconfined metal hydride electrodes, novel scaffolds that are electrically conductive but inert to Li ought to be explored.

Promising Complex Metal Hydrides as Solid-State Electrolytes
Since the discovery of the first ionic conductor PbF2 and Ag2S by Michael Faraday in the 19th century, further conductors have been described throughout the centuries [45,46] and material scientists have achieved great progress in novel ionic conductors with a higher conductivity and better applicable properties. These compounds are also promising for all-solid-state batteries (ASSBs). There are many different families of ionic conductors for ASSBs which may offer high ionic conductivity near RT: LIthium Super Ionic CONductor (LISICON), argyrodites, lithium-nitrides, lithium-hydride family, garnet type, perovskites, etc. [47]. Generally, the ionic conductivity of Li + in the materials can be improved by chemical substitution and material design [48]. The transportation mechanism of Li-ion in super ionic conductors has been reported as well [49][50][51][52][53][54]. Though the general understanding of the fundamental mechanism leading to high Li-conductivity is not well established, it is clear that for a specific material with high ionic conductivity, the availability of ion diffusion channels or conduction channels is a vital prerequisite. Metal hydrides and hydride complexes exhibit high hydrogen mobility. Therefore, they possess diffusion channels, which might be used to transport other ions, such as alkaline-earth cations, finding possible application as super ionic lithium conductors [55][56][57][58][59][60].

Lithium Borohydride
In the field of energy applications, LiBH4 with its high gravimetric hydrogen content was originally studied as a material suitable for solid-state hydrogen storage (see Chapter 3.2 and Reference [56]). Due to the high mobility of hydrogen in LiBH4, it was suggested that lithium mobility could also be promoted, which makes the material act as a Li ionic conductor. The advanced mobility of lithium ion in LiBH4 was first confirmed by Matsuo et al. in 2007 [59]. The crystal structure of LiBH4 is shown in Figure 4. It transforms from a low-temperature (LT), orthorhombic phase to a hightemperature (HT) hexagonal phase at ~117 °C. The LT-orthorhombic and HT-hexagonal phases exhibit conductivities of 8.6 × 10 −8 and 10 −3 S cm −1 at 20 and 120 °C, respectively [59]. An explanation can be found in the LT phase, where it was reported that Li + ions are blocked by [BH4] − tetrahedral anions. The diffusion of Li + ions needs to be facilitated by the rotation of [BH4] − tetrahedrons, which needs a high activation energy. On the other hand, in the HT phase, Li + ions are not blocked by the tetrahedrons. Li + ions diffuse along the crystallographic axes a and b easily while rotation of the tetrahedrons also facilitates the migration of Li + [61][62][63]. While transforming from the LT to the HT phase, the activation energy for the Li + ion mobility decreases from 0.69 to 0.53 eV [62]. Moreover, LiBH4 is an electronic insulator in both structures. Theoretical calculations show a large band gap of ~6.7 eV [60]. While powders are most commonly used for such investigations, a similar behaviour is reported for LiBH4 melt-frozen films. Trück et al. [64] described an auspicious method of film preparation by spin coating and investigating the ionic conductivity of crystalline LiBH4 films. The authors have reported the phase change from LT to HT modification at 110 °C with an ionic conductivity of the HT phase of 10 −3 S cm −1 .

Figure 4.
Conductivity of LiBH4 at non-conductive low temperature phase (orthorhombic structure) and at conductive high temperature phase (hexagonal structure). The blue spheres are lithium ions and the light blue tetrahedrons are BH4 anions with a yellow sphere (boron) inside them.
With such high lithium conductivity for the HT phase, the electronic insulator LiBH4 becomes a very interesting candidate for solid electrolytes in ASSBs. Several successful achievements have been attained using different electrode materials combined with LiBH4 as solid electrolyte [65,66].

Lithium Nitride and Lithium Hydride
The interest in Li3N as an ionic conductor was based on its hexagonal crystalline structure made of a Li2N − layer and a pure Li + ionic layer, resulting in the free movement of Li + ions in a two-dimensional pathway within the lattice [67,68]. Thus, the ionic conductivity of Li3N reaches 1.2 × 10 −4 S cm −1 at 27 °C [67]. However, this compound decomposes at low potential of 0.44 V limiting its application in ASSBs [68,69]. Nevertheless, when Li is partially replaced by hydrogen, forming lithium imide Li2NH, the electrochemical stability of the compound is improved compared to Li3N [68]. It was shown experimentally that Li2NH has high ionic conductivity at RT with 2.54 × 10 −4 S cm −1 (according to activation energy Ea = 0.63 eV). This high conductivity originates from the presence of H + in the Li2N − layer. N-H bonds form, weakening the Li-N bonding and facilitating the diffusion of Li + . The decomposition voltage of Li2NH is 0.7 V, which is higher than that of Li3N but still too low for a solid electrolyte [70].
Lithium imide (Li2NH) has been studied extensively as a prospective reversible hydrogen storage material [71,72]. Nearly four decades ago, Boukamp et al. [73] reported, for the first time, high Li-ion conductivity in Li2NH (>10 −4 S/cm at RT) and since then, it has been reconsidered as a promising solid-state electrolyte in the Li-ion battery. Recent interest in Li2NH as a practical, cost effective, high-density electrolyte comes from several advantages. For example, its benefits are its lightweight (molar weight = 28.9 g mol −1 ), abundant elements (especially nitrogen), its thermal stable nitrogen sublattice, appropriate suitable anion dynamics for cation diffusion, uncomplicated synthesis methods, etc. Nevertheless, the understanding of the electrochemical properties of Li2NH is still debated in literature, especially ionic conduction, leading to several crystal structures and controversial results on resolved structures [70].
Recently, the connection of the cubic and the orthorhombic phases of Li2NH has been elucidated by Paik et al. [70], especially, in terms of the structural evolution, showing that the formation of the aforementioned phases is part of a continuous structural ordering [70]. The orthorhombic Li2NH phase was identified to undergo the order−disorder phase transformation. Although no rapid change in ionic conductivity has been reported from this phase transformation, which is confirmed by the ionic conductivity estimated for these structures between 27 and 127 °C . The ionic conductivity data also help to reveal a likely cation diffusion mechanism. Ionic conductivity in cubic Li2NH is reportedly higher than in the orthorhombic Li2NH. Furthermore, contrarily to a prior prediction [73], the electrochemical stability window in the cubic Li2NH is wider for the Li/Li2NH interface (>5 V).
Structural differences between the orthorhombic α-Li2NH phase and cubic β-Li2NH phase can be found in Figure 5 and have been determined by different characterization methods such as Raman and XPD (Figure 5b,c) [70]. However, the structural evolution of the aforementioned structures may be described as a transformation of tetragonal LiNH2 into α-and β-Li2NH via the formation of nonstoichiometric Li-NH phases, Li(1−x)NH(2−x) (0 < x < 1) [70]. This mechanism for transformation is based on the proposed study of the dehydrogenation model for LiNH2 by David et al. [74]. According to their model, the filling of vacant Li sites in the lattice of LiNH2 initiates a structural transition from the Li-poor to Li-rich phase and thus, from tetragonal into cubic Li2NH phase and converting a poor Li-ion-conductor into a better one, respectively. Theoretical calculations suggest that [NH2] − groups block Li + diffusion in LiNH2 and indeed, LiNH2 is almost an ionic insulator at RT with a high Li + ion activation energy of 1.05 eV [68]. The ionic conductivity for β-Li2NH has been reported to be 5 × 10 −4 S cm −1 , compared to ~10 −4 S cm −1 in α-Li2NH [70], in a temperature range between 27 and 107 °C . The endothermic phase change from α-to β-phase at 77-87 °C did not yield a drastic change in ionic conductivity. The increase is rather a result of cation and anion ordering.

Binary Phases of Hydrides with Lithium Halides
Similarly to the idea of replacing one Li by hydrogen in Li3N to stabilize the compound, lithium halides were added to LiBH4 to form a solid solution, expecting a HT LiBH4 structure stabilized at RT. For this reason, mixtures of LiBH4 with LiCl, LiBr, and LiI were investigated [75][76][77][78][79][80]. Solid solutions were obtained by high-energy ball milling or annealing of LiBH4 with different ratios of LiI under Ar and promising conductivities were achieved. An array of solid solutions of Lix+1I(BH4)x was studied by Orimo and co-workers, where the obtained Li4I(BH4)3 phase reached super ionic conductivity as high as 4 × 10 −4 S cm −1 at RT [75,77,78]. XPD studies confirmed the formation of a single phase without any impurities [75]. Additional DFT calculations showed that in the presence of LiI, lithium defects were easily formed at RT in the solid solution (energy formation of 0.44 eV). This was explained by the fact that energy barriers which were found between stable defect sites are low (0.2 to 0.3 eV), giving rise to a high defect mobility [81].
Following the same principle, solid solutions of LiNH2 and LiI at different ratios were studied [82][83][84]. The ionic conductivity of Li3(NH2)2I has been reported to be 1.7 × 10 −5 S cm −1 at 27 °C (Ea = 0.58 eV). However, the EIS spectrum of the compound showed a significant ionic resistance of both the bulk material and grain boundaries. The grain boundary resistance was attributed to the mechanical property of the compounds and the preparation method [82].
The heat treatment of LiNH2 and LiCl (molar ratio = 1:3) resulted in the formation of rhombohedral and cubic polymorphs of Li4(NH2)3Cl, and at lower LiCl ratios, Li7(NH2)6Cl formed [85]. These compounds were reported to have faster hydrogen desorption kinetics than the halidefree system. LiNH2 has the same decomposition temperature but additional NH3 was detected in addition to H2 [85]. The halide anion is believed to increase hydrogen mobility in the system. In addition, Li sites in the structure are not completely occupied which might increase Li + mobility and thus result in a higher ionic conductivity [85].
Two binary single phases Li2BH4NH2 and Li4BH4(NH2)3 were obtained, when the complex hydrides LiBH4 and LiNH2 were ball milled in ratios of 1:1 and 1:3, respectively. A conductivity of 2 × 10 −4 S cm −1 was measured for both ratios at RT [86]. At 97 °C, the conductivities increased up to 6 × 10 −2 S cm −1 and 10 −3 S cm −1 , respectively. Similarly to what was observed in pure LiBH4, EIS of the mixture exhibits no significant grain boundary resistance [86].
LiBH4-Li2NH: Li5(BH4)3NH-thorough characterization of a cluster complex hydride A new complex hydride with the formula Li5(BH4)3NH has been synthetized by ball milling of LiBH4 and Li2NH in a 3:2 molar ratio and by post annealing [72]. An orthorhombic phase was already observed by Blomqvist et al. [87], but could not be isolated or formed pure. Recently, the crystal structure of Li5(BH4)3NH was solved in space group (s.g.) Pnma [72] with unit cell parameters a = 10.2031(3), b = 11.5005(2) and c = 7.0474(2) Å at 77 °C. DFT and synchrotron radiation X-ray powder diffraction (SR-XPD) measurements followed the model previously established for the 1:1 composition of LiBH4:Li2NH by Hewett et al. [88], while solid-state nuclear magnetic resonance (NMR) measurements confirmed the chemical shifts calculated by DFT. The latter calculations underlined the ionic character of this lithium-rich compound, revealing a potential for use in LiBs.
Li5(BH4)3NH is a mixed anion salt comprised of borohydride and imide anions and is based on packing of a small complex anion BH4 − and big complex cation [Li5(NH)] 3+ . The latter corresponds to higher coordinating power of imide compared to amide [72]. In Orimo et al. [89], the reported complex of Li4(BH4)(NH2)4 is described by the replacement of amide anions by imide anions. Li4(BH4)(NH2)4 is based on packing of small cation Li + and complex anions BH4 − and NH2 − , resulting in close to cubic closed packing of anions. Figure 6a shows the crystal structure of Li5(BH4)3NH which contains a particular structural feature. The imide anion is coordinated by a square pyramidal cluster of five Li atoms. With five cations per four complex anions, the compound is Li rich and the coordination of the borohydride anions is five-fold. Nevertheless, the latter is more irregular than the coordination of imide. All Li atoms are a fragment of immensely deformed tetrahedral coordination comprising of one NH2 − anion and three BH4 − . The anion [Li5(BH4)4(NH)] − has an overall C2v symmetry when surrounded by four BH4 anions which has been determined from geometry optimization. The HOMO and HOMO-1 orbitals are shown in Figure 6b for the anion [Li5(BH4)4(NH)] − which are the px and py N atomic orbitals and both have the same energy. This specific arrangement proposes that the surrounding [BH4] − anions play a crucial role by stabilizing the core of [Li5(NH)] 3+ . With the exception of the covalent N-H bond, the aforementioned square pyramidal [Li5(NH)] 3+ core is of a pure ionic nature. It polymerizes into the Li5(BH4)3NH crystal structure while keeping its complex cation character balanced by three anions of BH4 − . Al3Li4(BH4)13 is the other known example of a complex cation which was found with [Li4(BH4)] 3+ as tetrahedral cation [90].
Wołczyk et al. [72] reported a comparison of the ionic conductivity data for LiBH4, Li2NH and Li5(BH4)3NH. The RT ionic conductivity of the Li5(BH4)3NH compound reaches 10 −6 S cm −1 with an activation energy of 0.73 eV. Showing one order of magnitude higher conductivity than LiBH4 does at RT and three orders of magnitude lower than Li2NH. The observed inconsistencies in the Li-ion conductivity in the three orthorhombic complex hydrides are elucidated in the following and explained by ionic conduction mechanisms. Noritake et al. have reported that the rotational motion of the translational static BH4 − anions improves the mobility of Li + ions in case of LiBH4 [91] while charged vacancies or Frenkel pair defects are considered in the case of Li2NH [68]. For the orthorhombic Li5(BH4)3NH compound, the Li-ion conductivity may be affected by the two anion configurations, which are closely related to the determined crystal structure. It is worth pointing out that the Li-ionic conductivity of Li5(BH4)3NH is comparable to some of the newly reported binary complex hydrides, including LiBH4-LiX solid solutions (X = Cl, Br, I) [75] and Li4(BH4)(NH2)3 [92]. This suggests a connection of Li-based complex hydrides containing [BH4] − and [NH] 2− anions which are potential choices to support a high Li-ionic conduction at RT [75].

Argyrodite Structure Materials
Since the first solid ionic conductor, researchers have worked on the improvement of ionic conductivity by replacing the cation [93] and anion [94] in the original substance. Following the high mobility of Ag + in mineral argyrodite Ag8GeS6, Deiseroth et al. discovered that substituting Ag + by other cations in the structure led to a very high cationic conductivity [93]. Subsequently, Li6PS5X compounds (X = Cl, Br or I) adopting the argyrodite structure were synthesized and the authors reported high conductivities [93,95]. For X = Br, the ionic conductivity reaches 10 −3 S cm −1 [96,97] which is comparable to some liquid electrolytes. Ionic conductivity (σ) of other halide argyrodites (Arg-X) decreases with the radius of the halide anion, i.e., σArg-Br > σArg-Cl > σArg-I.
The halide argyrodites with a high lithium ion conductivity are a new group of lithium super ionic conductors. The relationship between the ionic conductivity and the radius of the halide anions suggests that anion substitution of the halide can modify the lattice parameter and improve the conductivity. In the frame of studying the practical applications, the [BH4] − anion was chosen as a promising substituting anion which can potentially enhance the Li + conductivity of these argyrodites. Dao et al. [98] synthesized the argyrodite Li6PS5I5/6(BH4)1/6 and reported a significant improvement of ionic conductivity of 6 × 10 −4 S cm −1 at RT, two orders of magnitude higher than the original compound [99]. The crystalline structure of the cubic argyrodite 4 ̅ 3 remained; only the lattice parameters of the material was varied due to the difference of ionic radii. Surprisingly, while the conductivity of Li6PS5I5/6(BH4)1/6 was ameliorated, that of Li6PS5Br5/6(BH4)1/6 was reduced compared to the pure halide argyrodite, which can be explained by the local geometry constraints of the Li diffusion pathway. The conductivity of the Li6PS5BH4 synthesized by mechanochemical reaction is as high as 1.9 × 10 −3 S cm −1 at RT [100]. The combination of the borohydride with sulphide and thio-sulphate compounds was also found to be appealing, resulting in a high ionic conductivity, a good electrochemical stability and compatibility with many electrode materials [101,102].

Application of Metal Hydrides in Solid-State Cells
After a careful exploration of the use of metal hydrides as electrode and electrolyte materials, it is important to review the latest reports on their use in all-solid-state cells. Half-cells and full-cells have been reported in the literature and are discussed here, revealing the importance of metal hydrides for future advancements in all-solid-state batteries [103].
Complex metal hydrides, including Mg2FeH6, have poor cycling performances at RT when standard liquid electrolyte is used [112]. Huen et al. [113] studied, for the first time, Mg2FeH6 in solidstate batteries using LiBH4 electrolyte. At high temperature, the first Coulombic efficiency and retained capacity of Mg2FeH6 are improved significantly. In addition, the decomposition of Mg2FeH6 into MgH2 was discovered during discharge. However, it should be noted that Mg2FeH6 in ASSBs was tested at 120 °C and it is difficult to conclude whether the formation of MgH2 is temperaturedependent. Since the conversion reaction of Mg2FeH6 is more complete, this chapter also provides new insight into the reaction mechanism, especially during charge.
Resuming the work on yMgH2 + (1 − y)TiH2 (with molar ratio 0.2 ≤ y ≤ 0.8) mixtures, Dao et al. [114] investigated the properties of these composites at 120 °C using LiBH4 as solid electrolyte. As shown in Figure 7, half-cells were cycled at different C-rates. Compared to RT work [33], Dao et al. found a higher reversibility for all the mixtures. Composite 0.7MgH2 + 0.3TiH2 exhibits the highest initial reversibility (~90%) among the electrodes, which is in agreement with previous findings. Moreover, after ten cycles at C/50, this mixture delivers the highest capacity (1250 mAh g −1 ). This is attributed to better kinetics of the composite y = 0.7 compared to other yMgH2 + (1 − y)TiH2 electrodes. On the other hand, the Ti-rich composite y = 0.2 was found to be reversible, although at RT, this composite did not show the reformation of both hydrides. Furthermore, even if the initial reversibility is around ~45%, in the subsequent cycles, it exhibits the highest capacity retention, suggesting that at 120 °C TiH2 possesses a better cycling stability than MgH2. However, when a faster C-rate is applied, all electrodes show a decrease in capacity due to kinetic limitations. Nevertheless, these electrodes still provide capacities above 300mAh g −1 , even at 1C, with a Coulombic efficiency generally above 90%. All Mg-rich electrodes (y ≥ 0.5) are able to retrieve their capacities when C/50 is restored as working condition, whereas y = 0.2 shows a poor recovery, suggesting some degradation within the electrode due to the fast cycling. The authors suggest that the volume changes occurring during discharge/charge could lead to voids or cracks within a solid-state half-cell and thus, a loss in contact between the particles. Certainly, achieving good contact in an all-solid-state battery is a difficult task due to the fact that the solid electrolyte cannot easily reach every particle like a liquid electrolyte can. Moreover, the accommodation of volume changes could be hampered when using hard materials. Hence, electrode preparation and formulation still remain challenging topics for practical application of all-solid-state batteries. Table  1 summarizes recent results on the utilization of metal hydrides in all-solid-state batteries.  [108] 25Al2O3 − 75MgH2│80Li2S − 20P2S5│Li 3300 50% at 9 580 (17.6%) [106] Finally, the composites of MgH2 + TiH2 cycled with LiBH4 as solid electrolyte at 120 °C have displayed good electrochemical performance, which could suggest a suitable application in all-solidstate batteries in the future. Still, to accomplish the implementation in a complete cell, it will be necessary to find appropriate lithium-based cathode materials with comparable capacities while the challenge of chemical compatibility and stability against the solid electrolyte need to be faced.

A Full Solid-State Li-ion Cell
With the aim of building a full "all-solid-state battery" cell, López-Aranguren et al. [115] coupled the 0.8MgH2 + 0.2TiH2 mixture with sulphur as the counter electrode. Indeed, these materials show similar theoretical capacities (1718 vs. 1672 mAh g −1 ) and sulphur had already shown good results as electrode in cells with LiBH4 as a solid electrolyte [116]. The galvanostatic cycling of the allsolid-state battery is reported in Figure 8. As displayed in Figure 8a, the hydride electrode is fully lithiated during the first charge, similarly to the previous work in half-cells with both liquid and solid electrolytes. Within the first cycle, it shows a limited reversibility (~55%); however, in the subsequent charge/discharge, the battery exhibits a good capacity retention. At faster C-rates, a decrease in capacity is found though a Coulombic efficiency close to 100% is kept even at C/10. When C/50 is re-applied, the capacity regains previous values (~800 mAh g −1 ), exhibiting a good capacity retention in the subsequent cycles.
Comparing these data with those of Dao et al. [114], it is possible to notice that the full battery delivers a capacity slightly lower than the composite 0.8MgH2 + 0.2TiH2 in a half cell alone, suggesting that this limitation could be caused by the cathode material. Indeed, sulphur suffers of poor electric conductivity and sluggish kinetics [117], therefore, its theoretical capacity has never been achieved. Nonetheless, the complete all-solid-state battery of López-Aranguren et al. has shown the feasible coupling of hydrides and sulphur with promising results.

Conclusions
Generally, metal hydrides have shown improved performance when used in Li-ion battery systems at higher temperatures thanks to the fact that mass transportation within the electrode is enhanced compared to RT cycling. From the discussed findings, it is suggested that practical use of these active materials (electrodes and electrolytes) in Li-ion batteries might come initially for high temperature applications. However, the cell assembly still requires improvements, mainly in assuring good contact between particles in order to avoid the deterioration caused by volume changes during cycling.

Na-Based Closo-Borates for Na Batteries
As mentioned previously, several issues affect the current use of LiBs, including the low natural abundance and its non-ubiquitous distribution in the Earth's crust which could be a source of political instability in the future [118]. Furthermore, it is already the case that rare elements involved in the LiBs industries, such as Co, are generating ecological problems and exacerbating social inequalities in developing countries [119].
Consequently, other chemistries for post-Li secondary batteries have been proposed. Sodium (Na) is a first reasonable choice due to its alkali nature, meaning that the overall chemical processes are virtually the same compared to LiBs. Compared to Li, Na is heavier, with a larger ionic radius and with a lower redox potential (M Na = 22.99 g mol −1 ; R Na + =1.13 Å ; E0 Na = −2.71 V vs. standard hydrogen electrode (SHE) and M Li = 6.94 g mol −1 ; R Li + =0.73 Å ; E0 Li = −3.04 V vs. SHE). Because of this, the established electrode materials for LiBs, in general, feature differently for Na-ion batteries (NiBs). However, the strength of NiBs will lie in cheap, robust and large-scale installations where output voltage and energy density are not ultimately crucial. As for LiBs, NiBs also rely on organic carbonate-based solvents for the liquid electrolyte and so comparable safety issues affecting LiBs remain valid for NiBs.
In the context of finding suitable solid-state electrolytes, the family of hydroborates, commonly known as borohydrides, has shown interesting properties in terms of ionic conductivity and density while they also possess advantageous mechanical properties [59,120]. Even so, safety concerns remain when dealing with the [BH4] − anion. The larger cluster molecules and anions that are the decomposition products of the initial [BH4] − , so called closo-borates, are extremely stable and show a high ionic conductivity. The latter occurs after a structural phase transition that characterises their high-temperature phase. The electrochemical performance of these promising hydrides for Na-based batteries is discussed in more detail hereafter.

Na-Based Closo-Borates Solid Electrolyte
The Na2B12H12 Polymorphism The search for Na-based solid-state electrolytes by Brighi et al. was determined by the reduction of the high conductive phase transition temperature to below RT. The efforts were focused on the mixed [B12H12] 2− /[CB11H12] − system. The reason lies in the wide oxidative limits of such anions (4.1 and 5.1 V vs. Na + /Na, respectively) [121] and in their geometrical affinity, giving them the possibility to form a solid solution in which the Na-vacancy content can be tuned according to the anion ratio, leading to a high-symmetry disordered-structure, known to be a crucial parameter in ionic conductivity [122].
Four distinct polymorphs of Na2B12H12 were reported when studying the temperature-induced phase transitions [123,124], which also included a cubic closed packed (ccp) branch [125]. The in situ SR-XPD data in Figure 9a show the polymorphism as a function of temperature.
The RT monoclinic phase, in which all ions are ordered, splits at approximately the same temperature (~242 °C ) into a body centered cubic (bcc) and ccp branch, which is representative of an equilibrium state characterized by a very small energy difference separating the two polymorphs. While ccp is stable in the s.g.
3 ̅ , the bcc branch evolves transferring thermal energy into the lattice dynamic, increasing the disorder and the symmetry, causing the transition of 3 ̅ → 3 ̅ [126,127].
The coexistence of the two lattice packing variants is supported by DFT. Figure 9b shows the calculated ground state lattice energy as a function of unit cell volume for different packing types (shaded regions). Despite the global minima, represented by the monoclinic phase (s.g. P21/n), the bcc and ccp structures are very close in energy and are separated by 20 Å 3 /formula unit (f.u.).
Surprisingly, this matches the experimental observations quite well with the ccp density being higher than bcc (280 and 262 Å 3 /f.u. at 262 °C ). In all simulated structures, sodium ions are off-centered from tetrahedral interstitial sites, forming a disordered structure, which is a sign of facile Na diffusion [128].
The mentioned high degree of Na disorder plays a crucial role for ionic conductivity, which jumps from 10 −4 to nearly 0.1 S cm −1 at 267 °C for Na2B12H12 after the first phase transition [123].
Despite the operating temperature being far above the desired one for RT applications, a minor modification of the anion cage considerably improves the performance of such a material. Indeed considering the heterovalent substitution of [B12H12] 2− with [CB11H12] − , the RT orthorhombic polymorph of NaCB11H12 (s.g. Pca21) transforms to the high symmetry ccp-type lattice at 107 °C, with a resulting Na + conductivity of 0.12 S cm −1 [129].

Nax+2y(CB11H12)x(B12H12)y Solid Electrolyte
As presented in the previous section, Na2B12H12 represents an interesting backbone to host high-Na mobility. Indeed, Figure 9b shows how closely related different packing types like bcc and ccp are, suggesting that a little modification could stabilise one of those packing variants. In this context, the partial replacement of the divalent [B12H12] 2− for the monovalent [CB11H12] − would leave the anionic size unchanged but would affect the Na content, i.e., the ion-ion interaction. Three nominal compositions of xNaCB11H12 + (1 − x)Na2B12H12 (x = 1/3, 1/2 and 2/3) were mixed by ball milling to obtain Na2−x(CB11H12)x(B12H12)1−x; the structural characterization was obtained by SR-XPD by Brighi et al. [130]. The structure with sample composition x = 1/2 ( Figure 10), was solved in the s.g. I23 with lattice parameter a = 7.98 Å at RT and with a bcc anion sublattice. Na is disordered around the tetrahedral sites. A tetrahedral setting should per definition be lower in energy than an octahedral one for the Na + cation [131], when coordinated by closo-borates. However, the Na-Na repulsion in the bcc stabilizes off-centered tetrahedral positions due to the shorter distance of 2.83 Å and 3.99 Å between tetrahedral and octahedral sites, respectively. The suggested model shown for the composition x = 1/2 in Figure 10 agrees with this hypothesis. The I23 structure remains stable at RT for composition x = 2/3 while for x = 1/3, this configuration becomes unstable as the temperature increases. For the latter, a first phase transition to the s.g.
3 ̅ was observed at T = 63 °C and at T = 193 °C. First, the dynamic anion disorder increases and secondly, the symmetry increases again owing to the re-appearance of the body-centered symmetry in s.g.
3 ̅ (as followed by the polymorphism of the bcc branch of Na2B12H12) [124]. The Arrhenius plot in Figure 11a summarizes the ionic conductivity (measured by EIS) [130] of the mixed-anion compounds compared to the respective precursors ( Figure 11b) [123,129,132]. At RT, for x = 2/3, the Na + mobility is in the mS cm −1 range, and above 60 °C, approaches the superionic regime with >10 −2 S cm −1 . In this temperature range, a low activation energy Ea = 314 meV was determined from σ = σ0 T −1 exp(−Ea/kT), known as the Arrhenius relation. When the composition reaches the value of x = 1/2, the ionic conductivity remains constant while the activation energy slightly increases to Ea = 367 meV.
This trend is established for higher Na-contents (composition x = 1/3) with a conductivity drop of almost one order of magnitude at RT and an increase in the activation energy (Ea = 409 meV).
For the latter composition, at nearly T = 90 °C, where the symmetry changes to s.g. 3 ̅ , the ionic conductivity progresses notably, confirming the transition observed by SR-XPD. This tendency of Na mobility can be explained in terms of vacancy population that increases by the substitution for a monovalent anion, as in the case for x = 2/3 and 1/2. Hence, the same conduction mechanism suggested for the high-temperature phase of Na2B12H12, i.e., cation jumps between T-sites of bcc anion packing, applies to these compounds as well, similarly to other bcc superionics such as α-Ag2S and α-AgI [122]. The electrochemical stability of composition x = 2/3 was investigated with cyclic voltammetry (CV) at 60 °C (scan rate of 0.5 mV s −1 ) and is presented in Figure  12.
The cell is composed as follows: Na|Na2−x(CB11H12)x(B12H12)1−x|Na2−x(CB11H12)x(B12H12)1−x + carbon black|Au with Na as self-referenced counter electrode and Au as the working electrode. There is, however, an additional layer, composed by a mixture of solid electrolyte and carbon black that was used to amplify the low current signal by increasing the interfacial contact and improving the signal-to-noise ratio. This last modification allowed us to detect an anodic peak with the onset at 4.1 V vs. Na + /Na (Figure 12 inset), which keeps this compound in a suitable electrochemical window of 0-4 V vs. Na + /Na.  Cyclic voltammetry at a scan rate of 0.5 mV s −1 for composition x = 2/3. Au and Na disks were used as working and counter/reference electrodes, respectively, for the voltage region −0.1 < V < 1.5 vs. Na + /Na, indicated by the red curve. The black curve refers to the complementary CV, where a graphite layer was added to the solid electrolyte body (1.5 < V < 5 vs. Na + /Na). Sodium plating (negative current density) and stripping (positive current density) are represented by the ohmic behaviour in the vicinity of the Na + /Na redox potential (0 V). Reprinted from Brighi, M., F. Murgia, Z. Łodziana, P. Schouwink, A. Wołczyk, and R. Cerny, A mixed anion hydroborate/carbahydroborate as a room temperature Na-ion solid electrolyte. Journal of Power Sources, 2018. 404. Copyright (2018), with permission from Elsevier.
To test the long-term reversible plating/stripping vs. metallic Na, a pellet of Na4/3(CB11H12)2/3(B12H12)1/3 (x = 2/3) was sandwiched between two Na disks. Figure 13 shows the electrochemical behaviour of such cell at RT for over 500 h and two figures as magnifications on regions 20-25 and 300-305 h for better comparison. After an open circuit voltage (VOC) period of 5 h, a current density of ±25.5 μA cm −2 was alternatively applied for 30 min. The repeated plating/stripping occurs with a steady and limited polarization, starting from ±8.5 mV that further decreases, stabilising around ±6 mV towards Na + /Na. The electrolyte was stable towards Na metal for more than 500 operating hours.

Conclusions
The future class of solid electrolyte needs to fulfil several requirements, such as fast ionic conductivity, low density, non-toxicity, high thermal and electrochemical stability, but also, from a processing point of view, it should be highly ductile. Borohydrides are not the Holy Grail of all-solidstate batteries, but at the same time, satisfy many of the above-mentioned needs. In this context, another mixed-anion developed by Duchêne et al., Na2(B10H10)0.5(B12H12)0.5 [133], was used as proofof-concept for a 3 V all-solid-state NiB, showing high performances. The mixed-anion Na2−x(CB11H12)x(B12H12)1−x described here was shown to be stable against metallic Na operating more than 20 days with stable interfaces under Na plating/stripping. A further characterization spanning the entire range of current density up to 1 mA cm −2 is still necessary to determine whether dendrite penetration can be avoided (since in general having a solid material is not a sufficient condition) [134][135][136]. Nevertheless the achievements summarized herein put Na2−x(CB11H12)x(B12H12)1−x as a valid competitor with other solid electrolytes like oxides and thiophosphates.

Advances in Hydrogen Storage Materials
In the last decade, substantial effort has been concentrated on using hydrogen as a fuel and energy carrier. Finding the safest conditions for large volumes of hydrogen has been, nevertheless, a challenge. Compared to the conventional compressed and cryogenic hydrogen storage, the storage of hydrogen in the solid-state chemically bonded to other compounds to form the hydrides and hydride-based systems appears to be more suitable because the hydrogen pressure and temperature employed in this storage approach are milder than those required for compressed and cryogenic hydrogen storage. High volumetric and gravimetric hydrogen capacity, fast hydrogenation and dehydrogenation, full reversibility, low thermodynamic stability and low cost are the main parameters to be considered when studying a material with regards to its potential as a hydrogen storage source. In this part of the review, recent advances and perspectives on the different hydride categories, such as metal hydrides, complex hydrides, reactive hydride composites and rare earth complex hydrides, as well as their properties kinetically tailored by several approaches, i.e., the use of additives and confinement into a nanoporous scaffold, are discussed.

Pure Metal Hydrides (Mg, Pd, Ti)
One of the most promising materials belonging to the pure metal hydrides is magnesium (Mg) and its hydride (MgH2) [137,138]. Despite its many advantages, the formation of this hydride creates an exothermic reaction with an enthalpy of ∆H0 = −75 kJ mol −1 of H2 and an entropy of ∆S0 = −133 J mol −1 of H2 [139]. The high stability of this hydride consequently raises problems in reversibility while the decomposition temperature is too high for desired industrial applications (≈300 °C instead of the desired ≤100 °C) [140].
Alloying with transition metals (TM) and changing of the composition as well as nanostructuration are a few of the methods reported to improve the properties of hydrides [141][142][143][144]. The use of palladium (Pd) as a catalyst in the MgH2 formation has been shown to be essential [145][146][147][148][149]. To date, a system has not yet been found that monitors the industrial requirements and considerable effort is still invested for improving this material [150,151].
While powders are often investigated, thin films offer a more precise tool when focusing on the kinetics and the mechanism of diffusion in a system. An additional requirement for a quantitative evaluation of transport coefficients is an appropriate characterization of thermodynamic driving forces. In Mg, the equilibrium pressure of hydrogen has been reported since 1960 and an equilibrium pressure of 12-22.6 bar at temperatures of 380-410 °C has been identified [152][153][154]. Significant work on thin films has been conducted by Mooij et al. [155,156], who established a nucleation and growth mechanism while exploring hydrogenation at low pressures of 0.7 mbar in 10 nm thin Mg films that offer practically a 2D reaction geometry. By optical transmission microscopy, the growth of hydride islands was detected, which helped to establish a nucleation control. However, already at slightly higher pressures of 1.0 mbar, the islands could not be resolved any longer caused by a very high nucleation rate. X-ray diffractometry, electron and optical microscopy, Sievert apparatus and theoretical modeling are the main tools for hydride investigations. In systems, such as Mg and MgH2, the resistivity changes when hydrogen enters the metal. Such a setup is unavoidably challenging for high hydrogen pressures and wide ranges of temperature measurements. In 2005, Ingason et al. [157] reported resistance isotherms in a Mg system during hydrogen uptake. Moreover, Gharavi et al. [158] reported the resistance change in Mg thin films, additionally coated with a thin Pd layer, while focusing on the texture changes upon hydrogen loading and not on the kinetics of the system. Practical restrictions often hinder a simultaneous analysis of all parameters (temperature, pressure and time) with the use of one single apparatus. Hadjixenophontos et al. [159] suggested a way to fill that gap while studying the hydride formation by ex situ XPD and in situ resistance measurements in a wide range of parameters and demonstrated a linear to parabolic kinetic transition. By subsequently defining a resistivity model to their experimental results that considers this transition, they could quantitatively evaluate the kinetic parameters of diffusive transports through the hydride. They reported the linear growth constant obtained in the early stages at 200 °C to be equal to κ = 6.6 × 10 −9 cm·s −1 . Measurements in a wide temperature range from RT to 300 °C yield an Arrhenius behaviour with an activation energy Ea = 28.1 kJ·mol −1 [159] and comparisons can be found in [160][161][162]. With supplementary investigations and calculations, this low activation energy, the described small grain size of 17 nm and the comparison to theoretical predictions of bulk diffusion led to the conclusion that grain boundary diffusion is the main transport mechanism [159].
Grain boundaries and detailed microstructure changes in thin films are better visible in TEM micrographs. Hadjixenophontos et al. [163] showed that during hydrogenation of Mg, the hydride is immediately formed at the interface between the coating of Pd catalyst and the Mg thin film. The Mg hydride grows in a layer-like reaction towards the substrate. FIB used for preparation of cross sectional observations destroys the hydride phase but the microstructure change is nonetheless clear with an alteration from a columnar grain structure of freshly deposited pure Mg thin film to an equiaxed grainy structured film when the hydride is formed. Electron microscopy has been reported to irreversibly damage the hydride phase [164]. It can be used to follow the hydrogenation kinetics and the MgH2 layer growth not only in thin films but in powders as well.
Pd does not only bring advantages to the system but also some difficulties. Looking at the phase diagram of the Mg-Pd system [146] at fairly high temperatures of hydrogen loading (200 °C) and unloading (300 °C), intermetallic phases Mg5Pd2 and Mg6Pd are formed. Kumar et al. [165] reported the hydrogenation of Pd/Mg films at almost ambient temperatures (50-150 °C) that rapidly declined in the third cycle despite the fact that Pd appears stable at those conditions. This observation suggests that the Pd/Mg interface may not be stable and hinders the hydrogen transport. Reddy et al. [166] studied the formation of Pd-Mg intermetallics in thin films (at 200 °C) and presented the Mg5Pd2 and Mg6Pd phases by XPD. Further cycling of their films was not discussed. Hadjixenophontos et al. [159] demonstrated, also in thin films, that further de-hydrogenation and re-hydrogenation were possible regardless of the intermetallic reactions at the Mg/Pd interface. The formed phases were detected by TEM micrographs to form a very dense layer close to the surface and consumed the Pd layer over time.
Replacement of Pd on top of Mg films is a bit more challenging since many elements are miscible with Mg, but due to the high cost of Pd, alternatives are of high interest. Researchers have demonstrated that it is possible to improve the performance of Mg and its hydride by adding transition metals [140] such as Iron (Fe) [167], Chromium (Cr) [168], Nickel (Ni) [144,169], Titanium (Ti) [143,[170][171][172], Niobium (Nb) [173] and many others. Further investigations have been completed on merging the benefits of carbon materials for tuning MgH2 hydrogen properties [174]. Recently, MgCNi3 has also shown an important catalytic additive behaviour with its anti-perovskite structure [175]. Titanium, while being promising and immiscible with Mg, needs to be handled with care since the native oxide formed on top has been shown to control the hydrogenation. Hadjixenophontos et al. [164] reported a linear kinetic regime observed when hydrogenating TiO2/Ti/Mg films and showed a growth constant at 300 °C and 20 bar H2, of κ = 10 −10 cm s −1 . Additional measurements in a wide range of temperatures (25 °C-300 °C) and pressures (from 5 × 10 −5 bar up to 100 bar H2) revealed a plateau rate at about 1 bar H2. The authors suggested that above this pressure, a full coverage of the oxide surface from hydrogen atoms occurs [164]. Additionally, the diffusion through the oxide is very slow and this hinders further hydrogen movement into the sample. Attributed to this oxide effect, hydrogenation performance decreases at higher pressures.
Based on the fact that light materials perform significantly better for hydrogen storage applications, the ternary system Li-Mg-Si has been reported to form promising intermetallic phases. Pavlyuk et al. [176] gave a complete structural report on seven ternary compounds that can be found in this system while additionally testing their hydrogenation properties. The ternary alloy with the Li-rich phase Li70Mg10Si10 is shown to absorb the highest amount of hydrogen (8.8 wt.%) with a desorption temperature between MgH2 and LiH.
Conclusively, Mg and its hydride stay under the radar of research investigations due to the dominant benefits they offer. However, the system lacks efficiency for industrial applications and a great effort is given on improving it or finding alternative systems that are more complex.

Amide and Imide Based Systems for H2 Storage
Complex metal hydrides made from light elements, such as lithium, boron, nitrogen, or aluminium very often have particularly high hydrogen densities while poor thermodynamic and kinetic properties as well as limited reversibility are drawbacks. The chemistry of complex metal hydrides is exceptionally diverse. Many novel materials have been provided. In this respect, amides of alkali and alkaline-earth metals in combination with metal hydrides have the potential to meet the needs of on-board hydrogen storage for electric vehicles powered by fuel cells due to their high hydrogen capacity and tuneable thermodynamics [177]. A large number of amide-hydride and amide-complex hydride materials have been developed and studied in recent years [178]. Nevertheless an amide-based composite with suitable hydrogen capacities, favourable thermodynamics and decent kinetics is still a challenging mission. The first task is to accomplish a favourable thermodynamic stability, to realize hydrogen storage at moderate temperatures. Secondly, ammonia (NH3) is an unwanted by-product which degrades the fuel cell by a loss of contact between the Pt/C-catalyst and the membrane [179]. Therefore, preventing NH3 release needs to be addressed and filters are necessary to prevent the aforementioned membrane degradation. Thirdly, amide-hydride composites suffer from sluggish kinetics for the hydrogenation and dehydrogenation reactions. The barriers for the kinetics are primarily a result from the mass transport and from reactions, which occur at the interface. Reducing particle size for instance by ball milling or nanoconfinement and mixing additives into the hydrogen storage materials are all effective ways to improve kinetics [180].
Research in this field started in 2002, with the first characterization of the hydrogen storage properties of the LiNH2-LiH composite system [181]. Li3N was reported to reversibly store more than 9 wt.% of hydrogen in a two-step process, which can be described according to the following reactions in Equations (2) and (3): The net reaction for the complete absorption process thus becomes: Despite its high gravimetric capacity, the reaction enthalpy of the Li-N-H system has been reported to still be too high, measured to be 66 kJ mol −1 H2 only for the first step [181], while the reported reversible hydrogen capacity has later been revised to 5.2 wt.% [182].
Many attempts to modify the composite system in order to decrease the reaction enthalpy have followed. The most successful approach was the partial replacement of Li with Mg [183][184][185][186][187]. In particular, of the many different compositions explored within the Li-Mg-N-H system, the Mg(NH2)2 + 2LiH composite showed the lowest sorption temperatures and a promising low reaction enthalpy (38.9 kJ mol −1 H2) [188,189]. The thermodynamics of the system allowed predicting 1 bar of hydrogen pressure at a temperature as low as 90 °C. Experimentally however, this result has never been achieved due to kinetic limitations.

Insights into Alkali-Based Amides and Imides Including Boron
In the first part of this chapter, insight into binary systems that combine lithium borohydride, amide and imide are presented, discussing their improved thermodynamics, new structures, phase diagrams and hydrogenation properties. The second part gives an overview of recent insights into the ammonolysis of alkali and alkaline-earth metal amides, in the hydrogen storage and structural properties of K-Mg-N-H, KNH2-KH, Rb-Mg-N-H and Rb-N-H systems. The chapter ends with a comprehensive summary.

LiBH4-LiNH2
It has been evidenced that composite systems containing lithium borohydride (LiBH4) [190], or mixture of borohydrides, or borohydrides and reactive hydride composites, have improved hydrogen storage properties, especially when a liquid phase is involved [191]. Furthermore, variating composition and crystal structure can lead to a fine-tuning of kinetics and thermodynamics of hydrogen storage properties as a function of temperature, pressure and phase stability. The full thermodynamic optimisation and assessment of LiBH4 opens the way for new studies of more complex systems (i.e., binary mixtures, eutectic borohydrides) [192].
The LiBH4-LiNH2 system has been extensively investigated in a wide composition range by means of different synthesis routes (i.e., ball milling, annealing treatment and quenching from liquid phase). Wołczyk et al. [71] reported a detailed examination of the phase stability combining experimental techniques, ab initio calculations and CALPHAD assessment to fully determine the equilibrium pseudo-binary LiBH4-LiNH2 phase diagram ( Figure 14). In detail, this system consists of two compounds: Li2BH4NH2 has been reported to be metastable and Li4BH4(NH2)3 presents peritectic melting [71]. DFT calculations on LiBH4, LiNH2, Li2BH4NH2 and Li4BH4(NH2)3 gave good agreement between experimental [196,197] and optimized structures [71]. The optimized geometries of Li2BH4NH2 and Li4BH4(NH2)3 were found to have lower energy than the mixtures LiBH4 + LiNH2 and LiBH4 + 3LiNH2, by 6.6 and 13.9 kJ mol −1 , respectively. Previously, Herbst et al. [198] reported a lower energy of reaction equal to 6 kJ mol −1 for Li4BH4(NH2)3; however, the new determination by Wołczyk et al. fits reasonably well with the value obtained by Siegel et al. [199] at 0 K (11.3 kJ mol −1 ). Calculations of Li2BH4NH2 geometries have not been reported in the literature yet [200]. It is believed, however, that the Li4BH4(NH2)3 is much denser. The thermodynamic assessment of the system shows that the formation reaction of Li2BH4NH2 from LiBH4 and Li4BH4(NH2)3 has a slightly more positive energy value, thus being a metastable compound.
While neglecting entropy variations, the Gibbs free energy of Li4(BH4)(NH2)3 has been determined through ab initio calculations for the reaction in Equation (5) [199]: The Gibbs free energy of the liquid phase, including new experimental temperature and enthalpy of melting for pure LiNH2 obtained in Reference [71], is equal to Tm = 357 °C and ΔHm = 14.3 kJ mol −1 . Considering a regular solution model to describe the liquid mixture between LiNH2 and LiBH4, a negative interaction parameter, Ωliquid = −13.0 kJ mol −1 , was obtained from the assessment. As already reported previously and according to recent experimental information [203,204], the Li2BH4NH2 compound is metastable and therefore, it is not present in the stable phase diagram.
The reactions of LiNH2 and complex hydrides, including LiBH4, show a number of interesting features. It was reported that more than 11 wt.% hydrogen can be desorbed (endothermic reaction) from a mixture of 2LiNH2 + LiBH4 in the temperature range 250-350 °C [205]. In conclusion, these materials have an outstandingly rich chemistry, which is based on their structural flexibility and additionally, a wide range of elemental compositions [194].

LiNH2-Li2NH
The first to study binary systems based on LiNH2-Li2NH was by Chen et al. [181,206] in 2002. Originally, they suggested that H2 release could take place at the LiNH2/LiH interface via an acidbase reaction between H δ− of LiH and H δ+ of Li amide. This mechanism, however, would require a very close contact between the two phases [207]. Furthermore, Miceli et al. [208] studied the binary LiNH2-Li2NH system. The reaction rates and the diffusivity of different species in LiNH2 and Li2NH collected from their ab initio simulations provided for a microscopic scenario for the solid-state reaction in Equation (6): According to these ab initio calculations, the transformation path depends on the surface-tovolume ratio of the crystallites of LiNH2. NH3 formation is stimulated by the presence of Li Frenkel pairs, as proposed by David et al. [74]; however, NH3 formation is more likely to take place at the surface rather than in the bulk. Thus, it is expected that for small LiNH2 particles, the transformation might be controlled by the surface. In this case, in the early stage of decomposition, the rate-limiting step is the formation of NH3 supported by Frenkel pairs at the LiNH2 surface. Once a shell of Li2NH is formed, the rate-limiting step becomes the diffusion of H + in the Li2NH product. Transfer of H + /Li + across the LiNH2/Li2NH moving interface and the formation of NH3 from NH2 − groups in Li2NH are not rate limiting [208]. Shaw et al. [207] proposed a kinetic model in which a shell of Li2NH grows from the surface while the core of LiNH2 shrinks. However, for larger crystallites with small surfaceto-volume ratio, NH3 formation would take place in the bulk. In the presence of Frenkel pairs, this would happen at a higher pace due to the larger number of defects. However, the formation of Li2NH in the bulk of LiNH2 via the accretion of Li2NH nuclei is anticipated by the diffusion of NH2 species which might lead to the formation of the sub-stoichiometric phases reported by David et al. [74]. Miceli et al. [208] reported that the size dependence of the decomposition pathways might not be restricted to LiNH2 but could be compared to other complex hydrides [208].

Insights into the Structure and Reaction Mechanism of Metal Amide-Metal Hydride Composite Systems
The effort for tailoring the amide-hydride composite system of Li and Mg (Li-Mg-N-H system) resulted in the discovery of two additives, KH and LiBH4 [209,210]. In the KH-doped system the desorption temperature was lowered by 50 °C and it was possible to perform the dehydrogenation/re-hydrogenation at 107 °C under equilibrium conditions [210]. The addition of LiBH4 allowed to desorb 5 wt % of H2 at 140 °C and re-absorb the same amount of hydrogen at 100 °C [209]. A comprehensive summary on the effect of these additives and their practical application for hydrogen storage can be found in a recent review [177]. The next sub-chapters, concerning the ammonolysis of alkali and alkaline-earth metal amides and the K-Mg-N-H, KNH2-KH, Rb-Mg-N-H and Rb-N-H systems, will present an overview of recent discoveries on some more fundamental aspects of amide-hydride systems.

Ammonolysis of Alkali and Alkaline-Earth Metal Amides
The first in situ SR-XPD study of the ammonolysis process of lightweight alkali and alkalineearth metal hydrides showed clear evidence of the formation of previously unknown intermediates [211]. In fact, while for LiH and NaH, the formation of LiNH2 and NaNH2 starts to occur almost at RT, for KH, the formation of the RT polymorph of KNH2 (monoclinic, P21/m) is not observed. Instead, isolated Bragg peaks appear almost at RT, which resemble the pattern of the cubic KNH2 polymorph. However, the cubic polymorph of potassium amide should be stable only above 75 °C [212]. By increasing the temperature, the Bragg peaks of LiNH2, NaNH2 and the cubic KNH2-like phase increased in intensity due to the continuation of the ammonolysis process, while the intensity of LiH, NaH and KH peaks decreased, as could be expected, hypothesizing that the ammonolysis process occurs according to the following reactions in Equations (7)- (9): NaH + NH3 → NaNH2 + H2 KH + NH3 → KNH2 + H2 In fact, for the Li-based system, as well as for the Na-based system, the Bragg peaks of the solid phases involved in the process show a gradual shift to lower 2θ-values when the temperature increases, as it usually occurs owing to the thermal expansion of the crystal lattice. However, for the K-based system, the peaks of the cubic KNH2-like phase gradually shift to higher 2θ-values when the temperature is raised above 160 °C, implying a counter-intuitive contraction of the cubic cell upon heating. Due to this observation, the authors suggested the possible formation of new K-N-H species with intermediate compositions [211].
Concerning the alkaline-earth metals, the most interesting piece of evidence, in view of hydrogen storage applications, is the partial formation of the high-pressure polymorph of γ-MgH2 (tetragonal, P42/mnm) by ball milling pre-treatment of β-MgH2 (orthorhombic, Pbcn) and its role in facilitating the ammonolysis process at lower temperatures. In fact, during the ammonolysis experiment, the formation of Mg(NH2)2 (starting at about 230 °C) occurs mostly with the consumption of the γ-polymorph rather than β-MgH2. This fact might be helpful in order to optimize the synthesis process of Mg(NH2)2, which is an important component in many of the most promising amidehydride systems.

K-Mg-N-H System
The interaction of KH and Mg(NH2)2 has been shown to have a central role in lowering the desorption temperature of the KH-doped Mg(NH2)2 + 2LiH system [213]. Moreover, in the attempt to characterize the K-Mg-N-H phases involved, a new phase was discovered, a mixed amide-imide phase: KMgNH2NH [214,215]. However, none of the other bimetallic phases previously reported (K2Mg(NH2)2NH, K2Mg(NH)2 and KMgN) were found as reaction product, by ex situ XPD [214,216,217]. In order to clarify this aspect, the K-Mg-N-H system was studied by Santoru et al. [218] by in situ SR-XPD, confirming that no bimetallic phases were formed, in addition to K2Mg(NH2)4 and KMgNH2NH. For the KNH2 + MgH2 composite the K-Mg-N ratio was calculated in order to release all possible hydrogen forming the speculated bimetallic nitride, KMgN. However, even above 300 °C, this phase could not be formed. Mg3N2 was formed instead and the K-N-H containing phase melted and decomposed rather than interacting with Mg3N2 to form a bimetallic nitride. The Mg(NH2)2 + KH composite led to the formation of KMgNH2NH. In both systems; however, KH and KNH2 reacted in an unexpected way, where with the progress of the reactions, their diffraction peaks gradually shifted in 2θ-scale rather than simply increasing or decreasing in intensity [218]. This behaviour was similarly reported for the ammonolysis process of KH, implying unexpected expansions and contractions of the cubic cell [211].

KNH2-KH System
A new study by Santoru et al. [219] closely examined the KNH2-KH system and confirmed the hypothesis previously reported, indicating the possibility of forming phases with intermediate amide-hydride compositions. In situ SR-XPD measurements were carried out under inert atmosphere on samples with different molar ratios of KNH2 and KH. This series of experiments proved that in any case the two starting components react, forming two cubic phases that become more and more similar, until their Bragg peaks merge into a single set of reflections ascribable to a single cubic phase. In other words, an anionic exchange occurs between the cubic structures of the amide and hydride phase, until a phase with a single composition is reached at 270 °C. Moreover, it was revealed that the higher the initial content of potassium amide is, the larger the unit cell of the cubic phase formed. In the same study [219], NPD and 1 H solid-state (SS)-NMR measurements furnished additional proof of the formation of a metal amide/hydride solid solution. In view of these findings, the reaction of KNH2 and KH can be generalized as Moreover, the ammonolysis process of KH is better described by the following chemical reaction:

Rb-Mg-N-H and Rb-N-H Systems
The continuous effort to improve the hydrogen sorption properties of the already promising KH-doped Li-Mg-N-H system resulted in the testing of Rb-based additives [220][221][222][223]. This choice was prompted by the fact that K and Rb are both alkali metals with similar electronegativity and the ability to form ternary amides with Mg [220]. Rb-based additives were shown to be even better than the corresponding K-based ones in lowering the hydrogen desorption temperature of the composite. Moreover, the formation of a new RbMgNH2NH phase responsible for a lower reaction enthalpy and lower activation energy, was inferred by stoichiometric considerations and Fourier transformed infrared data, but not directly proven by diffraction methods [222]. For these reasons, an effort was made to characterize the intermediates possibly involved in the desorption process from the structural point of view and RbMgNH2NH was successfully synthesized, as confirmed in Santoru et al. [224]. Moreover, XPD and NPD measurements allowed to solve its crystal structure as an orthorhombic system, with s.g. Pnma being isostructural with KMgNH2NH [215,224]. Larger unit cell parameters were found, which could be expected taking into account the larger cation [224]. It is worth noticing that the use of CsH has been experimented recently [225][226][227][228] leading some authors to suggest the possible formation of a CsMgNH2NH phase [227,228]. It would be interesting for further studies to focus on isolating this compound in order to solve its structure and compare it to that of RbMgNH2NH/KMgNH2NH while determine the respective similarities and differences.
Considering the novelty of these findings and the attested formation of K(NH2)xH1−x phases, both in the desorption process of the K-Mg-N-H system and during the synthesis of KNH2 by ammonolysis of KH, the study expanded on the possibility of forming a Rb-based amide/hydride solid solution. A combination of in situ SR-XPD and 1 H SS-NMR confirmed that cubic phases with intermediate amide/hydride compositions, described as Rb(NH2)xH1−x, were indeed formed by thermal treatment. Moreover, the Rb(NH2)xH1−x phases were formed even at RT after hand-grinding the pure RbNH2 and RbH with an agate pestle and mortar [224].

Conclusions
During the exploration for novel materials for the storage of hydrogen, research interests have been focused on light metal borohydrides and amides as two characteristic chemical complex hydrides with very high hydrogen capacities. Mixed anion complex hydrides were discussed here, with a focus on the LiBH4-LiNH2-Li2NH system, which is shown to form a series of amide phases. Investigation on Li2NH itself brought interesting results in terms of new structure determination and good ionic conduction as shown in Chapter 2.1.
Amide-hydride systems have shown to be particularly interesting, not only as promising candidates as hydrogen storage systems but also because they continue to provide novelty even in their fundamental aspects. In particular, the study of the ammonolysis process of alkali metals showed marked differences in how KNH2 is formed as compared to LiNH2 and NaNH2. In contrast to LiH and NaH, KH reacts with NH3, forming a novel amide-hydride solid solution. K-Mg-N-H and Rb-Mg-N-H systems present a striking similarity in their ability to form an isostructural bimetallic amide-imide phase which is involved in the lower reaction enthalpy and lower activation energy of a RbF-doped Mg(NH2)2 + 2LiH system. The similarities in the chemistry of K and Rb is also evidenced in the formation of a rubidium amide-hydride solid solution with very similar properties to the Kbased one and which can be formed even at RT. Since the kinetic barrier of metal amide/hydride composite arises from mass transport and reactions occurring at the interface, it is reasonable to suspect that the lower activation energy of K-and Rb-doped systems might be due, at least in part, to the possibility of forming amide/hydride phases with a continuous range of intermediate compositions. Nonetheless, looking at the structural similarities seems to be a reasonable strategy for finding new additives to improve hydrogen storage systems.

Eutectic Metal Borohydride Systems
The concept of eutectic in metallurgy specifies the formation of a homogenous liquid mixture from a mixture of substances in a defined composition at a given eutectic temperature [229]. The eutectic temperature is the lowest temperature among the melting points of all separate constituents. This phenomenon is not only observed in alloys but also in molten-salt mixtures. These molten-salts mixtures are generally good conductors for electric current, which is a consequence of their highly dissociated ionic species in the molten status [230]. They are useful as reaction media [231] as well as in extractive metallurgy [232]. Recently, Harries et al. and Paskevicius et al. proposed the employment of molten-salts and eutectic-borohydrides mixtures in solar thermal energy [233] and H2 storage [234], respectively.
The low-melting-point borohydride mixtures are often denoted "eutectic metal borohydrides", while these mixtures can lead to hydrogen desorption at relatively low temperatures [235,236]. Furthermore, the low melting point is beneficial for the nanoconfinement approach discussed in Chapter 3.5. [237][238][239]. Furthermore, these low melting point mixtures have been investigated as ionic liquids, aiming to provide convenient and fast re-fueling of hydrogen in electric vehicles powered by fuel cells [240,241].
The formation of eutectic borohydride mixtures from physically mixing pure borohydrides is summarized in Table 2, showing a H2 release at lower temperatures than individual constituents [190]. The following chapter will focus and present experimental investigations of the reported systems, as well as their thermodynamics and hydrogen storage properties.
The use of eutectic borohydrides is strongly related to the stability of the liquid phase, as, for instance, alkali-metal borohydrides often melt before H2 desorption starts. The dehydrogenation of borohydrides from the liquid state may lead to less [B12H12] 2− in the reaction products [242]. Thus, the characterization of their thermodynamic properties and decomposition mechanism is of fundamental importance [243][244][245].
In the literature, it has been shown that binary combination of LiBH4, NaBH4 and KBH4 undergo eutectic melting [245,247] or a thermal minimum [248]. Starting from available experimental values and ab initio calculation, the thermodynamic assessment and optimisation of binary systems allow the modelling of solid and liquid phases, leading to a description of the pseudo-ternary phase diagrams using the CALPHAD approach [267].
In the following section, binary and higher combination of borohydrides will be discussed in detail concerning their experimental studies and their thermodynamic assessment. Phase diagrams and hydrogen storage properties will be presented.

LiBH4-NaBH4
Among the known eutectic borohydrides, the eutectic mixture of LiBH4-NaBH4 has relatively low costs while the high theoretical gravimetric hydrogen storage capacity is preserved with 15 wt.%. To date, two eutectic compositions have been reported based on experimental and thermodynamic simulation methods: 0.62LiBH4-0.38NaBH4 with a melting point of 224 °C and 0.71LiBH4-0.29NaBH4 with a melting point of 216 °C [235,246,247].
In 1961, Adams [268] proposed a partial pseudo-binary LiBH4-NaBH4 phase diagram, describing the system as eutectic and reporting experimental liquidus points for mixtures which are rich in LiBH4. However, the experimental method used for their determination was not reported. In 1971, Semenenko et al. [269] reported the pseudo-binary phase diagram by coupling thermographic and X-ray studies on annealed samples. The system was characterized without any eutectic point, but it was described by the formation of solid solutions, including a miscibility gap and a minimum melting temperature at a similar composition, as previously proposed by Adams, i.e., eutectic (0.62LiBH4-0.38NaBH4) [269]. It is worth noting that solid solutions of LiBH4-LiI exhibits a fast-ion conduction, owing to the stabilisation of the hexagonal structure of LiBH4 at RT [75,77,79,257,[260][261][262]264,270], as described in Chapter 2.1.
The LiBH4-NaBH4 system ( Figure 15) was recently studied experimentally and assessed by Dematteis et al. [247], showing the formation of solid solutions on both Li and Na rich side with an eutectic melting at 216 °C for the composition 0.70LiBH4-0.30NaBH4. In this system, the polymorphic transition of o-LiBH4 to h-LiBH4 structure was observed at 95 °C, which is 20 °C lower than that for pure LiBH4 [193].

LiBH4-KBH4
The LiBH4-KBH4 system (noted as LiK) was reported to be eutectic by Ley et al. [245] and Huff [271] with a melting temperature of 105 °C for the composition 0.72LiBH4-0.28KBH4. It has a theoretical H2 capacity of 13.2 wt % [235]. A similar phase diagram with the same trends, but with a different eutectic composition, was reported by Adams [268]. No solid solution has been observed in known LiK systems; however, a bimetallic compound LiK(BH4)2 has recently been reported, which decomposes at 96 °C into LiBH4 and KBH4 [245]. Figure 15b reports all experimental literature data and the calculated phase diagram of the system assessed by the CALPHAD method by Dematteis et al. [249].

NaBH4-KBH4
To date, two experimental studies have been reported for the NaBH4-KBH4 [248,269] system (noted as NaK). Jensen et al. [248] showed the formation of a solid solution with full solubility above 200 °C and a minimum melting temperature of 458 °C for the composition 0.68NaBH4-0.32KBH4. Below 200 °C, a miscibility gap has been evidenced by Jensen, but it was not reported from the previous study by Semenenko et al. [269]. The solid solution can be quenched at RT and the de-mixing kinetic of the solid solution at RT was studied by NMR [248]. The binary system was fully assessed by Dematteis et al. [249] and its phase diagram is presented in Figure 15c.

LiBH4-NaBH4-KBH4
The thermodynamics of LiBH4-NaBH4-KBH4 ternary system has been fully described combining experimental and theoretical investigations by Dematteis et al. [249]. Using the CALPHAD method combined with ab initio calculations, they determined the end-members and optimized LiBH4-NaBH4, LiBH4-KBH4 and NaBH4-KBH4 pseudo-binary phase diagrams data ( Figure 15). The new exact ternary eutectic composition was determined and confirmed to be 0.66LiBH4-0.11NaBH4-0.23KBH4 with a melting temperature of 102 °C, contrarily to that reported by Huff [271]. Figure 15d presents the liquidus projection of the eutectic melting in the LiBH4-NaBH4-KBH4 ternary phase diagram.

Other Systems
The formation of eutectics has been demonstrated in binary and ternary systems, and the hydrogen release often occurs, involving a liquid phase, typically above 200 °C and at characteristic peak temperatures (Tpeak) that depend on the interaction among the cations and the complex anion of [BH4] − in the liquid phase. The only binary system that does not show a low temperature liquid is the Mg(BH4)2-Ca(BH4)2 system, recently characterized by Dematteis et al. [272]. The dehydrogenation process is usually very complex with multi-step reactions and is subject to the conditions used, such as temperature, heating-rate, and atmosphere.
For the first time, the existence of a quinary liquid borohydride phase was described by Dematteis et al. [274]. Above 250 °C, a five-component liquid borohydride was synthesized by an equimolar mixture of LiBH4-NaBH4-KBH4-Mg(BH4)2-Ca(BH4)2 obtained via eutectic reactions involving bimetallic borohydrides. The dehydrogenation reaction of the five-component liquid borohydride occurs in a complex multistep reaction as for the ternary and quaternary systems. Dehydrogenation temperatures are not significantly improved with respect to the pure borohydrides. Furthermore, not all the initial borohydrides have shown decomposition during treatment up to 500 °C.

Thermodynamic Properties of Eutectic Borohydride Systems
The enthalpies of mixing the liquid phase for the assessed systems based on borohydrides [247,249] shows negative or low positive values (Table 3), suggesting an attractive interaction among different borohydrides in the liquid phase. Experimental values of the enthalpy of mixing are difficult to measure and the lack of information for such important thermodynamic properties leaves a gap in the full description of thermodynamics of eutectic borohydride systems. However, by the CALPHAD approach, eutectic borohydride systems can be assessed considering different contribution to the excess Gibbs free energy. They can be theoretically described in the whole temperature, pressure and compositional range. In the case of an ideal solution model, the enthalpy of mixing for a solution is equal to zero. If an ideal behaviour for both solid and liquid phases is considered for a binary A-B system, the enthalpy of mixtures is the weighted average of that of the pure components. Therefore, the enthalpy of melting varies linearly, as a function of composition, between those of the pure components. Neglecting the temperature dependence of the enthalpy, the deviation from the ideal behaviour of the liquid phase can be outlined by a simple thermodynamic cycle if the solid solutions are taken as ideal. In fact, the enthalpy of mixing for the liquid phase (ΔHmix) can be estimated from the difference between the enthalpy of melting of the AB mixture (ΔHmAB) and of the weighted average of that of the pure components (ΔHmA and ΔHmB): where xA and xB are the molar fractions of the A and B components, respectively. The regular solution model defines the enthalpy of mixing as a function of composition for a binary A-B system by Equation (13): where Ω is the interaction parameter, which can be easily estimated according to Equation (14): By the integration of the differential scanning calorimetry (DSC) signal of the melting reactions in mixtures of borohydrides, the enthalpy changes from solid to liquid phase can be measured. Thus, possible interactions in the liquid phase can be evidenced and the value for the enthalpy of mixing can be estimated. For binary systems showing thermal minima (e.g., NaBH4-KBH4), the solid solution can be considered as ideal. In case of a eutectic melting (e.g., LiBH4-NaBH4 and LiBH4-KBH4), the solid phases can be considered fully immiscible so that the enthalpy of the solid mixture is obtained by averaging that of the pure components. Then the CALPHAD method allows the assessment of these parameters using a least square fit and taking into account all available data on pure compound from the available database or new experimental data [235]. For the LiBH4-NaBH4, the assessed value of Ω (Table 3) is negative, confirming the presence of a strong interaction in the liquid phase. For the LiBH4-KBH4 system, the assessed value is strongly negative. Finally, for the NaBH4-KBH4 system, ΩASS presents a low positive value. A deviation from the ideal cubic solid solution can explain the assessed value. The assessed enthalpy of mixing for the liquid phase is reliable, because they take into account possible variation of enthalpy as a function of temperature and possible deviations for ideal solid solutions. * A temperature dependency and sub-regular model were used to describe the LiBH4-NaBH4 system. The reported values have been recalculated in this review for a system described by a regular model.
As presented, in borohydride mixtures, all binary multi-cationic systems investigated so far show a stabilisation of the liquid phase, forming eutectic mixtures with negative or low positive enthalpy of mixing of the liquid phase [249]. Meanwhile, binary multi-anionic system of borohydrides and halides evidenced a strong interaction in the solid phases, while no eutectic point has been reported yet. It can be concluded that in borohydride systems, the cationic and anionic sublattices play a different role in the interaction in solid and liquid phase upon mixing. Their thermodynamics are strongly related to the structures, coordination number, temperature, enthalpy and entropy of the transitions.
The stability of a phase is related to the reciprocal interaction among the ionic sublattices and electronegativity of the ions involved in the structure [276]. In the case of anionic substitutions, the formation of a solid solution is favoured, which may indicate that the anionic sublattices of the borohydrides are characterized of equivalent sites, which can host randomly different anions with a similar size. Nevertheless, the cation sublattices are more related to their characteristic coordination number, limiting the formation of solid solutions, but allowing destabilizing the sublattices towards a liquid phase. The stabilisation of the liquid phase at low temperatures can partially be understood considering a variation of the charge distribution in the physical mixture of different borohydrides. Short-range order and entropy could play a role as well in the formation of a multi-cation liquid phase, which should be further investigated.

Hydrogen Sorage Properties of Eutectic Metal Borohydride Systems
The dehydrogenation mechanism of the 0.62LiBH4-0.38NaBH4 mixture in flowing Ar has been systematically investigated by Liu et al. [237]. A weight loss of 10.8 wt % hydrogen was recorded after heating to 650 °C through two major decomposition routes with the precipitation of LiH, Li2B12H12 and B (287 °C to 520 °C), and Na and B (520 °C to 650 °C). The re-hydrogenation attempts showed that the hydrogen content decreased from 5.5 wt % to 1.1 wt % and to 0.8 wt % with no evidence of a reformation of LiBH4. It was suggested by Liu et al. [277] that this poor cyclic stability could be improved by nanoconfinement.
The eutectic 0.725LiBH4-0.275KBH4 melts at 105 °C. Subject to the temperature profile, it releases up to 8.9 wt % hydrogen in Ar [278]. Like the LiBH4-NaBH4 mixture, it does not release hydrogen during melting. The dehydrogenation steps are firstly involving the formation of LiH, Li2B12H12, B and H2 from 290 °C to 700 °C, and secondly K, KH, B and H2 from 550 °C to 700 °C. This mixture has a poor cyclic stability as the reversible hydrogen content decreased dramatically from 5.8 wt % to 1.0 wt % from the second cycle [278], which cannot be improved through nanoconfinement [238].
Dematteis et al. [191] investigated the hydrogen storage properties of eutectic borohydrides mixed with Mg2NiH4 to form reactive hydride composites (RHC). This approach evidenced enhanced kinetic and thermodynamics, and reversible hydrogen release under moderate pressure and temperature conditions [180]. In fact, the hydrogen release properties of mixtures, such as Mg2FeH6-M(BH4)x (M = Li, Na, K, Mg, Ca) have recently been explored [279]. Mg2NiH4-M(BH4)x systems (M = Li, Na, Ca; x = 1,2) have been investigated as well. These show an improvement of the hydrogen decomposition properties and a better cyclability [280][281][282][283][284][285]. The decomposition reaction might form boride species such as MgNi2.5B2. The latter species are boron donors to reversibly form the borohydride during the re-hydrogenation process. Dematteis et al. [191] also observed a lower decomposition temperature of the RHC mixture if LiBH4 is in a eutectic mixture with borohydrides that contain stable single charged metal cation, while a leftover of the more stable borohydrides was detected after cycling. In the eutectic mixture, the presence of double charged metal cation (i.e., Ca(BH4)2 and Mg(BH4)2) leads to a complete decomposition of the borohydrides. If compared the hydrogen release temperatures of the pure borohydride and pure Mg2NiH4, its decrease is promoted by the presence of a stable liquid at low temperature.

Conclusions
The present chapter focused on the thermodynamic and hydrogen storage properties of several known low melting point borohydride systems, such as LiBH4-NaBH4, while halide substitution has been described elsewhere [190,286]. Given by the nature of liquid phase, these systems can start to deliver hydrogen at much lower temperatures when compared with their pure constituents. The reorganization of crystal structure after fusion may, in principle, lead to a weakening (or strengthening) of chemical bonds as a consequence of changing chemical environment, which can affect the decomposition. Usually, the dehydrogenation of these systems goes through a complex path with the formation of a series of B-contained phases. One of the important issues is to identify the eutectic temperature for such multi-components systems. With the help of simulation methods, i.e., ab initio and thermodynamic assessment methods, the eutectic temperature could be refined for the known systems or predicted for new systems. Prior to this study, only few hydrides and LiBH4 were fully described with the CALPHAD method. A limited database was available for metal hydrides and complex hydrides. Within the ECOSTORE project, pure borohydrides, binary and ternary combination of eutectic borohydrides of interest were fully characterised and assessed. However, a detailed description of the structure and thermodynamics of the liquid phase in borohydrides is still missing and should be investigated, possibly with total scattering methods. Experimental values of thermodynamic properties of eutectics and liquid borohydrides would be of interest to enhance the understanding of the role of the liquid in the thermal decomposition or hydrogen release reactions. They can be determined by accurate calorimetric measurements coupled with volumetric measurements.
The critical assessment and the incorporation of data into databases for calculations improve the understanding of various industrial and technological processes as well as helping in the prediction of properties of multicomponent systems. The CALPHAD method is a powerful tool for a complete description of the thermodynamic properties of borohydride mixtures. An estimation of the decomposition reactions in the investigated system can be obtained with the current optimised database. However, it should be coupled with experimental evaluation of decomposition products. As a matter of fact, the studies presented in this chapter critically collect previous studies and combine them with new experiments to finally describe the thermodynamic properties in all temperature, pressure and composition ranges.

Kinetic Tailoring of 2LiBH4 + MgH2/2LiH + MgB2 with Cost Effective 3TiCl3· AlCl3
Hydrogen storage in complex metal hydrides (e.g., borohydrides [BH4] − , alanates [AlH4] − , and/or amides [NH2] − ) is a fascinating field, ever since Bogdanović and Schwickardi published their seminal article reporting the possibility of the re-hydrogenation of sodium alanate in the presence of a Tibased catalyst [287]. Since this discovery in 1996, the field of complex metal hydrides has expanded tremendously and a great number of complex metal hydrides for hydrogen storage system have been  [288][289][290], reaction kinetics involving the breaking and the formation are usually sluggish. As discussed previously, for a decade, several strategies have been developed to accelerate the hydrogen sorption properties of complex hydride materials. Firstly, the combination of two or more single hydrides to form new compounds may lower the stability or reaction enthalpy of the system and promote its reversibility while minimizing the loss of the hydrogen capacity of the original materials [151,285,[291][292][293][294]. While amide examples were discussed previously, other reactive hydride composites, such as MgH2 + 2NaBH4, Mg2NiH4 + Ca(BH4)2, and MgH2 + 2LiBH4, are a tactic to destabilize metal hydride compounds by formation of corresponding metal hydride and MgB2 upon thermal decomposition. The latter reaction product was shown to be a desired product as it aids in re-hydrogenation by providing a pathway for the borohydride formation [295]. Secondly, the doping with suitable additives, i.e., transition metal-based additives, which provide kinetically favored nucleation sites for the formation of reaction products for the proceeding of the hydrogen uptake and release at moderate temperature and pressure conditions. Several additives such as Ti, Zr/V/Ce containing additives, Sc2O3, Nb2O5 or halides (NbF5, TiCl3) greatly improved the dehydrogenation-hydrogenation rates of lithium-based reactive hydride composites [296][297][298][299][300][301][302][303][304][305][306]. Thirdly, the confinement of the storage materials in adaptive scaffolds for improved hydrogen sorption properties [307] leads to the increase of the materials surface area, reduces the hydrogen diffusion distance and provides adjacent contact between reactants [308][309][310][311][312][313].
In this chapter, the second approach will be addressed in more detail for the kinetic improvement of Li-RHC system. The Li-RHC system (i.e., 2LiBH4 + MgH2/2LiH + MgB2, denoted as Li-RHC) has attracted much attention for hydrogen storage due to its appealing theoretical hydrogen storage capacity of 11.5 wt % and calculated reaction enthalpy of 45.9 kJ mol −1 H2 [291,294]. This system shows the possibility to obtain fully reversible metal borohydrides at evaluated temperature and hydrogen pressure conditions. Vajo et al. [294] reported the possibility to hydrogenate 2LiH and MgB2 including a catalyst (2-3 mol % TiCl3) following the reaction scheme: 2LiBH4 + MgH2 ↔ 2LiH + MgB2 + 4H2. Further efforts have been made to investigate the RHC NaH + MgB2, with a theoretical hydrogen capacity of 7.8 wt % for a 2:1 molar equivalent of NaH and MgB2. The suggested reaction products upon hydrogenation are 2NaBH4 and MgH2. However, it was shown that this procedure had intermediate reaction products such as NaMgH3, which decreased the hydrogen capacity. The reaction is plagued with slow hydrogen sorption kinetics and with a low experimentally obtained hydrogen absorption capacity of only 3.8 wt % [314][315][316][317][318][319]. Recently, Le et al. [320] reported a study of the effect of 3TiCl3·AlCl3 on the kinetic sorption properties of Li-RHC system. In their work, they showed remarkably improved dehydrogenation kinetics of Li-RHC in the presence of the 3TiCl3·AlCl3 additive (Figure 16), i.e., the incubation for the formation of MgB2 is considerably reduced by 10 h. The increase of the molar amount of 3TiCl3·AlCl3, from 7.5 to 30, leads to minimizing the incubation period between the two reaction steps. However, because of the increased amount of 3TiCl3·AlCl3, the amount of released hydrogen considerably declines from about 10 wt % to 8 wt %. It was shown above that the use of the selected additive clearly promotes the sorption kinetics of the system (Li-RHC). Figure 17 shows the measured reversible hydrogen capacity of the studied materials in both the hydrogenated and dehydrogenated states for 20 hydrogenationdehydrogenation cycles. The additive 3TiCl3·AlCl3 presents a beneficial effect on the hydrogen storage of the material prepared in dehydrogenated state (Figure 17b-Li RHCd) rather than on the material prepared in the hydrogenated state (Figure 17a-Li RHCa). Clearly, the Li-RHCd + 7.5(3TiCl3·AlCl3) material shows the highest measured hydrogen storage capacity, i.e., ~9.5 wt % and a rather stable cycling behaviour, with a loss of capacity smaller than 1.2 wt % over 20 cycles. The reduction of the hydrogen storage capacity for most of doped Li-RHC composites is less than 1.2 wt.%, except for the system Li-RHCd + 3.7(3TiCl3·AlCl3), containing a small amount of additive, which shows the largest drop of capacity (>2 wt %) in the cycling period. This could be attributed to the unsatisfying dispersion of the additive over the host material Li-RHC. Moreover, Li-RHCd + 7.5(3TiCl3.AlCl3) displays the shortest kinetic performance (hydrogenation-dehydrogenation times of ~30 min) in both hydrogenation and dehydrogenation processes in comparison to the material prepared in the hydrogenated state Li-RHCa + 7.5(3TiCl3·AlCl3), reported in [320]. In additional reports [296][297][298][299][300][301][302][303][304][305], the formation of TM borides was strongly suggested from the interaction between the Li-RHC with TM-based compounds. For the Li-RHC doped with 3TiCl3·AlCl3, the calculations of phase composition are referred to in the study of Le et al. [320]. The calculation showed that the formation of TiB2 and AlB2 (Equation (15)) over milling is thermodynamically more favorable than the formation of AlTi3 (Equation (16)). However, the characterization by XPD and 11 B NMR [320] showed that no observation of transition metal borides is detected and the presence of free Mg as proposed in Equation (15) is not observed. Thus, the formation of TiB2 and AlB2 may be kinetically constrained and the formation of AlTi3 is likely to take place. 0.072LiH(s) + 0.024MgB2(s) + 0.06(3TiCl3·AlCl3)(s) → 0.072LiCl(s) + 0.024Mg(s) + 0.006AlB2(s) + 0.018TiB2(s) + 0.036H2(g) G1bar, 25 °C = -10.9 kJ mol −1 (15) 0.072LiH(s) + 0.06(3TiCl3·AlCl3)(s) → 0.072LiCl(s) + 0.006AlTi3(s) + 0.036H2(g) G1bar, 25 °C = -7.6 kJ mol −1 (16) Conclusively, the presence of Ti/Al containing particles in the Li-RHCa + 7.5(3TiCl3·AlCl3) was examined by the high-resolution transmission electron microscopy (HR-TEM) technique by Le et al. [320]. Figure 18 shows the nanosized particles of hexagonal AlTi3 and cubic AlTi2 which are observed in the material after ball milling. During cycling, the XPD data depicted in Figure 18b and 18c show hexagonal AlTi3 and cubic AlTi2 which are still detectable as well as cubic AlTi3. As observed, the most frequently identified nanoparticles belong to the hexagonal AlTi3 phase. In addition, the transition metal borides (TiB2 and AlB2), Mg-Al alloys and Mg-Al-B compounds have not been found via HR-TEM. The sizes of the hexagonal AlTi3 after milling and cycling are estimated in the range of 1 to 10 nm and 1 to 5 nm, respectively. These observations are in accordance with the anomalous small angle X-ray scattering (ASAXS) results [320].
It is well known that for metal alloy systems, the nucleation and growth of precipitates is strongly influenced by crystallographic and morphological features [321]. The presence of a phase with particular crystallographic structure plays a critical role in the nucleation, growth and distribution of certain precipitates. The similarities of crystal features between the new phase and the parent phase lead to a lower energy at the coherent interface by decreasing the interface lattice misfit. Zhang et al. [322] pointed out the criteria for the edge-to-edge matching model and stated that the strain energy between the substrate and nucleus is significantly minimized when the misfit of the interatomic spacing of their close-packed or nearly close-packed atom rows is <10% and their interplanar mismatch is < 6%. As for the precipitates of the metal alloys, in the RHC system, the crystallographic features of the in situ formed nanoparticles and their location seem to play an important role to the processes of nucleation and growth of new phases. Bösenberg et al., Bonatto Minella et al., Puszkiel et al., and Le et al. [296,320,323,324] used these above-mentioned criteria to explain the influence of the additives on the kinetic properties of the RHC systems. They independently reported that the parent phases having similar crystal structure as the newly developing phase can act as heterogeneous nucleation sites for nucleus formation. Several transition metal borides such as TiB2, NbB2 and VB2 can provide coherent interfaces to improve the heterogeneous nucleation and growth of hexagonal MgB2 [296,[298][299][300][301][303][304][305]325]. As an example, the illustration of the growth of MgB2 during dehydrogenation, adapted from [296], is given in Figure  19. For this case, Li-RHC mixed with 3TiCl3·AlCl3 results in the in situ formation of AlTi nanoparticles. The necessary requirements for an effective heterogeneous nucleation are reported in Refs. [296,322,323]. In detail, those are firstly a low interfacial energy amongst the nucleation agent and the newly developing phase. Secondly, a sufficient high amount of the nucleation agent. Thirdly, the appropriate distribution of the nucleation agent. To obtain the minimized lattice mismatch, the atom rows in the close-packed or nearly closed-packed planes arranged to meet edge to edge at the interface. A simple approach to identify the close-packed planes is by XPD. The close-packed planes are those with the highest X-ray diffraction intensity in the diffraction patterns. The relative difference in the d-value between two close-packed planes is known as d-value mismatch. There is a critical value of the d-value mismatch, below which the planes containing the atom rows in the two phases can match well, so that the interfacial energy is minimized. A reasonable reference value for the critical d-value mismatch is 6% or smaller [322].
Opposed to the reports of Bösenberg et al., Bonatto Minella et al., and Puszkiel et al., [296,323,324], in Le et al., AlTi alloys have been identified as a product of the interaction between Li-RHC and 3TiCl3·AlCl3. The d-value mismatches of AlTi alloys and MgB2 have been calculated, referred to [320], that is, MgB2{1011}//hexagonal-AlTi3 {2021} = 3.3%; MgB2 {1010}//cubic-TiAl3 {2020} = 31.9% and MgB2 {1010}//cubic-TiAl2 {2020} = 35.0%. These results point out that the hexagonal AlTi3 phase has a similar crystal structure to MgB2 with a d-value mismatch less than 6%. Additionally, based on the observed kinetic improvement in Li-RHCd + 7.5(3TiCl3·AlCl3) and according to the ASAXS [320] and HR-TEM results, the proper amount of AlTi3 nanoparticles appears to be well dispersed and is located on the formed MgB2 upon cycling. From these observations, it is indicated that the hexagonal AlTi3 alloy plays an efficient heterogeneous nucleation for formation of newly developing phase (MgB2). Furthermore, the presence of AlTi3 facilitates the decrease of MgB2 particle size, leading to a faster consumption of MgB2, suppressing the MgB2 agglomeration and providing a more active surface for the interaction between material and hydrogen gas.  In conclusion, the investigations on tailoring the sorption kinetics with additives of the reactive hydride composites, involve the understanding of the kinetic and thermodynamic behaviours of the studied system. The findings reported in this chapter lay the ground for continuing research and for transferring the developed approaches to similar RHC systems. A deeper investigation of possible synthetic pathways for producing TM-based nanoparticles is expected to lead to a better tailoring and to further improvement of the material kinetic properties.

Role of Nanoconfinement in Enhancing the Properties of Hydrogen Storage Materials
Despite the high hydrogen content of many light metal hydrides and complex hydrides, the high stability of these compounds precludes their application for hydrogen storage purposes, commonly due to the high temperatures and low reversibility of the dehydrogenation reactions. As already mentioned before one of the strategies to address this issue involves the nanoconfinement of hydrides in porous matrices with pore size in the order of few nanometres, destabilizing the confined material and promoting the hydrogen exchange [326]. With this approach, it is possible to reduce drastically the macroscopic size of the material by introducing and trapping it in small cavities. The confined hydride is physically prevented from agglomerating by the pore walls, allowing the formation of nanostructured material in form of size-controlled nanoparticles. The term nanostructured material referred to a material with a microstructure characterized by length scale in the order of a few nanometres, typically between 1 and 10 nm, that maybe extended into one, two or three dimensions, as in nanorods, layers or coating and nanoparticles, respectively. The decrease of the particle size leads to an increase of the surface to volume ratio. However, creating surfaces costs energy, if the surface energy of the hydride is larger than that of the metal, this operation will reduce the enthalpy of the formation of the system. For small particles, the energetic contribution given by the surface energy cannot be ignored and the reduced ΔH implicates a substantial destabilisation of the whole system [327][328][329]. In addition to the aforementioned destabilisation, the size reduction of the particles directly affects the kinetics of hydrogenation/dehydrogenation, increasing the rate of hydrogen dissociation on the surface and the diffusion of hydrogen into the structure of the material [313,330].
High-energy ball milling [254] is still the most common method to obtain such small particles, with grain sizes down to 3-25 nm with a relatively controllable particle size distribution and reduced equipment costs [331][332][333]. This first step in a "top-down patterning to guide bottom-up assembly" [334] is followed by more complicated approaches involve bottom-up mechanisms of nucleation and particles growth in solution or vapor that requires good control of the reaction conditions. Such methods are, however, not easy to scale up and are mostly used in the case of TM [335][336][337][338]. However, the good achievements reached with ball milling and other synthesis of hydrides nanoparticles are soon nullified by hydrogenation-dehydrogenation cycles that contribute to the aggregation and reaggregation of the particles via repeated sintering [293,339,340].
The confinement of hydrides requires porous materials with certain characteristics in order to be viable for application. In first instance, the scaffold needs a proper porous structure. A high pore volume is a fundamental aspect to allow the confinement of sufficient amounts of active material. Common values of total pore volume are near to 1 mL g −1 where higher values are even more appreciated [341][342][343]. However, not all pores are good for the confinement of hydrides and the synthesis of nanoparticles. The size and connectivity of the pores are crucial aspects to take into consideration. If the cavities are excessively large, the host particles would have a behaviour comparable to that of the bulk. Moreover, capillary forces that drive the introduction into the pores of material in a liquid phase are substantial between a diameter of 2 and 50 nm [344]. Specifically, the International Union of Pure and Applied Chemistry (IUPAC) nomenclature refers to materials with a pore size between 2 nm and 50 nm as "mesoporous materials" [345].
The structural material must be light in order to maintain an overall acceptable gravimetric density of hydrogen. For this reason, carbon and silica are commonly favoured over other materials in the preparation of porous materials such as metal oxides. The latter are often chosen for catalytic applications, for the preparation of heat exchangers and biomedical applications [346][347][348][349]. Moreover, a good scaffold should be chemically inert with hydrides and have a high thermal stability as the common operation temperatures of metal hydrides are excessively high for scaffolds based on organic and polymeric structures. For this reason, both carbon and silica scaffolds are usually preferred, even if silica, due to many functional groups on the surface may react with the host material. For example, in the case of complex metal hydrides, LiBH4 reacts with silica, forming irreversibly lithium silicates [41,350,351]. Carbonaceous scaffolds, however, can be heated at high temperatures without structural collapses and with an inert behaviour.
Both silica and carbon porous materials also satisfy the characteristics necessary for production on a large scale: the low price of production, involving cheap precursor materials and simple production methods.
In this section, nanoconfinement approaches will be discussed and related to complex metal hydrides.

Nanoconfinement Approaches
The confinement of the active material in pores of few nanometer can be performed following different approaches, limited by the physical chemical characteristics of the material. Wet infiltration and melt infiltration are the two main methods to introduce hydrides in a porous scaffold. The first method involves the solubility of the hydride in a solvent that, in the form of solute, can deeply penetrate into the porous structure of the host material. The solvent can then be removed by evaporation, leaving the active material trapped and widely dispersed in the porous structure, as reported by Peru et al. and others [352][353][354][355]. However, this method suffers from synthetic restrictions due to the solubility of the hydrides to nanoconfine, limiting the amount of material that is possible to solubilize and infiltrate, requiring more impregnations in order to load the scaffold properly [353]. Moreover, the strong interactions among solvent and solute hydrides can make the complete extraction of solvent required for the infiltration difficult, in particular, in the cases when the solvent generates stable coordination complexes with the solute [356,357]. The second method, melt infiltration, is adopted on hydrides or mixtures of hydrides with a decomposition temperature higher than their melting point. Ionic hydrides such as LiH, NaH or MgH2 generally have high melting points. Much lower temperatures are required in order to melt complex metal hydrides such as alanates, borohydrides and amides. The melt infiltration of these compounds is simple and possible at relatively low temperatures with the application of a hydrogen backpressure in order to prevent the material decomposition with consequent hydrogen release [344,351]. Without the use of a solvent, the infiltration is faster and easier to perform. Moreover, the melt infiltration allows a more efficient load of the porous material since there are no limits due to the solvent solubility.
As mentioned previously, due to their high hydrogen content, borohydrides are among the most interesting compounds for hydrogen storage. However, these complex hydrides are characterized by elevated stability that prevents them from practical use. For this reason, the nanoconfinement approach may be one of the best solutions to destabilize and make borohydrides and other systems previously discussed viable materials for hydrogen storage applications.

Confined Borohydrides
As mentioned on several occasions in this review, lithium borohydride seems promising for solid-state hydrogen storage. However, in bulk form, its stability allows only a slow hydrogen desorption. The presence of LiH among the reaction products reduces the overall hydrogen amount that one could obtain by thermal decomposition. Moreover, the high stability of the products makes it impossible to rehydrogenate LiBH4 under reasonable pressure and temperature conditions [89,357]. A certain degree of destabilisation that improves the LiBH4 sorption properties can be achieved by nanoconfinement [358]. LiBH4 introduction in a porous scaffold has been performed both through chemical impregnation using tetrahydrofuran (THF) as solvent [359,360], and melt infiltration at about 280-300 °C under hydrogen or inert gas pressure [361][362][363]. The infiltrated LiBH4 shows a clear increase of the hydrogen exchange rate of several orders of magnitude [361] and the desorption temperature decrease drastically. The onset temperature of composite materials with carbon is shifted to about 200 °C, while the main desorption peak is generally centered between 300 and 350 °C [359,362,[364][365][366]. Moreover, the confined LiBH4 shows enhanced reversibility under mild conditions, partially preserving its hydrogen capacity after few cycles. The reason for this improvement is the suppression of B2H6 release that, in turn, results from smaller pores preventing the loss of boron through B2H6 during dehydrogenation [366,367].
The investigation of the nanoconfinement effect is not limited to the pure LiBH4. The presence of a low melting temperature eutectic mixture of borohydrides allowed the introduction in porous scaffold of LiBH4 at even lower temperatures than 280 °C and further destabilisation due to the interaction of different cations in the molten phase [235,289]. In most of the eutectic mixtures examined, LiBH4 constitutes the major fraction of the mixture, giving an overall high hydrogen density. NaBH4 is very attractive for hydrogen storage but like LiBH4, it releases hydrogen only at very high temperatures with the main desorption starting above 450 °C and proceeding with a multistep decomposition above 500 °C [368][369][370]. Evidence of its destabilisation and drastic reduction of desorption temperatures have been reported for nanoconfined composites. Pure NaBH4 melts at about 500 °C [235], which allows for the melt infiltration under hydrogen atmosphere [369]. However, the stability in different solvents allowed the infiltration through various media such as liquid NH3 [371], diglyme [352], n-propylamine [372] and even water at high pH values [352]. The desorption temperatures of nanoconfined NaBH4 found in the aforementioned literature variate widely by the method, the materials adopted for the infiltration and by the scaffold. The composite materials obtained by melt infiltration in porous carbon show that the onset temperature shifted to below 300 °C while the main desorption takes place at 450 °C [369]. For the same materials, but impregnated with an alkaline aqueous solution, the values for the onset and main hydrogen desorption dropped further to 200 °C and 350 °C, respectively [369]. Similar temperatures for the main desorption were obtained by using NH3 or diglyme as solvents showing initial low hydrogen release even below 150 °C [352,371]. Even better performances were achieved confining NaBH4 among graphene layers by solvent impregnation with n-propylamine; in this case, the dehydrogenation starts below 40 °C with a main desorption at 200 °C showing a stable reversibility over a few sorption cycles [372].
The loading limitations due to the NaBH4 solubility and the solvent elimination from the sample encouraged the improvement of the melt infiltration approach. The use of recently studied eutectic mixtures of LiBH4 and NaBH4 allows the introduction in a porous scaffold of a larger amount of hydrides [237,247]. A composite material obtained by nanoconfinement of a mixture of borohydrides achieves a kinetic enhancement by both the presence of the carbon scaffold and multiple cationic species. The introduction in a carbon aerogel allows the reduction of the hydrogen desorption temperatures of 107 °C, bringing the main desorption of the composite LiBH4-NaBH4 in carbon aerogel to 333 °C [246].
Among the alkali borohydrides, KBH4 seems not to be very suitable for hydrogen storage applications, mostly due to the high stability and as consequence the high temperature of hydrogen release, ~680 °C [235]. However, KBH4 combined with LiBH4 generates a low melting point system. The eutectic composition, with the 0.275 mol % of KBH4, melts at about 105 °C, allowing the extremely low temperature melt infiltration of LiBH4 and KBH4 in porous scaffolds [245]. In carbon aerogel with an average pore diameter of 25 nm, the kinetics are improved and the system starts to release hydrogen at about 190 °C, with a major desorption at about 400 °C [238].
Another complex hydride with the potential to fulfill the requirements for hydrogen storage materials is Mg(BH4)2 due to its extremely high content of hydrogen of 14.9 wt.%. Mg(BH4)2 shows a variety of crystal structures, (α-, β-, γ-, δ-, ζ-and two as-yet unsolved structures β'-and ε-Mg(BH4)2) and its transitions are rich. For instance the porous γ-Mg(BH4)2 [373] experiences two endothermic phase changes, the first at ~150 °C to the ε-phase, and the second at ~180 °C to β', as reported by Heere et al. and others [374,375], before the high temperature β-phase transition (~220 °C) occurs. In contrast, α-Mg(BH4)2 transitions directly into the β phase at ~220 °C. The complexity of decomposition pathways, structures and dynamics are reported in [376,377].
In general, the hydrogen desorption for the Mg(BH4)2 bulk material occurs between 290 °C and 500 °C in different steps [378,379], while the nanoconfinement showed a general improvement of the hydrogen sorption behaviour. The infiltration of Mg(BH4)2 is performed via solvent impregnation, commonly using diethyl ether. In a carbon scaffold with pores smaller than 2 nm, Mg(BH4)2 shows hydrogen desorption starting at about 150 °C under vacuum and peak temperatures at 257 °C and 317 °C [380]. Even better performances were recorded by infiltrating the Mg(BH4)2 in a porous scaffold doped with transition metals [381]. The use of a nickel-doped carbon drastically reduces the temperature of hydrogen release with the dehydrogenation starting at 75 °C and reaching the maximum desorption rate at 155 °C [382]. A melt infiltration approach is possible by using an eutectic mixture of LiBH4 and Mg(BH4)2. The two borohydrides with a molar ratio of about 1:1 show a very low melting point of 180 °C, allowing the introduction in porous scaffolds of a large amount of LiBH4 and Mg(BH4)2 [383]. The presence of carbon accelerates the hydrogen desorption rate; meanwhile the encapsulation in small pores inhibits the formation of closo-borates, [B12H12] 2− , enhancing the cyclability [384]. The eutectic mixture infiltrated in carbon aerogel shows low desorption temperatures, with a first desorption peak in coincidence with the melting temperature of the mixture (180 °C) and a major dehydrogenation at about 250 °C against the bulk mixture, where the main dehydrogenation events occur in the temperature range of 230-400 °C. Moreover, the 8.3 wt % of hydrogen uptake is maintained after four sorption cycles in comparison to the approximately 3 wt % of the bulk [239].
The last example in this chapter is calcium borohydride. This complex hydride decomposes in two main steps, the first at about 320 °C and the second at 550 °C, with an overall hydrogen capacity of 9 wt.% [385]. The introduction of pure Ca(BH4)2 in a carbon aerogel structure was performed by solvent impregnation with methyl tert-butyl ether (MTBE) [355]. Once nanoconfined, the hydride starts to release hydrogen at about 100 °C, with a main desorption peak at about 280 °C while showing a hydrogen capacity of 2.4 wt % for 18 cycles [355]. Further studies have been performed on eutectic mixtures of Ca(BH4)2 with LiBH4 and their nanoconfinement. This eutectic mixture, with the composition determined to be 0.68LiBH4-0.32Ca(BH4)2, melts at about 200 °C, allowing a low temperature melt infiltration in a porous material [236,252]. The infiltration in different carbon scaffolds showed the gradual decrease of the main dehydrogenation peak temperature with the reduction of the pore size, as shown in Figure 20 [386]. An even lower desorption temperature has been registered for high surface area CO2 activated aerogel carbon, with a difference between bulk and infiltrated material of 95 °C and the main desorption peak occurring at 298 °C compared to 393 °C for the bulk [246]. However, the improvements registered for the first desorption are gradually lost within five hydrogenation cycles [387].

Conclusions
Even bringing consistent improvements to various complex hydrides, the analyzed nanoconfined systems still do not have the characteristics required for practical applications. However, the study of new infiltrated materials opens new possibilities and lead us closer to the realization of a solid-state hydrogen storage application.

Rare Earth Borohydrides
The International Union of Pure and Applied Chemistry (IUPAC) defines rare earth (RE) elements (E) as the 15 elements belonging to the lanthanides group, plus yttrium and scandium, as they are usually found within the same ore. REE have atomic numbers (Z) of 57 to 71, 39 and 21, respectively. The lanthanide group, situated in the 4f block of the periodic table, has a decreasing ionic radius with an increasing atomic number. La (Z = 57) has the largest ionic radius (1.172 Å ) and Lu (Z = 71) the smallest ionic radius (1.001 Å , based on the ionic radius of RE 3+ cations in octahedral environment) [388].
When referring to REE, society usually infers this term to be related to their abundance, distribution and prices. In fact, the word 'rare' does not mean that they are not abundant; 'rare' simply refers to the difficulty scientists had in the 19th century in separating the REE from its ore. 'Earth' is an ancient term for oxide and finds its origin in the fact that REE permanently form oxides [389]. Their abundance in the earth crust is comparable to copper and lead. More specifically, the light REE (La-Sm) are more abundant than the heavy REE (Eu-Lu). Their abundancy is between two and five orders of magnitude higher than precious Au or Pt [389].

Synthesis of Rare Earth Borohydrides (REB)
Rare earth borohydrides, pure and in composite mixtures, as well as in bimetallic compounds, are the focus of this chapter. A general overview of the metal borohydrides and their derivatives can be found in [190]. Nevertheless, the high thermal stability of the alkali and alkaline-earth borohydrides has attracted more attention to less stable compounds such as RE borohydrides and much attention has been paid to solvent-free as well as halide-free synthesis [390]. Generally, the approaches implemented for the synthesis of REB can be divided into solvent-free synthesis and solvent-based synthesis, as discussed below.

Solvent Free Synthesis of REB
The initial synthesis method for REB compounds was based on the reactions in Equation (17) - (18) and carried out by ball milling of a RECl3 with an alkali borohydride (Li, Na, K). LiBH4 has been proven to be the best precursor for this reaction [391][392][393].
However, the products obtained with this method are dependent on the ionic radius of the RE metal. REB compounds with four different structure types are formed [391,392]. Larger cations (La 3+ -Sm 3+ ) form cubic spinel like LiRE(BH4)3Cl structure with s.g. 4 ̅ 3 Equation (17) (18)). The latter was derived from NPD and 3 ̅ is the DFT optimized s.g. for β-RE(BH4)3. For Sm, α-and β-Sm(BH4)3 and small amounts of LiSm(BH4)3Cl form after ball milling [254]. Heating the sample to T ~ 180 °C, reduced Sm 3+ to Sm 2+ and Sm(BH4)2 is formed. Attempts to synthesize europium borohydride with the ball milling method were unsuccessful and any known Eu-containing phases could not be identified in the XPD pattern after milling. Yb forms both α/β-Yb(BH4)3 and LiYb(BH4)4 and heating the sample to T ~ 100 °C forms Yb(BH4)2 based on reaction shown in Equation (19). The smallest cation (Lu 3+ ) acts as the strongest Lewis acid and forms LiLu(BH4)4 compound in s.g. 4 ̅ 2 .
RE(BH4)3  RE(BH4)2 + 0.5 B2H6 (g) + 0.5 H2, RE = Sm, Yb The structure types of most compounds of the Ln-series based on solvent free synthesis are summarized in Figure 21 with respect to cations ionic radii.  (18) are contaminated with LiCl which acts as a 'dead mass' and decreases the active hydrogen content. Moreover, the formation of REB with larger cations was not possible via the solvent-free method. Therefore, solvent methods were developed to obtain pure rare earth borohydrides. In the latter, precursors are first reacted in a solvent i.e., toluene/ether which acts as a reaction medium to form RE(BH4)3 as per Equation (20): After solvent removal, a second solvent i.e., dimethyl sulfide (DMS) is added to the mixture forming complex solvate RE(BH4)3S(CH3)2 that is soluble in DMS and allows the filtration of LiCl. Using this method, RE(BH4)3, RE = La, Ce, Pr have been synthesized and allowed the formation of Eu(BH4)2 for the first time [398][399][400]. However, Payandeh et al. [401] found that the samples obtained are often contaminated with LiBH4 due to the partial solubility of LiBH4 in DMS. This contamination was observed for the first time in a La(BH4)3 compound which formed Li3K3La2(BH4)12 upon reaction with KBH4.
The third method for synthesis of REB is based on the reaction of an ionic hydride REH3 and a borane donating compound, e.g., S(CH3)2BH3 based on Equations (21) and (22).

REH3(s) + 3 S(CH3)2BH3(l)
Using this method, RE(BH4)3, RE = La, Ce and Eu(BH4)2 were synthesized and two high temperature polymorphs of Eu(BH4)2 were identified [400,402]. The crystal structures of RE(BH4)n (n = 2,3) obtained by either solvent free or solvent method are summarized in Figure 22 and in the following section, the crystal structures of trivalent REB are presented and the relationship between the different structure types is discussed.  [391,392,395,396,399,400,[402][403][404][405][406][407] are presented. The ionic crystal radius were taken from [388]. The high-temperature polymorphs are displayed with empty signs. The figure was adapted from [408] with permission from The Royal Society of Chemistry.

Crystal Structures of Monometalic REB
In Figure 23, three major structure types of RE(BH4)3 are shown. All of them are of the cubic rhenium oxide (ReO3) structure type in which Re atoms sit at the corner and O atoms at the edge. A wide selection of structures can be derived by coupled rotations of groups of octahedra.
3 ̅ and 3 ̅ polymorphs are distorted ReO3 structures but the 3 ̅ phase is an ideal cubic ReO3 structure. The ReO3 structure type is very closely related to the perovskite, SrTiO3, without the Sr atoms placed in the middle of the cube and therefore, having an empty void in this position, as shown in Figure  23b by the reddish sphere [409].

Pbcn F2dd Pbca
In all of the structures, RE atoms are coordinated to six BH4 − in an octahedral geometry and all the BH4 − groups are bridging between the RE 3+ centers. Pr(BH4)3 is the only REB that shows all three polymorphs with different symmetries, i.e., 3 ̅ , 3 ̅ and 3 ̅ . At RT, Pr(BH4)3 crystalizes in the s.g.
3 ̅ . Upon heating, the structure first unfolds and β-Pr(BH4)3 as ideal ReO3 structure ( 3 ̅ ) forms, but this polymorph is unstable and immediately transforms to r-Pr(BH4)3 with a smaller unit cell volume; this transformation is assigned to the large voids in the β-polymorph shown in Figure  23b by the reddish sphere. These voids allow the bending of the Pr-B-Pr bonds based on the rotation of Pr(BH4)6 octahedra. As a consequence, the Pr-B bond increases and the Pr-Pr bond distance decreases, which leads to contraction of the unit cell volume and stabilisation of r-phase ( 3 ̅ ), which was found by Payandeh et al. [408].
However, performing the same experiment under elevated hydrogen pressure (p(H2) = 40 bar) changes the polymorphic transition mechanism and β-Pr(BH4)3 as porous polymorph is stabilized as the major phase at 190 °C. Figure 22 shows that the crystal structures of the REB are dependent on the cation size, where praseodymium is a borderline element. When the cation size is smaller than Pr, the structure collapses and rhombohedral phase ( 3 ̅ ) forms at RT which is due to formation of large voids in the structure, which collapse. However, when the metal cation is smaller such as neodymium, 3 ̅ stabilizes as the RT phase.

Crystal Structures of Bimetallic REB
The structures of bimetallic rare earth borohydrides are more diverse and are presented in Table  4. While this chapter only discusses the selected compounds, further details can be found in [410]. Bimetallic LiRE(BH4)3Cl, RE = Ce, Pr, Nd, Gd and LiLa(BH4)3X, X = Cl, Br, I compounds have received significant attention due to their high ionic conductivity of ~10 −4 S cm −1 at RT. These compounds crystalize in cubic unit cell with a spinel-like structure and s.g. 4 ̅ 3 (No. 217), as has been reported by Payandeh et al. [411] and others [63,391]. In this structure Z = 8 and La, B, and X atoms fully occupy the 8c, 24g, and 8c Wyckoff sites. However, Li-ions occupy the 12d sites and in order to maintain the charge neutrality, the occupancy should be 2/3. In this structure, anions, [La4X4(BH4)12] 4− , as isolated tetranuclear anionic clusters exist and these contain a distorted cubane-like La4 × 4 core. Furthermore, those are charge-balanced by Li + cations (Figure 24b). Li ions, which are coordinated tetrahedrally to four BH4 − groups via the tetrahedral edge and lanthanum atoms octahedrally coordinate to three halide ions and three borohydride groups via the faces (η3) and form a distorted octahedron. Partial lithium occupancy leads to a high lithium-ion conductivity for these types of compounds. Moreover, it has been found that the conduction channel can be tuned by changing the anions in the structure and the optimized size of channel is suggested when bromide is used as anion. However, unlike LiLa(BH4)3X, X = Cl, I samples, the bromide-containing sample is not stable and slowly decomposes to La(BH4)3 and LiBr over time. CuAlCl4 not specified, [391] The NaRE(BH4)4, RE = La, Ce, Pr, compounds crystalize in a new structure type with orthorhombic unit cell and s.g. Pbcn, (Figure 25a). In this structure, both Na and RE atoms have an octahedral coordination to BH4 − units. Two distinct types of BH4 -anions with coordination number of three exist in this structure. The first one bridges between two RE atoms and one Na atom. The second one bridges between two Na atoms and one RE atom, which creates a zig-zag chain of edge sharing Na(BH4)6 octahedra connected with edge sharing La(BH4)6 octahedra along the c-axis. The latter as a new type of structure for REB built from brucite-like layers of octahedra (hcp of anions) was reported by Payandeh et al. [412] with half of the octahedral sites empty. Similar octahedral chains have been observed for α-PbO2 (zig-zag chains) and rutile (straight chains).
The second group of NaRE(BH4)4, RE = Y, Er, Yb, Sc, with smaller size of rare earth cations crystalize in the s.g. Cmcm (Figure 25b). The structure consists of RE(BH4)4 − anions and Na + cations, forming a distorted trigonal prism around the complex anion. RE atoms have tetrahedral coordination to BH4 − units while Na atoms have octahedral coordination to BH4 − and from edgesharing K(BH4)6 octahedral chains along c-axis, see Figure 25b [403]. Figure 25c presents the unit cell volume per formula unit V/Z of the bimetallic sodium rare earth borohydride. Bimetallic borohydrides with bigger rare earth cations (La, Ce, Pr) crystalize in s.g. Pbcn and have a smaller unit cell volume whereas borohydrides with smaller rare earth cations (Y, Er, Yb) crystallize in s.g. Cmcm or C2221 with a larger unit cell volume. NaGd(BH4)4 forms an amorphous yet unstable phase. In s.g. Pbcn, both cations have the coordination number six, which leads to denser packing and smaller unit cell volume for these structures. Moreover, the unit cell volume of bimetallic borohydrides can be related to the unit cell volume of REB. As shown in Figure 22, bigger rare earth cations (RE = La, Ce) crystalize in the rhombohedral unit cell with s.g. R3 ̅ and V/Z in the range of 154.0-155.2 Å 3 . However, smaller rare earth cations (RE = Nd-Yb) crystalize in a cubic unit cell with s.g. Pa3 ̅ and larger V/Z in the range of 183-153.4 Å 3 . Therefore, when rare earth borohydrides (RE = La, Ce) with rhombohedral unit cell (R3 ̅ ) react with NaBH4, bimetallic borohydrides with smaller unit cell volume and Pbcn symmetry form [412]. However, the reaction between REB with cubic unit cell (Pa3 ̅ , RE = Nd-Yb) and NaBH4 forms bimetallic compounds with space group symmetry C2221 or Cmcm and larger unit cell volume. Pr(BH4)3 is the only rare earth borohydride with a cubic unit cell at RT (Pa3 ̅ ) that undergoes a polymorphic transition to a rhombohedral phase (R3 ̅ ) at T ~190 C. This transition is accompanied by a significant contraction in the unit cell volume. The reaction between NaBH4 and Pr(BH4)3 occurs at elevated temperatures and therefore, the formation of NaPr(BH4)4 compound with s.g. Pbcn is suggested to be a reaction between NaBH4 and the rhombohedral polymorph of Pr(BH4)3 (R3 ̅ ) [401].

Reactive Hydride Composites with REB
In this section, the hydrogen storage properties of selected RE(BH4)3 are investigated. Previous reports showed that in Er(BH4)3, synthesized by mechanochemical milling of ErCl3 + LiBH4, up to 20% of the initial released hydrogen could be re-hydrogenated [426]. However, an explanation of reaction products was not given. In order to gain better insight into the reaction mechanism, Er(BH4)3 was synthesized based on solvent methods. Surprisingly, for this pure Er(BH4)3 no reabsorption of hydrogen could be observed after initial decomposition and re-hydrogenation, and only amorphous decomposition products ErB4 and ErHn n = 2, 3 were expected to form.
However, Heere et al. [406] showed that by mixing 50 wt % LiH into these decomposition products and applying re-hydrogenation, ErH3 and LiBH4 were formed, monitored by in situ SR-XPD ( Figure 26). In summary, the re-hydrogenation can only be achieved in combination with a lithium source. Therefore, 20 wt % re-hydrogenation of Er(BH4)3 observed by the aforementioned report is attributed to the contamination with LiBH4 from solvent-free synthesis. The latter forms LiH during decomposition and enables the re-hydrogenation reaction, as shown in Equation (23).
ErB4(s, amorph) + 4 LiH(s) + 7.5 H2(g) Additionally, two composites were investigated, Er(BH4)3-6LiH and 3LiBH4-Er(BH4)3-3LiH. The reactants were ball milled for 1 h and in both samples, ErH2 forms after milling which suggests a reaction between Er(BH4)3 and LiH based on Equation (24) In the next step, absorption was conducted for 12 h at 340 °C and p(H2) = 100 bar. The formation of LiBH4 and ErH3 from the XPD pattern after absorption suggests the hydrogen uptake based on the Equation 23.
The hydrogen release and uptake of the 3LiBH4-Er(BH4)3-3LiH composite was studied by Heere et al. by a Sieverts' apparatus and the results are presented in Figure 27 [427]. This composite has a hydrogen storage capacity of 9 wt %, but, only 6 wt % of hydrogen can be released at a moderate temperature due to the formation of stable LiH. Figure 27 shows that 4.2 wt % of H2 is released in the first desorption. The second and third desorption cycles released 3.7 and 3.5 wt % H2, respectively, which are equivalent to 88.1% and 83.3% of the initial hydrogen release [427]. In conclusion, the hydrogen storage properties in these mixtures are attributed to the reformation of LiBH4 based on the reaction of REB4 and LiH. The pure REB are not promising candidates for hydrogen storage properties and more favourable results can be obtained by mixing rare earth borides/hydrides with alkali metal borohydrides such as LiBH4, which facilitates their rehydrogenation at milder hydrogenation conditions.

Conclusions
Research on rare earth borohydrides is further increasing due to their potential as energy materials for applications including hydrogen storage and solid-state ion conductors for battery applications. Li ionic conductivities of 10 −4 S cm −1 in LiRE(BH4)3X (RE = La, Ce, Gd; X = Cl, Br, I [63,411,413]) have been observed, which is about three magnitudes higher than conductivities for o-LiBH4 at RT [63]. Similar conductivity values have been observed for the h-LiBH4-LiX with X = Cl, Br, I [75,80]. Recently, ternary borohydrides including new garnet type materials, Li3K3Ce2(BH4)12 and Li3K3La2(BH4)12 [414], have been explored by Brighi et al. as Li ion conductors with a conductivity value of 10 −5 S cm −1 . All those σ values are in the range of their corresponding oxide garnets, such as Li5La3Ta2O12 with σ = 10 −6 S cm −1 at RT [428].
REB show further interesting properties for applications within the optical and the magnetic field. Marks et al. [429] investigated RE(BH4)2(THF)2 (RE = Eu, Yb) and showed that the RE = Eu compound exhibits a bright blue luminescence. Magnetic properties have been investigated and magnetic entropy change of 54.6 J kg −1 K −1 at a magnetic field change of 9 T has been reported for K2Gd(BH4)5 [404,417,430].
This chapter reviewed recent accomplishments in selected RE borohydrides. The synthesis routes for solvent-free, halide-free and solvent-and halide-free RE borohydrides were discussed, along with the properties of mono-and bi-metallic RE borohydrides and RE borohydrides in composite mixtures, while a special focus was given to hydrogen storage properties and reaction pathways.

Final Conclusions and Outlook
The review gives an overview of recently achieved milestones within the research of metal hydrides with respect to promising applications in all-solid-state batteries and complex metal hydrides for solid-state hydrogen storage.
Generally, under the framework of this project, the performance of metal hydrides were improved at higher temperatures thanks to the fact that mass transportation within the electrode was enhanced compared to RT cycling, which suggests that electrodes and electrolytes from metal hydrides might be initially realized for high temperature applications. The application of metal hydride anodes and hydride complexes in LiBs and particularly in all-solid-state cells were presented. The mechanism and internal behaviour of the hydride anodes in LiBs were also elucidated. Interestingly, the first full all-solid-state lithium metal hydride-lithium sulfur cell was successfully assembled and cycled with a promising performance. New metal hydride complexes were also discovered and studied with high lithium ionic conductivity and stability for application as solid electrolytes, i.e., the borohydride substituted argyrodite Li6PS5X1−y(BH4)y, and a cluster complex hydride LiBH4-Li2NH: Li5(BH4)3NH. Other potential complex hydrides, including sodium closo-borates, as solid-state electrolytes were shown, which are viable alternatives satisfying many needs. In this context a mixed-hydroborate-anion developed by Duchêne et al., Na2(B10H10)0.5(B12H12)0.5 [133], was used as proof-of-concept for a 3 V all-solid-state sodium ion battery. In the future, new cage structures will be investigated such as the recently published nido-borate (i.e., NaB11H14) by Payandeh et al. [431], which can be synthesized directly from rather cheap NaBH4 and have been employed also in combination with closo-borates. Other working ions such as Mg are also promising, as a new compound synthesized from Mg(BH4)2 and ethylenediamine has been shown to have an exceptionally high conductivity of up to 6 × 10 −5 S cm −1 at 70 °C in the solid state [432]. A follow up with different organic complexes reported details on conductivities with a particular focus on the dynamics of Mg(BH4)2-diglyme0.5 [433], while Ruyet et al. [434] revealed a high conductivity of 3 × 10 −6 S cm −1 at 100 °C for a solid-state Mg conductor based on Mg(BH4)(NH2). The achievements summarized herein put the aforementioned complex hydrides as valid competitors to other solid electrolytes, like oxides and thiophosphates, while their investigation will remain a highly interesting topic.
At the same time, hydrogen as energy carrier will be used more dominantly in the future. Japan will employ hydrogen during the Olympic Games 2020 and will show how a hydrogen economy can be implemented in all parts of life. From generating "green and renewable" hydrogen via electrolysis, through the transport of people in hydrogen powered cars and buses, to warm water production for the sustainable use of heat from fuel cells. New applications are vital for strongly required infrastructures such as hospitals, which, in the case of earthquakes and tsunamis, are not allowed to use Diesel powered generators (in Japan). Already today, back-up systems in hospitals work with fuel cells and hydrogen is stored within metal hydride alloy storage units, which supply energy and warm water, when centralized energy supply is failing. Therefore, an energy storage system as a role model for our intermittent "green and renewable" energy sources could be realized from metal hydrides. as well. All partners and students made it a very vibrant innovative training network. Last but not least, the two most important personalities responsible for the project coordination, Klaus Taube, and our secretary Sigrid Wulff have to be thanked deepfully, as this project and its success would not have been possible without them!