Missing pieces of the puzzle or about some unresolved issues in solid state chemistry of alkali metal aluminohydrides

The paper discusses some unresolved issues in the solid state chemistry of alkali metal aluminohydrides (alanates) relevant to high-capacity hydrogen storage.

Analysis of experimental data available in chemical and materials science literature suggests a one-step mechanism for the thermal decomposition of both pure and Ti-doped aluminohydrides.

Most likely, the presence of a titanium hydride phase in the catalyst is responsible for the catalytic effect of Ti-additives.

Furthermore, ball-milling promotes chemical and phase transformations of solid alanates by enhancing mass transfer in the material and creating high-pressure spots where pressure-driven chemical reactions can take place.


For almost a century, aluminum hydride and complex alkali metal aluminohydrides (also called alanates) have been known as powerful reducing agents, propellants, and precursors for aluminum coatings.1

From time to time, they have also been evaluated as possible hydrogen storage media but removed from the list of potential candidates due to irreversibility of pure hydrides.2

However, the situation changed when it became clear that conventional hydrogen storage systems are unable to provide storage capacities in excess of 6 wt.% of H2 and operate around 80–100 °C, i.e. at the working temperatures of modern Proton Exchange Membrane (PEM) fuel cells.3,4

In 1997, B. Bogdanovic and M. Schwickardi5 used the ability of alkali metal hydrides to react with aluminum and hydrogen in the presence of titanium additives6,7 and developed a new type of reversible hydrogen storage media that were operational at moderate hydrogen pressures and temperatures.2NaH + 2Al + 3H2 → 2NaAlH4 (Ti catalyst)In the years that followed, Ti-doped alanates have developed into materials capable of reversibly storing at least 4% of hydrogen by the weight below 125 °C.8

Also, the decomposition temperature of lithium aluminohydride has been reduced to room temperature by combining a metal additive (TiCl4) with mechanical processing—an approach that allowed tuning of hydrogen release from alanate-based materials.9,10

Despite the clear advances of the last years, the demonstrated hydrogen storage capacities of metal-doped alanates remain unsatisfactorily low, which explains the recent pessimism regarding the target number of 6+ wt.% of H2 in NaAlH4-based systems.3,10,11

In addition, solid state events occurring in alkali metal aluminohydrides remain poorly understood10,12,15 so that the puzzle portraying their conduct in the condensed phase still misses a number of crucial pieces.

In the paper below, we will briefly review three major issues in the solid state chemistry of aluminohydrides—thermal decomposition, solid state transformations in the presence of titanium salts, and mechanically induced solid state reactions—and propose a non-conventional interpretation of known experimental facts, which may help in locating those missing pieces.

Simultaneously, this paper is intended to attract the attention of the research community to the ideas which have been left out of mainstream research during the last decade.

Thermal decomposition of alkali metal aluminohydrides

Since the early 1970s the thermal decomposition of alkali metal aluminohydrides MAlH4, where M = Li, Na, K or Cs, has been described as a multi-step process, which begins with melting of the aluminohydride and proceeds through an intermediate formation of an alkali metal hexahydroaluminate, M3AlH6 to the metal hydride (MH), aluminum and hydrogen [eqns.

(1)–(3)]13.MAlH4(s) → MAlH4(liq)3MAlH4(liq) → M3AlH6(s) + 2Al(s) + 3H2(g)2M3AlH6(s) → 6MH (s) + 2Al(s) + 3H2(g)Since the transformation of the tetrahedral [AlH4] ion into the octahedral [AlH6]3− ion is associated with complex structural re-arrangements, melting of hydrides is believed to be a vital step in the decomposition process.14

Although very attractive, this mechanism has never been unambiguously confirmed.

It was formulated by Dilts and Ashby based on the results of thermal gravimetric (TGA) and differential thermal analyses (DTA) (Table 1).

However, both techniques were unable to provide information about chemical and phase composition of the sample.13

Furthermore, TGA and gas-volumetric (GV) measurements reported later turned out to be surprisingly ‘inaccurate’, departing from theoretically predicted numbers on hydrogen evolution by 10% and more.10,13–15,23

Also, recent in situ X-ray and neutron diffraction studies have merely confirmed the presence of a M3AlH6 phase in decomposing samples without addressing their exact composition.15,16

The later is extremely important, however, since deviations in the content of M3AlH6 in the decomposing sample from the values predicted by eqn. (2) clearly indicate shortcomings of the theory.

Finally, traditional mechanism completely failed to explain how and why metal-doped lithium and sodium aluminohydrides can quickly decompose at temperatures much lower than their melting points.5,8–10,18,19

It is no surprise, therefore, that when quantitative analytical techniques, such as the solid state nuclear magnetic resonance spectroscopy, were finally applied to the studies of alkali metal alanates, the inconsistency of eqns. (1)–(3) became quite obvious.

Thus, investigation of solid lithium aluminohydride using the solid state 27Al nuclear magnetic resonance spectroscopy (the solid state 27Al NMR)18–20 revealed a gradual thermal decomposition of the pure LiAlH4 in the solid state without melting (Fig. 1).

This observation also agrees with neutron diffraction studies performed independently.16

The same investigation has also showed that actual concentration of Li3AlH6 phase in the sample during solid state decomposition is considerably lower than is predicted by eqn. (2).

Remarkably, a very similar observation was also made when melted LiAlH4 was investigated using a liquid-state27Al NMR spectroscopy.21

An interesting fact has been reported by Grochala and Edwards who tried to correlate decomposition temperatures (Tdec) of Li, Na and K aluminohydrides with their ΔH°dec.10

Unexpectedly, researchers failed to find a reasonable fit for ΔH°dec derived from eqn. (2).

On the contrary, Tdec and ΔH°dec perfectly agreed with one another when ΔH°dec were obtained from eqn. (4) (Table 2), that is, from the equation representing an alternative route of the thermal decomposition of alkali metal aluminohydrides.2MAlH4(s) → 2MH (s) + 2Al(s) + 3H2(g)MAlH4(s) + 2MH (s) → M3AlH6(s)M = alkali metalIt is worth noting that Dilts and Ashby13 also considered eqns. (4) and (5) as possible scenario for the thermal decomposition of alkali metal aluminohydrides.

They even presented some experimental evidence supporting decomposition of LiAlH4 directly to LiH and Al, but finally went for an alternative model.

The reason for their decision was that solid state reaction (5) looked unlikely at that time.13

Another mechanism of the thermal decomposition of alkali metal aluminohydrides can be found in the recent paper by Walters et al.,17 who believe that thermal transformations of sodium aluminohydride begin with its dissociation into sodium hydride and aluminum hydride.

Further, NaH reacts with starting NaAlH4, giving rise to Na3AlH6; AlH3 decomposes into Al and H2 [eqns.

(6)–(11)].MAlH4 → MH + AlH3NaH + NaAlH4 → [Na2AlH5]NaH + [Na2AlH5] → Na3AlH6Na3AlH6 → 3NaH + AlH3AlH3 → Al + 3H2H → H2Indeed, AlH3 is thermally sensitive and quickly decomposes into Al and H2 between 110 and 150 °C making the sequence eqns. (6)–(10) feasible.10,22

On the other hand, AlH3 has never been detected in alkali metal aluminohydrides, neither by scattering techniques nor by NMR, nor by any other direct analytical technique.

Thus, if aluminum hydride forms during the decomposition process, it should be a part of the transition/intermediate state rather than an independent constituent.

The experimental data collected to date are completely consistent with the thermal decomposition of alkali metal aluminohydrides in the solid state.

Any discrepancies between theory and experiment disappear if the decomposition process is described as a one-step event [eqn.

(4)] without intermediate formation of alkali metal hexahydroaluminates.

However, a hexahydroaluminate can form in the decomposing material according to eqn. (5) in amounts contingent upon the rates of all competing processes.

The concentration of M3AlH6 in the sample may even agree with eqn. (2) if reaction (5) is much faster than reaction (4), that is, when hydride MH is consumed faster than MAlH4 decomposes.

Otherwise, the amounts of hydrogen released on separate decomposition stages would deviate from those predicted by eqns. (2) and (3).

Finally, when decomposition of MAlH4 is tracked by DTA/TGA or gas volumetric techniques, M3AlH6 forming in the side-reaction (5) collapses at elevated temperatures [eqn.

(3)] thus creating a perfect illusion of the multi-step process.

An assumption10,20 that the thermal decomposition of alkali metal aluminohydrides is a one-step process leads to several conclusions which may considerably influence future research in the area.

According to the fundamental principle of micro-reversibility of chemical reactions, the forward reaction and the backward reaction pass through the same activated state.25

Therefore, in an ideal case, the re-charging of an alanate-based material could follow the path reverse to its decomposition, that is, re-charging of a decomposed alanate should produce MAlH4 directly from MH, Al and H2.

In reality, however, solid state reactions such as transformations of metal hydrides involve a set of elementary processes, which are located in various zones of the solid; for instance, adsorption, surface reactions, diffusions and some other processes.

An example is shown in Fig. 2, where the solid state reaction AH → A + H (gas) proceeds differently from the reverse process.

The molecules of gas H are desorbed from the surface of AH in the case of the direct reaction (also through cracks in the layer of A formed) while they must be absorbed on the phase AH and have to penetrate it before they reach the surface of A in the reverse process.

On the other hand, this controversy can be resolved once the layer of AH is removed from the surface of A by means of a mechanical treatment such as ball-milling.

Second, the formation of alkali metal hexahydroaluminates during hydrogen absorption and desorption cycles should be attributed to the reverse reaction (3) and to the reaction (5), i.e. side-processes, which reduce effective storage capacities of hydrogen absorbers.

Since alkali metal hexahydroaluminates are thermodynamically more stable than corresponding aluminohydrides, they can build up in the material during charge–discharge cycles—a phenomenon recently reported by Srinivasan et al.26

Finally, the maximal storage capacity of an alanate-based material can only be reached if the reversible reaction (4) becomes dominant over any other process.

This may be achieved using an appropriate treatment and a catalyst that selectively enhances reaction rates in eqn. (4) and decelerates other adverse processes.

Indeed, a number of transition metals promote solid state decomposition of complex aluminohydrides; titanium is the most efficient among the known catalysts3–5,7–12,18,23,26,27.

Catalytic transformations of alkali metal aluminohydrides and the nature of the metal catalyst

Since the late 1990s, Ti-doped alkali metal aluminohydrides have attracted considerable attention as potential ultra-high capacity hydrogen storage media.3,4,8,10,11,19,27

Currently, a substantial part of hydrogen storage research is dedicated to titanium-doped lithium and sodium alanates and the number of publications dealing with these hydrogen absorbers is constantly growing.10,27

Thus, it has been found that TiCl4, TiCl3 and some other Ti derivatives catalyze hydrogen release and uptake by sodium and lithium alanates and considerably reduce their decomposition temperatures.5,8–12,15,18,19

In addition, it turned out that Ti-doped NaAlH4 can reversibly store about 4 wt.% of H2 and maintain its storage capacity upon prolonged cycling.11,26,27

Furthermore, when metal catalysis is combined with mechanical processing, the decomposition of alkali metal alanates becomes tunable.10,19

Those and other related studies published recently have been summarized in several reviews,3,4,8,10,11,19,27 which can be looked up for further details.

Here, we would like to address only one issue that became a source of continuous controversies; namely, the nature of chemical substances responsible for catalytic effect of titanium in alkali metal aluminohydrides.

Almost all research groups active in the field, tried to resolve this problem.3,4,8,10,11,19,27

Several guesses regarding the nature of the Ti-catalyst had been made15,28,29 and rejected.30,31

As a result, deficiency of constructive knowledge was admitted on several occasions.4,10,12,15

In 2000, one of us reported the decomposition of LiAlH4 at room temperature upon ball-milling in the presence of a catalytic amount of TiCl4.9

Further experiments revealed that TiCl4 and LiAlH4 react under mechanical activation, producing microcrystalline LiCl, Al and an Al3Ti alloy.18,19

Assuming that Al3Ti might be responsible for the catalytic effect of TiCl4, we proposed the mechanism of Ti-catalyzed transformations in solid LiAH4:TiCl4 + 4LiAlH4 → Ti + 4LiCl + 4Al + 8H2Ti + 4Al → Al3Ti + Al2LiH + LiAlH4 → Li3AH6Subsequently, other researchers confirmed our hypothesis by repeating some of our experiments, and finding Al3Ti in the Ti-doped sodium alanate.12,27,30,32–35

For a while, it appeared that Ti-catalyzed transformations of Li/NaAlH4 were finally understood.

Unfortunately, additional studies revealed poor catalytic activity of both the tetragonal DO22-type Al3Ti alloy and the meta-stable cubic L12 phase forming upon its ball-milling.18,36

The behavior of Al3Ti in alkali metal alanates makes little sense unless one assumes the presence of another phase in the catalytically active material.

Should Al3Ti synergically enhance the activity of this phase, the amount of hydride sufficient for generating the catalytic effect might fall below the detection limits of conventional analytical techniques, so that the careful consideration of indirect evidence would be necessary to fully assess the process.

Fortunately, a long history of aluminohydride-related chemical research and extensive studies into the Al–Ti system have created experimental knowledge that is quite sufficient for developing a credible hypothesis.

Attempts to prepare transition metal aluminohydrides of a general formula M(AlH4)n, where M is a n-valent transition metal, have been made a number of times.1,37

They also included reactions between LiAlH4 and TiCl4 or TiBr4, which produced titanium alanate, Ti(AlH4)4, stable below −80 °C.

Between −80 and 20 °C, the material releases up to 70% of its entire hydrogen content leaving behind titanium hydride (TiH2), titanium and aluminum [eqns.

(16)–(18)].TiX4 + 4LiAlH4 → Ti(AlH4)4 + 4LiX  −110 °CTi(AlH4)4 → TiH2 + 4Al + 7H2  >−80 °CTi(AlH4)4 → Ti + 4Al + 8H2  >−80 °CX = Cl, BrNano-sized titanium hydride inclusions were also found in the Al3Ti alloy treated with gaseous hydrogen.

Although Al3Ti is a very poor hydrogen absorber, it reacts with hydrogen, producing a titanium hydride phase finely distributed in the material.38

A nano-crystalline Al3Ti–TiH2 composite also forms during ball-milling of Al and Ti powders in a hydrogen atmosphere.39

The particle size in such composite is effectively reduced due to the presence of TiH2 and grains in the particles become nano-size with the progression of ball-milling.

A high resolution transmission electron spectroscopy unambiguously confirmed the presence of TiH2 phase at the boundaries of Al3Ti grains.

Authors attribute the formation of Al3Ti–TiH2 to comparable heats of formation of TiH2 and Al3Ti (−144 vs. −146 kJ mol−1).39

Though, due to its higher stability, Al3Ti forms as a principal component.

Surprisingly, possible participation of the titanium hydride species in Ti-promoted transformations of alkali metal aluminohydrides has not been fully explored.

At the same time, the catalytic effect of TiH2 on the decomposition of LiAlH4 and NaAlH4 is well-known.12,19

Furthermore, metallic titanium and titanium micro-powders, which easily form hydride phases in the presence of hydrogen, proved to be active catalysts in hydrogen desorption and absorption by sodium alanate.27,40

Titanium’s ability to split hydrogen molecules into atoms once H2 is absorbed on a Ti-containing surface provides an additional support for the formation of a titanium hydride phase in Al–Ti alloys exposed to hydrogen.41–43

Thus, experimental data collected to date point towards a titanium hydride phase as the most likely reason for catalytic activity of the in situ generated Al3Ti.

However, since titanium forms a broad variety of hydride phases TiHx in the composition range 0 < x < 2,44 it is premature to speculate about its nature until this phase has been fully identified.

The later task may become quite challenging because concentration of meta-stable hydride phases in the sample can fall below detection limit of contemporary analytical techniques or they may decompose while exposed to high-energy irradiation.

In summary, Ti-catalyzed transformations of alkali metal aluminohydrides can be presented as a sequence of chemical processes (Fig. 3), which begin with interactions between Ti-dopant and the host material and produce an Al3Ti alloy containing a catalytic hydride phase.

The catalyst promotes solid state transformations of MAlH4 into MH, Al, and H2; MH formed further reacts with remaining alanate giving rise to M3AlH6, which, in turn, thermally decomposes at an elevated temperature.

Currently, there is no reason to assume that thermal decomposition of alkali metal alanates should depend on the nature of M+.

However, owing to relatively high thermodynamic stability of NaAlH4 and KAlH4 and low reactivity of LiH, decomposition rates and amounts of M3AlH6 formed in decomposing LiAlH4, NaAlH4 and KAlH4 may vary10,13,14,22,24.

Mechanical activation of solid state processes in alanate-based materials

Even a brief look at the previous chapters is sufficient to realize the complexity of solid state processes, which take or can take place in alkali metal alanates.

Extensive structural changes and mass transfer, which accompany most of those transformations, are conspicuously restrained in solids.

Therefore, any treatment enhancing freedom of molecular movement in solid hydrogen storage materials should considerably improve their operation.

High-energy ball-milling, where steel or ceramic balls rapidly move in a tightly closed vial, smashing and mixing solids trapped inside, offers a unique opportunity to enhance mass transfer in solid materials without changing their physical state (see also Fig. 2).9,10,19,45–50

While crushing, dispersing, and mixing solids by mechanical milling have been known for millennia, the chemical effect of mechanical processing is hardly explored.

Unfortunately, once again, practical applications in the field of mechanically-induced chemical transformations surpassed theoretical knowledge leaving behind numerous unresolved issues and unanswered questions.

It appears, for instance, that mechanochemical reactions are not exclusively temperature-driven processes as suggested by some researchers.51

Thus, lithium alanate and a temperature sensitive ammonium carbonate, (NH4)2CO3, maintain their integrity even during long-term ball-milling in a Spex mill.23

However, once a catalytic amount of TiCl4 is added to LiAlH4, it quickly decomposes upon ball-milling in the same milling equipment.

Furthermore, it takes only a few hours of milling to prepare Li3AlH6 from LiAlH4 and LiH,23 or to perform a variety of solvent-free chemical reactions otherwise occurring exclusively in solution.45–50

Theoretical analyses of ball-milling in a commercial Spex-type unit revealed a moderate temperature effect of mechanical processing.

At the same time, the pressure generated in the solid, trapped between two colliding balls, can rise to an extremely high level of several GPa52,53 and facilitate genuine ultra-high pressure activated processes.54–56

Unfortunately, a lack of reliable experimental data limits our current understanding of specific chemical effect of mechanical processing.

Therefore, establishing the nature of mechanical activation in molecular and ionic solids requires considerable research effort and is an issue to be resolved in the near future.


In conclusion, analysis of experimental data available to date clearly shows that:

(1) The conventional mechanism of the thermal decomposition of alkali metal aluminohydrides does not satisfactory explain experimental facts acquired by different researchers during the last three decades.

Decomposition of alkali metal aluminohydrides can occur in the solid state as a one-step process without intermediate formation of corresponding hexahydroaluminates.

The presence of the latter in decomposing samples can be attributed to the side-reaction between one of the decomposition products, alkali metal hydride, and the host material, alkali metal alanate.

(2) Experimental data reported in chemical and materials science literature suggests that the catalytic effect of titanium on the thermal decomposition of alkali metal alanates can be attributed to a titanium–hydrogen phase present in the Al3Ti catalyst, which forms upon the reduction of Ti-dopants by alanates during doping process.

(3) Little is known about mechanical activation of chemical transformations in solids.

It is quite clear, however, that besides enhancing the mass transfer, ball-milling has a specific chemical effect, the exact nature of which is to be explored.