Recent Progress and New Perspectives on Metal Amide and Imide Systems for Solid-State Hydrogen Storage

Abstract

Hydrogen storage in the solid state represents one of the most attractive and challenging ways to supply hydrogen to a proton exchange membrane (PEM) fuel cell. Although in the last 15 years a large variety of material systems have been identified as possible candidates for storing hydrogen, further efforts have to be made in the development of systems which meet the strict targets of the Fuel Cells and Hydrogen Joint Undertaking (FCH JU) and U.S. Department of Energy (DOE). Recent projections indicate that a system possessing: (i) an ideal enthalpy in the range of 20–50 kJ/mol H₂, to use the heat produced by PEM fuel cell for providing the energy necessary for desorption; (ii) a gravimetric hydrogen density of 5 wt. % H₂ and (iii) fast sorption kinetics below 110 °C is strongly recommended. Among the known hydrogen storage materials, amide and imide-based mixtures represent the most promising class of compounds for on-board applications; however, some barriers still have to be overcome before considering this class of material mature for real applications. In this review, the most relevant progresses made in the recent years as well as the kinetic and thermodynamic properties, experimentally measured for the most promising systems, are reported and properly discussed.

1. Introduction

The constant growth in world population coupled with the rapid industrialization of the developing countries could lead to a drastic increase in the number of light-duty vehicles from now to 2050. Considering that the majority of the vehicles are currently based on the internal combustion engine (ICE) technology, the transportation sector is strongly dependent on petroleum-derived fuels which represent one of the biggest causes of carbon dioxide emission. In 2008, passenger cars were responsible for 17% of CO₂ emissions in the European Union. For these reasons, clean and environmentally-friendly energy alternative carriers to the fossil fuels, capable of stabilizing global CO₂ level while sustaining the global mobility demand are desired. In this context, the “green” hydrogen fuel cell vehicles (FCVs) represent promising alternatives to the gasoline ICE. In order to achieve an extensive commercialization of hydrogen-powered cars, some obstacles have still to be overcome. Among them, the pressing issue is for on-board hydrogen storage in safe and efficient manner and, in this frame, hydrogen storage in solid state is considered as one of the most promising solution, as verified by the numerous studies all over the world focused on finding materials meeting the requirements for on-board applications established by the European Fuel Cell & Hydrogen Technology Platform and the US Department of Energy (DoE). For example, in proton exchange membrane fuel cells (PEMFC), hydrogen is supposed to be delivered under temperature below 90 °C, which implies the use materials with ∆H_des between 20–50 kJ/mol H₂ and ∆S_des between 100–130 J/(mol H₂ K) in order to reach 1 bar equilibrium hydrogen pressure.

Recently, many systems based on solid-state hydrogen storage materials have been tested in laboratory or on mobile prototypes. However, the parameters evaluated in laboratory tests were usually not suitable, in terms of gravimetric capacity and kinetics, for most light-duty passenger vehicles.

Since its discovery in 2002, the LiNH₂-LiH system [1] and in general amides and imides, have attracted more and more attentions as promising hydrogen storage materials, becoming one of the principal candidates for onboard applications on light-duty vehicles [2]. As shown in Figure 1, during the last 15 years, more than 300 among peer-reviewed papers and books have been published in the field of solid-state hydrogen storage, concerning this class of complex hydrides characterized by N-H bonds.

From the picture in Figure 1, it is also possible to evince a constant and still actual fervent interest in amides/imides compounds for hydrogen storage purposes. This specific attention can be motivated by the fact that these systems can store sufficient amount of hydrogen in terms of gravimetric and volumetric densities, while possessing promising thermodynamic and kinetic properties with respect to the other classes of complex hydrides [2].

From a general and scientific point of view, most of the metal amide/imide systems exhibit endothermic hydrogen desorption with partial or full reversibility. Among them the Mg(NH₂)₂-2LiH composite possess high reversible hydrogen storage capacity (i.e., 5.0 wt. %) together with suitable thermodynamic properties (i.e., ΔH_des ≈ 39 kJ/mol H₂ and ΔS_des = 112 J/mol H₂ K), optimal parameters for on-board application in fuel cell vehicles (FCVs). However, for technological applications, an important barrier is the poor kinetic rate during the dehydrogenation process. This is mainly due to the high activation energy, which can be opportunely modulated by different approaches, including the addition of catalysts and the reduction the particle size. Also in this context, significant progresses have been made in the last years on the amides/imides systems, reducing the activation energy to ~80 kJ/mol, considered appealing for technological devices.

In this section, a detailed overview combined with the most recent progresses for different amides will be provided. Particular emphasis will be put on those systems with attractive features for practical applications.

2. Li-N-H System

One of the most interesting system, at present, is represented by the LiNH₂ + LiH mixture. When mixed, these substances show a reversible hydrogen storage capacity of 6.5 wt. % combined with a working temperature below 300 °C (285 °C) and appealing thermodynamic properties (ΔH_des = 45 kJ/mol) [1]. On the contrary, pure lithium amide and lithium hydride, separately, decompose at temperatures above 300 °C and 550 °C, respectively [3]. In the first desorption mechanism proposed, the driving force is represented by the release of hydrogen as a consequence of the direct reaction between the H⁺ in LiNH₂ and H⁻ in LiH to form H₂ [4]. The second desorbing mechanism involves the decomposition of LiNH₂ to ½ Li₂NH and ½ NH₃ (Equation (1)), while, in the second step, ½ LiH quickly reacts with ½ NH₃ forming ½ LiNH₂ and ½ H₂ (Equation (2)) [5]:

LiNH₂ ↔ ½ Li₂NH + ½ NH₃

(1)

½ LiH + ½ NH₃ ↔ ½ LiNH₂ + ½ H₂

(2)

Experimentally, in the temperature range of 180–400 °C this system desorbs hydrogen together with a large amount of ammonia, which should be avoided because it is the main responsible for the catalyst poisoning phenomenon in fuel cells [6,7,8]. Furthermore, it has been proved that the ammonia concentration brutally increases as the number of ab-desorption cycles increases, affecting the gravimetric capacity of the whole system [9]. Kiobayashi and coauthors reported that the starting hydrogen storage capacity of 5 wt. % in the Li-N-H system decreases up to 2 wt. % after 200 cycles tests at 300 °C. This significant reduction of the gravimetric capacity can be, in fact, explained by the loss of nitrogen due to the ammonia release during cycles [9]. For these reasons, many attempts have been made to minimize the ammonia evolution and to drastically improve the poor sorption kinetic performance.

The working temperatures are, in fact, still too high to be practical for a PEM fuel cell, although the activation energies estimated by both Kissinger (54 kJ/mol) and Arrhenius-plots (56 kJ/mol) do not result so high with respect to the observed kinetic constraint [10]. In this regard, when TiCl₃ (1 mol. %) is added to the system, almost the 80% of hydrogen is desorbed (5.5 wt. %) within 30 min at 200 °C and without ammonia release [7,11]. TiCl₃ acts as a catalyst favoring the transfer of ammonia to lithium hydride and thus improves the hydrogen desorption kinetics of the LiNH₂-LiH system. However, the activation energies calculated for the TiCl₃-doped system (95 kJ/mol, for 1 mol. % TiCl₃) resulted higher than that reported for the un-doped one [10]. In the latter, the small pre-exponential factor has the major effect, therefore it is possible to expect that in the pristine system the reaction occurs through different rate controlling steps. Homogeneous dispersion of Ti(III) species preserves the active surface of reactants and favors the reduction of LiNH₂/LiH particles sizes by ball milling [10,11]. This last cited procedure has also an important effect on desorption behavior of the Li-N-H system. As reported by Lu et al. [12], the local Li electronic environment is strongly influenced by mechanical milling, which improves the hydrogen kinetic properties of the LiNH₂-LiH composite. Mechanical milling, in fact, is an effective technique to reduce particle size, increase surface area and facilitate the formation of an intimate mixture between the two reagents, LiNH₂ and LiH. For example, Shaw and coauthors [13] proved that the hydrogen desorption temperature of Li-N-H in a molar ratio 1:1.1 can be decreased by high-energy ball milling. In this specific case, the onset dehydrogenation temperature to reach 1 wt. % of released H₂, is reduced from 300 °C in the unmilled material to 50 °C when the powders are subjected to 180 min of mechanical treatment. Furthermore, for the milled powders, no trace of ammonia gas is detected upon desorption. Activation energy seems to be also influenced by ball milling, although completely different values have been reported for the starting system by Shaw et al. with respect to a previous work of Matsumoto et al. [10].

More recently, again Shaw and coauthors [14] showed that milling at liquid nitrogen temperature (−196 °C) is effective in improving the kinetic performance of the powders during hydrogen desorption: the average diffusion rate increases by 450% by milling at liquid nitrogen temperature. Interestingly, the powders upon mechanical treatment at −196 °C show crystallite sizes of 18.3 and 21.5 nm for LiNH₂ and LiH, respectively, much higher with respect to the values estimated for the systems milled at room temperature. Therefore, the significant improvement of the desorption kinetics cannot be ascribable to the particle size reduction but, most likely, to the large amount of lattice and surface defect accumulated during milling at low temperature. Another important issue which affects the kinetic properties of this system, is represented by the hydrolysis/oxidation of LiH into LiOH [15]. This reaction sequestrates the active amount of LiH able to convert NH₃ to LiNH₂ and H₂ with a reduced hydrogen storage desorption performance. Varin et al. [15,16] studied also the hydrogen storage properties of the LiNH₂-LiH systems as a function of the milling time. They measured that the onset DSC (Differential Scanning Calorimetry) peak corresponding to the endothermic desorption step in the 1:1 mixture decreases from 325 °C to 235 °C when the sample is milled 25 h, while in the powders further milled for 100 h the DSC signal is peaked at 245 °C. This is in agreement with the fact that during the milling the Brunauer-Emmett-Teller (BET) specific surface area (SSA) increases monotonically with increase of milling time up to 25 h (16.0, 26.4, 56.0 and 59.6 m²/g for the system milled at 0, 1, 5 and 25 h). On the other hand, upon 100 h of mechanical treatment, the value of SSA decreases to 45.6 m²/g, probably due to powders agglomeration induced by the cold-welding process commonly produced in high-energy ball milling.

Another strategy adopted to improve the kinetic and thermodynamic properties of the Li-N-H system is represented by the addition of a second element [16,17] or compounds [16,18,19,20,21]. Nayebossadri [17] reported that elemental Si and Al can effectively improve the dehydrogenation rate of the Li-N-H system. In particular, the addition of Si decreases the dehydrogenation end-temperature by 200 °C compared to the pristine material. The destabilization effect can be attributed to the interaction between LiH and Si, which forming the Li₂Si phase increases the concentration of the H⁻ ions. Varin and coauthors [16], demonstrated that the system LiNH₂ + 1.2LiH + graphite (5 wt. %) can absorb reversibly 5 wt. % of H₂ at 325 °C. The inclusion of graphite in the Li-N-H matrix, guarantees an homogeneous heat transfer into the powders and, most important, seems to avoid the undesired hydrolysis/oxidation of LiH normally occurring in the ball milled mixture [16]. Ti- and B-based nitrides additives are also effective to produce a remarkable improvement in the dehydrogenation rate and desorption temperature of the Li-N-H system [19]. High concentration of TiN (35 wt. %) seems to have a significant impact on the kinetic process lowering the activation energy from 163.76 kJ/mol reported for the unmilled material to a promising value of 67.8 kJ/mol.

Structural studies on the Li-N-H system, by using synchrotron facilities as X-ray source, have been performed in order to clarify the reaction mechanism between LiNH₂ and LiH [22,23]. David and coauthors, noted that the non-stoichiometric intermediates reported as Li_1+xNH_2−x species are formed during desorption and absorption of the Li-N-H system, in accordance with the Frenkel defect model [24].

In the Li-N-H system, the poor sorption kinetics in both ab-desorption processes could be ascribable to a reduced Li⁺ ion mobility (Li⁺ ion conductivity at room temperature is 10⁻⁸ Ω⁻¹ cm⁻¹) [25]. In 2010, Orimo and coauthors [26] have successfully synthesized a new complex hydride Li₃(NH₂)₂I exhibiting lithium fast-ion conductivity of 1 × 10⁻⁵ Ω⁻¹ cm⁻¹, 1000 times higher than LiNH₂. Along this way, Anderson et al. [25,27,28] conducted a systematic study on the effects of different halides, LiX and MgX₂ (X = Cl, I, Br) on the hydrogen sorption properties of LiNH₂ and Li₂NH. The new phases, Li₄(NH₂)₃Cl, Li₃Mg_0.5(NH₂)₃Cl and Li₆Mg_0.5(NH₂)₆Br, when heated with LiH are able to release hydrogen more rapidly than pristine LiNH₂ and LiNH₂ + LiH system, with undetectable NH₃ emission. The general desorption reaction is reported in Equation (3) [25]:

Li_(1+n)(NH₂)X_n + LiH → Li_(2+n)(NH)X_n + H₂

(3)

Preliminary conductivity analysis on these new amide-halide phases proved that the systems with high ionic conductivity show the quickest hydrogen desorption kinetics during heating [25,29]. More recently, two new halide-amide phases, Li₇(NH₂)₆Cl (SG = R3. a = 9.7367 Å. b = 8.9307 Å) and Li₆Mg_0.5(NH₂)₆Cl (SG = R3. a = 9.756 Å. b = 8.9448 Å) have been synthesized and characterized as potential hydrogen storage materials [30]. Both phases, when heated with LiH (1:1), released hydrogen at lower temperatures (150 °C and 200 °C), with no emission of ammonia gas and faster kinetics if compared with the pristine Li-N-H system (220 °C). In addition, rehydrogenation of the dehydrided compounds, in the presence of LiH, is possible under mild conditions of 90 bar H₂ at 300 °C.

Furthermore, as highlighted by Chen et al. [31], another important aspect for improving the hydrogen sorption kinetics is related with the sample morphology and desorption conditions. For example, in the Li₂NH₂Br phase, LiNH₂ is confined in the Br cage resulting in less mobility of Li⁺ ion. For this reason, the desorption of the Li₂NH₂Br + LiH system follows the NH₃ mediated mechanism although the two components have been intensively ball milled. In other cases, when the LiNH₂-type and LiH phases get an intimate mixture, the migrations of Li⁺ and H⁺ plays a crucial role. Leng et al. [32] showed significant improvements in the desorption properties of the Li-N-H system when mixed with different amounts of MgCl₂. From this study emerges that the hydrogen desorption temperatures do not decrease linearly with the increase of MgCl₂ amount. Interestingly, for small amount of MgCl₂ (1 mol. %), the main hydrogen desorption temperature is reduced to 267 °C, similar to the TiCl₃-doped system (265 °C). Increasing the MgCl₂ amount up to 4 mol. %, the formation of a solid solution is observed without any improvement in the desorption temperature, while for content of MgCl₂ higher than 25 mol. %, a reaction with LiNH₂ takes place forming Mg(NH₂)₂. Recent works published by Gennari et al. [33,34] introduced AlCl₃ as additive and interesting benefits have been reported in terms of ab/desorption rates, cycles stability and hydrogen gravimetric capacity in the Li-N-H system. In a first work [33], it was reported that, for low amount of AlCl₃, Al³⁺ is incorporated into the LiNH₂ structure which can reversible store 4–5 wt. % of H₂ at 275 °C. Furthermore, the ammonia gas is completely suppressed during desorption reaction. The system was investigated also for increasing amount of AlCl₃ (0.03, 0.08 and 0.13 mol) [34]. Interestingly, the dehydrogenation rates of the composite milled with 0.08 mol and 0.13 mol of AlCl₃ are three-fold and six-fold faster, respectively, than the pristine system. In particular, upon 5 h of intensive ball-milling of the starting materials LiNH₂, LiH and (0.13 mol) AlCl₃, a disordered Face Centered Cubic (FCC) solid solution, not yet reported in previous work, is observed. After heating at 300 °C under hydrogen (7 bar), new trigonal and cubic amide-halide (chloride) phases, isostructural with Li₄(NH₂)₃Cl [25], are observed. As reported for the previously cited examples, also in the case of the AlCl₃-doped Li-N-H system, the improvement of the hydrogen desorption kinetics is accompanied by a significant reduction of the ammonia release. In this specific case, the preparation of the new cubic and trigonal Li-Al-N-H-Cl phases consisted in ball milling and thermal treatment performed under different atmospheres (argon and hydrogen) of LiNH₂ and AlCl₃ (0.11 mol) [35]. The whole optimized process is resumed in Figure 2. The desorption reaction of the amide-chloride phases depends on the addition of LiH which completely suppresses the ammonia release and avoids the multi-steps decomposition pathways observed in the post milled LiNH₂-AlCl₃ composite. The desorbed product, an imide FCC phase (a = 5.172 Å), is re-hydrogenated under moderate temperatures and represents a reversible system. However, despite the progress made to optimize the whole process, the crystal structure of the two active amide-chloride phases, remains unsolved.

The performance, in terms of reversibility and maximum desorption temperatures, of the Li-Al-N-H-Cl system results comparable and in some case better with respect to the most promising doped LiNH₂-LiH systems reported in the current literature [18,32,36,37]. For example, after the fourth ab/des cycle the AlCl₃-doped composite is able to desorb roughly 92% of the theoretical amount of H₂, significantly higher than the 67% reported for the potassium fluoride KF-doped mixture at the same working temperature (300 °C) [33,36]. A comparison of the maximum desorption temperatures for different doped systems is reported in Figure 3. As emerged, the dopants show a positive effect reducing the desorption temperatures in the LiNH₂-LiH system and favoring the release of hydrogen under more moderate conditions. Among them, AlCl₃ results to be the most efficient, decreasing the working temperature by 70 °C with respect to the pristine composite.

Gennari et al. [38] also tested the contribution of MgH₂, CaH₂ and TiH₂ in enhancing the kinetic properties of the LiNH₂-1.6LiH system. Addition of TiH₂ does not show significant improvement, while for the CaH₂- and MgH₂-based systems positive effects have been reported. For example, the LiNH₂-1.6LiH-0.2CaH₂ systems shows a dehydrogenation rate three time faster at 300 °C than the un-doped materials with about 3.8 wt. % of H₂ stored. Further improvement in terms of working temperature and kinetic rate have been recently reported by Lin and coauthors, by using CeF₄ as efficient catalyst for LiNH₂-LiH. This system, when mixed with 10 wt. % of CeF₄ can desorb most of its hydrogen content (~5 wt. %) under milder condition (220 °C) compared with the un-doped mixture. CeF₄ seems to act as a catalyst reducing the activation energy to 114 kJ/mol and suppressing the formation of ammonia gas [39].

The operating temperature of Li-N-H can be reduced by modifying the thermodynamic properties of the system, and, in this context, the substitution of LiH with other metal hydrides represents one of the strategy widely adopted in the recent past [40,41,42,43,44,45,46]. Among the most relevant results, it is worth to highlight that the combination of LiNH₂ and LiBH₄ (2:1) lead to the desorption of a significant amount of hydrogen, around 10 wt. %, at 350 °C. The decomposition of this binary system passes through the formation of a ternary amide, Li₃BN₂H₈, which was the object of an intensive study due to the poor reversibility and kinetics constrains [41,42]. For example, Liu and coauthors reported that the addition of Co(OH)₂ to the 2LiNH₂-LiBH₄ mixture significantly improved its hydrogen storage properties. At 200 °C the doped system was able to desorb 9.1 wt. % of hydrogen in only 15 min, while a partial reversibility (1.1 wt. %) was achieved at 350 °C [45]. The activation energy of the doped system was estimated to be 97 kJ/mol, 25% lower with respect to the pristine composite (E_a = 129 kJ/mol). Reversibility under milder conditions (200 °C) was achieved when CaH₂ is added to LiNH₂. However, the recharged amount (2.7 wt. % H₂) and the poor kinetic rate (1.1 × 10⁻⁴ wt. %/s) represent a strong limitation for this binary system [40].

These new and important results have to be considered in order to improve the hydrogen storage properties of the Li-N-H system, at the moment insufficient for any practical application in the field of the on-board technology. In fact, the thermodynamic data reported in literature show that the equilibrium temperature at H₂ pressure of 1 bar is higher than 200 °C for all the systems analyzed up to now. Such high equilibrium temperature sets this system far from the DoE targets for automotive purposes. For these reasons, further efforts have to be addressed to exploit this system in transportation sector and, in this way, understand the local structure, ion mobility and kinetic constrains to dehydrogenation of the intermediate species Li_1+xNH_2−x represent the next challenges in this field [47].

3. Li-Mg-N-H System

With respect to the Li-N-H system discussed in the previous section, thermodynamic destabilization can be further achieved by partially replacing Li with Mg, with a significant reduction of the desorption enthalpy down to ca. 40 kJ/mol. The reaction between LiNH₂ and MgH₂ (or Mg(NH₂)₂ and LiH) for hydrogen storage was reported for the first time in 2004 by five different groups [48,49,50,51,52]. They investigated several factors in the reaction including different ratios between amide and hydride and ab/desorption conditions. Xiong et al. [52] heated lithium amide and magnesium hydride in a 2:1 molar ratio up to 350 °C, and, similar to the Li-N-H system, only hydrogen, without any ammonia, is desorbed. It is found that the hydrogen desorption temperatures are lower than in the LiNH₂ + LiH system. From ex-situ powder X-ray Diffraction (XRD) performed on the sample upon heating, new peaks are observed. This new sequence of Bragg reflections does not match with any of the previously identified in the Li-N-H compounds, and it is finally associated with the Li₂MgN₂H₂ phase. This ternary imide crystallizes with a cubic lattice (a = 10.035 Å) and its formation is summarized in the following reaction:

2LiNH₂ + MgH₂ → Li₂Mg(NH)₂ + 2H₂

(4)

The solid product of Equation (4) can be re-hydrogenated under 90 bar of H₂ pressure at 180 °C. Upon cycling of the ternary imide, the XRD pattern showed that the products are represented by Mg(NH₂)₂ and LiH rather than the original starting materials LiNH₂ and MgH₂ (Equation (5)):

2LiNH₂ + MgH₂ → Li₂Mg(NH)₂ + 2H₂ ↔ Mg(NH₂)₂ + 2LiH

(5)

At the same time, Luo [51] also published further work where the absorption and desorption steps of the 2LiNH₂ + MgH₂ and LiNH₂ + LiH systems are compared. The Li-Mg-N-H system starts to desorb hydrogen at lower temperature (100 °C) with respect to the Li-N-H system. Important differences also emerge in the P-C-T curves: the Li-Mg-N-H composite shows a hydrogen equilibrium pressure of 50 bar at 220 °C against the 1 bar of H₂ at 280 °C reported for the Li-N-H mixture. Furthermore, the Li-Mg-N-H sample can be cycled 9 times with no degradation in desorption capacity. Chen et al. [53] investigated in detail the first 4 cycles of the ball-milled 2LiNH₂ + MgH₂ system. In the first dehydrogenation cycle, the desorption rate is significantly slower than the subsequent three runs, ascribable to the fact that, in the first one, there is an interaction between the LiNH₂ and MgH₂. The sorption mechanism, summarized in Figure 4, has been further investigated by Sickafoose et al. [54] In the first step of the isotherm, one hydrogen atom is inserted into the Li₂MgN₂H₂ phase, forming Li₂MgN₂H₃. The other hydrogen atoms react with Li₂MgN₂H₃ in a second step to produce the new species LiH and Mg(NH₂)₂.

Luo et al. [55,56] determined the NH₃ emission in the desorption step of the 2LiNH₂ + MgH₂ system. The results indicate that the NH₃ concentration is around 180 ppm at 180 °C and 720 ppm at 240 °C, restricting the cycling and then the applicability. Markmaitree et al. [57] compared the reaction kinetics and the NH₃ emission of the 2LiNH₂ + MgH₂ and LiNH₂ + LiH systems. The 2LiNH₂ + MgH₂ mixture seems to possess a higher activation energy and NH₃ emission than the LiNH₂ + LiH mixture. In addition, the samples ball milled for longer time present reduced amount of NH₃ released during the desorption step. This aspect can be explained taking into account the smaller particle size and larger surface area of the produced MgH₂, which favor the interaction with NH₃. When the system is heated up to 210 °C for 5 h, MgNH, Li₂Mg(NH)₂ phases and residual LiNH₂ and MgH₂ are present in final mixture. On the other hand, after heating for 10 h, LiNH₂ and MgNH are completely disappeared, while a small amount of MgH₂ is detected. The main phase is represented by the ternary imide Li₂Mg(NH)₂ which suggests that the reaction mechanism involves the decomposition of lithium amide to imide and NH₃ followed by the reaction of MgH₂ with NH₃ as reported in Equation (6):

MgH₂ + 2NH₃ → Mg(NH₂)₂ + 2H₂.

(6)

The reaction between MgH₂ and NH₃ takes place with a low rate [58]. The magnesium amide forms and then decomposes to MgNH at a reasonable rate from 250 °C with the release of NH₃ [57]. Further ammonia is produced in the last step, then reacts with residual MgH₂ and the reaction cycle continues. The products from this cycle are MgNH and Li₂NH which could react to form Li₂Mg(NH)₂ as suggested in Equation (7):

MgNH + Li₂NH → Li₂Mg(NH)₂

(7)

However, XRD analysis does not support evidence of Mg(NH₂)₂ in any step, probably because the reaction reported in Equation (6) proceeds with a slow rate. Furthermore, Mg(NH₂)₂ could not be observed by FTIR because of its amorphous nature upon ball milling [59].

Rijssenbeek et al. [60] investigated the hydrogen sorption path of the 2LiNH₂ + MgH₂ mixture by in situ X-ray diffraction and neutron diffraction. These studies confirm that the dehydrogenated product is a ternary lithium-magnesium imide, Li₂Mg(NH)₂, which undergoes two structural transitions from an orthorhombic structure to a primitive cubic structure (β-Li₂Mg(NH)₂) at intermediate temperature (350 °C) followed by a face-centered cubic crystal structure (γ-Li₂Mg(NH)₂) at high temperature (500 °C). Complete hydrogen absorption is observed for α-Li₂Mg(NH)₂ with the formation of Mg(NH₂)₂ and LiH. During the dehydrogenation phases of the cycling process, NH₃ is still detected, although in lower amount than the first desorption. It is important to highlight that using Mg(NH₂)₂ and LiH as starting materials, the three phases of Li₂Mg(NH)₂ can be formed, although α-Li₂Mg(NH)₂ is difficult to detect and the β-Li₂Mg(NH)₂ phase appears at a lower temperature.

Concerning the experimental thermodynamic properties, Yang et al. [61] estimated a value of ΔH = 41.6 kJ/mol, close to that expected by theoretical calculations [50]. However, although the temperature of hydrogen release is approaching 90 °C, the reaction is kinetically limited. Therefore, to improve the kinetics at temperatures lower than 200 °C, the reaction needed to be catalyzed with suitable dopants or additives. During the last years, many attempts have been made to improve the sorption kinetics of the 2LiNH₂ + MgH₂ system and, in this context, a large number of papers have been published [62,63,64,65,66,67,68,69,70,71,72,73,74]. Luo et al. reported faster absorption kinetics by doping with less than 4 mol. % of KH. In these experiments, the difference in the absorption rates of the doped and undoped systems is particularly noticeable at 180 °C and 115 bar, while it becomes less pronounced at higher temperature and pressure (i.e., 200 °C and 155 bar, respectively) [73]. The studies conducted separately by Yang [75] and Sudik [76] proved that the addition in stoichiometric amount of LiBH₄ has a significant impact on the hydrogen storage properties of the Li-Mg-N-H system. Ulmer et al. [71] verified that co-doping with LiBH₄ and KH can significantly improve the de/absorption kinetics of the 2LiNH₂ + MgH₂ system. Interestingly, co-doping with LiBH₄ and ZrCoH₃ is even more effective: a mixture with composition 2LiNH₂ + 1.1MgH₂-(0.1LiBH₄ + 3 mass% ZrCoH₃) desorbs 3.2 wt. % under 1 bar H₂ at 180 °C within 100 min and can be re-hydrogenated in only 7 min. The study also demonstrates that this system can be processed efficiently in batches up to 200 g (vibrational mill) and utilized in a tank prototype, showing similar performance to the material processed at lab-scale (planetary mill) [71].

More recently, Goudy et al. [65] showed that RbH added in catalytic amount is significantly better than KH for enhancing the hydrogen sorption properties of the 2LiNH₂ + MgH₂ system, while the desorption temperature remains almost the same for both dopants. The kinetic findings prove that the RbH-doped system releases hydrogen two and 60 times faster than KH-doped and undoped systems, respectively. In another study, the same authors determine that the desorption rates of the K-, Rb- and Cs-doped 2LiNH₂ + MgH₂ system increase in the order: undoped < CsH < KH < RbH. Modeling of the kinetic data using a shrinking-core model suggests that the desorption process is diffusion controlled in the first stage of the reaction [66].

To better increase the kinetic reaction and suppress NH₃ emission in efficient manner, Demirocak et al. [64] doped the Li-Mg-N-H mixture with single-walled carbon nanotubes SWCNT/20%Ru catalyst, obtaining an effective elimination of the ammonia gas. Anderson et al. [63], demonstrated that addition of halides (CaCl₂ and CaBr₂) on the Li-Mg-N-H system can improve the desorption kinetics. In particular, these halides seem to favor the metathesis reaction already during the milling. The lowest dehydrogenation onset temperatures (<100 °C) are achieved in the 2LiNH₂-MgH₂-0.15CaX (X = Cl₂ and Br₂) system. In particular the CaBr₂-doped system presents the lowest desorption onset and peak temperatures (80 °C and 135 °C, respectively) and an activation energy of 78.8 kJ/mol. Combined effect of ball milling and molar ratio LiNH₂:MgH₂ on the sorption and structural process were studied by Varin et al. [77,78]. The LiNH₂ + nMgH₂ (n = 0.55, 0.6 and 0.7) system is partially converted, by a metathesis reaction promoted by high-energy ball milling, into Mg(NH₂)₂ and LiH. The apparent activation energy of 71.7 kJ/mol is observed for the LiNH₂ + 0.7MgH₂ system milled for 25 h, while a further decrease down to 65.0 kJ/mol is obtained when catalyzed by n-Ni composite. The Van’t Hoff analysis shows that the equilibrium temperature at 1 bar H₂ is 70.1 °C for this mixture. Additionally, the LiNH₂ + 0.7MgH₂ mixture is fully reversible and desorbs/absorbs 3.6 wt. % H₂ at 175 °C.

A mixture of Mg(NH₂)₂ and LiH is regarded as an equivalent of the 2LiNH₂ + MgH₂ system because of the metathesis reaction between the two hydrides-amides pairs:

2LiNH₂ + MgH₂ → Mg(NH₂)₂ + 2LiH.

(8)

The Mg(NH₂)₂ + 2LiH system presents some advantages compared to the 2 LiNH₂ + MgH₂ composite. For example, as well explained in the previous section, after a cycle of hydrogen desorption/absorption, the 2LiNH₂ + MgH₂ mixture becomes a Mg(NH₂)₂ + 2LiH system. Considering that this step is accompanied by release of ammonia and kinetic constraints, starting from the Mg(NH₂)₂ + 2LiH system is more advisable. One of the first studies on Mg(NH₂)₂ + 2LiH was reported by Xiong et al. [79] in 2005. The reactants, ball-milled together for 48 h, upon dehydrogenation at 250 °C release a considerable amount of hydrogen (5.0 wt. %) with the formation of the Li₂Mg(NH)₂ phase. The system presents a very appealing reaction enthalpy of 38.9 kJ/mol H₂, but, on the other hand, the activation energy estimated is relatively high (E_a = 102 kJ/mol). Xiong et al. concluded that to decrease the activation energy of this system it is necessary to favor the interaction between H⁺ and H⁻ of the amide and hydride, respectively, or to weaken the Li-N and N-H bonds. Xiong et al. [80] also investigated the effect of molar ratios of Mg(NH₂)₂ and LiH, on the desorption properties. The 1:2 mixture presents remarkable differences and better desorption properties with respect to the 1:1 or 1:3 ratios. Furthermore, lower content of LiH favors the desorption of NH₃. These results were also confirmed by other works [81,82,83,84]. For example, Hu et al. [82], highlighted that for lower content of LiH, severe ammonia release should be expected, while an excess of LiH affects, in negative, the gravimetric capacity of the system. Liu et al. [85] investigated the effect of sodium compounds on the Li-Mg-N-H system. The hydrogen desorption kinetics of the Na-containing systems are significantly improved and NH₃ emission reduced. The activation energies of the three samples, Mg(NH₂)₂ + 2LiH, Mg(NH₂)₂ + 1.6LiH + 0.4NaH and 0.8Mg(NH₂)₂ + 0.4NaNH₂ + 2LiH, are 105.5, 97.7, and 92.5 kJ/mol, respectively. In spite of lower activation energies, the sodium-containing systems possess, as expected, decreased hydrogen gravimetric capacities with respect to the un-doped material. Sudik et al. [86], instead, investigated the effect of Li₂Mg(NH)₂ (5 wt. %, 10 wt. % and 15 wt. %) on the sorption properties of the Mg(NH₂)₂ + 2LiH systems. The system milled with 10 wt. % Li₂Mg(NH)₂ desorbs hydrogen at a 40 °C lower temperature. The desorption curves reveal a two-step hydrogen release as for the un-doped material. Additionally, the activation energy of the Li₂Mg(NH)₂-coontaining sample is lowered by 13% from 88.0 kJ/mol to 76.2 kJ/mol. Liu et al. [87] synthesized Li₂MgN₂H₂ by annealing a mixture of Mg(NH₂)₂-2LiNH₂ and investigated its size-dependent hydrogen storage performance. Markedly enhanced kinetics of hydrogen absorption/desorption are achieved with a reduction in the particle and grain size. Janot et al. [88] compared the hydrogen storage performance of the two reactions Mg(NH₂)₂ + 2LiH and 2LiNH₂ + MgH₂. For the 1:2 MgH₂-LiNH₂ system, it evolved ammonia under dynamic vacuum while the ball-milled 1:2 Mg(NH₂)₂-LiH mixture desorbs 5.0 wt. % of hydrogen in 25 min at 220 °C and it can be cycled at 200 °C with a total capacity around 4.8 wt. %. Further efforts have been devoted to the kinetic improvement of the Mg(NH₂)₂ + 2LiH system by using catalysts or additives [89,90,91,92,93,94]. Wang et al. [91] proposed graphite-supported Ru nanoparticles as efficient additive in the Mg(NH₂)₂ + 2LiH system: after mixing, a considerable enhancement in the ab/dehydrogenation kinetics for more than 10 cycles have been observed. Srivastva et al. [92] proved that the addition of vanadium-based catalysts can catalyze efficiently the desorption reaction of the Li-Mg-N-H system. In particular, the sample mixed with VCl₃ presents a hydrogen desorption temperature lower than the pristine material. This mixture starts to release hydrogen at 50 °C ad its desorption rate is enhanced up to 38%. Similar performance has been obtained in the Mg(NH₂) + 2LH catalyzed by CaH₂. The desorption starts at temperature of 78 °C and the activation energy estimated for the first dehydrogenation step decreases from a value of 133.8 ± 4.1 kJ/mol in the pristine material to 105.1 ± 3.2 kJ/mol when CaH₂ is added [95].

Significant improvements, in terms of sorption properties and ammonia suppression were achieved by Chen et al. in 2008, either adding LiBH₄ or by partially replacing LiH with KH [93,96]. The composite Mg(NH₂)₂ + 2LiH + 0.1LiBH₄ can desorb and fully re-absorb 5 wt. % of H₂ at temperatures of 140 °C and 100 °C respectively, with a rate three times faster than the undoped material. The system Mg(NH₂)₂ + 1.9 LiH + 0.1 KH starts to release hydrogen at temperatures as low as 80 °C. Moreover, complete absorption and desorption can be carried out in equilibrium conditions (Pressure-Composition-Isotherm (PCI) measurements) at 107 °C, revealing a H₂ pressure of 2.5 bar. The hydrogenation can be performed to a considerable extent (ca. 75%) in just 12 min with a hydrogen pressure of 30 bar and a temperature equal to 143 °C: a real breakthrough, considering that for the undoped system 20 h are necessary in order to absorb the same amount of hydrogen. Moreover, differently from the undoped sample, in the KH-containing system the ammonia released was almost undetectable below 200 °C. Several studies tried to identify and characterize the bimetallic intermediates that can be formed from the interaction of KH and Mg(NH₂)₂ [97,98,99,100,101]. Interesting results, in terms of hydrogen sorption kinetics, can be achieved using different K-sources, such as KF and KOH, however, the active species seems to be always KH [102,103,104,105,106]. Comparable performance was obtained also with the use of RbF: this was mainly ascribable to the similar structures of KMgNH₂NH and RbMgNH₂NH formed during the dehydrogenation step [107,108]. Interestingly, the addition of both KH and RbH is a simple but effective strategy to further enhance the sorption rate and reversibility of the system [89]. The nominal Mg(NH₂)₂ + 2LiH + 0.04KH + 0.04RbH system is able to store 5.2 wt. % of hydrogen reversibly at 130 °C (dehydrogenation)–120 °C (hydrogenation) with a very fast kinetic (43 times faster than pristine material). Furthermore, a noticeable cycling performance has been reported: around of 93% of hydrogen storage capacity remains after more than 50 cycles. Also worth mentioning are the results obtained by using CsH in place of RbH, as single dopant [109] or as co-dopant (with KH) [110] in the Mg(NH₂)₂-2LiH system.

In a very recent work, Guo et al. synthesized uniform Li₂Mg(NH)₂ nanoparticles embedded into carbon nanofibers, which can ab/desorb efficiently (50 cycles) 2 wt. % of hydrogen at 130 °C. This high performing system exhibits an enthalpy of dehydrogenation of 35.7 kJ/mol H₂, slightly lower if compared with that observed in the bulk system (44.1 kJ/mol H₂). This thermodynamic destabilization was ascribed to both the interaction between the Li₂Mg(NH)₂ compound and the carbon matrix and the reduction of the particle size to nanometer scale [111].

4. Li-Mg-N-H-Borohydride Systems

Recently, the Li-Mg-N-H-borohydride system has attracted increasing interest. In fact, the use of borohydrides as additives alters not only the kinetics but also the thermodynamics of Li-Mg-N-H system. Inspired by the richness of the chemistry observed in the binary systems 2LiBH₄ + MgH₂ [112], 2LiNH₂ + LiBH₄ [41,113] and 2LiNH₂ + MgH₂ [51,52], a ternary mixture of 2LiNH₂ + MgH₂ + LiBH₄ was investigated by Yang et al. in 2007 [75]. In their following works [114,115], an optimum composition of this ternary system was found to be 6LiNH₂ + 3MgH₂ + LiBH₄ by combinatorial synthesis and screening techniques. A self-catalyzed reaction mechanism for this ternary composition was proposed: (1) during milling LiNH₂ reacts with LiBH₄ to form Li₄BN₃H₁₀ (Equation (9)); (2) Li₄BN₃H₁₀ interacts with MgH₂ producing Li₂Mg(NH)₂ (Equation (10)); (3) the formed Li₂Mg(NH)₂ functions as seeds for the reaction reported in Equation (5) leading to a general improvement of the hydrogen storage properties [116]. The beneficial effect of “seeding” in improving the dehydrogenation kinetics of Equation (5) has been confirmed by the same group [86]. The drawback of this ternary system is that the reversible H₂ capacity is a little lower than that of the pristine system. In fact, the maximum reversible H₂ content for the optimum composition is only 3.5 wt. %:

3LiNH₂ + LiBH₄ → Li₄(BH₄)(NH₂)₃

(9)

2Li₄(BH₄)(NH₂)₃ + 3MgH₂ → 3Li₂Mg(NH)₂ + 2LiBH₄ + 6H₂

(10)

The conversion of LiNH₂ and MgH₂ into Mg(NH₂)₂ and LiH may occur during the dehydrogenation process (Equation (5)). Hu et al. [117], compared the dehydrogenation performance, thermal effects and chemical changes of 1Mg(NH₂)₂ + 2LiH + 1LiBH₄ and 2LiNH₂ + MgH₂ + LiBH₄. This investigation showed that MgH₂ and LiNH₂ converts to Mg(NH₂)₂ and LiH at 120 °C in the presence of LiBH₄. Similar results can also be found by changing the ratios of MgH₂ in the 2LiNH₂ + LiBH₄ + x(MgH₂) system. During ball milling, MgH₂ reacts with LiNH₂ to form Mg(NH₂)₂ and LiH together with Li₄(BH₄)(NH₂)₃, as reported by Equation (9) [76].

Following these works, Hu et al. [96], introduced LiBH₄ as an additive for Mg(NH₂)₂ + 2LiH system. Mg(NH₂)₂ + 2LiH + 0.1LiBH₄ shifts its hydrogen desorption onset and maximum peak to lower temperatures, and more than 5 wt. % of H₂ can be reversibly released and stored at 140 and 100 °C, respectively. Furthermore, its theoretical dehydrogenation temperature is 70 °C at 1 bar of hydrogen pressure, which is about 20 °C lower than the pristine system. There are three main reasons for the achieved improvement. Firstly, LiBH₄ facilitates the recrystallization of Mg(NH₂)₂ after milling, the as formed N-atom matrix in the crystalline of Mg(NH₂)₂ shares a similar sublattice structure with the desorption product Li₂Mg(NH)₂, similar to the “seeding” effect reported by Yang et al. [86] thus leading to an improvement of dehydrogenation kinetics [118]. Secondly, Li₄(BH₄)(NH₂)₃ possess a high Li+ ion conductivity and low melting temperature, which increase the transportation of mass and enlarge the contact between reactants. In addition, being this phase cyclically formed and consumed during the dehydrogenation step, it participates in the capture and release of LiNH₂, one dehydrogenation product of Mg(NH₂)₂ + 2LiH. Thirdly, LiBH₄ alters not only the kinetics but also the thermodynamics of Mg(NH₂)₂ + 2LiH [96,118]. Varying the molar ratio of Mg(NH₂)₂, LiH and LiBH₄, the optimum molar ratio of 6:9:1 was found [119]. The stabilization of LiNH₂, the product of 6Mg(NH₂)₂ + 9LiH, by adding LiBH₄ (Equation (11)) leads to a theoretical dehydrogenation pressure of 1 bar at ca. 64 °C in the 6Mg(NH₂)₂ + 9LiH + LiBH₄ system:

6Mg(NH₂)₂ + 9LiH + LiBH₄ ↔ 3Li₂Mg₂(NH)₃ + Li₄(BH₄)(NH₂)₃ + 9H₂

(11)

To understand the reason for the high dehydrogenation plateau (PCI curves) at the starting stage of dehydrogenation of 6Mg(NH₂)₂ + 9LiH + LiBH₄, a systematic investigation of the influence of the reactants ratios was carried out [120,121,122]. Increasing the ratio of LiBH₄ in the system 6Mg(NH₂)₂ + 9LiH + xLiBH₄ to 12, the dehydrogenation enthalpy decreases to 24 kJ/mol H₂ (Figure 5a), allowing to achieve the theoretical equilibrium hydrogen pressure of 1 bar below room temperature. The reason behind this interesting property is the role played by the addition of high amount of LiBH₄ which strongly influences the dehydrogenation pathway. PCI curves of 6Mg(NH₂)₂ + 9LiH + 12LiBH₄ show the existence of two plateaus, one at high pressure and another having a sloppy appearance (Figure 5b). The overall dehydrogenation reaction for the PCI process has been proposed to be as follows: firstly, dehydrogenation occurs from the reaction between Mg(NH₂)₂ and LiH, forming MgNH, LiNH₂ and H₂; then the as-formed LiNH₂ surrounded by LiBH₄ produces a mixture of LiNH₂-2LiBH₄ (Equation (12)). This is the reason for the presence of the high-pressure plateau in the PCI curves (Figure 5b). Secondly, LiNH₂ reacts with LiH to form H₂ and Li₂NH with the help of LiBH₄, leading to the H₂ desorption in the sloppy pressure region (Equation (13)) [122]:

2Mg(NH₂)₂ (s) + 2LiH (s) + 4LiBH₄ (s) ↔ 2MgNH (s) + 2[LiNH₂ + 2LiBH₄] (l) + 2H₂ (g)

(12)

2MgNH (s) + 2[LiNH₂ + 2LiBH₄] (l) + LiH (s) ↔ 2MgNH (s) + 2[(Li₂NH + 2LiBH₄)-(LiNH₂ + 2LiBH₄)] (l) + H₂ (g)

(13)

The volumetric characterization of this system shows a significant release of hydrogen around 98 °C, close to the operating temperature of PEM fuel cell. However, very fast kinetics, suitable for practical applications, are possible only at 143 °C.

Similarly to the reaction between LiNH₂ and LiBH₄, Li₂NH can also react with LiBH₄ forming several different compounds [123,124,125]. Interestingly, these compounds have extremely appealing Li⁺ ion conductivity. The ionic conductivity of such compounds is a key factor in improving the kinetic properties of the Mg(NH₂)₂ + LiH system [126,127]. Examples of such compounds in the Mg(NH₂)₂ + LiH + xLiBH₄ system are Li₂(BH₄)(NH₂), Li₃(BH₄)(NH₂)₂, and Li₄(BH₄)(NH₂)₃. The Li⁺ conductivity of Li₂(BH₄)(NH₂) is around 2 × 10⁻⁴ S cm⁻¹ at room temperature (RT). Upon heating to 105 °C, its conductivity reaches to 6 × 10⁻² S cm⁻¹. Li₄(BH₄)(NH₂)₃ also possesses a high Li⁺ conductivity of 2 × 10⁻⁴ S cm⁻¹ and 1 × 10⁻³ S cm⁻¹ at RT and 100 °C, respectively [128].

Recently, various transition metal salts [129,130,131,132,133,134,135] were tested as additives to further improve the hydrogen properties of the LiBH₄-doped Li-Mg-N-H system. For example, Li₃N and YCl₃ have been used as co-additives for 6Mg(NH₂)₂ + 9LiH + xLiBH₄. The co-additives improve the hydrogen storage capacity and the de/absorption kinetics. 4.2 wt. % of hydrogen was charged in only 8 min under isothermal conditions at 180 °C and 85 bar of hydrogen pressure. Furthermore, by increasing the H₂ pressure above 185 bars, the absorption process can be performed under milder temperature conditions below 90 °C. These hydrogenation capacity and absorption kinetics can be maintained constant for more than 10 cycles [135]. The reaction mechanism depicted in Figure 6, proposed for the hydrogenation step, reveals that hydrogen is firstly dissociated by the action of the additive and then interacts with the ternary imide Li₂Mg(NH)₂ forming the binary system Mg(NH₂)₂-LiH. The diffusion of hydrogen and the mass transfer is ruled by the liquid matrix Li₄(BH₄)(NH₂)₃.

The effect of some nanoadditives such as Ni, Co, Fe, Cu, and Mn have been investigated by Srinivasan et al. [129] for the 2LiNH₂ + 1.1MgH₂ + 0.1LiBH₄ system, previously introduced in the Section 3. As the result of this investigation, it was found that some of these additives (Co, Ni) reduce the de-hydrogenation temperature of more than 75 °C; some others (Cu, Fe) increase significantly the de-hydrogenation rate.

However, one of the most efficient additives for the 2LiNH₂+1.1MgH₂+0.1LiBH₄ system still remains ZrCoH₃ [130]. The addition of 3 wt. % of ZrCoH₃ to this system leads to absorb 5.3 wt. % of H₂ under 70 bar of H₂ pressure in 10 min and desorb 3.8 wt. % of H₂ under 1 bar H₂ pressure in 60 min at 150 °C. X-ray absorption fine structure (XAFS) study suggests that the chemical environment of both Zr and Co atoms and the crystal parameters of ZrCoH₃ remain unchanged during de/re-hydrogenation processes. These results point out to the fact that ZrCoH₃ works as a catalyst in this ternary composition [131]. 2LiNH₂ + 1.1MgH₂ + 0.1LiBH₄-3 wt. %ZrCoH₃ has been considered as one of the most promising materials for possible real scale applications owing to its excellent hydrogen storage properties [136,137,138]. A prototype tank to feed a 1 kW HT-PEM stack for an Auxiliary Power Unit (APU), was built using this ternary system as storage material. The gravimetric capacity of this tank was roughly 2.1 wt. %, fully reversible at the temperature range of 160–180 °C [138].

Other borohydrides such as Mg(BH₄)₂ [139,140,141] and Ca(BH₄)₂ [142] were also added to the system, and similar effects to that of LiBH₄ were observed. For example, the Mg(NH₂)₂ + 2LiH + 0.1Mg(BH₄)₂ sample can reversibly de/re-hydrogenate ~4.5 wt. % of H₂ at 140 and 120 °C. Mg(BH₄)₂ and LiH convert to LiBH₄ and MgH₂ during ball milling, subsequently, the as formed MgH₂ reacts with Mg(NH₂)₂ to form MgNH. Upon heating, LiBH₄ support the recrystallization of Mg(NH₂)₂ and reacts with LiNH₂ to form Li₄(BH₄)(NH₂)₃. Meanwhile, MgNH may act as a seeding agent for Li₂Mg(NH)₂ leading to a fast dehydrogenation of Mg(NH₂)₂ and LiH as the consequence of the structural similarity. Thus, the as formed LiBH₄ and MgNH synergistically influence the hydrogen storage performances of the Mg(NH₂)₂ + 2LiH system [140]. Recently, Li₄(BH₄)(NH₂)₃ has also been tested as an additive to alter the hydrogen storage properties of the Li-Mg-N-H system due to its low melting point and fast Li⁺ ion conductivity [143,144,145,146].

5. Ternary Transition Metal Amide-Hydride System

Various amide-hydride systems with different kinetics and thermodynamics have been developed by the combination and/or replacement of several amides and/or hydrides. Meanwhile, doping with transition metal/salt such as Sc, Ti, V, Ta, Ni et al. has demonstrated to be an effective way to improve the hydrogen storage properties of amide-hydride system. The hydrogen sorption properties of LiNH₂ + LiH can be significantly improved by doping it with nanosized Ti-based additives. In fact, this system can desorb roughly 5.5 wt. % H₂ within a temperature range of 150 to 250 °C.

Considering the advantages achievable by following both ways of improving the hydrogen storage performance, recently we introduced several ternary alkali metal transition metal amides as hydrogen storage materials after mixing them with LiH. For some of the investigated systems, outstanding absorption kinetics have been observed. As an example, the product of the decomposition of K₂[Zn(NH₂)₄] + 8LiH can be fully hydrogenated within 30 s, as reported in Figure 7, at 230 °C and 50 bar of H₂ [147]. Similar phenomenon was also observed in K₂[Mn(NH₂)₄] + 8LiH under the same temperature and hydrogen pressure conditions [148].

The details of the reaction pathway of K₂[Zn(NH₂)₄] + 8LiH system were investigated by means of in situ X-ray diffraction studies (Figure 8) [149]. This study showed that K₂[Zn(NH₂)₄] + 8LiH converts to 4LiNH₂ + 4LiH + K₂ZnH₄ during milling; upon heating, 4LiNH₂ + 4LiH + K₂ZnH₄ releases H₂ in multiple steps. The final dehydrogenation products are KH, LiZn₁₃, and Li₂NH. During absorption, KH reacts with LiZn₁₃ under 50 bar of H₂ to form K₃ZnH₅. This phase most likely is the key to enhance the re-hydrogenation process; thanks to the existence of K₃ZnH₅, the absorption reaction takes place in only 30 s at ca. 220 °C [149]. Inspired by these works, a large variety of ternary metal transition amides can be produced opening new perspectives for the development of systems which can ab/desorbs hydrogen at lower temperatures more suitable for practical applications.

6. Sodium-Based Composites

In the last 10 years, the ever-increasing cost of lithium on a worldwide scale has encouraged the search for lithium-free metal amides systems with high-performing properties for on-board applications, which can replace the most expensive Li-based amides. NaNH₂ and its composites, in particular, have attracted a lot of attentions because its considerable high amount of hydrogen, low cost and interesting thermodynamic properties [150,151,152,153,154].

Due to its high catalytic activity in the decomposition of NH₃, pure sodium amide, has been considered as potential material also for other applications different from solid-state hydrogen storage. In this context, its decomposition path has been topic of debate. Recently, two different decomposition paths have been presented. Jain et al. [151], proposed that NaNH₂, after melting at 200 °C, decomposes to sodium, ammonia and nitrogen (>400 °C), without passing for its corresponding nitride and/or imide phase as, on the other hand, shown by LiNH₂ and Mg(NH₂)₂. Recently, different results have been obtained by Miyaoka and coauthors, where the thermal decomposition of sodium amide was studied under different experimental conditions [154]. From this specific study emerged that the decomposition reaction products depend on the pressure conditions. The authors discovered a new intermediate, an imide-like phase with a cubic structure, which decomposes to sodium, at 400 °C, under vacuum, while, for the closed system, the partial pressure of hydrogen promotes the formation of sodium hydride [154].

A significant reduction of the hydrogen desorption temperatures, has been successfully obtained for composite systems. For example, Pecharsky et al. [153] reported that the 2NaNH₂ + 3MgH₂ system released hydrogen in the temperature range of 130–400 °C (vs. 186 °C–400 °C for 2LiNH₂ + 3MgH₂) following the Equation (14):

2NaNH₂ + 3MgH₂ → Mg₃N₂ + 2NaH + H₂

(14)

A total release of 5.1 H₂ wt. % was finally achieved at 400 °C. The decomposition path summarized in Equation (14) proceeds with a series of competitive solid-state process including the formation of Mg(NH₂)₂ at low temperature (>140 °C), which suggests that a metathesis reaction occurred. Only 2.1 wt. % of hydrogen could be re-absorbed at 395 °C under 190 bars of hydrogen, probably because, at this high temperature, NaH is liquid and/or decomposes to Na. However, any absorption was achieved at lower temperatures. New insights on the desorption pathway of NaNH₂/MgH₂ were obtained by Sheppard and collaborators [152]. 3.3 wt. % of hydrogen was in fact desorbed between 70 and 335 °C in the equimolar milled system with the formation of two new magnesium-containing imide and nitride phases during the dehydrogenation. Despite the lower desorption temperature with respect the 3:2 system, the kinetic desorption remains a not yet solved limitation for this mixture even after the addition of catalysts such as TiCl₃.

Effect of high-energy ball milling on the NaNH₂/MgH₂ systems with different stoichiometry, 2:1 and 2:3, have also been evaluated [155,156,157]. The mechanically activated metathesis reaction occurred for the 2:1 system, ball milled for a prolonged milling time. Without releasing of hydrogen gas. Conversely to the thermally activated process, which exhibits solid intermediates, only the formation of Mg(NH₂)₂ and NaH phases were observed upon the extended mechanical treatment (50 h). This represented a valid strategy to prepare the Mg(NH₂)₂-NaH system in nanostructured conditions. However, in the same work, it emerged that the desorption properties cannot be interesting for practical applications [156]. This evidence was also confirmed by a previous work, published by Chen et al. devoted to the study of the Mg-Na-N-H system [158]. Mg(NH₂)₂ was mixed with different amounts of NaH and the hydrogen storage properties tested. Working temperatures lower than 200 °C were observed but only 2.17 wt. % of hydrogen could be reversibly desorbed, which implies a scarce interest from a practical applications point of view.

One of the most important benefit recognized to the addition of metal hydrides to sodium amide, is represented by the ammonia suppression during the thermal decomposition step, as reported by Jain et al. [159] In this work, NaH has been shown to efficiently suppress the emission of ammonia gas when added to NaNH₂. Furthermore, a clear picture of the mechanism involved in this process has been reported for an extended number of amide hydride systems: the dehydrogenation process occurred with the formation of double-cation amides, which decrease the hydrogen desorption temperatures (except for the CaH₂-containing system) [159].

A further way to destabilize NaNH₂ is represented by the addition of complex hydrides, such as NaBH₄, Ca(BH₄)₂ and Mg(BH₄)₂ in stochiometric amount [150,158,160,161,162,163]. For example, the starting reagents NaNH₂-NaBH₄ (2:1) when subjected to intensive ball milling, react producing Na₃(NH₂)₂BH_4. This new phase desorbed hydrogen (6.85 wt. %) gas forming Na₃BN₂ [162]. The desorption path is characterized by two main steps: the first in the temperature range 70–170 °C, while the second one between 190–420 °C. At higher temperatures, Na₃BN₂ decomposes to Na. The activation energy of this system was estimated to be 159.6 kJ/mol, and it can be significantly reduced to 70 kJ/mol by the addition of Co-B catalyst [163]. A comparable kinetic enhancement was also reached for the un-doped 2:1 system prepared by liquid phase ball milling (E_a = 76.4 kJ/mol) instead of the classical dry milling, paving the way for the exploration of a new efficient synthetic route for this kind of composites [161]. Synergetic effect of the liquid phase ball milling preparation and the addition of Co-Ni-B dopant, contributed to significantly improve the kinetic performance of the dehydrogenation step further lowering the activation energy to 68.2 kJ/mol and the thermodynamic property of the system. In fact, 5.05 wt. % of hydrogen was successfully desorbed at temperature below 200 °C [163].

Concerning the addition of Mg(BH₄)₂ and Ca(BH₄)₂, a very recent and detailed structural investigation was performed for different compositions [160]. Both the systems desorbed by the emission of undesired ammonia gas at low temperature while the formation of NaBH₄ occurred through a metathesis reaction between NaNH₂ and M(BH₄)₂ (M = Mg, Ca). Hydrogen gas evolution started only at temperature higher than 350 °C for the 1:2 system and above 450 °C for the equimolar 1:1 composites, making these systems not properly suitable for on-board applications.

Finally, different studies have also been focused on the opportunity to improve the hydrogen storage properties of NaNH₂ by combining it with LiAlH₄ as effectively occurred for LiNH₂ [164,165,166,167]. In this context, Xiong et al. proved that during ball milling a solid-state interaction between NaNH₂ and LiAlH₄ took place, leading to the formation of the tetramide Li₃Na(NH₂)₄ which seems to favor the desorption at lower temperatures [166]. Recently, Jensen and coauthors synthesized, by high-energy ball milling the lithium sodium tetramide and corroborated its key role together with LiNa₂(NH₂)₃, in the destabilization of different sodium-based systems [165]. For example, LiNH₂-NaH and NaNH₂-LiH desorbed 0.9 wt. % of hydrogen from 180 to 340 °C with a faster kinetic if compared with the most explored LiNH₂ + LiH system. Unfortunately, as emerged from in situ synchrotron X-ray powder diffraction (XRPD) studies, imides and nitrides cannot be formed under moderate temperatures, which implies the irreversibility of these systems.

7. Concluding Remarks

The technological challenges connected with the exploitation of solid materials for storing efficiently hydrogen are primary sources of interest, and the opportunity to use metal amides-based composites in this specific area captures the attention of an increasing number of scientists. From the overview offered in Section 2, it is clear that the Li-N-H system, despite of the encouraging results recently obtained, possess several limitations, in particular from the thermodynamic point of view, which are in part mitigated when Li is partially replaced by Mg, forming the Li-Mg-N-H system. The latter, in fact, shows more suitable thermodynamic properties and superior kinetics when compared with the first one. To get light on the promising properties of these systems, the thermodynamic and kinetics features of the most appealing mixtures, experimentally achieved, have been summarized in

Table 1. Thermodynamic and kinetic data of some Li-Mg-N-H systems.

Composition	ΔH kJ/(mol H2)	ΔS J/(mol K)	Ea kJ/mol	Hydrogen Content—wt. % (Desorption Working Temperature and Time)	Ref.
2LiNH2 + MgH2	65.0	-	119.0	2.8 (210 °C, 250 min)	[66]
	65.8		119.0	-	[65]
1.9LiNH2 + 1.1MgH2 + 0.1KH	42.0	-	87.0	5.4 (210 °C, 250 min)	[65]
				5.0 (210 °C, 50 min)
1.9LiNH2 + 1.1MgH2 + 0.1RbH	42.7	-	86.8	5.5 (210 °C, 50 min)	[65]
1.9LiNH2 + 1.1MgH2 + 0.1CsH	45.7	-	109.1	5.0 (210 °C, 100 min)	[66]
				4.4 (210 °C, 50 min)
2LiNH2 + MgH2 + 0.1LiBH4	-	-	156	2.5 (180 °C, 180 min)	[145]
2LiNH2 + MgH2 + 0.1LiBH4 + 4 wt. % ZrFe2	54.0	159.6	135	4 (180 °C, 180 min)	[145]
2LiNH2 + MgH2 + 0.1LiBH4 + 3 wt. % ZrCoH3	45.0	124.0	131.8	4.1 (150 °C, 450 min)	[131,136]
			79.9	4.5 (150 °C, 350 min)	[134]
Mg(NH2)2 + 2LiH	38.9	112.0	102.0	5.0 (200 °C)	[79]
	41.6		112.2	2.6 (140 °C, 400 min)	[140]
6Mg(NH2)2 + 9LiH	40.1	113.3	-	2.8 (146 °C, 500 min)	[119]
				4.4 (81 °C) *
6Mg(NH2)2 + 9LiH + 3/2LiI	33.3	99.9	-	2.4 (146 °C, 500 min)	[119]
				3.0 (60 °C) *
6Mg(NH2)2 + 9LiH + 3LiBr	31.9	99.5	-	1.8 (146 °C, 500 min)	[119]
				2.7 (48 °C) *
6Mg(NH2)2 + 9LiH + LiBH4	35.8	106.2	-	4.0 (146 °C, 500 min)	[119,135]
				4.0 (174 °C-15 cycles)
				4.2 (64 °C) *
6Mg(NH2)2 + 9LiH + 3LiBH4	-	-	109.0	3.8 (143 °C, 220 min)	[121]
6Mg(NH2)2 + 9LiH + 6LiBH4	-	-	86.0	3.3 (143 °C, 220 min)	[121]
6Mg(NH2)2 + 9LiH + 12LiBH4	24	92	76.0	2.6 (98 °C, 140 h)	[121,122]
				2.6 (108 °C, 40 h)
				2.6 (143 °C, 8 h)
				2.7 (RT) *
6Mg(NH2)2 + 9LiH + LiBH4 + 2 wt. % YCl3	-	-	130.0	1.9 (140 °C, 800 min)	[135]
6Mg(NH2)2 + 9LiH + LiBH4+5 wt. % Li3N	-	-	134.0	1.9 (140 °C, 800 min)	[135]
6Mg(NH2)2 + 9LiH + LiBH4 + 2 wt. % YCl3 + 5 wt. % Li3N	-	-	128.0	2.1 (140 °C, 800 min)	[135]
Mg(NH2)2 + 2LiH + 0.1LiBH4	36.5	105.6	-	4.3 (140 °C, 400 min)	[96]
Mg(NH2)2 + 2LiH + 0.2LiBH4	38.5	-	101.4	3.7 (140 °C, 400 min)	[140]
Mg(NH2)2 + 2LiH + 0.1Mg(BH4)2	36.7	-	99.7	4.1 (140 °C, 400 min)	[140]
Mg(NH2)2 + 2LiH + 0.2MgNH	-	-	105.5	3.1 (140 °C, 400 min)	[140]
Mg(NH2)2 + 1.9LiH + 0.1KH	42.0	-	-	4.8 (163 °C, 50 min)	[93]
Mg(NH2)2 + 2LiH + 0.08KF	39.4	-	-	4.3 (150 °C, 150 min)	[104]
Mg(NH2)2 + 2LiH + 0.08RbF	39.3	-	-	3.6 (130 °C, 350 min)	[107]
Mg(NH2)2 + 2LiH + 0.04KH + 0.04RbH	38.0	-	-	3.8 (130 °C, 350 min)	[89]
Li2Mg(NH)2@CNFs	35.7	100.2		2.0 (130 °C, 50 min)	[111]

* theoretical dehydrogenation temperature over 0.1 MPa equilibrium hydrogen pressure.

.

Depending on the dopant, the 2LiNH₂ + MgH₂ system shows different behaviors. In comparison to the un-doped material, the addition of additives based on KH; RbH and CsH, (Figure 9), lead to significant thermodynamic and kinetic improvements. The introduction of LiBH₄ contributes to decrease the temperature of hydrogen release from 210 °C to 180 °C, however, the gravimetric capacity of the system is affected. An important enhancement is produced when this composite is doped with ZrFe₂ and ZrCoH₃, as clearly emerged by the bars in Figure 10. The system containing ZrCoH₃ has been extensively investigated and, up to day, among the metal amide-based family, it is probably the only system tested under near practical conditions [137].

Interestingly, the working temperature can be further decreased inverting the system cations, i.e., Mg(NH₂)₂-LiH. This composite presents several advantages such as reduced or almost suppressed ammonia emissions, working temperatures of about 150 °C and good gravimetric capacity if compared with the previous composites (Figure 10). For this system, the enthalpy is reduced from ~50 kJ/mol H₂ of the LiNH₂/MgH₂-based mixtures to ~40 kJ/mol H₂. However, as also evidenced in Table 1, the data concerning the activation energy of the dehydrogenation process are quite scarce and more efforts have to be made to obtain a more detailed picture of this system.

In order to better discriminate among the most appealing systems which approach to the targets fixed by DoE, two main parameters, gravimetric density and working temperature properties have been selected and depicted in Figure 11 for several materials. Other systems not discussed in this review, such as NH₂NH₂, LiBH₄ ∗ N₂H₄, metal borohydrides, etc., have been also included in the graph, for comparison purpose. Form these data it is possible to evince that, by now, there is no amides-based system which can completely fits the specific targets fixed by DoE for 2020 (olivine panel in Figure 11). However, it is important to highlight the important progresses made in the last years in term of effective working temperature. In particular, the ternary system Mg(NH₂)₂ + LiH + LiBH₄ shows a working temperature close to the desired room temperature. LiBH₄ is proved to have an extraordinary effect in tuning the thermodynamic properties of the Li-Mg-N-H system. Likewise, nanoparticles of Li₂Mg(NH)₂ encapsulated into carbon nanofiber matrix possess suitable temperature to be candidate as promising material for practical application. However, the hydrogen release under these operative conditions is not enough to satisfy the expected requirement.

Furthermore, despite the reaction enthalpy of 24 and 35.7 kJ/mol H₂, results quite satisfactory, both systems are kinetically hindered. In fact, in the ternary composites desorbs reversibly 2.8 wt. % of hydrogen below 100 °C only after 140 h. The encapsulated system can reversibly release similar amount of hydrogen with faster kinetic (50 min) but under higher temperature (130 °C). For this reason, at the moment, kinetic improvement represents the first objective to be accomplished for these systems and many efforts have to be addressed in this direction as future perspective. In this context, solid-state catalysts, using K-and Rb-based additives, or ZrCoH₃, can represent a valid strategy, together with a suitable preparation technique, for reaching the expected results for both these systems.

Another challenge is represented by scaling-up the production of the specific systems. This implies to approach two main important aspects: (i) the preparation techniques and (ii) cost of materials. If Li₂Mg(NH)₂@CNFs presents a restricted number of solid reagents involved in its preparation, which reduces the whole cost of production, the soft chemistry approach used in this synthesis is, of course, less practicable for producing large amounts of product with respect to the traditional ball milling. Likewise, the cost issue of the starting materials has to be taken in consideration. The commercial prices of the involved materials are reported in

Table 2. Commercial prices of materials involved in the preparation of the metal amides-based composites.

Materials	LiNH2	LiH	MgH2	LiBH4
Purity (%)	95 (Alfa Aesar) 95	97 (Alfa Aesar)	95% (Albemarle)	95 (Acros Organics)
Price (€/kg)	348	483	600	7980

. The actual costs of reagents represent a further barrier for a rapid commercialization of these systems. However, a valid chance, in prospective, can be represented by using cheap raw materials obtained from Mg and Li industrial wastes, which can allow to halve the production cost of several amides.