Thermal and mechanically activated decomposition of LiAlH4

Article abstract of DOI:10.1016/j.materresbull.2007.05.018

Thermal stability of as-received LiAlH₄ and milled LiAlH₄ has been investigated. The thermal decomposition mechanism of as-received LiAlH₄ depends on the temperature-time history. Apparent activation energies and enthalpie

Full text of DOI:10.1016/j.materresbull.2007.05.018

Materials Research Bulletin 43 (2008) 1263–1275
www.elsevier.com/locate/matresbu
Thermal and mechanically activated decomposition of LiAlH4
b
a
a
J.R. Ares a, , K.-F. Aguey-Zinsou , M. Porcu , J.M. Sykes ,
*
M. Dornheim b, T. Klassen b, R. Bormann b
a Department of Materials, Oxford University, Parks Road OX1 3PH, Oxford, UK
b Institute for Materials Research, GKSS-Forschungszentrum, Max-Planck-Str. 1, D-21502 Geesthacht, Germany
Received 4 May 2006; received in revised form 13 February 2007; accepted 20 May 2007
Available online 24 May 2007
Abstract
Thermal stability of as-received LiAlH4 and milled LiAlH4 has been investigated. The thermal decomposition mechanism of as-
received LiAlH4 depends on the temperature–time history. Apparent activation energies and enthalpies of the reactions have been
obtained. During milling treatment, the high temperature and pressures locally induced by shocks lead to LiAlH4 mechanically
decomposition. The decomposition temperatures of LiAlH4 and Li3AlH6 are both reduced by ꢀ60 8C due to particle size reduction
produced by mechanical milling. Besides, the activation energy of the decomposition reaction of LiAlH4 decreases as compared to
as-received LiAlH4. Moreover, a layer of oxide (ꢀ5 nm) at the surface of the milled alanate Li3AlH6 is observed. This layer could
have a drastic influence on decomposition H-kinetics.
# 2007 Elsevier Ltd. All rights reserved.
Keywords: A. Hydrides; C. Differential scanning calorimetry; C. Electron microscopy; D. Energy storage
1. Introduction
During the last 20 years, interest on hydrogen as a fuel for vehicles has increased substantially. One of the main
tasks to be fulfilled is to develop a suitable on board H-storage system. On board H-storage systems need to be
safe, low cost, with high gravimetric hydrogen capacity and rapid hydrogen absorption/desorption kinetics at
ambient temperature and atmospheric pressures. At present,ꢁseveral storage materials are being investigated [1–3]
and, among them, complex hydrides containing a AlH4 group (alanates) seem to be the most promising
candidates due to their high gravimetric hydrogen capacity. In spite of this, alanates were not considered as
candidates for H-storage until Bogdanovic and Schwickardi in 1997 demonstrated hydrogen reversibility under
moderate pressures by adding TiCl3 to sodium alanate [3]. After this work, investigations focused on improving
the H-properties of sodium alanate by mechanically milling NaAlH4 with different additives [4–6] have been
done.
The progresses made with NaAlH4 have led to consider it as a model-system for other alanates with higher
H-capacity (10.6 wt%), such as LiAlH4. LiAlH4 was synthesised in the late 1940s and is usually used as a reducing
agent. Only few groups explored the structural [7] and thermal properties [8–10] of LiAlH4 until 10 years ago. But,
since the discovery of the reversibility with NaAlH4, investigations with LiAlH4 as a potential H-storage material have
´ ´ ´
* Corresponding author. Present address: Laboratorio de Materiales de Interes Energetico. Dpto. Fısica de Materiales. Facultad de Ciencias C-IV.
UAM. 28049. Madrid (Spain). Tel.: +34 91497477; fax: +34 914978579.
E-mail address: joser.ares@uam.es (J.R. Ares).
0025-5408/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2007.05.018

1264
J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
greatly increased. Hydrogen desorption kinetics has been improved by milling LiAlH4 with additives [11–15] and
some degree of reversibility (1.8 wt%) was achieved by Chen et al. [15]. However, at present, the mechanism of
absorption/desorption of hydrogen and the role of additives are not well understood.
The first step to understand the possible catalytic role of the additives during desorption of hydrogen in LiAlH4
should be investigate the decomposition mechanism of LiAlH4. As in all alkaline-alanates [16], the thermal
decomposition of LiAlH4 is summarised in the following two-step process:
LiAlH4 ! Li3AlH6 þ Al þ H2 ð5:3 wt% HÞ ðT ¼ 150 ꢁ 175 ꢂCÞ Reaction I
(1)
(2)
1
3
2
3
1Li3AlH6 ! LiH þ Al þ H2 ð2:6 wt% HÞ ðT ¼ 180 ꢁ 220 ꢂCÞ Reaction II
1
1
3
3
2
However, this reaction route is more complex when it is analysed carefully. For instance, whereas a mechanism in
which LiAlH4 melts before it is decomposed into Li3AlH6 has been suggested by Dymova et al. [17] and supported by
DTA, TG and DSC measurements [10,17], the decomposition of LiAlH4 at temperatures lower than the melting point
has been detected by ‘‘in situ’’ XRD and NMR [18,19].
In order to explain these results, the existence of intermediate reactions in (1) and (2) have been proposed. Wiench
et al. [19] suggest that LiAlH4(s) decomposes directly into LiH(s) and Al(s) expelling hydrogen, and that LiH(s)
recombines with LiAlH (s) to produce Li AlH . Both reactions are thermodynamically feasible (DGꢂ < 0) [20], and,
r
4
3
6
for instance, recombination of LiH(s) with LiAlH4(s) to obtain Li3AlH6(s) is often made by mechanical milling
[11,20]. However, as far as we know, LiH(s) has never been detected. This situation might be explained by the very fast
kinetics of the recombination reaction, which implies that detection of LiH(s) could be difficult. A second intermediate
path, based on results obtained for NaAlH4 decomposition, has also been proposed [21]. The aluminium segregation at
the surface of the particles [22,23] suggests that certain intermediate Al-species (AlH3) could diffuse across the
particle to release hydrogen. Despite AlH3 has still not been detected, its decomposition temperature, between 70 and
150 8C [24], makes the process feasible.
On the other hand, milling treatment leads to a decomposition of LiAlH (LiAlH is metastable (DGꢂ < 0) [20]).
r
4
4
Balema et al. [11] have observed that LiAlH4 decomposition occurs at long milling times. Balema suggests that
impurities providing from milling enhance the decomposition of LiAlH4 into Li3AlH6 and Al at long milling times. In
fact, it has been reported that the presence of a great amount impurities reduce the thermodynamic stability of LiAlH4
[25]. However, due to LiAlH4 metastability, decomposition during milling could be also mechanically activated by the
shocks during milling and/or by increasing vial temperature during milling treatment [26]. In first case, milling
treatment could form new phases (for instance high pressure phases) [27,28] and subsequently, have an influence on
the decomposition of LiAlH4.
Balema et al. [11] have also been observed that milling improves the desorption H-kinetics of the LiAlH4.
Andreasen et al. [29] have recently attributed that enhancement to a reduction of crystallite size of LiAlH4 which
improves the species transport. However, the influence of parameters, such as particle size, vacancies density on
H-kinetics has not been investigated yet and they may drastically affect to H-kinetics as occurs in other H-storage
materials like MgH2 [30].
The main purpose of this article is to clarify the thermal and mechanical decomposition mechanism of LiAlH4 as
well as the influence of parameters, such as particle size, crystallite size on its decomposition. To this aim, thermal
analysis has been carried out to obtain the enthalpies and apparent activation energies of the decomposition reactions.
Thermal measurements have been complimented by a structural and morphological characterisation by XRD, SEM
and TEM in order to investigate the influence of particle and crystallite size on the decomposition process of LiAlH4.
Impurities providing for milling have been measured by ICP technique. Infrared spectroscopy (IR) has also been used
to characterise the evolution of Al–H environment during mechanical milling.
2. Experimental
The starting material LiAlH4 (>95%) was purchased from Sigma–Aldrich. The chemical synthesis of Li3AlH6 was
performed according to the method described in literature [31]. To prevent any oxidation of samples and starting
powders, all operations were carried out in an argon-atmosphere glove-box (<5 ppm O2). As-received LiAlH4 was
milled inside the argon glove-box with a Fritsch P7 planetary mill. Milling was performed at a vial rotation speed of
230 rpm and with a ball to powder ratio of 10:1. Materials were milled for 30 min, 1.5, 7.3, 14, 35, and 70 h.

J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
1265
Elemental analysis of samples was done using an ICP-MS Perkin-Elmer Elan 6000. Samples were dissolved in
concentrated HCl followed by the addition of distilled water. Then, they were analysed according to the normal
procedure [32].
Structural analyses were done at room temperature by using a X-ray diffractometer Bruker D8 Advance equipped
with Cu Ka radiation. Sample holders were filled inside the glove-box and covered with a Kapton foil to avoid any
oxidation during the X-ray measurement. Diffraction patterns were analysed by the Rietveld method [33] using the
Fullproff software.
The morphology of the samples was analysed by scanning electronic microscopy (SEM) with a Zeiss DSM 962,
operating at 10 kV. Samples were transferred to the SEM under argon atmosphere by a glove bag. High-resolution
transmission electron microscopy (HRTEM) analysis were performed on a JEOL 3000F FEG (point resolution 0.16 nm)
operated at 300 kV. All operations have been carried out inside a nitrogen glove-box and the powders were dropped on an
amorphous carbon-coated TEM copper grid by using hexane. LiAlH4 was found to be stable under the electron beam.
Fourier transformation infrared (FT-IR) spectra were recorded with a Brucker Equinox 55 equipped with an
attenuated total reflection (ATR) accessory. The spectral resolution was 4 cmꢁ1. Measurements were done under an
argon atmosphere.
Thermal analyses were conducted with a DSC 404 C Pegasus from Netzsch. All measurements were carried out
under a flow (50 ml/min) of high purity argon (99.999%). Sample mass was typically 10 mg. Heating runs were
performed at different rates from 50 to 450 8C. Thermal desorption measurements (TDM) were performed with a
home-made instrument under vacuum.
3. Results and discussion
3.1. Thermal decomposition of as-received LiAlH4
As-received LiAlH4 is formed of crystals with a well-defined monoclinic morphology as displayed in the SEM
micrograph (Fig. 1). The crystals exhibit a bi-modal distribution in size centred around ꢀ50 and ꢀ150 mm. An average
initial crystallite size greater than 300 nm is obtained by Rietveld method of X-ray diffraction pattern.
Fig. 2a–e shows the DSC curves of as-received LiAlH4. Two different decomposition processes are observed
depending on heating rate (HR). At HR above 1 8C/min, three peaks (I, II and III) are detected as shown Fig. 2a–c.
Whereas endothermic peak I does not depend strongly on the HR, the position of endothermic peak II and exothermic
peak III are shifted to lower temperatures when the HR is decreasing. Besides these peaks, a small exothermic peak is
sometimes detected at ꢀ150 8C [8,9] that is related to a reaction between LiAlH4 and hydroxide traces present on the
sample. Endothermic peak I does not depend on HR. During this process there is no hydrogen release as shown by
Fig. 1. SEM micrograph of as-received LiAlH4.

1266
J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
Fig. 2. (a–e) DSC curves of as-LiAlH4 at different heating rates. The peaks of the reactions are marked. In (b), thermal desorption curve is shown
with a dash line. Inset in (d) shows that step IIb is composed by two peaks.
TDM (Fig. 2b, dashed line) and the value obtained of reaction enthalpy (13 kJ/mol) corresponds to that of LiAlH4(s)
melting [10]. However, the exothermic peak II and the endothermic peak III exhibit a drastic dependence on HR. The
relation of TDM peaks area is ꢀ2 which confirms that peak II is associated with the decomposition of LiAlH4(l) and
the formation of solid Li3AlH6 with H2 release (5.3 wt%) while peak III is related to the decomposition of Li3AlH6 into
LiH with H2 release (2.6 wt%). Reaction enthalpies obtained for the two peaks (ꢁ14 and 15 kJ/mol, respectively) are
in agreement with data previously reported [9].
On the other hand, when HR is below 1 8C/min, only two endothermic peaks (IIb and III) are detected (Fig. 2d–e).
The position of the peaks also depends on HR and no melting is detected. The process of melting and solidification has
been substituted by an endothermic reaction (peak IIb) with an enthalpy of 1.6 kJ/mol. This value agrees with
previously data obtained by computer calculations corresponding to the decomposition of LiAlH4(s) into Li3AlH6(s)
[34–36] and it is similar to an experimental data reported in Ref. [9]. As concern the higher temperature peak (III), the
reaction exhibits a reaction enthalpy of 16 kJ/mol similar to that obtained at faster rates, which suggests that the same
reaction occurs, i.e. decomposition of Li3AlH6 into LiH and Al without melting.
However, inset of Fig. 2d shows that peak IIb is double, which indicates that the decomposition reaction of LiAlH4
exhibits intermediate stages. This stage may be related to induction or nucleation of a new phase (Li3AlH6 and Al) and
the diffusion of some species, respectively [37] or due to intermediate reactions [38] (for instance, involving AlHx
species). Further investigation is necessary to clarify which processes occur during these stages.
The apparent activation energies, Ea, of the different processes were obtained by the non-isothermal Kissinger
method [39]. Fig. 3 shows the Ea of the different reactions in a Kissinger plot. As concerns peaks II and IIb, i.e.
decomposition of LiAlH4 into Li3AlH6, the smaller apparent activation energy observed at higher heating rates
(115 kJ/mol versus 162 kJ/mol) indicates that the release of hydrogen after melting is kinetically more favourable than

J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
1267
Fig. 3. Kissinger plot for the decomposition steps II, IIb and III. Heating rates varying from 0.5 to 4 8C/min. Apparent activation energies (Ea) of the
reactions are given by the slope of the fittings of equation ln(b/Tmax) = (ꢁEa/RTmax) + constant (Tmax = peak temperature, b = heating rate,
R = Boltzmann constant).
the release of hydrogen without melting. Melting suppression observed on decreasing HR is a phenomena observed
with highly energetic materials [40] and it is explained by the ‘‘near static conditions’’ (slow HR) that allow a slow
diffusion of species without melting, whereas higher HR leads to a more favourable reaction path which implies a
melting of LiAlH4 and, subsequently, a higher mobility of the atoms. These results may explain the decomposition of
LiAlH4 into Li3AlH6 below the melting point of LiAlH4 as well as the discrepancies between results obtained from
different techniques, such as NMR and XRD in situ analyses [18,19] and TG or DSC [10,17]. The value of Ea at lower
HR (162 kJ/mol) is in good agreement to the value reported by Kang et al. [41] (154 kJ/mol). At faster heating rates,
the obtained value of activation energy (115 kJ/mol) is slightly higher to this recently obtained by Blanchard et al. [37]
(102 kJ/mol) and Andreasen et al. [29,42] (81 and 89 kJ/mol).
With regard to the higher temperature peak (III), i.e. decomposition of Li3AlH6 into LiH and Al without melting.
The apparent activation energy exhibits a value of 86 kJ/mol as it is shown in Fig. 3 and it is not influenced by heating
rate. This value is between the value reported (107 kJ/mol) by Andreasen et al. [29] and this given by Kang et al.
(72 kJ/mol) [41].
3.2. Mechanically activated decomposition of as-LiAlH4
Fig. 4 shows the thermal decomposition of LiAlH4 milled for different times. Whereas LiAlH4 milled for short
times (ꢃ14 h) releases hydrogen through the two basic reactions (Eqs. (1) and (2)) cited in Section 1, LiAlH4 milled
for longer times (>14 h) decomposes through one reaction (Eq. (2)). The evolution of the desorption temperature (Td)
and amount of hydrogen released for both reactions as a function of milling time are shown in Fig. 5a and b,
respectively. Two stages are differentiated: at short milling time (<1.5 h) Td of both decomposition reactions
decreases by ꢀ60 8C as well as the amount of hydrogen stored for the first reaction (ꢀ0.5 wt%). At intermediate
milling time (1.5 h < t < 14 h) the temperature of both decomposition reactions remains constant and a slight
hydrogen release is again observed (ꢀ0.5 wt%) for the first reaction. Further milling leads to LiAlH4 decomposition
into Li3AlH6 and Al.
3.2.1. Decomposition of as-received LiAlH4 during milling
The influence of milling time on particle morphology of LiAlH4 before decomposition is shown in the SEM
micrographs displayed in Fig. 6a–c. For milling time below 1.5 h particles are refined. Whereas the sample milled
30 min shows still very large particles (Dp = 80 ꢄ 40 mm) that have not been completely milled, the sample milled
1.5 h exhibits already a finer and more homogeneous particle size distribution particle (Dp = 30 ꢄ 15 mm). At the end
of this stage, smaller particles tend to agglomerate. Then, when increasing milling times (>1.5 h), as shown in Fig. 6c,
LiAlH4 particles agglomerate more and particle size is not further reduced (Dp = 35 ꢄ 10 mm).

1268
J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
Fig. 4. Thermal desorption of LiAlH4 milled different times. A heating rate of 8 8C/min was used.
To determine microstructural (crystallite size) and crystallographic properties (phase identification and cell volume)
XRD patterns have been analysed by the Rietveld method. All patterns exhibit good refinements. Fig. 7a and b shows the
XRD patterns of sample milled 1.5 and 7 h, respectively. All the peaks were indexed to a monoclinic P21/c structure of
LiAlH4 but a small peak at 32.28 that corresponds to LiCl impurity. Table 1 shows that the crystallite size of LiAlH4
decreases from initial values of 300 ꢄ 50 to 90 ꢄ 30 nm in only 30 min of milling and further milling time has no
Fig. 5. (a) Influence of milling time on H-capacity of the two decomposition reactions. (b) Influence of milling time on temperatures of the two
decomposition reactions.

J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
1269
Fig. 6. SE micrographs (a) LiAlH4 milled 30 min, (b) LiAlH4 milled 1.5 h and (c) LiAlH4 milled 7 h.
Fig. 7. XRD patterns of LiAlH4 milled (a) 0.5, (b) 14, (c) 35 and (d) 70 h. The phases following phases were used to fit the patterns: monoclinic-
LiAlH4 (P21/c), rombohedric-Li3AlH6 (R-3), cubic-LiH (Fm-3m) and cubic-Al (Fm-3m). The continuous line is the fit. More intense peaks are
marked. The quality of the refinement is given by the weight profile factor (Rwp). All patterns exhibit 9% < Rwp < 14%.
Table 1
Values of crystallite size and cell volume of LiAlH4 phase milled different times
3
Vcell (A )
˚
LiAlH4, tmilled (h)
Dv (nm)
0
0.5
1.5
7
300 (50)
91 (30)
90 (20)
130 (30)
109 (40)
277.2 (1)
277.1 (1)
277.1(1)
277.0 (1)
276.6 (1)
14
Values are obtained by Rietveld analyses of the diffraction patterns.

1270
J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
Fig. 8. Backscattering micrographs of LiAlH4 milled (a) 1.5 and (b) 7 h. White dots correspond to Al.
influenceon crystallite size. As concernscrystallographic properties, the cell volume of LiAlH4 is not drastically affected
by milling treatment.
Although LiAlH4 is the only phase detected in the X-ray diffraction pattern, Figs. 4 and 5 show that LiAlH4 milled
for 1.5 h exhibits a slight diminution of H-capacity (0.5 wt%) compared to as-received LiAlH4 which suggests that
decomposition of LiAlH4 has already taken place. In fact, a small amount of Al is detected by BSE at the surface of
LiAlH4 particles, as displayed in Fig. 8a. This confirms that LiAlH4 is partially decomposed after 1.5 h of milling.
When milling time is increased up to 7.5 h, the amount of stored hydrogen for the first reaction is further decreased (by
1 wt%) compared to as-received LiAlH4 (Fig. 5a). As a consequence, a higher amount of Al at the particle surface is
observed in the corresponding BSE micrograph (Fig. 8b). Therefore, the decomposition of LiAlH4 occurs at very short
milling times. So, because the amount of impurities detected by ICP is negligible and the vial temperature during
milling (ꢀ50 8C) is lower than decomposition temperature of LiAlH4, decomposition of LiAlH4 is rather related to the
high pressures and temperature rise locally created during milling than thermal decomposition by increasing the
temperature of vial or impurities generated during milling.
Moreover, Fig. 9a shows the selected area electron diffraction (SAD) pattern of one particle of LiAlH4 milled 1.5 h.
The particle is mainly composed of nanocrystalline LiAlH4. However, in a HRTEM image of the particle (Fig. 9b),
besides the planes corresponding to LiAlH4 monoclinic phase, several planes within the crystalline part of the particle
can be indexed by a different phase. These planes may be indexed to the tetragonal high-pressure phase Li3AlH6 [43].
Formation of high-temperature–pressure metastable phases are commonly occurring during milling [44] and it has
been observed with other ionic hydrides, such as MgH2 [45]. However, with regard to high pressure Li3AlH6, the
actual lack of crystallographic and thermodynamical data [27] as well as its low quantity make complicated to discern
whether this phase is a intermediate stage during the decomposition reaction or simply the result of milling the normal
pressure phase Li3AlH6. Nevertheless, the local presence of the high-pressure phase suggests that the decomposition is
driven mechanically.
At longer milling times (35 h), the total decomposition of LiAlH4 is almost achieved. Fig. 7c shows the pattern
diffraction of the sample milled 35 h. Two new phases were found: Li3AlH6 indexed to a rhombohedral R-3 [46] and
Al, indexed to a cubic Fm-3m. Also, a small quantity of LiAlH4 phase is detected. With regard to microstructure, large
crystallite size values (ꢀ80 nm) of both phases are obtained by Rietveld analysis. Further milling (up to 70 h) leads to
the full decomposition of Li3AlH6 into LiH and Al as shown in Fig. 7d.
Fig. 10a shows the morphology of the sample milled 35 h in a SEM micrograph. After decomposition, particle size
distribution is roughly similar to that before decomposition. The particles exhibit a more homogeneous shape as a
result of the phase transformation. BSE micrograph of the sample is shown in Fig. 10b. Particle exhibits large amount
of aluminium on the surface that confirms that decomposition took place. Aluminium at the surface of the particles has
recently been detected during the thermal decomposition of LiAlH4 [22], which suggests similarities between thermal
and mechanical decomposition as well as the existence of a long-range transport mechanism of Al species during
mechanical decomposition. SAD measurements in Fig. 11a confirm that particles milled 35 h are formed by a
rhombohedral Li3AlH6 and Al. Moreover, a small layer (ꢀ5 nm) is detected on the surface by HRTEM as shown

J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
1271
Fig. 9. (a) SAD of a particle after milled 1.5 h. (b) HRTEM image showing lattice.
Fig. 11b. This layer was only observed on particles milled for 35 h, which indicate that is formed after the
decomposition of LiAlH4 into Li3AlH6 and Al. Besides, measurements of EDS in the particles reveal a higher ratio
O/Al into this layer than inside of the particle. Therefore, due to the large quantity of aluminium observed on the
surface, the surface layer could be due to the oxidation of aluminium from decomposition reaction. In fact, after
milling process, sample becomes more active due to larger surface/volume of the particles and oxygen could have been
introduced either during milling or handling the sample. A similar oxide layer was found by Andrei et al. [22] in
LiAlD4 milled with different additives. Although the role of this layer is beyond of the scope of this paper, it is feasible
to expect some influence of this layer on the decomposition processes of the alanate, such as species diffusion,
H-surface dissociation, etc.
The decomposition process of LiAlH4 was also followed by IR spectroscopy. Recently, this technique has been
used in as-received NaAlH4 and milled NaAlH4 [47]. In LiAlH4, AlH4ꢁ ion has a tetrahedral symmetry with active
infrared frequencies in two regions [48] Al–H stretching modes n3(F2) at 1600–1800 cmꢁ1 and Lꢁi–Al–H bending
modes, n4(F2) at 700–900 cmꢁ1. Nevertheless, due to a slight distortion in the solid state (AlH4 ions having a
Fig. 10. (a) SE and (b) backscattering micrographs of LiAlH4 milled 35 h. White dots correspond to Al.

1272
J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
Fig. 11. (a) SAD of LiAlH4 particle after 35 h of milling. (b) HRTEM image of one particle. An oxide layer is observed on surface.
flattened tetrahedral symmetry (D2d)), a splitting of the active bands in infrared is usually observed [49–52]. Fig. 12i
shows the IR spectrum of as-received LiAlH4. The bands observed correspond to the one previously reported [50,51]
for solid LiAlH4 with stretching modes at 1760 and 1610 cmꢁ1 and bending modes at 900 and 803 cmꢁ1
.
As concerns to Li3AlH6, solid Li3AlH6 has been synthesised according to the method described in [31] for
comparative purposes. [AlH6]3ꢁ has an octahedral symmetry and shows active infrared frequencies around [50]:
1400 cmꢁ1 for the stretching mode n3(F1) and 990 cmꢁ1 for the bending mode n4(F1). The IR spectrum of Li3AlH6 is
Fig. 12. IR spectrum of (i) as-received LiAlH4, (ii) Li3AlH6 synthesised, (a) LiAlH4 milled 1.5 h, (b) LiAlH4 milled 14 h and (c) LiAlH4 milled 35 h.

J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
1273
presented Fig. 12ii. The bands observed at 1390 and 1250 cmꢁ1 for the stretching modes and 1000, 950 cmꢁ1 for the
bending modes are in agreement with the ones reported by Bureau et al. [51].
When LiAlH4 is milled for a short time, a small shift (20 cmꢁ1) to high frequencies of the stretching modes is
observed (Fig. 12a). This shift may be related to the strain created during milling because a similar shift was observed
[28] by Raman spectroscopy for LiAlH4 under a static pressure (ꢀ5 GPa) similar to the one expected during
mechanical milling. Further milling (Fig. 12b and c) leads to a decrease in LiAlH4 peaks intensity until they disappear
and two new bands at 1375 and at 1220 cmꢁ1 are detected. Despite intensities are not very strong, their positions are
similar to those of Li3AlH6 synthesised [52] that confirmed the near total decomposition of LiAlH4 into Li3AlH6
during mechanical milling. Decomposition occurs without intermediate products detected by IR measures.
3.2.2. Thermal decomposition of milled LiAlH4
The evolution of the desorption temperature (Td) and amount of hydrogen released for both reactions as function of
milling time have been shown in Fig. 5a. At short milling times (<1.5 h) Td of both decomposition reactions decreases
by ꢀ60 8C whereas at intermediate milling times (1.5 h < t < 14 h) the temperature of both decomposition reactions
remains constant. Further milling lead to LiAlH4 decomposition as shown in Section 2.
Temperature reduction of both decomposition reactions could be facilitated by reduction of particle and crystallite
size and, in certain cases by defects and impurities introduced during milling. As concerns impurities, it has been
reported that impurities uptakes could act as a catalyst for the decomposition of LiAlH4 [11]. However, the amount of
impurities detected by ICP after milling LiAlH4 for 70 h is much lower (<0.1 wt% of iron impurity) than those
reported in literature [12,13]. Therefore, the amount of impurities arising from milling tools is not the main cause of
the diminution of Td. A higher vacancy density could also contribute in reducing the desorption temperatures as
suggested by Fitchtner et al. [53]. Nevertheless, in our case, structural parameter as cell volume stays practically
constant during milling as it was shown in Table 1. With regard to microstructure, a crystallite size remains constant
after 30 min of milling but temperature continues decreasing up to longer milling times. Thus, reduction of crystallite
size is not the main reason to explain the temperature reduction of both reactions. However, Fig. 6a and b shows that
LiAlH4 milled 30 min still has very big particles compared to the same sample milled 1.5 h, which exhibits a finer and
more homogeneous particle size distribution. Hence, a influence on Td is expected, i.e. smaller particles exhibit shorter
diffusion paths and then, decomposition temperature is lower than those with larger ones. Further milling leads
(>1.5 h) to agglomeration of the particles as shown Fig. 6c, and not further particle size refinement is observed.
Therefore, the temperature of both reactions remains constant, as shown Fig. 5b, suggesting that particle size controls
the decomposition temperature of both reactions.
Moreover, milling LiAlH4 for short times has an influence on its thermal decomposition mechanism. Fig. 13 shows
the DSC curve of LiAlH4 milled 1.5 h. No melting was detected and two endothermic peaks similar to those observed
for as-received LiAlH4 were observed. These peaks exhibit also similar reaction enthalpies: 1.8 and 20 kJ/mol
Fig. 13. DSC curve of LiAlH4 milled 1.5 h.

1274
J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
suggesting similar thermal decomposition processes to those observed in the decomposition of as-received LiAlH4 at
lower heating rates.
The activation energies of the two decomposition reactions of milled LiAlH4 (t = 1.5 h) were calculated by
Kissinger method. An activation energy of 100 ꢄ 10 kJ/mol was obtained in the first decomposition reactions and a
130 ꢄ 15 kJ/mol in the second decomposition reaction. Therefore, the activation energy of the first decomposition
reaction of milled LiAlH4 is smaller than this of as-received LiAlH4. The difference could be related to the diminution
of crystallite size produced by milling during the first 30 min. A smaller crystallite size would improve the
decomposition kinetics and it would reduce the activation energy. However, with regard to the second reaction, an
increase of the activation energy as compared to un-milled sample is found. This may be attributed to the influence of
aluminium oxides (previously detected by TEM) and/or high-pressure phases that could degrade the process of
decomposition of Li3AlH6.
4. Conclusions
The mechanism of thermal decomposition of LiAlH4 depends on HR. Whereas slow HR lead to an endothermic
decomposition of LiAlH4 into Li3AlH6 without melting and with intermediate processes, fast HR lead to melting
before decomposition of LiAlH4 into Li3AlH6 and diminution of the apparent activation energy. However, the apparent
activation energy of the subsequent decomposition of Li3AlH6 into LiH and Al does not depend on HR. Mechanical
milling of LiAlH4 leads to its decomposition into Li3AlH6 and Al, and after, Li3AlH6 decomposes of into LiH. The
process of decomposition is mechanically driven and Aluminium is detected at particle surface (during milling of
LiAlH4) suggesting the existence of a long-range transport mechanism of Al-species during mechanical
decomposition. Besides, HTREM measurements reveal the existence of an oxide layer on the surface of the
alanate and a formation of high-pressure phase due to milling treatment. Finally, milling LiAlH4 results in reduction of
both temperature reactions. This reduction is mainly attributed to shorter diffusion paths because of a diminution of
particle size during the milling treatment.
References
[1] B.A.M. Seayal, D.M. Antobelli, Adv. Mater 16 (2004) 9.
[2] L. Schlapbach, A. Zu¨ttel, Nature 414 (2001) 353.
[3] B. Bogdanovic, M. Schwickardi, J. Alloys Compd. 253/254 (1997) 1.
[4] G.J. Thomas, K.J. Gross, N.Y.C. Yang, C. Jensen, J. Alloys Compd. 330–332 (2002) 702.
[5] B. Bogdanovic, R.A. Brand, A. Marjanovic, M. Schwickardi, J. Tolle, J. Alloys Compd. 302 (2000) 36.
[6] D. Sun, T. Kiyobayashi, H.T. Takeshita, N. Kuriyana, C.M. Jensen, J. Alloys Compd. 337 (2002) L8.
[7] N. Sklar, B. Post, Inorg. Chem. 6 (1967) 669.
[8] M. McCarty, J.N. Maycock, V.R. Pai Verneker, J. Phys. Chem. 72 (12) (1968) 4009.
[9] P. Claudy, B. Bonnetot, J.M. Letoffe, G. Turck, Thermochim. Acta 27 (1978) 213.
[10] J.P. Bastide, B. Bonnetot, J.M. Letoffe, P. Claudy, Mater. Res. Bull. 20 (1985) 999.
[11] V.P. Balema, V.K. Pecharski, K.W. Dennis, J. Alloys Compd. 313 (2000) 69.
[12] V.P. Balema, J.W. Wiench, K.W. Dennis, M. Pruski, V.K. Pecharski, J. Alloys Compd. 1/2 (329) (2001) 108.
[13] M. Resan, H.D. Hampton, J.K. Lomness, D.K. Slattery, Int. J. Hydrogen Energy 30 (13/14) (2005) 1413.
[14] M. Resan, H.D. Hampton, J.K. Lomness, D.K. Slattery, Int. J. Hydrogen Energy 30 (13/14) (2005) 1417.
[15] J. Chen, N. Kuriyama, Q. Xu, H.T. Takeshita, T. Sakai, J. Phys. Chem. B 105 (2001) 11214.
[16] J.A. Dilts, E.C. Ashby, Inorg. Chem. 116 (6) (1972) 1230.
[17] T.N. Dymova, V.N. Konoplev, D.P. Aleksandrov, A.S. Sizareva, T.A. Silina, Russ. J. Coord. Chem. 21 (3) (1995) 165.
[18] H.W. Brinks, B.C. Hauback, P. Norby, H. Fjellvag, J. Alloys Compd. 351 (2003) 222.
[19] J.W. Wiench, V.P. Balema, V.K. Percharski, M. Pruski, J. Solid State Chem. 177 (3) (2004) 648.
[20] T.N. Dymova, D.P. Aleksandrov, V.P. Konoplev, T.A. Silina, N.T. Kuznetsov, Russ. J. Coord. Chem. 19 (1993) 607.
[21] K.J. Gross, S. Guthrie, S. Takura, G. Thomas, J. Alloys Compd. 297 (2000) 270.
[22] C.M. Andrei, J.C. Walmsley, H.W. Brinks, R. Holmestad, D. Blanchard, B.C. Hauback, G.A. Botton, J. Phys. Chem. B 109 (2005) 4350.
[23] C.M. Andrei, J.C. Walmsley, D. Blanchard, H.W. Brinks, R. Holmestad, B.C. Hauback, J. Alloys Compd. 395 (1/2) (2005) 307.
[24] G. Sandrock, J. Reilly, J. Graetz, W.-M. Zhou, J. Johnson, J. Wegrzyn, Appl. Phys. A 80 (2005) 687.
[25] N.N. Mattesva, A.I. Golanova, Russ. J. Appl. Chem. 73 (2000) 747.
[26] D.S. Easton, J.H. Scheinbel, S.A. Speakman, J. Alloys Compd. 398 (2005) 135.
[27] P. Vajeeston, P. Ravindran, A. Kjekshus, H. Fjellvag, Phys. Rev. B 69 (2004) 020104.
[28] A.V. Talyzin, B. Sundqvist, Phys. Rev. B 70 (2004) 180101.

J.R. Ares et al. / Materials Research Bulletin 43 (2008) 1263–1275
[29] A. Andreasen, T. Vegge, A.S. Pedersen, J. Solid State Chem. 178 (2005) 3664.
1275
[30] F.-K. Aguey-Zinzou, J.R. Ares, T. Klassen, R. Bormann, Mater. Res. Bull. 41 (2006) 1118.
[31] J. Mayet, S. Kovacevicard, J. Tranchant, Bull. Soc. Chim. 2 (42) (1973) 503.
[32] S.J. Hill, Inductively Coupled Plasma Spectrometry and its Applications, Sheffield Academic Press, UK, 1999.
[33] J. Rodriguez-Carvajal, Physica B 192 (1993) 55.
[34] S.-C. Chung, H. Morioka, J. Alloys Compd. 372 (2004) 92.
[35] O.M. Lovvik, S.M. Opalka, H.W. Brinks, B. Hauback, Phys. Rev. B 69 (2004) 134117.
[36] T.J. Frankcombe, G.-J. Kroes, Chem. Phys. Lett. 423 (1–3) (2006) 102–105.
[37] D. Blanchard, H.W. Brinks, B.C. Hauback, P. Norby, J. Muller, J. Alloys Compd. 404–406 (2005) 743.
[38] R.T. Walters, J.H. Scogin, J. Alloys Compd. 421 (1-2) (2006) 54–56.
[39] H.E. Kissinger, Anal. Chem. 29 (1958) 1702.
¨
[40] Th. Rieckman, S. Volker, L. Lichtblan, R. Schirra, J. Anal. Appl. Pyrolysis 58/59 (2001) 569.
[41] J.K. Kang, J.K. Lee, R.P. Muller, W.A. Goddard, J. Chem. Phys. 121 (21) (2004) 10623.
[42] A. Andreasen, J. Alloys Compd. 419 (1/2) (2006) 40.
[43] S. Konovalov, B. Bolychev, V. Genchel, Russ. J. Inorg. Chem. (Engl. Trasl.) 35 (1990) 327.
[44] C. Surynarayara, C. Prog. Mater. Sci. 41 (2001) 1.
[45] J. Huot, G. Liang, S. Boily, A. Van Neste, R. Shultz, J. Alloys Compd. 293–295 (1999) 495.
[46] H.W. Brinks, B.C. Hauback, J. Alloys Compd. 354 (2003) 143.
[47] S. Gomes, G. Renardin, H. Hagermann, K. Yvon, M.P. Sulic, C.M. Jensen, J. Alloys Compd. 390 (2002) 305.
[48] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordinated Compounds, Part A, fifth ed., J. Wiley, 1997,, p. 189.
[49] A.E. Shirk, F. Shriver, J. Am. Chem. Soc. 95 (18) (1973) 5904.
[50] B.M. Bulychev, V.N. Verbetskii, K.V. Semenenko, Russ. J. Inorg. Chem. 22 (11) (1977) 1611.
[51] J.C. Bureau, B. Bonnetot, P. Claudy, H. Eddaoudi, Mater. Res. Bull. 20 (1985) 1147.
[52] J.C. Bureau, Z. Amri, P. Claudy, J.M. Letoffe, Mater. Res. Bull. 24 (1989) 23.
[53] M. Fichtner, P. Canton, O. Kircher, A. Leon, J. Alloys Compd. 404–406 (2005) 732.