Metal Oxide Additives Incorporated Hydrogen Storage Systems: Formation of In Situ Catalysts and Mechanistic Understanding

Abstract

Hydrogen storage is a critical bottleneck to hydrogen economy. Presently none of the solid-state hydrogen storage materials (metal hydrides) reaches the capacity vs performance target (6.5 wt.% at 85 °C/5-12 bar, 1500 cycles) for the commercialization of light duty H₂ fuel cell vehicles. A few reversible hydrogen storage materials (e.g. MgH₂, LiBH₄/MgH₂ composite) possess adequate capacity, but their performance needs to be improved significantly. Metal oxide additives improve the hydrogen storage performance of metal hydrides, but the additive-hydride reaction mechanism remains not well understood. In this context, the present chapter discusses how various metal oxide additives interact with metal hydrides and facilitate the low temperature de/ab-sorption of hydrogen. The metal oxide additives may either directly catalyze the reaction without making any chemical changes or they catalyze indirectly by making active in situ products. In this chapter, various oxides and hydride combinations of the latter category are analyzed, and factors governing the improved hydrogen ab-/desorption performance are highlighted.

5.1 Introduction

Nanoparticles exhibit significantly improved electronic and surface chemical properties as compared to their bulk counterparts (Kelly et al. 2003; Sapra and Sarma 2004). Gold nanoparticles are the simplest example, for which size-dependent properties offer a wide diversity of potential applications, e.g., electronic circuits, sensors, probes, diagnostics, therapy, and catalysis (Daniel and Astruc 2004; Eustis and El-Sayed 2006; Haruta 2002). For a detailed explanation regarding the synthesis, characterization, and application of nanomaterials, the reader is directed to material-specific, in-depth studies reported in the literature (Bruce et al. 2008; Cao 2004; Chen and Mao 2007; Schwarz et al. 2004). In the present chapter, our interest is metal oxide nanoparticles, especially for catalytic applications. Since surface chemical features such as oxidation/reduction, gas ad-/ab-/desorption, complexation, and ion exchange properties determine the system performance, oxides are naturally the main choice as additives/catalysts and templates/supports in heterogeneous catalysis. There are various specific reasons why oxide surfaces are unique for catalytic applications: (i) the presence of cations and anions in a certain ratio that is relevant for influencing another species existing in chemical proximity, (ii) existence of bonding sites, (iii) existence of surface fields due to the coulombic nature of the ionic lattice, (iv) presence of charged adsorbed species, (v) presence of surface acidity or basicity, (vi) cationic/anionic vacancies, (vii) ability of cations to undergo oxidation/reduction reaction, (viii) ease of surface bridging with functional groups, (ix) high mobility of lattice oxygen and the possibility of oxygen-oxygen hopping, (x) creation of surface oxygen derived free radicals, etc. These properties are relevant for a wide variety of surface/bulk reactions, such as redox, oxidative Mannich reactions, condensation, deprotection, alkylation/hydroxylation, cycloaddition, dehydration, de-/re-hydrogenation, transesterification, organosynthesis, etc. Especially, in the case of metal-hydrogen interaction systems, when high-capacity metal hydrides exhibit thermodynamic and/or kinetic limitations, incorporation of suitable additives yield significantly improved hydrogenation/dehydrogenation performance (Zaluska et al. 2001; Zaluski et al. 1997).

A general universal classification of oxides is given in Fig. 5.1. Note that only transition metal oxides receive significant accolades for catalysis because the existence of partially filled d-shells of the metal ions and oxide ligands influences the neighboring atoms/molecules (Kung 1989). Usually binary/ternary metal oxides are routinely used in heterogeneous catalysis; however, mixed metal oxides, the oxygen-containing combinations of two or more metallic ions in a specific stoichiometry, are also useful additives under various circumstances (Gawande et al. 2012; Yuan et al. 2014). Especially for hydrogen storage systems, a wide variety of binary, ternary, and mixed metal oxides were employed as additives, and significant performance improvements have been reported. In the present chapter, we highlight how various metal oxides interact with the host hydride particles and mediate the reversible gas-solid surface interaction. The present chapter is organized as follows. Initially, by reciting the energy-wave vector/density of states (DOS) relationship for bulk, nano 3D, 2D, 1D, and 0D materials, the quantization effect implied by crystal dimension is highlighted. This section is followed by a brief summary to highlight a few classes of important metal oxides normally used for catalytic applications. Characteristics of oxide additives loaded hydrides belonging to few important categories, such as aluminates, binary hydrides, reactive hydride composites (RHC), and amide/imides, are reviewed. Catalysts existing without any structural/chemical changes throughout the reaction process are routinely reported in the literature. Therefore, in the present chapter, the main focus is on additives that undergo phase/chemical transformation to enhance the recycling performance of hydrogen storage systems. Some recent works performed by the authors in this line specifically on the promising hydrogen storage materials MgH₂ and NaAlH₄ are also highlighted in this respect.

Fig. 5.1

Classification of oxides derived from the metals and non-metals

5.2 Role of Size Effects

In heterogeneous catalysis, the size of the catalyst and the host particles are important factors influencing their interaction. Since sensitive electronic properties such as redox behavior/ionization and conductivity are strongly affected by domain size of the solid matter (particle/crystallite size) (Norris and Bawendi 1996; Zhang et al. 2009), for the understanding of readers, in the following, a brief explanation regarding the size-energy relationship is provided. A free electron existing in a bulk 3D space (crystal structure ) is free to move and no boundary conditions apply on the electron wave function. The energy-wave vector relation in this case is given by the relation,

$$ E=\frac{{\mathbf{\hslash}}^2{k}^2}{2m} $$

(5.1)

As there are no restrictions, wave vector k is continuous and correspondingly energy values are continuous as shown in Fig. 5.2. On the other hand, when an electron is forced to confine in a 3D space, due to the potential barrier (V) existing at the boundaries of confined space “a,” boundary conditions (“±a”) apply. Therefore, the corresponding wave equation becomes solvable only for specific values of k and the derived value of k is given by the relation ${k}_n=n\frac{\pi }{a}$. Note that when “a” is high, the k _n values are small; hence the E vs k profile looks continuous. However, for the small values of “a,” wave vector is bound to exhibit a substantial separation and since the wave vector is related with energy by a square function, the E−k diagram shows an energy quantization effect. The corresponding E−k diagram shown in Fig. 5.2 (top) is a simplest direct illustration that “confinement effect induces a circumstance called quantization.” This quantization, how differs with a nanostructure where special restrictions exist in one or more directions, i.e., 2D, 1D, and quantum dots (OD), can be explained with the help of density of states (DOS) and energy relationships. For a 3D structure (nano), the density of states is given by the relation,

$$ {\mathrm{DOS}}_{3\mathrm{D}}=\frac{8\pi \sqrt{2}}{{\boldsymbol{h}}^3}{m}^{\frac{3}{2}}\sqrt{E} $$

(5.2)

where m is mass of electrons (9.109*10⁻³¹ kg), h is Planck’s constant (6.626 × 10⁻³⁴ J.s), ℏ = h/2π, and E is the energy accountable for the concerned states. Note that the DOS_3D is directly proportional to the square root of energy as shown in the DOS_3D vs energy profile of Fig. 5.2. In this case, although the electrons are confined than in the bulk material due to quantization, they have reasonable freedom to choose the states. On the other hand, when the structure is 2D, further confinement is forced due to the DOS_2D-energy relationship as shown by the expression:

$$ {\mathrm{DOS}}_{2\mathrm{D}}=\frac{4\pi m}{{\boldsymbol{h}}^2} $$

(5.3)

Fig. 5.2

Top: energy-wave vector diagram (E-k diagram) for a typical bulk and nano-3D structure. Bottom: density of states (DOS) vs energy profiles for dimension-controlled solid-state materials. (Adapted with permission from Handelman et al. 2012. Copyright © 2012, Royal Society of Chemistry)

In this case, the DOS is not proportional to energy (or proportional to E ⁰); hence, the DOS vs energy relation follows a step profile as shown in the figure. It means that only at specific energy values a significant number of states available for charge carriers when one degree of freedom is restricted. This situation, unlike the nano 3D case, forces electrons to occupy only certain states, ensuring higher confinement. When the movement of electrons is restricted in two directions, the DOS_1D vs energy relation is given by,

$$ {\mathrm{DOS}}_{1\mathrm{D}}=\frac{2\sqrt{2{m}^{\frac{1}{2}}}}{\sqrt{E}} $$

(5.4)

In this case the DOS is inversely proportional to the square root of energy, which implies a sharp decrease of states after a specific energy value. For the 0D materials, i.e., quantum dots, the DOS_0D is a delta function as given in the following relation:

$$ {\mathrm{DOS}}_{0\mathrm{D}}=2\delta (E) $$

(5.5)

This relationship suggests that only with specific energy values, states can be accessed and everywhere except these is forbidden for electrons, ensuring high degree of confinement. The representation by Eqs. (5.1), (5.2), (5.3), (5.4) and (5.5), thus, provides a brief fundamental quantum mechanical explanation for readers regarding the origin of nanoconfinement. For a more detailed overview of mathematical expressions, the reader is directed to comprehensive quantum mechanical chapters discussed in the literature (Yu and Cardona 1996). In the case of metal oxides, apart from the naturally occurring size-dependent confinement effects (3D, 2D, 1D, or 0D structure), the surface chemical composition also significantly contributes to catalytic activity (Henrich and Cox 1996).

5.3 Metal Oxides for Catalytic Applications

Oxides commonly studied as catalysts/additives belong to the structural classes of corundum, rock salt, wurtzite, spinel, perovskite, rutile, and anatase structures. Table 5.1 summarizes these structure classes with a few examples. In fact there is no rule for the structural choice of oxides for catalytic applications. However, under certain circumstances, certain unique properties of a particular class of oxides look relevant for specific applications, and the catalyst is chosen accordingly.

Table 5.1 Summary of common metal oxides used for catalytic applications (Categorized based upon crystal structure )

Metal oxides are proven to be influential additives for almost every class of metal hydride systems. However, based upon the observations made for one hydride, the catalytic reaction mechanism cannot be generalized for every metal hydride systems. The oxide additive may be generated in situ or externally added, and the oxide may directly or indirectly catalyze the reaction. One interesting example for the in situ generated oxide additive that indirectly influences the metal-hydrogen interaction is lanthanum oxide existing in the LaNi₅ intermetallic/hydride system. The surface studies by spectroscopic methods clearly indicate a strong surface enrichment of oxidized La in LaNi₅ powder. When the intermetallic alloy LaNi₅ is exposed for a short time either in air or hydrogen gas (consisting oxygen impurities), oxidized La and Ni clusters are incorporated in the surface by the following reaction:

$$ 2{\mathrm{La}\mathrm{Ni}}_5+1.5\;{\mathrm{O}}_2\to {\mathrm{La}}_2{\mathrm{O}}_3+10\mathrm{Ni}. $$

(5.6)

A few authors have also identified the presence of La(OH)₃ and Ni clusters (Wallace et al. 1979). In either case, chemisorption of H₂ by Ni clusters was confirmed, as the studies identified a fermi surface at the top of the Ni 3D-derived states (Schlapbach 1981; Weaver et al. 1980). After breaking the H-H bond, the La oxide and Ni cluster interface act as a gateway for the diffusion of hydrogen inside the bulk LaNi₅. The same happens in the reverse direction during hydrogen release (Wallace et al. 1979). This is one of the simplest examples for in situ generated oxide additives and their indirect effect on H sorption performance. In the case of high-capacity complex hydrides and binary hydrides, an external additive needs to be incorporated as hydrogen is strongly bonded with metal atoms. A large number of metal oxide additives and interesting materials aspects have been presented in the literature. In the following sections, we present a few metal oxide additives loaded hydrogen storage systems that are significant for commercial applications.

5.4 Metal Oxide Additives for High-Capacity Hydrogen Storage Systems

A few most important metal oxide additives studied in recent times is summarized in Table 5.2. When applying an oxide additive for improving the performance of a hydride, one may come across any of the following observations:

• (i) The additive remains chemically unchanged and provides consistently better system performance.

• (ii) The additive reduces while mixing with hydride and improves the performance of the system.

• (iii) The additive remains intact in the first cycle (or the first few cycles) but reduces during repeated cycles, improving the performance of the hydride.

• (iv) The additive reduces during cycling runs but degrades the performance of the system.

• (v) The additive gets reduced and makes new in situ catalysts, and the stable in situ catalyst improves the performance of the system for the long-term.

• (vi) The additive makes a new in situ catalyst and provides better system performance, but the performance degrades due to slow compositional changes occurring to the catalyst. Whereas case (i) is straightforward, any of the situations (i)–(vi) can occur when an oxide additive is incorporated with a hydride.

Table 5.2 Summary of a few most successful metal oxide additives studied for important high-capacity hydrogen storage systems

In experimental context, our requirement is to obtain higher system performance, no matter whatever mechanistic category, among (i)–(vi), the additive/catalyst undertakes.

5.4.1 Metal Oxides Loaded Aluminates

Owing to the good H/M ratio, lightweight aluminates (MAlH₄ and M^′(AlH₄)₂, M = Li, Na, K and M^′= Ca, Mg) receive considerable interest for hydrogen storage applications (Orimo et al. 2007). Among all the aluminates, good reversibility is reported only for two aluminates, NaAlH₄ and KAlH₄ (Bogdanović and Sandrock 2011; Morioka et al. 2003; Pukazhselvan et al. 2012, 2015), and their working temperature range is identified to be 180–350 °C. By incorporating additives, better performance can be achieved, but the mechanism of hydrogen release/reabsorption is not well understood (Bogdanović et al. 2007). In order to successfully tailor new promising reversible complex hydrides, it is necessary to understand the exact mechanism of H de-/absorption promoted by additives. Since sodium alanate (NaAlH₄) is the best model system in the category of aluminates, for mechanistic studies, researchers prefer NaAlH₄ over other aluminates (later on the acquired knowledge can be extended to other aluminates). Since aluminum contains only three valence electrons for covalent bonding with hydrogen, the stable existence of Al-H covalently bonded [AlH₄]⁻ tetrahedral complex relies on its ionic bond strength with Na⁺ ion. The catalyst/additive should stimulate the antibonding between Na⁺ and [AlH₄]⁻ ions for the liberation of hydrogen atoms. The circumstance of the additive/catalyst may correspond to any of the cases (i)–(vi) mentioned above (Sect. 5.4), but antibonding will be stimulated only when the additive/catalyst possesses appropriate electronic features for depleting the stabilizing electron from the complex [AlH₄]⁻. It is interesting that, in the literature, additives falling across all the categories (i)–(vi) have been reported for NaAlH₄/LiAlH₄. A few of them are discussed below.

Lee et al. (2007b) performed a comparative study for 10 wt.% lanthanide oxides, La₂O₃, CeO₂, Sm₂O₃, and Gd₂O₃ mixed NaAlH₄ and suggested that all these oxides trigger higher kinetics at lower dehydrogenation temperatures as compared to neat NaAlH₄. XRD investigation suggested that except CeO₂ additive which transforms to cerium hydride, all other oxides remain unchanged. Among these oxides, Sm₂O₃ additive was found to be the best for improving the reaction kinetics , reversibility, and cyclic stability. Xuanhui et al. (Rafi ud et al. 2012) used Nb₂O₅, TiO₂, and Cr₂O₃ nanoparticle additives for NaAlH₄ and found that TiO₂ and Nb₂O₅ are better than Cr₂O₃. The authors observed the reduction of TiO₂ and Nb₂O₅ in the first cycle itself. Xuanhui et al. (Rafi ud et al. 2011) also tested the dehydrogenation behavior of Nb₂O₅ and Cr₂O₃ additives loaded LiAlH₄ and found that both additives improve the desorption of hydrogen at lower temperatures as compared to additive-free samples but Nb₂O₅ is the best among these additives. After ball milling with LiAlH₄ for 30 min, both Nb₂O₅ and Cr₂O₃ additives remain chemically unchanged. However, the powder tested after dehydrogenation measurements suggests that whereas Cr₂O₃ remains chemically unchanged Nb₂O₅ additive reduces during the first cycle dehydrogenation. Apparently, the existence of three different Nb variants, NbO₂, NbH, and LiNbO₃, was identified, but it is not exactly clear which product among these was responsible for catalytic activity. Notwithstanding, the authors suggested that the variable valence states of Nb incorporated in the powder may be a prime reason for the catalytic activity. Although such a claim requires further verification, it is interesting that the reduction of additive and the subsequent formation of multiple in situ additive(s) helps to destabilize the complex hydride. In another study, Li et al. (2013) used NiFe₂O₄ additive for LiAlH₄ and found that the additive interacts with LiAlH₄ during ball milling itself and forms LiFeO₂ and Al-Ni alloy (Al₄Ni₃). During dehydrogenation, these in situ products further interact chemically and produce LiAlO₂ and Al_1.1Ni_0.9 products. The authors suggested that the synergetic effect of these products is the reason behind the superior performance of the system.

In one of the authors’ earlier studies (Pukazhselvan et al. 2010), the hydrogen storage behavior of metal oxide nanoparticle additives such as TiO₂, CeO₂, La₂O₃, Pr₂O₃, Nd₂O₃, Sm₂O₃, Eu₂O₃, and Gd₂O₃ loaded NaAlH₄ was tested, and it was found that TiO₂ is the most effective additive. This observation is in agreement with the findings of Lee et al. (2008) who used 2 mol.% TiO₂ nanopowder for catalyzing NaAlH₄. The thermodynamic calculations of Lee et al. further revealed that titania reacts with NaAlH₄ to form an intermetallic phase TiAl₃ in the mixture. Usually pure NaAlH₄ releases 5.5 wt.% hydrogen through a two-step decomposition reaction, first step at 180 °C and second step at 230 °C, as given by the reaction (5.7):

$$ {3\mathrm{NaAlH}}_4\leftrightarrow {\mathrm{Na}}_3{\mathrm{AlH}}_6+2\mathrm{Al}+{3\mathrm{H}}_2\leftrightarrow 3\mathrm{NaH}+3\mathrm{Al}+4.{5\mathrm{H}}_2. $$

(5.7)

As per the thermodynamic calculations of Lee et al., 2 mol.% TiO₂ reacts with NaAlH₄ by the following reaction:

$$ 0.98{\mathrm{NaAlH}}_4+0.02{\mathrm{TiO}}_2\to 0.85{9\mathrm{NaAlH}}_4+0.07{3\mathrm{Na}}_3{\mathrm{AlH}}_6+0.01{1\mathrm{Al}}_2{\mathrm{O}}_3+0.01{3\mathrm{TiAl}}_3+0.04{4\mathrm{H}}_2 $$

(5.8)

The 1.3 mol.% TiAl₃ derived from 2 mol.% TiO₂ catalyzes the remaining 85.9 mol.% NaAlH₄ and 7.3 mol.% Na₃AlH₆ and lowers the H desorption temperature by at least 30–50 °C while also appreciably increasing the reaction kinetics . The existence of 1.1 mol.% alumina is believed to have no impact as alumina is not a fine catalyst for NaAlH₄ (Ma et al. 2012). Another study that directly employs mechanochemically synthesized TiAl₃ as catalyst proves that TiAl₃ exhibits a pronounced catalytic activity for NaAlH₄ (Lee et al. 2007a). It may herewith be recalled that the well-studied TiCl₃ additive also makes TiAl₃ by the following reaction:

$$ 0.98{\mathrm{NaAlH}}_4+0.02{\mathrm{TiCl}}_3\to 0.91{6\mathrm{NaAlH}}_4+0.02{5\mathrm{Na}}_3{\mathrm{AlH}}_6+0.010\mathrm{NaCl}+0.01{3\mathrm{TiAl}}_3+0.03{6\mathrm{H}}_2 $$

(5.9)

Various forms of Ti-based additives, Ti nanoparticles (Fichtner et al. 2003), chlorides/fluorides (Majzoub and Gross 2003), and oxides/alkoxides (Bogdanović et al. 2003; Pukazhselvan 2012), were tried for NaAlH₄, and formation of in situ Ti_xAl_y intermetallic alloy was identified for many of these cases. The chemical state of reduced titanium and its coordination in the Ti_xAl_y alloy were well studied by various researchers (Bogdanović et al. 2007; Léon et al. 2004), and it is concluded that Ti exists in zerovalent state that does not change by hydrogenation/dehydrogenation cycles. Felderhoff et al. (2004) identified slight structural changes for Ti_xAl_y alloy during charging/discharging cycles, but the zero valence of Ti is always maintained. Following these observations, Pukazh et al. tested a 10 mol.% TiO₂ loaded NaAlH₄ powder and explored further details regarding the reduction of titania and the Ti_xAl_y in situ additive (Pukazhselvan 2012). The results are demonstrated by XRD as shown in Fig. 5.3.

Fig. 5.3

XRD profiles of 10 wt.% titania additive mixed NaAlH₄ tested for various cycles. (a) 10 mol% TiO₂ (25 nm):NaAlH₄ ball milled (1 h) starting material. (b) Re-hydrogenated powder of the 1st cycle dehydrogenated material, (c) 2nd cycle dehydrogenated powder, (d) 6th cycle dehydrogenated (to 1st step), (e) 10th cycle hydrogenated, and (f) 15th cycle hydrogenated. (Reproduced with permission from Pukazhselvan 2012, Copyright © 2012 Hydrogen Energy Publications. Elsevier.

Note* The existence of Ti_xAl_y alloy in the Ti/Ti variant additives loaded NaAlH₄ powder was a chapter of intense debate in the literature. In the early days of research, its presence could not be confirmed due to two reasons: (i) The size of Ti_xAl_y crystallites is too small; hence when a small concentration of additive is used, its XRD peaks get hindered by the background noise (in some cases amorphous Ti_xAl_y is also reported). (ii) The peak position of Ti_xAl_y alloy is closer to the peak position of Al(111) peak; hence it is discernible only as a shoulder to the Al (111) peak (in the Ti_xAl_y alloy family, several alloy combinations show XRD reflections in the range 39–44° 2θ, and among these TiAl₃ is a widely known alloy whose peak position is too close to Al(111) for simple deconvolution. Using higher additive concentrations is therefore recommended for characterization studies; however, for hydrogen storage studies, the additive concentration should be as little as possible)

The existence of anatase titania peak in profile “a” of Fig. 5.3 suggests that although a partial reduction may be possible during mechanical milling, a major portion of the titania additive remains unreacted. Profile “b” suggests that reduction of TiO₂ takes place during the first de-/re-hydrogenation cycle, but a smaller part still remains unreacted at the end of first cycle. The XRD profiles corresponding to cycles 2 and 6 (profiles “c” and “d,” respectively) show no presence of TiO₂. It is clear from these results that titania reduces completely during the second cycle and the reduced Ti combines with Al (Ti_xAl_y phase peak position, 40.8°). Upon cycling the sample further, as revealed by comparing profiles “d,” “e,” and “f,” the formation of another phase of Ti_xAl_y alloy (Schoenitz et al. 2004) is also identified (position 39.9°, right-side shoulder to Al(111) peak. For clarity, this is highlighted in the inset). This observation possibly indicates the transformation of one Ti_xAl_y phase to another during repeated H ab-/desorption cycles. Further studies may provide more clarity for these observations. However, these observations are enough to conclude that in certain cases, metal oxides get reduced and make new in situ products, and the in situ product itself get chemically and/or structurally modified upon catalyzing the reaction. It is therefore a challenge for material scientists to identify the most promising composition for most active catalytic performance and retain the optimum catalyst concentration/structure during cyclic studies. Presently although the actual chemical species (Ti_xAl_y) responsible for catalytic activity is confirmed, further studies are required to know how it influences the ionic Na⁺ − [AlH₄]⁻ and covalent Al-H bonds. Extensive theoretical studies for understanding the electronic structure of Ti_xAl_y will throw further light on the bond breaking/making mechanism of NaAlH₄.

5.4.2 Metal Oxides Loaded Magnesium Hydride

The high capacity reversible binary hydride, MgH₂ (7.6 wt.% and 110 g/L), is another potential solid-state system for hydrogen storage. However, owing to the high enthalpy of formation (ΔH = −76 kJ/mol), MgH₂ requires temperature over 300 °C for the release of hydrogen at 1 bar equilibrium pressure (Pukazhselvan et al. 2012). Extensive research works have, thus, been conducted for enhancing the hydrogen storage performance of MgH₂ through various strategies (Kalidindi and Jagirdar 2009; Pukazhselvan et al. 2014b; Zhao-Karger et al. 2010; Zlotea et al. 2015), e.g., (i) size tailoring, (ii) nanoconfinement, (iii) chemical modification, (iv) additive loading, etc. Nevertheless, considering the context of the present chapter, we restrict our discussion to category “iv,” especially on metal oxide additives. Various metal oxides, transition metal oxides (Jung et al. 2006), rare earth oxides (Sadhasivam et al. 2013), mixed ternary oxide phases (Patah et al. 2009; Rahman et al. 2011), and rock salt oxides (Pukazhselvan et al. 2014a) were used by researchers for improving the dehydrogenation of MgH₂. It is widely agreed that metal oxide additives are the best for MgH₂, but the catalytic reaction mechanism of metal oxides added MgH₂ remains a chapter of intense debate. In this context, by referring some significant results reported in the literature, we attempt to provide a reasonable understanding regarding the mechanistic role of best metal oxide additives for MgH₂.

The formation enthalpy of MgO, −601 kJ/mol, is a much higher value than the formation enthalpy of MgH₂, i.e., −76 kJ/mol. Therefore, naturally the tendency of Mg oxidation is higher when the system contains oxygen impurities. Formation of MgO as a result of oxidation of Mg in the surface of Mg/MgH₂ is believed to be a serious contamination effect that impedes the Mg-H sorption interaction. Therefore, the additive/catalytic selectivity steers two considerations: (i) additives that incorporate surface oxide contaminants need to be avoided, and (ii) additives that offer clean surface and facilitating Mg-H interaction need to be deployed. In order to protect the Mg/MgH₂ surface from oxygen/moisture, some methods have been suggested in the literature, such as embedding MgH₂ on gas-selective polymers (e.g., PMMA) or encapsulating with graphene or graphene oxide layers. These surface protectors prevent the intrusion of larger molecules, such as O₂, N₂, and H₂O, while facilitating the permeation of hydrogen (Jeon et al. 2011; Kim et al. 2013). Recently Cho et al. (2016) have demonstrated a method to laminate Mg nanoparticles by using reduced graphene oxide (rGO) and suggested that it is a good method for preventing surface oxidation. As shown in the pictorial representation (Fig. 5.4a), Cho et al. showed that about 3.3 nm sized Mg nanocrystals can be laminated by large sheets of rGO by complexing GO with bis(cyclopentadienyl)magnesium and then reducing it by treating with lithium naphthalenide solution. As shown in Fig. 5.4b, the rGO-laminated Mg instantly absorbs (200 °C/15 bar H₂ pressure) over 6 wt.% H₂ and desorbs all hydrogen (300 °C/vacuum). Moreover, the results are consistent even after 3 months of exposure in air. The HRTEM images shown in Fig. 5.4c and d, respectively, correspond to rGO-laminated Mg and its hydrogenated counterpart. Figure 5.4e represents the overview of one portion of the rGO-laminated MgH₂ sample. The data shown in Fig. 5.4 clearly reveals that Mg/MgH₂ nanoparticles are well confined and stable over the conditions applied for recycling tests. Based upon DFT studies and Mg K-edge X-ray absorption spectra, Wan et al. (2017) have recently explored that when MgO nanoparticles are sealed by rGO, the outermost atomic layer of Mg gets slightly oxidized and forms a monolayer honeycomb MgO. Whereas the bulk MgO rock salt is known as a potential H diffusion barrier, the monolayer MgO formed as a result of interaction with rGO provides no barrier and in fact works similarly to a clean Mg surface. It is also in agreement with the work of Kobayashi et al. (Kobayashi et al. 1994) that suggested that thin MgO substrates exhibit suitable electronic structure for the chemisorption of hydrogen molecules.

Fig. 5.4

(a) Pictorial representation of Mg laminated by rGO, (b) hydrogen ab-/desorption kinetics profile of Mg crystals laminated by rGO sheets, (c) HRTEM image of as-prepared Mg-rGO laminates, (d) HRTEM image of Mg-rGO laminates after hydrogenation, and (e) overview of one portion of the rGO-laminated MgH₂ sample. (Reproduced from the work of Cho et al. 2016 Copyright © 2016, Springer Nature)

By applying density functional theory, Wan et al. calculated the H₂ dissociation energy profile for clean Mg surfaces and monolayer MgO covered Mg surfaces as demonstrated in Fig. 5.5. As seen, the H₂ dissociation energy is the same value (1.14 eV) for both surfaces. Nonetheless, once the H-H bond is broken, the MgO-layered system relaxes to a much lower energy state, roughly about 1 eV lower than the initial H₂ physisorbed state at a distance of 2.9 Å away from the surface. Note that when a few layers of such honeycomb structured MgO deposits over another, it results to the formation of a bulk rock salt MgO structure, which is well known as hydrogen diffusion barrier. Hence, from the understanding gained from this study, a valid question arises that whether the proven best oxide additives play a role to restrict the growth of the MgO and lead to a circumstance as presented in Fig. 5.5. Moreover, there are also convincing data that supports (Wu et al.) that the energy required to break the dihydrogen bond on the MgO surface varies depending on the sites where H₂ faces the MgO surface (Wu et al. 2009). In another work, Borgschulte et al. (2008) investigated the surface composition and chemical state of ball-milled MgH₂ by XPS and H/D exchange experiments. This study confirms the existence of MgO wrapped surfaces with cationic vacancies having catalytic tendencies. These studies give us two interesting pieces of information: (i) not all the MgO surfaces are inactive for Mg-H interaction, and (b) tuning the chemistry of the MgO surface may be a critical step in the process of catalysis for reducible metal oxide additives loaded MgH₂ system.

Fig. 5.5

Hydrogen molecular dissociation energy profile for clean Mg surface and monolayer MgO covered Mg surface. (Reproduced with permission from Wan et al. 2017. Copyright © 2017, American Chemical Society)

Proceeding with the above information, in the following, we reviewed a few important results reported for various metal oxide additives loaded MgH₂. In some of these studies, the existence of oxidized Mg surfaces was proven, but the catalytic contribution of this oxidized Mg surface was considered insignificant. Hanada et al. (2009) used 1 mol % of transition metal oxide nanoparticle additives such as Nb₂O₅, V₂O₅, and TiO₂ and showed that all these products after dehydrogenation show the coexistence of Mg and MgO without any traces of intact additives. The K-edge XANES spectra revealed the existence of the used additives in partially reduced state. Based upon these observations, the authors concluded that improved hydrogen sorption is due to the formation of reduced oxides. However, in this study it was not detailed how influential was the existence of an oxidized Mg surface. Oelerich et al. (2001b) investigated the hydrogen absorption/desorption behavior of MgH₂ by incorporating a range of metal oxide additives, such as Sc₂O₃, TiO₂, V₂O₅, Cr₂O₃, Mn₂O₃, Fe₃O₄, CuO, Al₂O₃, and SiO₂. By this comparative study, the authors categorized the additives from most effective to the least effective. For example, for hydrogen absorption, a nearly similar effect was observed for TiO₂, V₂O₅, Cr₂O₃, Mn₂O₃, Fe₃O₄, and CuO additives. On the other hand, for dehydrogenation, Fe₃O₄ and V₂O₅ show comparable effects, followed by Mn₂O₃, Cr₂O₃, and TiO₂ additives. It was shown that only 0.2 mol.% of the additive is sufficient for providing fast sorption kinetics. However, details regarding why certain oxides are better than others for MgH₂ were not explored.

Oelerich et al. (2001a) made a comparative hydrogen storage test for MgH₂ using various forms of vanadium additives, such as V, V₂O₅, VN, and VC. This study confirms that vanadium oxide is better than other forms of vanadium additives. Further studies revealed that V₂O₅ gets reduced during processing the powder toward H ab-/desorption measurements. Porcu et al. (2008) observed that the Nb₂O₅ additive reacts with MgH₂ and oxidizes a significant portion of MgH₂. Polanski et al. (Polanski and Bystrzycki 2009) performed comparative hydrogen storage measurements for Cr₂O₃, TiO₂, Fe₃O₄, Fe₂O₃, In₂O₃, and ZnO additives mixed MgH₂ and suggested that Cr₂O₃, TiO₂, Fe₃O₄, and Fe₂O₃ additives are better than indium and zinc oxides for improving the dehydrogenation of MgH₂. Later on, Polanski et al. (2011) used Cr₂O₃ nanoparticle additive for MgH₂, and an ab-/desorption study for 150 cycles revealed that reduction of Cr₂O₃ leads to the formation of Cr and MgO. A substantial capacity reduction was observed due to the loss of Mg as MgO and the consequent microstructural changes. Patah et al. (2009) observed that co-catalyzing MgH₂ by Nb₂O₅ and Cr₂O₃ provides better effect than when these additives were incorporated separately. The authors suggested the generation of diffusion paths as the reason for the result; however, no further information was added how the diffusion pathways are generated by mixing together the Nb₂O₅ and Cr₂O₃ additives. Croston et al. (2010) used titania additive for improving the dehydrogenation of MgH₂ and suggested that titania is reduced by its interaction with MgH₂. Huang et al. (2006) used oxides of iron, such as Fe₂O₃ and Fe₃O₄ as additives for MgH₂ and found that during heating, both iron oxides reduce to pure Fe. For both samples, there were no substantial differences in catalytic activity in terms of decreasing the desorption temperature and increasing the reaction kinetics . However, a slightly higher loss of capacity for Fe₂O₃ additive loaded sample as compared to Fe₃O₄ additive loaded sample was observed. The authors suggested the capacity, especially for MgH₂ + Fe₂O₃ composite, to be a result of the reduction reaction, as more magnesium is lost to MgO in this mixture. In these studies, the role of MgO is ignored, at times referring this as an unintended reaction product and a potential barrier layer incorporated in the system by consuming useful magnesium. Friedrichs et al. (2006c) observed that the Mg/MgH₂ samples exhibit a strong tendency for surface oxidation and an oxide passivation layer, size 3–4 nm, forms instantly in MgH₂/Mg samples upon the availability of oxygen impurities. This study further revealed that once the surface is neatly covered by a thin layer of MgO, further oxidation does not occur. These studies necessitate an in-depth study for exploring the details of oxide-MgH₂ chemical interaction and its role on the catalysis of the system. The information observed through these studies suggest that the MgO layer existing in oxide additives loaded MgH₂ cannot be ignored as a barrier layer as was thought previously.

Among all the metal oxide additives reported in the literature, Nb₂O₅ is the most widely investigated additive for MgH₂. Barkhordarian et al. (2003) revealed that Nb₂O₅ additive loaded MgH₂ at the reaction temperature of 300 °C releases about 7 wt.% hydrogen within 130 s and reabsorbs it within 60 s. Friedrichs et al. (2006a) used 15 nm- and 100 nm-sized Nb₂O₅ additives for MgH₂ and observed that, whereas both sizes are beneficial, smaller particles strongly enhance the kinetics of the reaction. In another study Hanada et al. (2005) showed that in the case of a 1 mol.% Nb₂O₅ admixed material, the required dehydrogenation temperature is reduced by at least 50 °C in the second cycle as compared to the first cycle. Note that the performance improvement during cycling measurements is a sign for the formation of modified in situ additives. Although all these results are in good agreement with each other, regarding mechanistic understanding, considerable anomalies exist in the literature. Aguey-Zinsou et al. (2007) suggested that Nb₂O₅ acts as lubricant, dispersing and/or cracking agent during milling and helping to further reduce MgH₂ particle size, while no chemical changes occur to Nb₂O₅ during cyclic measurements. Further exemplary studies, however, revealed that Nb₂O₅ in fact gets reduced chemically and a significant amount of oxidized Mg gets incorporated in the powder.

Presently it is clear that some of the best metal oxide additives reported for MgH₂, for example, Nb₂O₅ and TiO₂, involve a chemical interaction with MgH₂ to make active in situ catalysts. However, there are contradictory opinions in the literature regarding the exact chemical composition of the reduced in situ catalyst . Ma et al. reported that Nb nanocrystals of size ~10 to 20 nm are the apparent reaction product when ball milling MgH₂/Nb₂O₅ mixture and these nanocrystals may be responsible for the catalytic activity (Ma et al. 2013). Friedrichs et al. (2006a) suggested the formation of oxygen deficient magnesium niobate (MgNb₂O_3.67) in the Nb₂O₅ added MgH₂ samples. However, their further study suggested the composition to be Mg_xNb_0.8(1−x)O (Friedrichs et al. 2007). By an in situ X-ray diffraction study, Nielsen and Jensen identified that Nb dissolved MgO forms during the dehydrogenation of Nb₂O₅ additive loaded MgH₂ (Nielsen and Jensen 2012). Recently Pukazh et al. employed Nb₂O₅ and TiO₂ additives for MgH₂ and showed that metals substituted MgO rock salt products, typified by Mg_xNb_yO_x + y/Mg_xTi_yO_x + y form as a result of chemical interaction between the additives with MgH₂, and this rock salt is responsible for the catalytic activity (Pukazhselvan et al. 2016a, 2017a). Recent theoretical calculations of Sandhya et al. by DFT modeling also support these observations (Pukazhselvan et al. 2017b; Sandhya et al. 2016).

Pukazh et al. made composite mixtures of MgH₂+nNb₂O₅ (n = 0.083, 0.10, 0.123, 0.167, 0.25, 0.50, 1.0, and 1.5) and tested the mechanochemical reaction products for three sets of reaction times, such as 1 h, 10 h, and 30 h (totally 24 samples) (Pukazhselvan et al. 2016a). The best composition out of these 24 samples, i.e., MgH₂+0.167Nb₂O₅, was then studied by varying the reaction time, i.e., 2 min, 5 min, 15 min, 30 min, 45 min, 1 h, 2 h, 5 h, 10 h, 15 h, 20 h, 25 h, and 30 h (Pukazhselvan et al. 2016d). The corresponding XRD profiles of these samples are shown in Fig. 5.6. As seen at the initial stage (15 min to 5 h) of mechanochemical reaction, MgH₂ disappears gradually, and an unidentified phase (indexed by * symbol) forms in minor quantities. During 5–15 h of reaction, dominant existence of a monophase rock salt product with characteristic (111), (200), and (220) reflections can be noticed. Upon further increasing the milling time up to 30 h, a monophase with visibly unchanged XRD features after 25 h, having close crystallographic resemblance with MgO rock salt is observed. Further spectroscopic/microscopic studies suggested that the product existing in 30 h reacted material is actually Nb incorporated MgO, typified by a general formula, Mg_xNb_yO_x + y.. Samples taken from 1 h, 10h, and 30h reacted powders were used as additives (3 wt.%) for MgH₂, and the DSC profiles were recorded. The DSC profiles demonstrated by Fig. 5.7 show a systematic shift toward lower temperatures with a series, 30 h > 10 h > 1 h > no additive. This result shows that the rock salt product Mg_xNb_yO_x+y possesses a significant catalytic behavior for the low-temperature dehydrogenation of MgH₂. Following this study, Pukazh et al. explored a similar observation for titania added MgH₂ through an XRD/in situ XRD study performed for a 10 wt.% titania loaded MgH₂ (Pukazhselvan et al. 2017a). As it can be seen in the in situ XRD shown in Fig. 5.8, reduction of titania resulting to evolution of a new phase closer to the peak position of MgO can be observed (compare the MgO peak position from Fig. 5.6) upon increasing the temperature. Further characterization studies by XRD, HRTEM, and elemental chemical mapping techniques revealed the existence of homogeneously distributed Ti in a crystalline rock salt matrix (Mg_xTi_yO_x + y) which is similar to that observed for the Nb₂O₅ loaded MgH₂ system. Note that MgO and metals dissolved MgO (e.g., Mg_xTi_yO_x + y/ Mg_xNb_yO_x + y) exhibit almost identical crystallographic features; hence its existence is usually overlooked by many researchers. Sandhya et al. (2016) recently revealed that incorporation of Nb in the Mg site of MgO rock salt structure results in an improved electronic behavior and hence is very active for catalytic interaction with MgH₂ and H₂.

Fig. 5.6

Mechanochemically reacted MgH₂+0.167Nb₂O₅ powder for (a) 2 min, (b) 5 min, (c) 15 min, (d) 30 min, (e) 45 min, (f) 1 h, (g) 2 h, (h) 5 h, (i) 10 h, (j) 15 h, (k) 20 h, (l) 25 h, and (m) 30 h and (n) standard MgO rock salt. (Reproduced with permission from Pukazhselvan et al. 2016d, Copyright © 2015, John Wiley and Sons)

Fig. 5.7

DSC profiles of MgH₂ added with a 3 wt.% of 1h, 10h and 30h reacted additive (MgH₂+0.167Nb₂O₅). Symbols: (▵) as-received MgH₂; (♦)1 h reacted additive powder mixed with MgH₂; (▿)10 h reacted additive powder mixed with MgH₂; (◊)30 h reacted additive powder mixed with MgH₂. (Reproduced with permission from Pukazhselvan et al. 2016a. Copyright © 2015 Hydrogen Energy Publications, LLC. Elsevier)

Fig. 5.8

In situ XRD patterns of 10 wt.% TiO₂ added MgH₂ (5 h milling). Inset shows the XRD of the re-hydrogenated sample. (Reproduced with permission from Pukazhselvan et al. 2017a, Copyright © 2017 Elsevier)

From these understandings, a surface reaction model shown in Fig. 5.9 (Pukazhselvan et al. 2016c) seems to be a logically valid model for Nb₂O₅ loaded MgH₂. As shown in Fig. 5.9a, the commercial Mg/MgH₂ samples usually contain thin MgO rock salt layers in the surface. When the surface MgO rock salt layer is thick as shown in Fig. 5.9b, the system requires high operating conditions because of the high diffusion barrier caused by MgO. On the other hand, when Nb replaces few Mg spots (i.e., Mg_xNb_yO_x + y), the rock salt layer becomes catalytically active because of modifications in its electronic structure. Earlier in Fig. 5.6, it was shown that Mg_xNb_yO_x + y forms from the reaction between MgH₂ and Nb₂O₅ additive. Hence the MgH₂ particles sealed by Mg_xNb_yO_x + y shells as shown in Fig. 5.9c seems to be apparently a valid model for Nb₂O₅ loaded MgH₂. Note that the insertion of Nb, having a slightly higher ionic size than Mg, may also make fractures in the rock salt lattice. Moreover, since XPS study (Pukazhselvan et al. 2016b) proves the existence of a small amount of Nb-O secondary phase(s) other than Mg_xNb_yO_x + y, the presence of NbO_x/Mg_xNb_yO_x + y/Nb interface cannot be ruled out. The NbO_x/Mg_xNb_yO_x + y/Nb interface, fractures, and the modified electronic structure of rock salt phase may all contribute the high H₂ diffusion/surface interaction in the system. A widely accepted similar mechanism was earlier proposed by Wallace et al. for LaNi₅ intermetallic hydride (Wallace et al. 1979).

Fig. 5.9

Schematic representation for enhanced diffusion of hydrogen across the MgO shells, (a) thin MgO shell in commercial MgH₂ samples, (b) surface passivated Mg/MgH₂ particles due to thick MgO shells, and (c) active Nb dissolved MgO surface/interface layer. (Reproduced with permission from Pukazhselvan et al. 2016c. Copyright © 2016 Hydrogen Energy Publications. Elsevier)

5.4.3 Reactive Hydride Composites, Amide/Imides, and Other High-Capacity Systems

An oxide composition with a specific stoichiometry may be catalytically active for various metal hydride systems, but the mechanism of dehydrogenation through additive-host interaction is not necessarily the same for each system. For example, LiTi₂O₄ additive is found to be effective for improving the dehydrogenation of both MgH₂, LiAlH₄ and LiNH₂ + LiH systems, but the mechanism is believed to be substantially different for each system (Zhang et al. 2014b). It was reported by Zheng et al. that the crystal structure of LiTi₂O₄ additive play a key role in enhancing the H ab-/desorption kinetics of LiNH₂ + LiH hydrogen storage system. Since the diffusion rate of Li cations is high (108 cm²/s) for LiTi₂O₄ spinal structure (Johnson 1964), its admixing with LiNH₂ + LiH is expected to assist the overall cationic mobility (Li⁺) under lower temperatures. This “mobile” effect may apparently influence the hydrogen transport within the solid-state admixture. The authors did not comment whether the same holds true for LiAlH₄ and MgH₂. Particularly for MgH₂, since it is categorically a different hydride, further studies are necessary to ascertain whether the cationic mobility within the crystal structure of the additive provide bond breaking/making effects.

Chemical transformation of additive was suggested to be a main cause of catalysis in some cases. Recently, Puszkiel et al. (2017) employed lithiated titania (Li_xTiO₂) as additive for the promising reactive hydride composite (RHC), 2LiBH₄ + MgH₂. In this case, the additive plays a key role in “reaction route modification” that results to equilibrium condition at shorter reaction time. Note that LiBH₄ is an irreversible complex hydride that releases hydrogen through the following reaction (ΔH=75 kJ/ mol H₂):

$$ {\mathrm{LiBH}}_4\to \mathrm{LiH}+\mathrm{B}+1.{5\mathrm{H}}_2 $$

(5.10)

In order to restore LiBH₄, the LiH + B mixture requires to be placed under a hydrogen atmosphere at 600 °C and 350 bar pressure. Hence, LiBH₄ is generally categorized as an irreversible hydride. However, when LiBH₄ is reacted with MgH₂ in 2:1 molar ratio, a reversible reaction occurs as shown in (5.11):

$$ {2\mathrm{LiBH}}_4+{\mathrm{MgH}}_2\to 2\mathrm{LiH}+{\mathrm{MgB}}_2+{4\mathrm{H}}_2 $$

(5.11)

This RHC reaction exhibits enthalpy 46 kJ/mol.H₂ with a theoretical hydrogen capacity of 11.4 wt%, which is 29 kJ/mol.H₂ lower in comparison with pure LiBH₄. However, under standard reaction conditions with a few bars of overpressure, instead of evolving to MgB₂, the decomposition of LiBH₄ and MgH₂ occurs separately with the imminent formation of unintended intermediate products. For example, at 400–450 °C with an overpressure of 3–5 bar pressure, initially, MgH₂ decomposes to Mg and H₂. LiBH₄ proceeds with an unintended slow kinetic intermediate reaction (5.12) before ending with reaction products as given in (5.11),

$$ {2\mathrm{LiBH}}_4\to \left(1/6\right){\mathrm{Li}}_2{\mathrm{B}}_{12}{\mathrm{H}}_{12}+\left(5/3\right)\mathrm{LiH}+\left(13/6\right){\mathrm{H}}_2 $$

(5.12)

A few researchers suggest that the Li₂B₁₂H₁₂ formation can be suppressed by making the reaction at pressure higher than 10 bar. On the other hand, lithiated TiO₂ (Li_xTiO₂) additive loaded RHC proceeds to reaction (5.11) at low back pressures (3 bar) without any signs of Li₂B₁₂H₁₂ intermediate in the reaction. An experimental evidence indicating the modification of reaction route when adding lithiated titania additive with LiBH₄/MgB₂ RHC is demonstrated by Fig. 5.10. The profiles (a) and (b), respectively, show the isothermal dehydrogenation kinetics recorded at 425 °C at 3 bar H₂ pressure (temperature program: heating with 10 °C/min to the target temperature). As seen, in the RHC sample that contains no additive, among the total capacity (11.4 wt.%, normalized in the figure), about 60–70% is liberated in the first step with fast kinetics, whereas the remaining hydrogen releases with very slow kinetics. On the other hand, in the lithiated titania loaded RHC sample, among the total hydrogen capacity (8 wt.%, normalized in the figure), although about a 20% is released during heating the sample (probably due to a premature release from MgH₂ but is usually expected), comparatively a much better kinetics can be achieved. When samples taken from two different stages of desorption (marked in the figure) were tested by Raman spectroscopy, the existence of Li₂B₁₂H₁₂ complex can be proved only in the additive-free RHC sample (taken at state 2 of the sample). The Raman spectral data shown in Fig. 5.10 further reveals that transformation of boron atoms from [BH₄]^- complex to make MgB₂ occurs only in the additive loaded sample under identical experimental conditions. Further studies reveal that when a 5 mol.% lithiated titania (Li_0.59TiO₂) admixed 2LiBH₄ + MgH₂ is dehydrogenated, the additive initially reacts with LiBH₄ as shown in (5.13),

$$ {\mathrm{Li}\mathrm{BH}}_4+{\mathrm{Li}}_{0.59}{\mathrm{TiO}}_2\to {\mathrm{Li}\mathrm{TiO}}_2+0.59\mathrm{LiH}+\mathrm{B}+1.70{5\mathrm{H}}_2 $$

(5.13)

Fig. 5.10

(a) Isothermal dehydrogenation kinetics at 425 °C/3 bar H₂ pressure for 2LiBH₄+MgH₂ sample, (b) kinetics at 425 °C/3 bar H₂ pressure for 2LiBH₄ + MgH₂ + 0.5Li_xTiO₂ sample, (c) Raman spectra of both samples “a” and “b” taken from two reaction stages (marked as point 1/point 2 in both samples). (Adopted with permission from Puszkiel et al. 2017 Copyright © 2017, Royal Society of Chemistry)

Subsequently, formation of MgB₂ seeds occurs from the partly decomposed Mg and B, and it defies the formation of Li₂B₁₂H₁₂ intermediate. The MgB₂ seeds provide coherent interfaces to promote the heterogeneous nucleation and the further growth of hexagonal MgB₂. Interestingly, apart from contributing the formation of MgB₂, the additive also contributes for the formation of LiH through reaction (5.14).

$$ {\mathrm{Li}\mathrm{TiO}}_2+0.20{5\mathrm{H}}_2\to {\mathrm{Li}}_{0.59}{\mathrm{TiO}}_2+0.41\mathrm{LiH}. $$

(5.14)

When the MgB₂+LiH mixture is hydrogenated, formation of LiTiO₂ occurs again. The entire reaction system in the above experiments suggest that when Li_0.59TiO₂ additive was incorporated with the 2LiBH₄+MgH₂ composite, the additive undergoes a chemical transformation and reversibly mediates the entire reaction by suppressing the undesired reaction products. Note that there were metals, chlorides, fluorides, and various other classes of materials tested for 2LiBH₄+MgH₂ and it would be interesting to know how different the reaction mechanism of those systems are in contrast with the lithiated titania. Extensive theoretical investigations focused to extracting the transition states and the corresponding energetics will give interesting clues to understand the generalized mechanism of RHC reaction (5.11) improved by additives.

In another work, Fan et al. (2008) used Nb₂O₅ additive for catalyzing the RHC reaction (5.11) and observed that Nb₂O₅ transforms to NbH₂, initially during ball milling the LiBH₄/MgH₂ with Nb₂O₅ additive and subsequently by the dehydrogenation experiment. In this case a temperature reduction of at least 50 °C with significantly improved kinetics was reported, and there was no identification of intermediate phase(s) during the desorption measurements. Moreover, the product forms the RHC again, and the NbH₂ remains without further chemical modification. It obviously suggests that the way Nb₂O₅ catalyzes the RHC is different as compared to lithiated titania discussed above, but the role seems to be suppressing the intermediate phase(s). Reports in the literature suggest that LiBH₄ itself can be separately catalyzed by metal oxide additives for low-temperature hydrogen release. In this case the catalytic reaction is found to be redox-type chemical reaction, as depicted by LiBH₄+MO_x → LiMO_x+B+2H₂ (Yu et al. 2009). Nonetheless, the reaction is not reversible, and addition of MgH₂ is necessary for driving reversibility.

Another distinct example regarding the superior role of metal oxides as additives is the catalytically enhanced hydrogen evolution reaction in ammonia borane (AB) system (Bluhm et al. 2006). Ammonia borane is known as an irreversible high-capacity hydrogen storage system (hydrogen capacity, 20 wt.%). In order to release hydrogen from AB, one has to either thermally decompose it (thermolysis) or react it with water (hydrolysis). In the process of thermolysis, hydrogen is liberated in three steps at temperatures 100 °C, 150 °C, and 500 °C. By hydrolysis on the other hand, desorption can be achieved at lower temperatures, but the reaction kinetics is poor (<1 wt.%/day) due to high activation barriers. Lapin and D’yankova (Lapin and D’yankova 2013) showed that oxides of certain metals, especially Co₃O₄ substantially decrease the activation energy barrier from 100 kJ/mol.H₂ to 48 kJ/mol.H₂. The authors tested a number of metal oxide additives and noted that the catalytic activity of ZnO, CdO, TiO₂, MnO₂, Cr₂O₃, and WО₃ is negligible whereas Cu₂O, CuO, MoO₃, V₂O₅, Co₃O₄, and Fe₂O₃.nH₂O yield moderate improvement. These observations suggest that an active oxide additive known for one metal hydride may be passive for another metal hydride .

Lithium amide/imide is another promising hydrogen storage system for which metal oxides were proven to be effective for improving the reaction kinetics (Zhang et al. 2014a). Chen et al. was the first to report that lithium nitride (Li₃N) absorbs 10.4 wt.% of hydrogen reversibly (Chen et al. 2002) through a two-step reaction as shown in (5.15) and (5.16).

$$ {\mathrm{Li}}_3\mathrm{N}+{\mathrm{H}}_2\leftrightarrow {\mathrm{Li}}_2\mathrm{NH}+\mathrm{LiH} $$

(5.15)

$$ {\mathrm{Li}}_2\mathrm{NH}+{\mathrm{H}}_2\leftrightarrow {\mathrm{Li}\mathrm{NH}}_2+\mathrm{LiH} $$

(5.16)

The enthalpy of these two reactions, respectively, −165 kJ/mol.H₂ and −44.55 kJ/mol.H₂. Due to the low enthalpy, only the second reaction (capacity, 6.5 wt.%) is considered viable for hydrogen storage applications. However, one of the important issues is that usually LiNH₂ releases ammonia at 300 °C by following reaction (5.17).

$$ {2\mathrm{LiNH}}_2\to {\mathrm{Li}}_2\mathrm{NH}+{\mathrm{NH}}_3 $$

(5.17)

Lchikawa et al. (Ichikawa et al. 2004) showed that the release of NH₃ depends on how intimately LiH binds with LiNH₂. The poorly bound powder releases ammonia as per reaction (5.17), whereas the closely bound mixture releases hydrogen as per reaction (5.16). Therefore, one of the important roles of additive in this case is believed to be shielding the reactants so as to offer a better surface contact.

Later developments in lithium amide hydrogen storage system suggested that Li can be substituted by Mg and better performance and higher capacity can be achieved (Janot et al. 2007; Li et al. 2014). For further understanding regarding the hydrogen storage aspects of amide-imide systems, the reader is directed to references (David et al. 2007; Miwa et al. 2005). The work of Anton et al. (2011) on MgH₂ + LiNH₂ hydrogen storage system identified that by incorporation of defects through ball milling, the release of ammonia can be controlled and better reaction kinetics can be achieved. When the additives V₂O₅ and Fe₂O₃ were employed, the authors observed controlled release of ammonia and faster kinetics. From these observations, it was concluded that the role of oxide additives is to incorporate more defects in the system. Yuan et al. (2012) employed various metal oxide nanoparticle additives such as CeO₂, TiO₂, Fe₂O₃, Co₃O₄, and SiO₂ for improving the hydrogen storage performance of another mixed hydride system LiBH₄/2LiNH₂/MgH₂. Based upon the improved desorption performance, they have categorized the catalytic activity with a series, Fe₂O₃ > Co₃O₄ > CeO₂ > TiO₂ > SiO₂.

Numerous examples are available in the literature regarding the superior effect of metal oxides as additives for many other hydrogen storage materials. For further information, the reader is directed to material specific journal articles regularly updated in the literature.

5.5 Summary and Outlook

The above discussions suggest that the best metal oxide additive known for one metal hydride is not necessarily a best for another hydride and also the additive-hydride interaction cannot be generalized. Likewise, for a particular metal hydride, many oxide additives may be proven active, but the interaction need not remain the same. However, since bond breaking/making is the end result of each catalytic reaction, catalytically enhanced common bonding/antibonding mechanism do exist, and it should be well understood for optimizing any system for commercial applications. In this connection, for understanding the catalytic mechanism, the details regarding how a hydride chemically interacts with various oxide additives should be well studied. In the literature, by performing comparative studies, researchers categorize the effectiveness of oxide additives from most effective to least effective. On many occasions, whereas extensive characterization studies are performed on the most effective oxide additive, the least effective oxides are ignored without further scrutiny. Nonetheless, it is also necessary to identify what makes an additive least effective for formulating a general catalytic mechanism for the system. Moreover, note that any comparison of additives made in the literature without taking into account the size effect may not be accurate. There are materials known to be less active in bulk scale that become very active when scaled down to nano-order. This suggests that any comparative studies reported in the literature with no reference to size effects may have erroneous conclusions. As we decrease the size of the particles, the number of atoms in direct chemical contact with the surrounding species increases significantly. Apart from this, in the case of metal oxides, the exposed atoms in the surface usually do not reflect the correct stoichiometry of the compound as the surface cation/anion ratio is significantly different as compared to the overall ionic ratio and bulk stoichiometry. The distinct valence states and vacancies existing at the surface of oxide nanoparticles substantially influence their chemical proximity as compared to the bulk structure. When two non-reducible oxide samples with the same size and chemical composition but exhibiting different crystal structure are used as additives for a hydride, the result may not remain the same, because the surface cation/anion ratio is different for different crystal structures. Considering all these aspects, in order to understand the mechanism of a catalytic reaction, it is necessary to correctly test the interaction of various metal oxides with a hydride and understand what common observation connects all these oxides. Such a study will be helpful for making a generalized mechanism for understanding catalytically improved reactions. In this connection, the interesting literature data supports compiled in this chapter may guide readers for understanding the various possibilities of additive-hydride interaction.

5.6 Conclusions

The additives that catalyze a metal hydride by making in situ catalysts through a chemical reaction with the hydride are the most suitable for high enthalpy hydrides. Due to the formation of in situ catalyst, tuning the thermodynamics of the hydride is likely, at least a small extent. Maintaining the stability of the in situ catalysts is one challenge presently, but further research in this direction will bring revolutions for sustainable energy development.