Recent Development in Nanoconfined Hydrides for Energy Storage

Hydrogen is the ultimate vector for a carbon-free, sustainable green-energy. While being the most promising candidate to serve this purpose, hydrogen inherits a series of characteristics making it particularly difficult to handle, store, transport and use in a safe manner. The researchers’ attention has thus shifted to storing hydrogen in its more manageable forms: the light metal hydrides and related derivatives (ammonia-borane, tetrahydridoborates/borohydrides, tetrahydridoaluminates/alanates or reactive hydride composites). Even then, the thermodynamic and kinetic behavior faces either too high energy barriers or sluggish kinetics (or both), and an efficient tool to overcome these issues is through nanoconfinement. Nanoconfined energy storage materials are the current state-of-the-art approach regarding hydrogen storage field, and the current review aims to summarize the most recent progress in this intriguing field. The latest reviews concerning H2 production and storage are discussed, and the shift from bulk to nanomaterials is described in the context of physical and chemical aspects of nanoconfinement effects in the obtained nanocomposites. The types of hosts used for hydrogen materials are divided in classes of substances, the mean of hydride inclusion in said hosts and the classes of hydrogen storage materials are presented with their most recent trends and future prospects.


Introduction
The 21st century has been marked by tremendously important technological breakthroughs, yet the massive expansion of industrialization has led to a deepening scarcity and skyrocketing prices of fossil fuels and energy raw materials, concomitant with a continual atmospheric pollution [1]. In the context of ever-increasing energy demands and the serious downsides of using fossil fuels, hydrogen has emerged over the past decades as a true and relevant promise of a carbon-free, green energy source for the world. However, hydrogen has a very low boiling point (20.4 K) at 1 atm, which severely restricts its use in the native form, except in some high pressure, cryogenic tanks that pose themselves additional energetic costs and safety risks regarding charging, transport and storing [1]. To circumvent the downfalls of using molecular dihydrogen (H 2 ), scientists have turned their attention and research focus on hydrogen-containing compounds, in the form of metal hydrides and related materials, which in turn feature higher thermal stability, safer handling, no fuel loss upon storage and overall produce the cleanest energy known today. The fuel of the future should ideally produce no carbon-containing by-products, exhibiting time-and property-related endurance over 1500 dehydrogenation-rehydrogenation cycles, and most importantly, all of this while featuring a gravimetric weight percentage of at least 5.5 wt.% (DOE's target set for 2025) [1][2][3][4][5][6]. The use of fossil fuels will eventually be phased-out and an

Characterization Methods: Old, New, and Their Pitfalls
Traditionally, hydrogen storage materials follow a typical characterization protocol involving structural (XRD), elemental (XPS), morphological (SEM, TEM, N 2 sorption isotherms) and recording of hydrogenation data (PCI curves) [8]. Recently, a fundamental issue regarding elucidation of local environment of hydrogen in energy materials has revealed fast sample spinning 1 H NMR high-resolution spectroscopy as an appropriate tool to quantitatively characterize hydrogenated TiZrNi quasicrystals [30]. Kweon et al., showed by employing fast-spinning NMR spectroscopy that neutral hydrogen is surrounded by metal atoms shifting gradually from Zr to Ti and then Ni with increasing hydrogen content [30]. 1 H magic-angle spinning (MAS) NMR spectra has shown real promise for tuning electronic characteristics in a Ba-Ti oxyhydride, and could become a tool to investigate hydrogen occupation in the vicinity of the nuclei (negative Knight shift, indicative of interaction of conduction band electrons and probe nucleus) [31]. A potential downside indicative of interaction of conduction band electrons and probe nucleus) [31]. A potential downside of using this technique is the high sensitivity to sample temperature, which was shown to increase due to fast rotor spinning , with a direct effect on main peak width change. Thus, additional precautions need to be undertaken to account for the effect of sample temperature increase when using fast spinning NMR spectroscopy [31].
Correct understanding of interfacial phenomena occurring during hydrogen storage is now termed as hydrogen spillover effect (HSPE). First discovered in 1964, it describes the migration of hydrogen atoms produced by H2 decomposition on an active site, and it allows for a more insightful view on the dynamic behavior of hydrogen in energy storage materials [7]. While molecular orbital energy computations showed unfavorable energy for H atom spillover on non-reducible supports, recent studies have shown that HSPE is indeed possible on inert supports such as siloxanic materials (SiO2) [7]. This bears a direct effect on hydrogen storage materials such as metal hydrides confined in mesoporous silica supports, where the spillover distance is limited to very short distances of ~10 nm [7].
Interestingly, developing tools to characterize metal hydrides during hydrogenation cycles has led to a summary of soft (X-ray absorption, XAS; X-ray emission spectroscopy, XES; resonant inelastic soft X-ray scattering, RIXS, X-ray photoelectron spectroscopy, XPS) and hard (X-ray diffraction, XRD) X-ray techniques used to this end ( Figure 1) [32]. Soft X-rat techniques (100-5000 eV) are particularly appealing for tracking mechanistic behavior and intermediate product formation during hydrogenation studies, with direct influence over hydrogen storage capacity. XAS measurements for instance are bulk or surfacesensitive, and show 3d transition metal (TM) L-edges corresponding to transition of a 2p electron to an unoccupied 3d orbital, hence enabling monitoring of oxidation state changes during hydrogen release (+n...0) and uptake (0…+n) [32]. Similarly, TM-catalyzed alanates (2 mol%-catalyzed NaAlH4) showed in XAS measurements the Al and Na K-edge and Ti L-edge consistent with a Ti-like state throughout the hydrogen release/uptake cycles, but with clear differences in Al state, which may undergo various intermediate states (Al/NaAlH4/Na3AlH6) [32]. Quasi-elastic neutron scattering (QENS) studies have been undertaken to establish hydrogen dynamics in nanoscale sodium alanate NaAlH4 and showed that fitting QENS to a Lorentzian function can yield two dynamic states of hydrogen and concluded that even at 77 °C there is a high percentage (18%) of mobile hydrogen atoms in the nano-NaAlH4 [33]. As an alternative method to the conventional pressure-composition-temperature (PCT) method typically used to characterize thermodynamic parameters for hydride-based sys-tems, a less complex investigation method has been described for MgH 2 -based materials: thermogravimetric analysis (TGA) [34]. This method relies on cycling the hydride under a flowing gas of constant hydrogen partial pressure, and the TGA curves are further analyzed using the van't Hoff equation to obtain the absorption/desorption enthalpies, which in the case of VTiCr-catalyzed Mg/MgH 2 materials, showed good agreement with traditional PCT results [34]. Other recent research established a nano-Pd patched surface of Pd 80 Co 20 to afford one of the most sensitive optical hydrogen sensors (fast response of <3 s, high accuracy of <5%, and very low limit of detection of 2.5 ppm) [35]. Employing interpretable machine learning could also help formulate general design principles for intermetallic hydride-based systems being used to validate limited data from the HydPARK experimental metal hydride database and stressing the recommendation for experimental groups to report ∆H, ∆S, P eq , T and V cell [27].
Valero-Pedraza et al., have characterized the hydrogen release form ammonia borane nanoconfined in mesoporous silica by means of Raman-mass spectroscopy, which confirmed hydrogen release from AB at lower temperatures, fewer BNHx gaseous fragments in nanoconfined samples and a lack of polyiminoborane formation during thermolysis [36].
The study also pointed out to silica-hydride interactions, which were identifiable based on modifications in the Raman spectra [36].
However, analysis of the literature data also points out to several weaknesses in applying traditional characterization methods that have not yet been tuned for current nanosized materials [15,34,37,38]. For instance, AB (ammonia borane) hydrogenation studies showed many inconsistencies [38]. By assessing TGA data in the literature, Petit and Demirci urge caution when evaluating ammonia borane weight loss (and consequently hydrogen release), as this was found to be highly dependent on the operation conditions (semi-closed/open reactor) and were shown to erroneously indicate a different hydrogen release temperature onset and hydrogen wt.% [38].
Surrey et al., conducted a critical review of a paper discussing electron microscopy observation of elementary steps in MgH 2 release mechanisms [37]. In this work, they debunked the general assumption that TEM microscopy can be used, as such, without further testing methodology adjustment in the case of hydrogen storage materials such as MgH 2 . The issue was serious, as it led initial authors to misinterpret TEM observations, by disregarding the key aspect of electron beam induced dehydrogenation of MgH 2 [37]. In a cascade chain of errors, the beam-induced heat producing dehydrogenation also led to a false interpretation of SAD (selected area diffraction) data, which only showed hollow MgO shells deprived of Mg-core, an effect actually ascribed to the nanoscale Kirkendall effect. As a result, it was apparent that the sample actually measured did not even contain MgH 2 any longer [37].
In line with the issues raised above, Broom and Hirscher discussed the necessary steps for reproducible results in hydrogen storage research [15].

Bulk vs. Nanomaterials
After its first inclusion on the research outlook of scientists worldwide in 1996, nano-sized hydrides have known a wide expansion, mainly due to several important kinetic and thermodynamic improvements of nanoconfinement over their bulk counterparts [4,8,14,16,18,[21][22][23]27,28,. Over time, nanoconfinement has emerged as a reliable tool for tuning not only thermodynamic and kinetic behavior at nanoscale, but also for altering reaction pathways, lowering or even suppressing side-reactions and sideproducts, while also affording better size control of the particles over several hydrogen release/uptake cycles ( Figure 2).

Figure 2.
Main features of bulk and nanoconfined materials for hydrogen storage; exemplified for the case of an overly-studied hydride, MgH2. (inset reprinted/adapted with permission from Ref. [65]. 2022, Elsevier).

Types of Hosts
Confining LiBH 4 by a melt impregnation technique in nanoporous silica MCM-41 (1D, d pore < 2 nm) or SBA-15 (2D-ordered pore structure, d pore = 5, 7 and 8 nm) of different pore sizes reveals an interesting interfacial effect governing Li + and BH 4 − ion mobility [87]. Using solid-state NMR ( 1 H, 6 Li, 7 Li and 11 B), Lambregts et al., showed that, as a result of nanoconfinement, two distinct fractions of LiBH 4 coexist and this is a temperaturedependent equilibrium (Equation (2)): The high mobility LiBH 4 is located near silica pore walls, whereas LiBH 4 of lower mobility is located towards the pore's core; the theoretical wall thickness was estimated based on a core-shell model LiBH 4 @SBA-15, as t = r p (1 − f lower mobility . The dynamic layer thickness is temperature-dependent, and increases from 0.5 nm (30 • C) to 1.2 nm (110 • C). Here again the results of calorimetric data were found to overestimate the highlymobile LiBH 4 layer thickness (1.9 nm), pointing out the need for care when deriving the same parameter from different techniques [87]. While 6,7 Li NMR spectra was too complex for unequivocal deconvolution, 1 H and 11 B NMR spectra clearly show two components throughout the investigated temperature range (30-130 • C), consistent with the two LiBH 4 fractions of different ion mobility [87].
Melt impregnation of NaBH 4 in MCM-41 at 560 • C led to a drastic surface area decrease from 1110.9 m 2 g −1 (pristine MCM-41) to 3.5 m 2 g −1 (nanocomposite NaBH 4 @MCM-41), and to a 78% pore filling attested by pore volume decrease (1.02 cm 3 g −1 to 0.02 cm 3 g −1 ) [74]. Interestingly, some amount of sodium perborate NaBO 4 resulting from unavoidable oxidation of the borohydride with silanol (Si-OH) groups is the main additional phase detected by XRD, confirming no significant additional phases due to melt impregnation at >500 • C. The dehydrogenation onset peak for NaBH 4 was reduced by nanoconfinement from 550 • C (bulk) to 520 • C (nanocomposite) [74]. Due to the insulating nature of boron oxide phase (NaBO 4 ), the ionic conductivity did not improve the same way it does for LiBH 4 , and remained largely the same (7.4 × 10 −10 S cm −1 ). This 10-fold increase in ionic conductivity that only lasts up to 70 • C for the nanocomposite is attributed to the presence of larger dodecaborate ions B 12 H 12 2− whose distinct presence was signaled in 11 B NMR spectra by an additional sharp peak at −15.58 ppm (NaBH 4 @MCM-41) vs. −41.95 ppm (for pristine The high mobility LiBH4 is located near silica pore walls, whereas LiBH4 of lower mobility is located towards the pore's core; the theoretical wall thickness was estimated based on a core-shell model LiBH4@SBA-15, as = ( − . The dynamic layer thickness is temperature-dependent, and increases from 0.5 nm (30 °C) to 1.2 nm (110 °C). Here again the results of calorimetric data were found to overestimate the highlymobile LiBH4 layer thickness (1.9 nm), pointing out the need for care when deriving the same parameter from different techniques [87]. While 6,7 Li NMR spectra was too complex for unequivocal deconvolution, 1 H and 11 B NMR spectra clearly show two components throughout the investigated temperature range (30-130 °C), consistent with the two LiBH4 fractions of different ion mobility [87].
Melt impregnation of NaBH4 in MCM-41 at 560 °C led to a drastic surface area decrease from 1110.9 m 2 g −1 (pristine MCM-41) to 3.5 m 2 g −1 (nanocomposite NaBH4@MCM-41), and to a 78% pore filling attested by pore volume decrease (1.02 cm 3 g −1 to 0.02 cm 3 g −1 ) [74]. Interestingly, some amount of sodium perborate NaBO4 resulting from unavoidable oxidation of the borohydride with silanol (Si-OH) groups is the main additional phase detected by XRD, confirming no significant additional phases due to melt impregnation at >500 °C. The dehydrogenation onset peak for NaBH4 was reduced by nanoconfinement from 550 °C (bulk) to 520 °C (nanocomposite) [74]. Due to the insulating nature of boron oxide phase (NaBO4), the ionic conductivity did not improve the same way it does for LiBH4, and remained largely the same (7.4 × 10 −10 S cm −1 ). This 10-fold increase in ionic conductivity that only lasts up to 70 °C for the nanocomposite is attributed to the presence of larger dodecaborate ions B12H12 2− whose distinct presence was signaled in 11 B NMR spectra by an additional sharp peak at −15.58 ppm (NaBH4@MCM-41) vs. −41.95 ppm (for pristine BH4-) ( Figure 3) [74]. The organic-inorganic hybrid poly(acryalamide)-grafted mesoporous silica nanoparticles (PAM-MSN) have been evaluated as functionalized nanoporous hosts for tuning hydrogen release/uptake behavior in ammonia borane (AB), which started to desorb hydrogen in the said nanocomposite at a lower temperature with respect to pristine AB, which was further enhanced by functionalization of the mesoporous silica shell with car- Figure 3. Possible decomposition pathways for bulk NaBH 4 (a,b) and for melt-impregnated, nanoconfined NaBH 4 (c).

of 49
The organic-inorganic hybrid poly(acryalamide)-grafted mesoporous silica nanoparticles (PAM-MSN) have been evaluated as functionalized nanoporous hosts for tuning hydrogen release/uptake behavior in ammonia borane (AB), which started to desorb hydrogen in the said nanocomposite at a lower temperature with respect to pristine AB, which was further enhanced by functionalization of the mesoporous silica shell with carboxylic -COOH groups [88].
2D-ordered mesoporous silica of cylindrical pores (SBA-15) was successfully used by Yang et al., for enhancing the ionic conductivity of a mixed-anion borohydride, Li 2 (BH 4 )(NH 2 ). By following a melt infiltration procedure, the Li-ion conductivity was increased in Li 2 (BH 4 )(NH 2 )@SBA-15 to 5 × 10 −3 S cm −1 at 55 • C [89]. A marked kinetic improvement of hydrogen release (∆T = 70 • C) was recently reported by Rueda et al., by confinement of ammonia borane (AB) in silica aerogel by simultaneous aerogel drying and AB gas antisolvent precipitation using compressed CO 2 , and achieving a weight AB loading of up to 60 wt.% [90].

Gas Selective-Permeable Polymers
Attempts to restrict oxygen and moisture exposure of active hydrogenation sites in hydride materials have been made through the engineered approach of covering the hydride materials with a layer of H 2 -permeable polymer [88,127,156,[173][174][175]. This approach proved to be very successful, provided that the hydride coverage was indeed complete (Table 6).

MXene Type
Hydrogen Storage Material Nanoconfinement Method Ref.

Catalytic Effects of Doping the Host and/or Substitution of the Hydride Species
Improvements on hydrogen release/uptake cycles have often been explored in conjunction with utilization of catalysts used to either dope the host, or the hydride material. This strategy is based on formation of active sites for hydrogenation reaction to occur, or is sometimes ascribed to the formation of a reactive intermediate species [19,68,92,102,[111][112][113]117,125,128,151,160,161,163,[195][196][197]. In addition, cation substitution or anion substitution in complex hydrides has been employed to reduce energy barriers and improve overall recyclability of the hydride materials (Table 8).
Other approaches start from the organometallic precursor of the metal, which undergoes reduction (with H 2 or another reductant, such as LiNp) typically after impregnation into the porous host. (Equation (4))

Melt Infiltration
Melt infiltration of complex hydrides has widely been used to introduce the active hydride material into nanoporous hosts. This technique has the advantage of requiring no solvent (so it consists of less steps), but the hydride material must have a lower melting temperature, and the infiltration is carried out under H 2 pressure in order to avoid the onset of dehydrogenation reaction.

Solvent Infiltration
Solvent infiltration has become the method of choice as it achieves pore filling of the porous scaffold at temperatures that are near ambient, provided that a suitable solvent for the material has been identified. This is typically an issue, as solubility data on complex hydrides are rather scarce, and usually their solubility in ether-like solvents is limited [16].

Solvent-Assisted Ball-Milling
Nanoconfinement of hydride-based materials in nanoporous hosts has the potential advantage of bypassing the slow kinetics of their bulk counterparts, thus enabling a shorter refueling time, in pursuit of the DOE's current targets [5,6]. Very high surface area supports (MOFs, activated carbons) afford good hydrogen sorption capacities, but since the adsorption is mainly governed by physisorption, it is only relevant at 77 K. At this low temperature, a rough estimation (Chahine's rule) is that for pressures that would occupy all adsorption sites (exceeding 20 bar), the expected storage capacity is~1 wt.%/500 m 2 g −1 and scales proportional to the specific surface area [8]. Ball milling (with or without a solvent) can introduce the hydride material into the porosity of the employed scaffold. The process is energy-intensive and can proceed with an important increase in the local sample temperature, and therefore the process is carried out in steps (for instance, 20 min milling followed by a 10 min pause allowing controlled cooling).

Metal Hydrides and Their Recent Nanoconfinement Studies
Pristine metal hydrides have recently been comprehensively reviewed, and the results show promising trends upon nanoconfinement [213].

LiH
Alkali metal hydrides have been used for catalytic reactions, but have attracted attention due to their lightweight characteristics, as well as the high hydrogen gravimetric content. However, their high thermal stability makes them less attractive in their pure form; LiH, for instance, melts at 689 • C and decomposes at 720 • C into Li and H 2 (Equation (5)). Alkali metal hydrides have unusually high decomposition temperatures due to their saltlike nature (LiH, mp = 698 • C; NaH, mp = 638 • C; KH, mp~400 • C with K vaporizing in H 2 current). Given their high decomposition temperature, alkali metal hydrides require kinetic and thermodynamic destabilization (Table 10). Co(OH) 2 -Li@SiO 2 @Co(OH) 2 N/A αLiOH + 2αLi + + 2αe -= α Li 2 O + αLiH (0 < α < 1); High Li + storage in anode [217] Recently, a series of strategies have been utilized to produce nanosized LiH, but not all attempts dealt with hydrogen storage applications [114,133,198,[214][215][216], and some utilizing LiH-containing nanocomposites for their Li-storage capacity in a Co(OH) 2 -LiH novel anode material [217]. Even when dealing with potential hydrogen storage materials like LiH + MgB 2 , studies have focused on the phase-evolution process and XPS tracking thereof, rather than collection of hydrogen storage data [198]. Still, XPS data pointed to presence of LiBH 4 , Mg (3−x)/2 Li x (BH 4 ) x or Li-borate species present on account of multiple LiH-containing peaks identified [198]. At near-surface regions, LiBH 4 or mixed Li-Mg borohydrides can form at 100 • C below the threshold for hydrogenation of MgB 2 ; expectedly, LiBH 4 production scales with the LiH in the starting composite (Equation (6)) [198].
Sun et al., have shown that harnessing the plasmonic thermal heating effect of Au nanoparticles could lead to light-induced dehydrogenation of nanocomposites Au@LiH, which showed a 3.4 wt.% loss ascribed to dehydrogenation content [214]. The Au NPs dispersed on the surface of LiH, Mg or NaAlH 4 all showed marked improvements in hydrogenation studies. The preparation of Au/LiH composites involved LiH suspension in THF under sonication and overnight stirring at 500 rpm, after which a THF solution of HAuCl 4 was added and stirring continued for an additional 24 h, leading to the Au/LiH material after centrifugation and overnight drying by Schlenk line technique. Hydrogen absorption was carried out under 14.8 atm H 2 , while desorption was conducted under 0.2 atm pressure, utilizing Xe lamp illumination affording 100 • C local temperature [214].
Overcoming kinetic and thermodynamic barriers in the complex Li-N-H system (Equation (7)) led White et al., to study the Li 3 N effect on the LiNH 2 + 2LiH composite behavior [215]. On this occasion, a kinetic analysis showed the rate-limiting step is the formation of H 2 (g) at the surface of the core-shell structure Li 2 NH@Li 3 N [215]. Again, the use of TEM measurements was shown to be inappropriate for LiNH 2 materials, due to decomposition upon prolonged electron beam exposure. The equilibria shown in Equation (7) already occur upon the exposure of Li 3 N to 10 bar H 2 (200 • C, 2 h), but not at one bar H 2 , which only altered the α-to-β ration of Li 3 N [215].
Considering the gravimetric hydrogen densities required by DOE standards, LiH, MgH 2 and AlH 3 are the main binary systems proposed to date [216]. Silicon doping of LiH has shown a drastic reduction in decomposition temperature (∆T = 230 K), and could store up to 5 wt.% H 2 with release at 490 • C [216]. A nanostructured electrode of Co(OH) 2 and silica was recently employed in Li-conductivity studies and showed the formation of active LiH species, although the material was not investigated for its hydrogen storage properties [217].
A series of Li-based materials was investigated by Xia et al., who grafted on graphene LiH by in situ reduction in nBuLi with H 2 (110 • C, 50 atm), producing LiH@G. This nanocomposite LiH@G was further treated with B 2 H 6 or AB/THF, and novel LiBH 4 @G and LiNH 2 BH 3 @G nanocomposites were thus obtained (Equation (8)) [114].
The 2D LiH nanosheets were about 2 nm thick and afforded a 6.8 wt.% H 2 storage when loaded at 50 wt.% in the said graphene-based nanocomposite, which withstood structural integrity upon further hydride-to-borohydride transformation ( Figure 4) [114].
The morphology was tracked by SEM analysis and XRD diffraction, while hydrogenation data confirmed the modest 1.9 wt.% hydrogen storage by TGA ( Figure 5). This nanoconfinement approach in high surface area carbon (HSAG) of pore size 2-20 nm showed a high thermodynamic improvement, allowing for hydrogen release at 340 °C in LiH@HSAG rather than at the high 680 °C for pristine LiH [133].
Zhang et al., have dispersed TM-oxides (TiO 2 in particular) on amorphous carbon to achieve excellent, reversible hydrogen storage capacity, releasing in 10 min. at 275 • C, 6.5 wt.% hydrogen (85.5% that of pristine MgH 2 ) (Figure 7) [95]. Notably, the activation energies for desorption (E a,des ) and absorption (E a,abs ) have been considerably reduced compared to bulk magnesium hydride (Figure 7a). In a multi-fold enhancement strategy, the MgH 2 was first dispersed on carbon (MgH 2 + C), which showed modest improvements (<1 wt.% H 2 ) over MgH 2 bulk with no dehydrogenation in the same timespan (Figure 7c), TiO 2 was used as additive for MgH 2 to yield composites of MgH 2 + TiO 2 NPs, which surprisingly released~6 wt.% H 2 in 10 min [95]. Driven by these enhancements, nanocomposites of the type MgH 2 + TiO 2 SCNPs/AC were synthesized, which further improved hydrogen release/uptake: even at 50 • C, over the course of 20 min,~1.5 wt.% H 2 is released, whereas at 125 • C (~4.8 wt.%) and at 200 • C (6.5 wt.%) the kinetics is sped up considerably (Figure 7c-e). The rehydrogenation occurs within 5 min at 200 • C, and full recovery of the hydrogen storage capacity is achieved (6.5 wt.%). In addition, no appreciable hydrogen storage loss was recorded up to the 10th cycle (Figure 7f ) [95].  Using an FeCo nanocatalyst (mean size of 50 nm), Yang et al., synthesized composites MgH2 + nano-FeCo able to recharge to 6.7 wt.% hydrogen in one minute at 300 °C, and could desorb 6 wt.% (9.5 min, 300 °C) (Figure 8) [201]. In fact, even treatment under H2 backpressure at 150 °C produced 3.5 wt.% absorption in 10 min (Figure 8b). This highlights the importance of catalyst chosen, but also its morphology (nanosheets in the case of FeCo-nano). Plotting the Arrhenius equation also yielded the apparent activation energies: Ea,des = 65.3 ± 4.7 kJ mol −1 (60 kJmol-1 reduction from pristine MgH2), and the absorption energy Ea,abs = 53.4 ± 1.0 kJ mol −1 (Figure 8d). Gratifyingly, the FeCo-catalyzed magnesium hydride composite was able to rehydrogenate fully and was tracked over the course Using an FeCo nanocatalyst (mean size of 50 nm), Yang et al., synthesized composites MgH 2 + nano-FeCo able to recharge to 6.7 wt.% hydrogen in one minute at 300 • C, and could desorb 6 wt.% (9.5 min, 300 • C) (Figure 8) [201]. In fact, even treatment under H 2 backpressure at 150 • C produced 3.5 wt.% absorption in 10 min (Figure 8b). This highlights the importance of catalyst chosen, but also its morphology (nanosheets in the case of FeCo-nano). Plotting the Arrhenius equation also yielded the apparent activation energies: E a,des = 65.3 ± 4.7 kJ mol −1 (60 kJ mol −1 reduction from pristine MgH 2 ), and the absorption energy E a,abs = 53.4 ± 1.0 kJ mol −1 (Figure 8d). Gratifyingly, the FeCo-catalyzed magnesium hydride composite was able to rehydrogenate fully and was tracked over the course of 10 hydrogen release/uptake cycles (Figure 8h) [201].  Using a nanoflake Ni catalyst, Yang et al., have synthesized MgH2 + 5 wt.% Ni, composites able to store 6.7 wt.% hydrogen (des., 300 °C, in 3 min) (Figure 10). The absorption was also very fast, achieving 4.6 wt.% at 125 °C in 20 min, under 29.6 atm H2 [202]. The results also translate into much lowered activation energies (Arrhenius plot): Ea,des = 71 kJ The thermodynamic predictions that smaller size NPs will show the most important destabilization, Zhang et al., have produced ultrafine MgH 2 that was able to release and recharge hydrogen under ambient temperature, with a very high hydrogen storage capacity of 6.7 wt.% (Figure 9) [222]. This capacity was checked over 50 cycles, and showed virtually the same high-capacity behavior (Figure 9). The conditions employed for reversible behavior were 360 min at rt (6.7 wt.%), or 60 min at 85 • C (6.7 wt.%), under 30 bar H 2 . This unexpectedly high storage capacity (65.6 g H 2 /L) surpasses even DOE's requirement (50 gH 2 /L), and was possible solely on account of well-designed, size-restriction of MgH 2 to nanoscale [222].
Decomposition of n Bu2Mg typically used as an organometallic precursor to Mg/MgH2 NPs can follow two different steps, depending on the reaction temperature (Equations (11) and (12)). However small it might be, nanosized matter in general is also more reactive towards various gases and substrates, and Mg/MgH2 coupled system is no exception. Previous examples have overcome this downside by either pressing the nano-powders into pellets, or capping them with other reagents. There are however many reports where MgH2 has been introduced in the porosity of a carbonaceous host, such as the 3D activated carbon utilized by Shinde et al., to achieve a reversible hydrogen storage of 6.63 wt.% (Figure 11) [137]. Not only was the nanocomposite MgH2@3D-C storing hydrogen under relatively mild These results have been explained by means of the intermediate Mg 2 Ni intermediate, which is an intermetallic well-known in the Mg-Ni systems, and which absorbs rapidly H 2 to form Mg 2 NiH 4 . This functions as an effective "hydrogen pump" (Figure 10a) (Equation (10)) [202].
Decomposition of n Bu 2 Mg typically used as an organometallic precursor to Mg/MgH 2 NPs can follow two different steps, depending on the reaction temperature (Equations (11) and (12)).
However small it might be, nanosized matter in general is also more reactive towards various gases and substrates, and Mg/MgH 2 coupled system is no exception. Previous examples have overcome this downside by either pressing the nano-powders into pellets, or capping them with other reagents. There are however many reports where MgH 2 has been introduced in the porosity of a carbonaceous host, such as the 3D activated carbon utilized by Shinde et al., to achieve a reversible hydrogen storage of 6.63 wt.% ( Figure 11) [137]. Not only was the nanocomposite MgH 2 @3D-C storing hydrogen under relatively mild conditions 6.63 wt.% (five minutes, 180 • C), but the desorption was likewise fast (6.55 wt.%, 75 min, 180 • C), and perhaps more importantly, the nanoconfined MgH 2 was air-stable thanks to the protective carbon shell [137]. To the observed enhanced kinetics and improved thermodynamic behavior contribute decisively the transition metal dispersed into the 3D carbon: NI>Co >Fe. Running in a continuous regime, the nanocomposite was able to cycle for about 435 h (more than 18 days), without a palpable decrease in the hydrogenation storage capacity (Figure 11) [137].
While typically reduction in n Bu 2 Mg infiltrated into a nanoporous host to afford MgH 2 NPs is carried out in heterogeneous conditions (under H 2 pressure), Shinde used a mixed reductant system: TEA ((HOCH 2 CH 2 ) 3 N)/NH 2 NH 2 hydrazine to reduce Mg(II) to Mg(0) [137]. The synthetic procedure is nicely followed in Figure 11, and in this case, both scanning electron microscopy (SEM) and transmission electron microscopy (TEM) could be used for characterization, since the electron beam no longer hits directly the MgH 2 NPs; thus, the risk of in-situ decomposition during data acquisition is minimized (Figure 11). The hydrogen storage capacity exceeds 6 wt.% in case of Ni-NPs deposited in the 3D-AC (MHCH-5), confirming the beneficial and synergistic role of Ni when used in conjunction with MgH 2 . The plausible intermediate Mg 2 Ni forms the coupled system Mg 2 Ni/Mg 2 NiH 4 during hydrogenation, and this can be held responsible for the superior cycling behavior in case of MgH 2 @3D-AC (MHCH)-5(Ni), whereas this type of intermetallic is not common for Co or Fe [137].
The self-assembled MgH2 NPs are well embedded into the carbonaceous host, which plays a critical role in the overall performance of MHCH-5. It is implied, based on the thermal conductivity data (Figure 11h), that the carbon shell is important. The high thermal conductivity (70 W/mK), many times higher than that of MgH 2 NPs themselves, induces a lower temperature gradient in the sample and a high heat transfer coefficient, thus contributing to the exemplary behavior of the sample during hydrogenation cycling [137].
The reaction of LiH and AlCl 3 was shown to be greatly sped up by using a 0.1 molar TiF 3 , when the final product obtained after five hours milling under Ar pressure was a nanocomposite of composition α-AlH 3 /LiCl-TiF 3 [203]. Duan et al., have shown the critical role of TiF3 that acted as a seed crystal for α-AlH 3 . The pressure was also a crucial factor, as running the reaction under lower gas pressure only led to Al metal formation, without the envisioned hydridic phase (Equation (13)) [203].
However, thermodynamic data showed a Gibbs free energy for the expected α-AlH 3 formation of ∆G = −269 kJ mol −1 , therefore thermodynamically possible at 298 K [203]. Furthermore, tracking the reaction by solid-state 27 Al NMR spectra has shown the complex behavior of the reactive mixture ( Figure 12) (Equation (14)).
The kinetics are vastly improved, and raising the temperature above 120 °C allows for complete dehydrogenation in roughly 10 min (Figure 12).
The phase composition already shows formation of Al, consistent with the dehydrogenation reaction that had occurred. The report also highlighted the important role of the fluoride additive, as TiF3 reduced Ea of H-desorption to 52.1 kJ/mol [203].
Nanoconfinement of alane in a Cr-based MOF (MIL-101) with Al-doping has led to a nanocomposite able to store and recharge at 298 K (ambient) and 100 bar H2, 17.4 mg H2/g (equivalent to 1.74 wt.% H2) [40]. The introduction of alane inside the MIL-101 pores was made via solvent infiltration from a THF solution of AlH3. In fact, the pristine MOF MIL- The kinetics are vastly improved, and raising the temperature above 120 • C allows for complete dehydrogenation in roughly 10 min (Figure 12).
After five hours of ball milling under Ar pressure and dehydrogenation at 160 • C for 600 s, the final composite ( Figure 13) shows nanosized AlH 3 (mean size of α-AlH 3 was 45 nm, without traces of agglomerates). Furthermore, tracking the reaction by solid-state 27 Al NMR spectra has shown the complex behavior of the reactive mixture ( Figure 12) (Equation (14)).
The phase composition already shows formation of Al, consistent with the dehydrogenation reaction that had occurred. The report also highlighted the important role of the fluoride additive, as TiF3 reduced Ea of H-desorption to 52.1 kJ/mol [203].
The phase composition already shows formation of Al, consistent with the dehydrogenation reaction that had occurred. The report also highlighted the important role of the fluoride additive, as TiF 3 reduced E a of H-desorption to 52.1 kJ/mol [203].
Nanoconfinement of alane in a Cr-based MOF (MIL-101) with Al-doping has led to a nanocomposite able to store and recharge at 298 K (ambient) and 100 bar H 2 , 17.4 mg H 2 /g (equivalent to 1.74 wt.% H 2 ) [40]. The introduction of alane inside the MIL-101 pores was made via solvent infiltration from a THF solution of AlH 3 . In fact, the pristine MOF MIL-101 (3148 m 2 g −1 , 2.19 cm 3 g −1 and 2.5-3 nm pores) was shown to store 0.55 wt.% H 2 under the same conditions. The hydrogen release profile from the investigated samples shows the improvement of nanoconfinement of AlH 3 in MOF pores over the hydrogen release performance (Figure 14) [40].  The gravimetric storage capacity (17.4 mg H2 g −1 composite) was rather low considering DOE's goals, due to the inability to increase Al-doping of the framework without crystallinity loss, and the role of AC additive became apparent in order to enhance hydrogen interaction with confined Al NPs [40].
In an attempt to improve upon previous results, Duan switched the nano-host to MWCNT (multi-walled carbon nanotubes) of high pore textural characteristics (550 m 2 g −1 , 6-8 nm diameter) and obtained by ball-milling xMgH 2 + AlH 3 (x = 1-4) nanocomposites MgH 2 /AlH 3 @CNT of crystal size 40-60 nm that released 8.  The gravimetric storage capacity (17.4 mg H2 g −1 composite) was rather low considering DOE's goals, due to the inability to increase Al-doping of the framework without crystallinity loss, and the role of AC additive became apparent in order to enhance hydrogen interaction with confined Al NPs [40].
The Al metal produced in the first dehydrogenation stage of the composite ( Figure 16) will react with MgH 2 not yet dehydrogenated, to yield an intermetallic phase of Al 12 Mg 17 , which was confirmed by XRD data (Equation (15) The Al metal produced in the first dehydrogenation stage of the composite ( Figure  16) will react with MgH2 not yet dehydrogenated, to yield an intermetallic phase of Al12Mg17, which was confirmed by XRD data (Equation (15)).
Wang et al., showed the potential of nanosizing by introducing (injection in HSAG of Et2O solution of freshly-made AlH3 from metathesis of LiAlH4 and AlCl3) [44]. Considering the 14 wt.% loading with AlH3 in the composite AlH3@HSAG (by ICP-OES), the expected hydrogen capacity was 1.4 wt.%. However, only 15% of the Al behaved reversibly and thus only an overall 0.25 wt.% storage could be attributed to the nanoconfined AlH3 [44]. Interestingly, during sample preparation, the composite was heated at 65 °C under Ar to yield α-AlH3 polymorph and minimize spontaneous decomposition of AlH3 [44]. Either way, the reduction in dehydrogenation onset to ~60° (60…270 °C with a peak at 165 °C) shows the effect of nanosizing, effectively reducing hydrogen release by 50 °C [44].
The reactions involved in the mechanistic proposal of the authors also allowed computation of the apparent activation energies (by Kissinger plot), which were of 97.3 kJ mol −1 for MgH 2 and 61.4 kJ mol −1 for AlH 3 (Figure 16c).
Wang et al., showed the potential of nanosizing by introducing (injection in HSAG of Et 2 O solution of freshly-made AlH 3 from metathesis of LiAlH 4 and AlCl 3 ) [44]. Considering the 14 wt.% loading with AlH 3 in the composite AlH 3 @HSAG (by ICP-OES), the expected hydrogen capacity was 1.4 wt.%. However, only 15% of the Al behaved reversibly and thus only an overall 0.25 wt.% storage could be attributed to the nanoconfined AlH 3 [44]. Interestingly, during sample preparation, the composite was heated at 65 • C under Ar to yield α-AlH 3 polymorph and minimize spontaneous decomposition of AlH 3 [44]. Either way, the reduction in dehydrogenation onset to~60 • (60 . . . 270 • C with a peak at 165 • C) shows the effect of nanosizing, effectively reducing hydrogen release by 50 • C [44].
The EELS spectra of AlH 3 @CTF-biph and AlH 3 @CTF-bipy confirm that both contained aluminum, thus AlH 3 introduction in the CTF-based frame was achieved. However, inherent oxidation had also occurred so the Al 2 O 3 presence was also recorded by EELS data [51]. Although alane introduction into CTF-biph and CTF-bipy porosity was confirmed by N 2 sorption isotherms (Figure 18), there was no reversibility in the case where CTF-biph was used as host [51].

TM-Hydrides
While main group metal hydrides are attractive due to metal abundance and low atomic weight of the metal (so higher wt.% H2 storage capacity), some TM (transition metals) have also been recently investigated by employing nanosizing effects (Table 13) [79,97,169,200,212,216,234]. The simplest and most classical model system to study TM-H interaction is the Pd-H system [200,234]. While the gravimetric storage capacity is too low for it to be considered for vehicular applications, the nature of Pd…H interaction has shed new light on thermodynamic predictions in Pd NPs forming PdHx, estimating cluster expansion, phase boundaries Pd/Pd…H, phase transitions (>400 K) and interfacial free energies by using DFT method [200,234]. Pd is often thought of as being able to absorb H2 like a sponge, reversibly absorbing more than 1000 times its own volume. In short, interaction of H2 with palladium comprises of H-H dissociation in atomic [H], diffusion of [H] into Pdbulk, where it occupies the free interstitial sites in fcc lattice of Pd, forming either an α-phase PdHx (x < 0.03, rt) or the hydridic β-phase PdHx (x > 0.03) [200]. The catalytic role of Ph-hydride has been recently harnessed in a complex Pd hydride CaPdH2, for semihydrogenation of CnH2n-2 (alkynes) to CnH2n (alkenes) [79]. Figure 19. The XRD pattern (0.9AlH 3 -0.1Li 3 N) dehydrog (a), the hydrogen release profile under isothermal conditions (100 • C) of (1 − x)AlH 3 -xLi 3 N (x = 0, 0.05, 0.1, 0.15) (b), and the calculated apparent activation energy (c). Reprinted/adapted with permission from Ref. [206]. 2022, Wiley-VCH GmbH. Figure 19b shows the isothermal dehydrogenation of (1 − x)AlH 3 -xLi 3 N (x = 0.05, 0.1, 0.15) at 100 • C, confirming a decrease in H 2 wt.% with the content of Li 3 N. The XRD pattern confirms that the sole dehydrogenation product of the composite is metallic Al (Figure 19a). The onset of dehydrogenation was conveniently reduced to 66.8 • C (0.95AlH 3 -0.05Li 3 N), thus approaching an operating regime suitable for FCEs. The beneficial role of lithium amide was confirmed by the apparent E a which is strongly reduced (Figure 19c) [206].

TM-Hydrides
While main group metal hydrides are attractive due to metal abundance and low atomic weight of the metal (so higher wt.% H 2 storage capacity), some TM (transition metals) have also been recently investigated by employing nanosizing effects (Table 13) [79,97,169,200,212,216,234]. The simplest and most classical model system to study TM-H interaction is the Pd-H system [200,234]. While the gravimetric storage capacity is too low for it to be considered for vehicular applications, the nature of Pd . . . H interaction has shed new light on thermodynamic predictions in Pd NPs forming PdH x , estimating cluster expansion, phase boundaries Pd/Pd . . . H, phase transitions (>400 K) and interfacial free energies by using DFT method [200,234]. Pd is often thought of as being able to absorb H 2 like a sponge, reversibly absorbing more than 1000 times its own volume. In short, interaction of H 2 with palladium comprises of H-H dissociation in atomic [H], diffusion of [H] into Pd bulk , where it occupies the free interstitial sites in fcc lattice of Pd, forming either an α-phase PdH x (x < 0.03, rt) or the hydridic β-phase PdH x (x > 0.03) [200]. The catalytic role of Ph-hydride has been recently harnessed in a complex Pd hydride CaPdH 2 , for semi-hydrogenation of C n H 2n−2 (alkynes) to C n H 2n (alkenes) [79].

Conclusions and Outlook
The urgency of a green, renewable and sustainable fuel to replace fossil fuels is more stringent today than ever. The metal hydrides constitute materials that possess intrinsically high gravimetric and volumetric hydrogen storage capacities, but their sluggish kinetics and poor thermodynamics still constitute an obstacle for the wide acceptance of their use in the fuel of the future. However, various strategies have been recently explored, and perhaps the most returns derive from basic shifts in thinking: oriented growth of MgH2 on catalytically active substrates; size-reduction in metal hydrides to few nm when thermodynamic destabilization works best; or usage of new class of catalysts of 2D-structure (MXenes)-they have all showed unexpectedly good results. There is clearly room for improvement in the fascinating field of metal hydrides, and research efforts ought to concentrate on improving nanoparticle system design, careful consideration of the incorporating matrix and selected hydrogenation/dehydrogenation catalysts, from both an economic and a feasibility point of view. Given the raw material scarcity but also reactivity and particular characteristics of some complex hydrides (like volatility of Al(BH4)3, or extreme toxicity of Be(BH4)2 etc.), the optimal hydrogen storage material will likely be based on magnesium nanoconfined in a carbonaceous host and/or catalyzed by Ti-based The most stable reversible capacity during cycling was achieved for 0.95 MgH 2 −0.05 TiH 2 nanocomposite, which shows fast kinetics and does not fall below 4.8 wt.% even after 20 cycles ( Figure 21). Additionally, no Mg-ETM-H ternary phases were observed [169].

Conclusions and Outlook
The urgency of a green, renewable and sustainable fuel to replace fossil fuels is more stringent today than ever. The metal hydrides constitute materials that possess intrinsically high gravimetric and volumetric hydrogen storage capacities, but their sluggish kinetics and poor thermodynamics still constitute an obstacle for the wide acceptance of their use in the fuel of the future. However, various strategies have been recently explored, and perhaps the most returns derive from basic shifts in thinking: oriented growth of MgH 2 on catalytically active substrates; size-reduction in metal hydrides to few nm when thermodynamic destabilization works best; or usage of new class of catalysts of 2D-structure (MXenes)-they have all showed unexpectedly good results. There is clearly room for improvement in the fascinating field of metal hydrides, and research efforts ought to concentrate on improving nanoparticle system design, careful consideration of the incorporating matrix and selected hydrogenation/dehydrogenation catalysts, from both an economic and a feasibility point of view. Given the raw material scarcity but also reactivity and particular characteristics of some complex hydrides (like volatility of Al(BH 4 ) 3 , or extreme toxicity of Be(BH 4 ) 2 etc.), the optimal hydrogen storage material will likely be based on magnesium nanoconfined in a carbonaceous host and/or catalyzed by Ti-based catalysts (such as TiO 2 , TiO, or MXenes). The realistic application of metal hydride systems is conditioned by a number of factors: (i) the discovery of a material that displays a reliablyreversible behavior in hydrogenation studies; (ii) consistent performance across hundreds of H 2 -absorption/desorption cycles; (iii) lower activation energies and consequently faster absorption/desorption kinetics and improved thermodynamics; (iv) consistently fast kinetics for fast refueling; (v) thermodynamic stability and material integrity to afford safe storage in a fuel tank; (vi) reasonable resistance to air and/or moisture; (vii) synthesis route moderately easy and preferably comprising of few steps; (viii) access to sufficient raw materials and limit amount of CRM (critical raw materials) used; (ix) reliable scaling-up of the lab demonstrator to a multi-KW tank capable to drive a vehicle for 500 km or more; (x) strong safety precautions and technological parameters implementation to afford a tank capable to store, release and withstand high H 2 pressures (of more than 100 atm). Within this framework, the EU directives to limit CRM usage is expected to drive the research towards more-abundant metal sources such as Mg or Al (Mg was also included in the list of CRM from 2020, although currently it can be obtained in enough quantities). Noble metal catalysis (like Pd) will probably not become a commercial way of speeding up hydrogen delivery or the recharging of hydride-based fuels due to the associated cost. Other catalysts like MXenes can be produced on a larger scale, but the Ti-based material could also face soon shortages. Nanoconfinement still offers general improvements across the board for hydride-based materials, but the choice of host is limited-among the classes of hosts presented in the current review, the most promising are carbonaceous frameworks and MOFs. Carbonbased materials can be tailored morphologically for hydride inclusion, and their cost is modest; however, this must be considered with care since a zero-carbon policy might imply soon that carbon should not be used as a host any longer. Even though it releases no CO 2 in the atmosphere; there will be an associated cost with treatment of the end-of-life C-based fuel, and so the carbon footprint will not be negligible.
Considering these material, performance, safety and cost restrictions, the final choice for a viable, sustainable hydride-based material is a delicate one and only validation through a scaling-up proven in an operational environment could confirm whether it can be used on a large-scale tank for vehicular applications and afterwards adopted by industry. The ultimate goal is, without a doubt, to approach as much as possible the reversible, theoretical hydrogen capacity, and this is a joint venture of all the above considerations.