Decomposition behavior of unmilled and ball milled lithium alanate (LiAlH4) including long-term storage and moisture effects

Article abstract of DOI:10.1016/j.jallcom.2010.05.059

A comprehensive study of the decomposition behavior of as received and mechanically (ball) milled LiAlH₄ has been carried out using differential scanning calorimetry (DSC), X-ray diffraction (XRD) and volumetric hydrogen desorption in a Sieverts-type apparatus. Alfa Aesar LiAlH₄ powder investigated in this work has the average particle size of 9.9 ± 5.2 μm as compared to 50-150 μm for Sigma-Aldrich LiAlH₄ investigated by Ares et al. [9]. High energy ball milling reduced the particle size of the present LiAlH₄ to 2.8 ± 2.3 μm. In general, comparing the results of our microstructural studies with those reported by Ares et al. [9] it is clear that the morphology, microstructure and chemistry of LiAlH₄ can be very dissimilar depending on the supplier from which LiAlH₄ powder was purchased. We do not observe a partial decomposition of LiAlH₄ during milling up to 5 h under high energy impact mode. The observed melting of LiAlH₄ in a DSC test is a very volatile event where the liquid LiAlH₄ starts foaming and flowing out of the alumina crucible. After completion of solidification and desorption at temperatures above melting the powder resembles a lava rock. A thermal sectioning in DSC tests at pre-determined temperatures and subsequent XRD studies show that LiAlH₄ starts decomposing into Li ₃AlH₆ immediately after melting. Li₃AlH ₆ seems to be already solidified before it starts decomposing in the next stage. All volumetric desorption curves at the 120-300 °C range clearly exhibit a two-stage desorption process, Stage I and II. As received LiAlH ₄ is able, in a fully solid state, to desorb at 120 °C under pressure of 0.1 MPa H₂ (atmospheric) as much as 7.1 wt.%H₂ within ～259,000 s (～72 h), i.e. ～93% of the purity-corrected H ₂ content from the reactions in Stage I (LiAlH₄(s) → (1/3)Li₃AlH₆(s) + (2/3)Al(s) + H₂) and Stage II ((1/3)Li₃AlH₆(s) → LiH + (1/3)Al + 0.5H ₂). The apparent activation energy for Stage I and II for unmilled LiAlH₄ is equal to ～111 and ～100 kJ/mol, respectively. For the ball milled LiAlH₄ the apparent activation energy for Stage I and II is slightly lower ～92.5 and ～92 kJ/mol, respectively. The water absorption up to 11.7% due to exposure to air for 1 h does not change in any drastic way the hydrogen desorption rate of ball milled LiAlH₄ in Stage I. Flammability tests show that the ball milled LiAlH₄ powder does not self-ignite on contact with air but can only be ignited by scraping the cylinder walls with a metal tool and then the powder burns with an open flame.

Full text of DOI:10.1016/j.jallcom.2010.05.059

Journal of Alloys and Compounds 504 (2010) 89–101
Contents lists available at ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Decomposition behavior of unmilled and ball milled lithium alanate (LiAlH4)
including long-term storage and moisture effects
R.A. Varin∗, L. Zbroniec
Department of Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1
a r t i c l e i n f o
a b s t r a c t
Article history:
A comprehensive study of the decomposition behavior of as received and mechanically (ball) milled
LiAlH4 has been carried out using differential scanning calorimetry (DSC), X-ray diffraction (XRD) and vol-
umetric hydrogen desorption in a Sieverts-type apparatus. Alfa Aesar LiAlH4 powder investigated in this
work has the average particle size of 9.9 5.2 ␮m as compared to 50–150 ␮m for Sigma–Aldrich LiAlH4
investigated by Ares et al. [9]. High energy ball milling reduced the particle size of the present LiAlH4
to 2.8 2.3 ␮m. In general, comparing the results of our microstructural studies with those reported by
Ares et al. [9] it is clear that the morphology, microstructure and chemistry of LiAlH4 can be very dissim-
ilar depending on the supplier from which LiAlH4 powder was purchased. We do not observe a partial
decomposition of LiAlH4 during milling up to 5 h under high energy impact mode. The observed melting
of LiAlH4 in a DSC test is a very volatile event where the liquid LiAlH4 starts foaming and flowing out of the
alumina crucible. After completion of solidification and desorption at temperatures above melting the
powder resembles a lava rock. A thermal sectioning in DSC tests at pre-determined temperatures and sub-
sequent XRD studies show that LiAlH4 starts decomposing into Li3AlH6 immediately after melting. Li3AlH6
seems to be already solidified before it starts decomposing in the next stage. All volumetric desorption
curves at the 120–300 ◦C range clearly exhibit a two-stage desorption process, Stage I and II. As received
LiAlH4 is able, in a fully solid state, to desorb at 120 ◦C under pressure of 0.1 MPa H2 (atmospheric) as much
as 7.1 wt.%H2 within ∼259,000 s (∼72 h), i.e. ∼93% of the purity-corrected H2 content from the reactions in
Stage I (LiAlH4(s) → (1/3)Li3AlH6(s) + (2/3)Al(s) + H2) and Stage II ((1/3)Li3AlH6(s) → LiH + (1/3)Al + 0.5H2).
The apparent activation energy for Stage I and II for unmilled LiAlH4 is equal to ∼111 and ∼100 kJ/mol,
respectively. For the ball milled LiAlH4 the apparent activation energy for Stage I and II is slightly lower
∼92.5 and ∼92 kJ/mol, respectively. The water absorption up to 11.7% due to exposure to air for 1 h does
not change in any drastic way the hydrogen desorption rate of ball milled LiAlH4 in Stage I. Flammability
tests show that the ball milled LiAlH4 powder does not self-ignite on contact with air but can only be
ignited by scraping the cylinder walls with a metal tool and then the powder burns with an open flame.
© 2010 Elsevier B.V. All rights reserved.
Received 25 April 2010
Received in revised form 5 May 2010
Accepted 7 May 2010
Available online 20 May 2010
Keywords:
Solid state hydrogen storage
Hydrogen storage materials
Desorption temperature and kinetics
Lithium alanate (LiAlH4)
Ball milling
X-ray diffraction (XRD)
Differential scanning calorimetry (DSC)
1. Introduction
namic barrier for hydrogen desorption although a kinetic barrier
may still remain. The latter could be dealt with by using suitable
One of the most critical issues to be solved before the full imple-
mentation of the hydrogen economy is hydrogen storage. It is
imperative that it has to be solved before a technically and eco-
nomically viable hydrogen economy can be established. Current
methods of hydrogen storage such as compressed and liquid hydro-
gen have serious drawbacks as storage media [1]. In a long run, the
most promising is hydrogen storage system based on solid hydrides
or their mixtures (composites) having hydrogen capacities exceed-
ing 6 wt.% and the enthalpies of decomposition/reaction within the
20–50 kJ/mol range [1]. A low enthalpy removes the thermody-
catalysts. Hence, vigorous research efforts are necessary in that
field.
Lithium alanate (LiAlH4) hydride is one of the most interest-
ing candidate hydrides for solid state hydrogen storage. LiAlH4 has
been classified by Graetz and Reilly [2], as belonging to the fam-
ily of “kinetically stabilized hydrides” which are characterized by
low reaction enthalpies and extremely high equilibrium hydro-
gen pressure at relatively low temperatures. They are metastable
though and do not spontaneously decompose owing to not-well
understood kinetic limitations. One of these kinetically stabilized
hydrides is LiAlH4 which has the following virtues: (a) a low density
which allows for a high theoretical hydrogen capacity (∼10.6 wt.%),
(b) low decomposition enthalpies translating to low decomposition
temperatures expected and (c) a relatively large theoretical hydro-
∗
Corresponding author. Tel.: +1 519 888 4567; fax: +1 519 885 5862.
E-mail address: ravarin@mecheng1.uwaterloo.ca (R.A. Varin).
0925-8388/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2010.05.059

90
R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
gen quantity of about 7.9 wt.% that could be liberated below 250 ◦C.
Some drawbacks will include the presence of exothermic decom-
position reaction, extremely high plateau pressure at relatively low
temperatures which makes the hydride practically irreversible and
relatively slow hydrogen desorption rates. The latter can be much
improved by adding catalytic metal chlorides. This discovery led to
massive research efforts on the effect of catalytic metal chlorides
resulting in a large number of publications which are reviewed
in [1]. However, it is important that the decomposition behavior
of pristine LiAlH4 must be clearly elucidated before one can try
to improve its hydrogen storage properties by catalytic additives.
This is an on-going research in our laboratory. The decomposi-
tion behavior of pristine LiAlH4 and the effects of ball milling on
its behavior have been investigated for some time [3–9]. How-
ever, some aspects of decomposition behavior of pristine LiAlH4
still remain poorly understood, such as, for example, the problem
of mechanically driven partial decomposition of LiAlH4 during ball
milling [9].
For the first time a thorough morphological and microstruc-
tural investigations of the as received and ball milled powders
are presented in this work. In order to assess the effect of milling
energy on the microstructure and decomposition behavior ball
milling was carried out under controlled milling conditions with
varying energy using a magneto-mill [1]. A thermal sectioning
in DSC at pre-determined temperatures combined with a phase
analysis using X-ray diffraction was used to monitor the evo-
lution of phases during thermal decomposition of as received
and ball milled LiAlH4. A volumetric hydrogen desorption anal-
ysis under a technological condition of atmospheric pressure of
0.1 MPa H2 has been applied to study the decomposition behavior
of as received and ball milled LiAlH4. A very wide decomposition
temperature range from 120 to 300 ◦C was studied. The appar-
ent activation energy of hydrogen desorption was estimated using
the JMAK (Johnson–Mehl–Avrami–Kolmogorov) theory of phase
transformations. By comparison with powders used in [9] it is
shown that profound morphological/microstructural differences
exist between LiAlH4 powders purchased from various suppliers
which can affect milling conditions and subsequent decompo-
sition properties which may lead to discrepancies between the
results reported in the literature. An assessment of the exposure
to moisture on the desorption behavior of LiAlH4 and its flamma-
bility/pyrophoricity are also reported.
Fig. 1. A schematic of the cross-section of the milling vial and the approximate
trajectories of the ball movement under IMP68 mode of milling in the magneto-mill
Uni-Ball-Mill 5.
The nanograin/crystallite size of phases residing in the milled powders was cal-
culated from the broadening of their respective XRD peaks. Since the Bragg peak
broadening in an XRD pattern is due to a combination of grain refinement and lat-
tice strains, it is customary to use computing techniques by means of which one
can separate these two contributions. The separation of crystallite size and strain
was obtained from Cauchy/Gaussian approximation by the linear regression plot
according to the following equation [13]:
ꢀ
ꢁ
ı2(2ꢁ)
Kꢀ
L
ı(2ꢁ)
tan ꢁsin ꢁ
=
+ 16e2
(1)
tan2
ꢁ
where the term Kꢀ/L is the slope, the parameter L is the mean dimension of the
nanograin (crystallite) composing the powder particle, K is constant (∼1) and e
is the so-called “maximum” microstrain (calculated from the intercept), ꢀ is the
wave length and ꢁ is the position of the analyzed peak maximum. The term
ı(2ꢁ) = B[1 − (b2/B2)] (rad) is the instrumental broadening-corrected “pure” XRD
peak profile breadth [13] where B and b are the breadths in radians of the same
Bragg peak from the XRD scans of the experimental and reference powder, respec-
tively. They were calculated by the software TracesTM v. 6.5.1 as the full-widths at
half maximum, FWHM, after background removal. A compound LaB6, the National
Institute of Standards and Technology (NIST) standard reference material (SRM) 660
was used as a reference for subtracting the instrumental broadening from FWHM. It
must be noted that when the FWHM of the instrumental line profiles were obtained
in this manner, the Bragg peaks for the LaB6 SRM were at different 2ꢁ angles than
those of the analyzed phases in the milled powders. The interpolated FWHM values
between angles for the SRM peaks were found using a calibration curve.
2. Experimental
As received commercial LiAlH4 (97% purity) from Alfa Aesar was used in this
work. Controlled Mechanical Milling (CMM) was carried out in ultra-high purity
hydrogen gas atmosphere (purity 99.999%: O2 < 2 ppm; H2O < 3 ppm; CO2 < 1 ppm;
N2 < 6 ppm; CO < 1 ppm; THC < 1 ppm) under ∼600 kPa pressure in the magneto-mill
Uni-Ball-Mill 5 manufactured by A.O.C. Scientific Engineering Pty Ltd., Australia
[1,10–12]. In this particular ball mill the milling modes with varying milling energy
can be achieved by using one or two strong NdFeB magnets, changing their angular
positions and changing the number of hard steel balls (25 mm in diameter each) in
a milling vial. In order to vary milling energy we used 1 magnet and 4 balls with the
ball-to-powder weight ratio (R) ∼40 (R40) as well as IMP68 mode with the ball-to-
powder weight ratio R40 or R132 and the rotational speed of milling vial ∼200 rpm.
Fig. 1 shows a schematic set-up for a strong impact mode IMP68 with two magnets
positioned at 6 and 8 o’clock, at the distance of 10 and 2 mm, respectively, from the
milling vial (working distance – WD) and 4 hard steel balls in the vial. During milling
for 15 min the vial was not cooled but during milling for 2 and 5 h the milling vial
was continuously cooled using a fan.
After loading with powder, an air-tight milling vial with an O-ring, equipped
with a pressure valve mounted in the vial’s lid, was always evacuated and purged
several times with ultra-high purity argon (Ar) gas (99.999% purity) before final
pressurization with H2. All the powder handlings were performed in a purged glove
box in ultra-high purity argon atmosphere.
The crystalline structure of powders was characterized by Bruker D8 diffrac-
tometer using a monochromated Cu K␣1 radiation (ꢀ = 0.15406 nm) produced at
an accelerating voltage of 40 kV and a current of 30 mA. The scan range was from
2ꢁ = 10◦ to 90◦ and the scan rate was 3◦min−1. Powder was loaded into an argon-filled
environmental holder in a glove box not allowing any exposure to the environment.
The images of powders were obtained using a high-resolution, field emission
SEM (FE SEM) LEO 1530. The average size of the powder particles was calculated
as the particle Equivalent Circle Diameter, ECD = (4A/ꢂ)1/2, where A represents the
projected particle area. The projected area was estimated by attaching loose powder
to a sticky carbon tape and taking pictures under secondary emission (SE) mode in a
LEO 1530. The SEM images were subsequently analyzed by the Image Tool software.
Around 1200 particles were evaluated.
The hydrogen desorption was evaluated using a second generation volumet-
ric Sieverts-type apparatus custom-built by A.O.C. Scientific Engineering Pty Ltd.,
Australia. This apparatus built entirely of austenitic stainless steel allows loading
of a powder sample into a tightly sealed austenitic stainless steel sample reactor
in a glove box under argon and subsequent transfer of the reactor to the main unit
without any exposure to the environment. The weight of the powder sample in the
desorption experiments was in the range of 20–30 mg. The calibrated accuracy of
desorbed hydrogen capacity is about 0.1 wt.%H2 and that of temperature read-
ing 0.1 ◦C. Before starting the desorption test, the inner tubing of the apparatus
and reactor were evacuated and purged 3 times with argon and finally with hydro-
gen. The moveable furnace of the apparatus was heated separately to the desired
test temperature and subsequently inserted onto a sample reactor inside which an
atmospheric pressure of 0.1 MPa H2 was kept. Hence, the beginning of the desorp-
tion test was in reality pseudo-isothermal before the powder sample temperature
reached the desired value. However, the calibrated time interval within which the
powder sample in the reactor reaches the furnace temperature is ∼400–600 s in the
100–450 ◦C range, which is quite negligible compared to the desorption completion
time. Therefore, one can consider the test as being “isothermal” for any practical

R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
91
Fig. 2. (a) Scanning electron backscatter micrograph of the as received LiAlH4 particles. (b) A fit of the as received LiAlH4 particle distribution histogram by an extreme
probability distribution function.
purposes at this range of temperatures. The amount of desorbed hydrogen was cal-
culated from the ideal gas law as described in detail in [1]. Hydrogen desorption
curves were also corrected for the hydrogen gas expansion due to the increase in
temperature.
The activation energy for desorption process was estimated from the obtained
Sieverts’ desorption curves at corresponding temperatures using the Arrhenius plot
of k values with temperature [1]:
3. Results and discussion
3.1. Morphology and microstructure of LiAlH4 before and after
ball milling
Fig. 2a and b shows the SEM micrograph of the morphology and
particle size distribution of the as received LiAlH4 powder, respec-
tively. The individual LiAlH4 particles have a “blocky” shape which
is most likely related to the monoclinic crystal lattice of LiAlH4
(space group P21c) [7]. The size distribution seems to be quite uni-
form. The particle size distribution in Fig. 2b shows that the best
fitting is obtained by the extreme value type I (maximum) dis-
tribution function [15]. The fitting by a log-normal function gave
slightly worse fit. From measuring ∼1200 particles in SEM the aver-
age ECD particle size with standard deviation (S.D.) is obtained as
9.9 5.2 ␮m. It is to be pointed out that the morphology of the
present as received LiAlH4 purchased from Alfa Aesar is very dif-
ferent than that of LiAlH4 powder investigated by Ares et al. [9] who
purchased their powder from Sigma–Aldrich. The latter as received
powder exhibited much larger particle sizes having a distinctly
bi-modal distribution centered around ∼50 and ∼150 ␮m.
/RT
k = k0e−E
(2)
A
where EA is the activation energy, R is the gas constant and T is the temperature.
The rate constant k was determined using the Johnson–Mehl–Avrami–Kolmogorov
(JMAK) equation [1]:
ꢃ
˛ = 1 − e−(k·t)
(3)
where ꢃ is the reaction exponent (the Avrami exponent) related to the transforma-
tion mechanism, taken as a free value characteristic for each individual temperature
[1,14] rather than a fixed value for all temperatures, and ˛ is the desorption fraction
at time t.
The thermal behavior of powders was studied by differential scanning calorime-
try (DSC) (Netzsch 404) of ∼6 mg powder sample with heating rate of 10 ◦C/min and
argon flow rate of 100 ml/min. The powder was transported to a DSC instrument in
a glass vial filled with Ar then quickly loaded into an Al2O3 crucible with a lid and
then into the air-tight DSC chamber. This operation took about 2–3 min and only for
that short period of time the powder could be in contact with air.
The moisture effects were studied by exposing a powder of LiAlH4 milled for
15 min under IMP68 mode to air with 47% relative humidity for 1 h. The mass of
the powder was measured before and after exposure from which the percentage of
absorbed H2O was calculated.
Finally, it must be pointed out that all the tests described in the following sec-
tions were carried out within a couple of days from the completion of ball milling.
The effect of ball milling on the particle morphology of LiAlH4 is
shown in the SEM micrographs in Fig. 3a and b for a LiAlH4 powder
milled under high energy IMP68 mode for 2 h. Measuring of ∼1200
particles at the magnification of 2 K gave the average ECD value of
2.8 2.3 ␮m. Fig. 3b shows a particle size distribution histogram for
Fig. 3. (a) Scanning secondary electron micrograph of LiAlH4 after ball milling for 2 h under high energy impact mode IMP68 with R132. (b) Particle size (ECD) distribution
for milled powder.

92
R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
Fig. 4. XRD patterns of LiAlH4 as received and ball milled under varying energy milling modes and milling times. JCPDS file numbers for phase identification are given in
parenthesis.
this powder. As can be seen, in this case the best fit line is obtained
by a classical log-normal function [15] rather than by the extreme
value type I (maximum) as the one for the as received powder. A
long ball milling of LiAlH4 usually leads to more agglomeration and
is not recommended. For example, the average ECD particle size
for the powder milled for 20 h under high energy shearing mode
(HES57) was observed to be 5.2 4.3 ␮m [1].
It must be noted that the average ECD particle size of ball milled
LiAlH4 obtained in this work is much smaller than the average par-
ticle size reported by Ares et al. [9] which after 30 min of milling
of their Sigma–Aldrich LiAlH4 was 80 40 ␮m and after 1.5 h of
milling it was still 30 15 ␮m. Comparing with the milling results
reported in [9] it is clear that the smaller initial as received particle
size of LiAlH4 powder in this work results in a greater particle size
reduction during milling. Obviously, also the milling technique may
be to a certain extent responsible for the final particle size reduc-
tion during milling. For comparison, Ares et al. [9] used Fritsch P7
planetary ball mill.
Fig. 4 shows the XRD patterns of LiAlH4 after ball milling with
varying milling energy (number of magnets and ball-to-powder
weight ratio R) compared to the XRD pattern of as received LiAlH4.
The microstructure of as received LiAlH4 comprises, besides LiAlH4
also a small amount of Al and LiOH–H2O. The former is most likely
an impurity and LiOH–H2O might be a result of inadvertent con-
tact of powder with air during handling despite the fact that all
handling was done in a purged glove box filled with ultra-high
purity Ar and powders for XRD tests were loaded into an envi-
ronmental holder under ultra-high purity Ar. Similar observation
was reported by Blanchard et al. [16] for LiAlD4 (deuteride) where
they also found LiOH–H2O. In contrast, as received LiAlH4 from
Sigma–Aldrich studied by Ares et al. [9] did not contain any Al impu-
rities but instead contained a small amount of LiCl impurity. Again,
this comparison shows that LiAlH4 obtained from different sup-
pliers can have not only different morphology/microstructure but
also different type of impurities which can be responsible for some
differences in the results reported in the literature as discussed
below.
Al and H2 during prolonged milling up to 110 h. It must, however, be
pointed out that Andreasen et al. [8] claimed that a partial decom-
position of LiAlH4 might have occurred during milling up to 10 h
although no clear evidence was presented to this claim. Ares et
al. [9] reported that LiAlH4 partially decomposed after ball milling
for barely 1.5 h which was reflected in a diminution of hydrogen
capacity as compared to as received LiAlH4 although XRD patterns
showed LiAlH4 as the only phase existing in the milled powder.
They also detected Al at the surface of particles. We do not confirm
a partial decomposition of LiAlH4 after milling up to 5 h as men-
tioned above. As shown in Fig. 4 in our Alfa Aesar powder Al seems
to be an impurity and is not a product of decomposition since we do
not observe a clear diminution of the overall hydrogen capacity for
LiAlH4 ball milled 15 min as compared to the as received powder
as will be discussed later.
Table 1 shows the comparison of grain/crystallite size of LiAlH4
before and after ball milling calculated from the broadening of
diffraction peaks (Eq. (1)). Excellent coefficients of fit to Eq. (1)
give a testimony to the accuracy of the method. The most inter-
esting finding is that the grain/crystallite size of as received LiAlH4
has an average size of ∼30 nm and after milling under high energy
mode IMP68 the grain sizes are, on average, larger than that of as
received LiAlH4. For powder milled using a continuous cooling of
the vial the grain sizes are slightly smaller than those formed for
non-cooled samples. Furthermore, low energy milling mode with 1
magnet does not change the grain size as compared to as received
material. Such a behavior suggests some possibility of localized
heating during high energy ball milling resulting in a grain growth
of LiAlH4. However, the amount of heat produced is still insufficient
to induce a partial decomposition of LiAlH4 during milling at least
to the extent we could observe from the capacity changes. The grain
size of present LiAlH4 is different than that reported by Ares et al.
Table 1
Grain size of unmilled and ball milled LiAlH4.
Milling mode
Cooling during
milling
Grain size (nm) R2
It can be seen in Fig. 4 that the milling mode or R parameter does
not affect the phase presence in any major way. Also, milling up to
5 h does not lead to a partial decomposition of LiAlH4. Our finding
is in accord with the results of Balema et al. [6] who reported that
LiAlH4 revealed good stability up to 35 h of high energy milling.
They also observed a partial decomposition of LiAlH4 into Li3AlH6,
As received
1 magnet–4 balls-R40–15 min No
IMP68–R40–15 min
IMP68–R132–15 min
IMP68–R132–2 h
IMP68–R132–5 h
–
28
23
64
78
43
52
0.968
0.992
0.981
0.935
0.972
0.988
No
No
Yes
Yes

R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
93
Fig. 6. Optical photo showing 6.2 mg of as received LiAlH4 powder solidified after
melting in an alumina crucible used for a DSC test. Heating rate is 10 ◦C/min.
The above reaction according to Block and Gray [3] was related
to the evolution of a small quantity of hydrogen.
The endotherm (2) is commonly related to the melting of LiAlH4
[1] according to
LiAlH4(solid) → LiAlH4(liquid)
(5)
The melting of LiAlH4 is a very volatile event. Fig. 6 shows an
optical photograph of a small amount of as received LiAlH4 powder
which was tested in DSC at 10 ◦C/min up to 500 ◦C. Immediately
after the beginning of melting the liquid LiAlH4 started foaming
and flowing out of the alumina crucible raising the alumina lid as
can be seen in Fig. 6. The solidified powder obtained after desorp-
tion resembled a lava rock. The melting enthalpy of reaction (2) is
quoted in the literature as 13 kJ/(mol LiAlH4) [9] and 16.6 kJ/(mol
LiAlH4) [17].
Fig. 5. DSC curvesforLiAlH4 asreceived and ballmilledundervarying energy milling
modes. (a) As received, (b) milled for 15 min under low energy shearing (LES) mode
with 1 magnet, 4 balls and R40 and (c) milled for 15 min under high energy milling
mode IMP68 and R40. Heating rate is 10 ◦C/min.
In a molten state LiAlH4 starts desorbing hydrogen in the fol-
lowing exothermic reaction [1]
[9] for their Sigma–Aldrich LiAlH4 whose average as received grain
(crystallite) size was greater than 300 nm and decreased to about
90 30 nm after 30 min of milling with further milling having no
influence on grain (crystallite) size.
It is clear from the comparison of microstructural investigations
of the present LiAlH4 with those reported by Ares et al. [9] that the
morphology, microstructure and chemistry of LiAlH4 can be very
dissimilar depending on the supplier from which LiAlH4 powder
was purchased and may affect the storage properties.
LiAlH4(l) → (1/3)Li3AlH6(s) + (2/3)Al(s) + H2
(6)
where l-liquid and s-solid. According to a general consensus
[1] reaction (6) proceeds with a theoretical hydrogen release of
5.3 wt.%. The enthalpy of reaction (6) is quoted as being equal
to experimentally observed value of −14 kJ/(mol H2) [9] and
−10 kJ/(mol H2) [18] or alternatively equal to the two calculated
values of 9.79 and 3.46 kJ/mol [19]. The latter two values apparently
correspond to an endothermic reaction rather than to an exotherm.
It seems that there is some confusion regarding the nature of reac-
tion (6).
3.2. Thermal decomposition and microstructural evolution
during DSC test
Endothermic reaction (7) is supposed to occur completely in a
solid state according to the following path [1]
Fig. 5 shows a comparison of DSC traces for as received LiAlH4
and that milled under varying energy of milling. In agreement
with the early studies of Block and Gray [3] and McCarty et al. [4]
on the thermal decomposition of LiAlH4, two distinct exotherms
(designated 1 and 3 in Fig. 5a) and three endotherms (designated
2, 4 and 5 in Fig. 5a) are observed. All DSC curves regardless of
milling mode exhibit these five thermal peaks. The first exotherm
1 was assigned by Block and Gray [3] to the reaction of the surface
aluminum-hydroxyl groups owing to the presence of impurities of
the following type
(1/3)Li3AlH6 → LiH + (1/3)Al + 0.5H2
(7)
It proceeds with a theoretical hydrogen release of 2.6 wt.%. Its
enthalpy is quoted as being equal to 15 kJ/(mol H2) [9] or 25 kJ/(mol
H2) [17]. It must be mentioned that at low heating rates in DSC
of 1–2 ◦C/min some researchers reported no sign of melting [4,9]
and the process of melting and solidification is substituted by one
endothermic reaction [9].
Reactions (8a) and (8b) occur at temperatures around
400–450 ◦C [1] according to the following path
> Al-OH + H–Al < → > Al–O–Al < + H2
(4)
LiH → Li + 0.5H2
(8a)

94
R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
Fig. 7. Isothermal sectioning of DSC test for LiAlH4 powder ball milled for 15 min under low energy shearing (LES) mode with one magnet, 4 balls and R40. (a) DSC test
stopped at 165 ◦C (after peak 1), (b) DSC test stopped at 190 ◦C (after peak 2), (c) DSC test stopped at 210 ◦C (after peak 3), and (d) DSC test stopped at 300 ◦C (after peak 4).
Heating rate is 10 ◦C/min.
or alternatively [19]
LiH + Al → LiAl + 0.5H2
(8b)
with the enthalpy of 140 kJ/mol [18] for a LiH decomposition and a
hydrogen release of 2.6 wt.%.
In order to shed more light on the nature of the above reactions
we did a thermal sectioning in DSC at pre-determined tempera-
tures. As shown in Fig. 7 each DSC test was stopped at a temperature
of 165, 190, 210 and 300 ◦C beyond each reaction peak 1, 2, 3 and
4, respectively. After stopping the test the powder sample from a
crucible was immediately taken for an XRD test. The corresponding
XRD patterns are shown in Fig. 8. The sample heated up to 165 ◦C,
beyond the peak (1) of reaction (4) (Fig. 7a), still shows the presence
of a majority phase LiAlH4 and minority impurity phases of Al and
LiOH–H2O (same as in the as received powder in Fig. 4). On the basis
of this pattern it is hard to verify whether or not reaction (4) indeed
occurred. The sample heated up to 190 ◦C, just beyond the peak (2)
Fig. 9. Enlargement of Al peaks (1 1 1) and (2 0 0) from XRD patterns for DSC thermal
sectioning test up to 190, 210 and 300 ◦C with overlapped XRD (1 1 1) and (2 0 0)
peaks of as received LiH.
Fig. 8. XRD patterns taken from the powder sample obtained in a DSC thermal
sectioning test up to 165, 190, 210 and 300 ◦C. JCPDS file numbers for phase identi-
fication are given in parenthesis.

R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
95
Fig. 10. Volumetric desorption curves for as received LiAlH4 under 0.1MPa H2 pressure (atmospheric) at (a) 120 ◦C up to 259,000 s (desorbed 7.1 wt.%H2), (b) 140 ◦C up to
53,000 s (desorbed 7.5 wt.%H2), (c) 150 ◦C up to 48,000 s (desorbed 7.7 wt.%H2) and (d) 160 ◦C up to 46,000 s (desorbed 7.6 wt.%H2).
of reaction (5) (Fig. 7b), shows a clear presence of the Li3AlH6 and
Al phases and no presence of LiAlH4. This means that LiAlH4 starts
decomposing very fast into Li3AlH6 immediately after melting and
a half way at a temperature between peaks (2) and (3), LiAlH4 is
already fully decomposed. This also implies that Li3AlH6 is already
solidified before it starts decomposing. Therefore, it seems that the
exothermic character of peak (3) (Figs. 5a and 7) may not only
arise due to the exothermic decomposition of LiAlH4 as reported
in the literature [1] but also due to the exothermic solidification
event of Li3ALH6, i.e. these two exo events possibly superimpose.
The sample heated to 210 ◦C beyond the peak (3) of the supposed
reaction (6) (Fig. 7c) shows in Fig. 8 very weak diffraction peaks
of Li3AlH6, strong diffraction peaks of Al and quite pronounced
diffraction peaks of LiOH. It is clear that in the temperature range
between 190 and 210 ◦C, i.e. between peaks (2) and (3), the major-
ity of Li3AlH6 is already decomposed. The XRD of the sample heated
up to 300 ◦C, i.e. beyond the peak (4) of the supposed reaction (7),
shows the presence of only Al, LiOH and possibly Li and no presence
of Li3AlH6 (observed Cu peak arises from the substrate in the envi-
ronmental holder). It is clear then that the endothermic peak (4)
is related to the decomposition of the small amount of remaining
Li3AlH6 rather than the bulk of this phase as it has been a commonly
held belief.
However, a completion of the decomposition of the remaining
Li3AlH6 in reaction (7) requires the formation of LiH whose peaks
are evidently not present in the XRD patterns in Fig. 8 after DSC
Fig. 11. XRD pattern of as received LiAlH4 powder desorbed at 120 ◦C for 259,000 s (desorption curve in Fig. 10a). JCPDS file numbers for phase identification are given in
parenthesis.

96
R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
inadvertently embedded in the internal rubber O-ring during load-
ing the powder into the XRD holder in a glove box the lower and
upper part of the holder might not be tightly sealed allowing some
minutiae air leakage into the sample chamber. From our work on
the properties of LiH which will be published separately we have
found that LiH is extremely sensitive to even minutiae quantities
of moisture in atmosphere immediately converting partially into
LiOH.
3.3. Volumetric desorption behavior, apparent activation energy
for hydrogen desorption and corresponding microstructural
evolution
Fig. 10 shows the volumetric desorption curves (Sieverts) of as
received LiAlH4 obtained at various, relatively low temperatures,
which are lower or only slightly higher than the temperature of
peak (1) in Figs. 5 and 7. All desorption temperatures are decidedly
below the melting peak (2) of LiAlH4. All desorption curves clearly
exhibit a two-stage desorption process as designated by Stage I and
II in Fig. 10a. The most striking discovery observed in Fig. 10a is that
as received LiAlH4, fully in a solid state, is able to desorb at 120 ◦C
under pressure of 0.1MPa H2 as much as 7.1 wt.%H2 after ∼259,000 s
(∼72 h), i.e. ∼93% of the purity-corrected H2 content from reactions
(6) and (7). This is just a matter of appropriately long desorption
time. So far, such a result has never been reported. In order to con-
firm this behavior we conducted a second desorption test which
resulted in a nearly the same desorption curve 2 as compared to
the curve 1 of the first test. The powder from sample 2 after des-
orption for ∼72 h was tested by XRD as indicated in the legend in
Fig. 10a. Since reaction (6) alone (now in solid state) would provide
only 5.3 wt.%H2, therefore, the amount of ∼7.1 wt.%H2 desorbed
at 120 ◦C clearly shows that desorption process must have passed
through reaction (6) fully and reaction (7) partially, i.e. Stage I and
Fig. 12. A schema showing the times of the start and end of the decomposition reac-
tion as well as maximum capacities in Stage I and II. H – instantaneous desorption
capacity, Hc,max – total maximum hydrogen desorption capacity.
test up to 210 and 300 ◦C. It has been argued in the literature (e.g.
[4,9]) that the Bragg peaks of Al and LiH superimpose to such an
extent that it is impossible to discern between them. In order to
verify this notion we enlarged (1 1 1) and (2 0 0) Al peaks from the
XRD patterns taken from the powders after DSC up to 190, 210
and 300 ◦C and superimposed on them the peaks of as received
LiH as shown in Fig. 9. It is clear that the Al and LiH peak posi-
tions are split to the extent that their differentiation should not
create any serious problem. As such Fig. 9 confirms that there is,
indeed, no LiH present after decomposition of Li3AlH6 in a DSC test
(Fig. 8). Despite the precautions taken some air might have leaked
into an XRD environmental holder hydrolyzing LiH in the powder
and converting it into LiOH whose peaks are clearly seen in Fig. 8
after DSC tests at 210 and 300 ◦C. If a few powder particles were
Fig. 13. The Arrhenius plots of rate constant k with temperature for calculating the activation energy of hydrogen desorption for as received LiAlH4. (a) Stage I and (b and c)
Stage II at a desorption curve.

R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
97
Fig. 14. Volumetric desorption curves under 0.1 MPa H2 pressure (atmospheric) at (a) 130 ◦C up to 78,000 s (desorbed 7.4 wt.%H2), (b) 140 ◦C up to 32,000 s (desorbed
7.5 wt.%H2), (c) 155 ◦C up to 17,000 s (desorbed 7.6 wt.%H2) and (d) 170 ◦C up to 23,000 s (desorbed 7.6 wt.%H2) for LiAlH4 ball milled for 15 min under high energy impact
IMP68 mode with R132.
II, without any melting of the sample as the desorption temper-
ature is much below the melting point observed in Figs. 5 and 7.
Increasing desorption temperatures to 140, 150 and 160 ◦C slightly
increases the quantity of desorbed hydrogen to 7.5, 7.7 and 7.6 wt.%
and substantially reduces the desorption time to ∼53,000, 48,000
and 46,000 s, respectively. The quantities of desorbed H2 (∼100%
of the purity-corrected H2 content) clearly indicate that at temper-
atures higher than 120 ◦C the desorption process must also have
passed through reaction (6) and nearly fully completed reaction (7).
These observations clearly show that a LiAlH4 powder can desorb
a nearly full purity-corrected quantity of hydrogen under 0.1 MPa
pressure of H2 due to the decomposition of both LiAlH4 and Li3AlH6
and remain in a solid state throughout the entire decomposition
process. An XRD pattern in Fig. 11 obtained from a powder decom-
posed at 120 ◦C confirms this notion. Very strong Al peaks and weak
peaks of Li3AlH6 are observed which indicate that the powder is
near the completion of reaction (7). A small shoulder is observed
near the foot of the (2 0 0) Al peak which corresponds to the exact
position of (2 0 0) 100% intensity peak of LiH. In addition, visual
observations of the granular nature of the decomposed powder
clearly confirm that no melting occurred during desorption at the
120–160 ◦C range.
In order to get rid of the initial slack (incubation period) in the
desorption curves, usually appearing at low desorption tempera-
tures (e.g. Fig. 10a), the zero time for calculations with Eq. (3) was
taken at the intersection of the tangent line to the linear portion of
Fig. 15. The Arrhenius plots of rate constant k with temperature for estimation of the activation energy of hydrogen desorption for ball milled LiAlH4. (a) Stage I and (b) Stage
II at a desorption curve.

98
R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
Fig. 16. Volumetric desorption curves under 0.1 MPa H2 pressure (atmospheric) at (a) 190 ◦C up to 2900 s (desorbed 6.7 wt.%H2), (b) 190 ◦C up to 38,000 s (desorbed 7.7 wt.%H2),
(c) 210 ◦C up to 19,300 s (desorbed 7.6 wt.%H2) and (d) 300 ◦C up to 21,400 s (desorbed 7.9 wt.%H2) for LiAlH4 ball milled for 15 min under low energy shearing (LES) with
one strong magnet and R40.
judgment which value is more correct just based on the value of R2.
A common sense must prevail.
the desorption curve and the time axis as shown schematically in
Fig. 12. Subsequently, for Stage I decomposition, start of the reac-
tion is counted from time t0 such that t1ꢀ − t0, tnꢀ − t0, etc. and a
maximum capacity for this stage is taken as Hcꢀ . For Stage II decom-
position, start of the reaction is counted from time tcꢀ such that
t1ꢀꢀ − tcꢀ , tnꢀꢀ − tcꢀ , etc. with a maximum capacity for this stage calcu-
lated as (Hc,max − Hcꢀ ) where Hc,max is a total hydrogen desorption
capacity. The free reaction exponent ꢃ for each temperature [1,14]
and the corresponding rate constant k in Eq. (3) are calculated
from a linear interpolation of the typical plots ln[−ln(1 − ˛)] vs.
ln t [1]. The instantaneous value of ˛ at any time t between t1ꢀ and
tnꢀ is calculated as ˛ = H(t)/Hcꢀ for Stage I and between t1ꢀꢀ and tnꢀꢀ as
˛ = H(t)/(Hc,max − Hcꢀ ) for Stage II.
The obtained values of apparent activation energy can be com-
pared with those reported in the literature. For Stage I (reaction
(6)) the following values were reported: ∼122 kJ/mol (Kissinger
method) [1], ∼99.6 kJ/mol (isothermal) [4], ∼82 kJ/mol (a single
desorption curve fitting) [8], ∼115 kJ/mol (Kissinger method) [9],
∼81 kJ/mol (Kissinger method) [20], and ∼102 kJ/mol (LiAlD4; syn-
chrotron X-ray diffraction method) [21]. By comparison, the value
of apparent activation energy obtained in the present work for
Stage I is slightly higher than that reported in [21] and slightly lower
than that reported in [1,9]. However, we do not support a conclu-
sion derived by Ares et al. [9] who claimed on the basis of apparent
activation energy values measured at low and high heating rates
in DSC that the release of hydrogen after melting is more favorable
than the release of hydrogen without melting. In our case hydro-
gen was released in Stage I (reaction (6)) without melting with the
apparent activation energy of ∼111 kJ/mol which compares very
well with the apparent activation energy of ∼115 kJ/mol reported
by Ares et al. [9] for the decomposition after melting. It seems clear
to us that there is no such a great difference between the apparent
activation energy of decomposition of LiAlH4 into Li3AlH6 in a solid
and molten state.
Fig. 13 shows the Arrhenius plots for the estimate of the appar-
ent activation energy for as received LiAlH4 in Stage I and II which
also shows excellent coefficients of fit to Eq. (2) giving a testi-
mony to the accuracy of the method. Fig. 13a shows that for Stage
I (reaction (6) in a solid state) the apparent activation energy is
∼111 kJ/mol. Fig. 13b and c shows that for Stage II (reaction (7))
quite dissimilar apparent activation energies are obtained when
the apparent activation energy is calculated either from 4 des-
orption curves at 120, 140, 150 and 160 ◦C or from 3 desorption
curves at 140, 150 and 160 ◦C. From 4 desorption curves a value of
∼100 kJ/mol is obtained vis-à-vis ∼63.5 kJ/mol for 3 curves. Such a
problem has been encountered from time to time in our research on
solid state hydrogen storage [1] which shows that one must exer-
cise an utmost care in the estimate of apparent activation energy
since the slope of the Arrhenius plot is very sensitive to very small
changes in the position of experimental data points (i.e. tempera-
tures taken for the estimate). Since in both cases the coefficient of
fit R2 in the Arrhenius plot is very good, it is hard to make a clear
For Stage II (reaction (7)) the following values of appar-
ent activation energy were reported: ∼153 kJ/mol (Kissinger
method) [1], ∼196 kJ/mol (isothermal) and ∼100 kJ/mol (Kissinger
method) [4], ∼90 kJ/mol (a single desorption curve fitting) [8],
∼86 kJ/mol (Kissinger method) [9], and ∼108 kJ/mol (Kissinger
method) [20]. Our present value of ∼100 kJ/mol compares
favorably with nearly all the literature reported values except
∼153 kJ/mol in [1] and ∼196 kJ/mol in [4] which seem to be

R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
99
amount of ∼7.5 wt.%H2 is achieved in a shorter time than the time
needed to desorb the same quantity of H2 from as received LiAlH4
(Fig. 10). The Arrhenius plots for the estimate of apparent acti-
vation energy of ball milled LiAlH4 are shown in Fig. 15. Fig. 15a
shows that for Stage I (reaction (6) in a solid state) the apparent
activation energy is ∼92.6 kJ/mol and Fig. 15b shows that that for
Stage II (reaction (7)) the apparent activation energy is ∼92 kJ/mol.
Both values are quite similar but slightly lower than the corre-
sponding values for as received LiAlH4 which is in accord with
slightly faster desorption rates observed in Fig. 14 as compared to
Fig. 10.
In order to investigate desorption behavior and microstruc-
tural evolution at higher temperatures a LiAlH4 powder ball milled
under low energy shearing (LES) was desorbed at 190 ◦C up to
2900 s with 6.7 wt.%H2 desorbed (Fig. 16a), at 190 ◦C up to 38,000 s
with 7.7 wt.%H2 desorbed (Fig. 16b), at 210 ◦C up to 19,300 s with
7.6 wt.%H2 desorbed (Fig. 16c) and at 300 ◦C up to 21,400 s with
7.9 wt.%H2 desorbed (Fig. 16d). The quantities of hydrogen des-
orbed at 190–300 ◦C clearly show that at these temperatures after
the corresponding time durations used for desorption, reactions (6)
and (7) were, indeed, completed. In this case, visual observations of
the lumped nature of the decomposed powder indicates that some
melting occurred during isothermal desorption in a Sieverts-type
apparatus at temperatures of 190 ◦C and higher.
After desorption the powders were tested by XRD and the cor-
responding patterns are shown in Fig. 17a. After desorption at
190 ◦C for 2900 s (pattern (1)) the microstructure comprises a small
amount of Li3AlH6 (very weak peaks) and Al. It indicates that reac-
tion (6) is near completion. After desorption at 190 ◦C for 38,000 s
and at 210 and 300 ◦C for corresponding times (patterns (2), (3)
and (4)) the microstructure contains Al and possibly LiH. In order
to elucidate the presence of LiH, the (2 0 0) Al peak was enlarged
and is shown in Fig. 17b. A small “shoulder” is observed at the foot
of the (2 0 0) Al peak which corresponds very well to the position
of (2 0 0) 100% intensity peak of LiH. Therefore, it is quite likely
that some amount of LiH is present in the microstructure indicat-
ing that, indeed, reaction (7) was completed. However, surprisingly
the (2 0 0) 100% peak of LiH has really low intensity. It is not clear if
this is related to an X-ray absorption mechanism specific to LiH or
its partial amorphization although we did not observe any amor-
phization of pristine LiH under high energy ball milling (a paper in
preparation for publication).
Fig. 17. (a) XRD patterns of powders desorbed in a Sieverts-type apparatus for
LiAlH4 ball milled for 15 min under low energy shearing (LES) with one strong mag-
net and R40 from Fig. 16. Pattern (1): 190 ◦C (Fig. 16a); pattern (2): 190 ◦C (Fig. 16b);
pattern (3): 210 ◦C (Fig. 16c); pattern (4): 300 ◦C (Fig. 16d). (b) Enlargement of the
(2 0 0) Al peak to show small shoulders at the position of the (2 0 0) 100% intensity
peak of LiH. JCPDS file numbers for phase identification are given in parenthesis.
much higher. Also, from this comparison it seems that the
value of ∼63.5 kJ/mol obtained here from desorption curves
at 3 different temperatures in Fig. 13c is rather unreasonably
low.
Fig. 14 shows the volumetric desorption curves (Sieverts appa-
ratus) for LiAlH4 ball milled under high energy impact IMP68 mode
obtained at relatively low temperatures, which are lower than
the melting peak (2) of LiAlH4 in Figs. 5 and 7. By comparison
with Fig. 10 at similar temperatures it seems that the desorbed
It must also be pointed out that the microstructural investiga-
tions by XRD presented in Fig. 11 do not support a notion that
the thermal decomposition of LiAlH4 in a solid state may follow a
Fig. 18. (a) Desorption curves of LiAlH4 ball milled for 15 min under high energy impact IMP68 mode with R40 and stored for 37 and 51 days at room temperature (RT) as
compared to desorption curve of powder after milling (0 days). (b) Desorbed hydrogen capacity of long-term stored ball milled powder of LiAlH4 vs. storage time at room
temperature (RT) after milling.

100
R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
desorption curve could reach saturation as can be clearly seen in
Fig. 18a. Therefore, one can conclude that the thermodynamic sta-
bility of ball milled LiAlH4 with much reduced particle size which is
composed of nanograins is quite good at least up to approximately
2 months of storage at room temperature in argon atmosphere.
The observed behavior is definitely in contrast to the storage of
catalyzed LiAlH4 which looses a substantial quantity of H2 within
a comparable storage time at room temperature (a paper in prepa-
ration for publication).
3.5. The effects of moisture absorption on microstructure and
hydrogen capacity
LiAlH4 is perceived as a very reactive hydride [1] particularly
very sensitive to contact with water or moisture. In order to have
some assessment how exposure to moisture can affect hydrogen
storage properties of LiAlH4 the powder was ball milled for 15 min
(to sensitize it even more by refining the particle size) under high
energy impact IMP68 mode with R132. Subsequently, some amount
of ball milled powder was weighed in a glove box and then exposed
to air in a fume hood with relative humidity of 47% for 1 h. After
completion of the exposure the powder was weighed again and the
relative mass increase due to water absorption came up to 11.7%.
Fig. 19 shows an XRD pattern of sample which absorbed 11.7%
water compared to an XRD of sample ball milled for 15 min under
high energy impact IMP68 mode with R132. As can be seen the XRD
pattern after water absorption is not changed in any measurable
way as compared to its ball milled counterpart. Fig. 20 compares
hydrogen desorption curve for ball milled sample of virgin LiAlH4
with the one for a sample which absorbed 11.7% water. A virgin ball
milled sample and a sample which absorbed 11.7% water desorbs
5 wt.%H2 within ∼11,000 and 11,500 s, respectively. It seems that
the water absorption up to 11.7% does not change in any drastic
way the desorption rate of ball milled LiAlH4 in Stage I, i.e. reaction
(6).
Fig. 19. XRD pattern of LiAlH4 ball milled for 15 min under high energy impact
IMP68 mode with R132 and XRD pattern of the same ball milled powder exposed
to 47% humid air for 1 h which resulted in water absorption of 11.7%. JCPDS file
numbers for phase identification are given in parenthesis.
direct decomposition path into LiH and Al as suggested by Wiench
et al. [22]. Our observations show the presence of an intermediate
Li3AlH6 phase in Fig. 11.
3.4. The effects of long-term storage on hydrogen capacity
Some researchers have reported that pristine LiAlH4 is inher-
ently thermodynamically unstable and spontaneously decomposes
into Li3AlH6 after long-term storage at room temperature
[18,23,24]. We decided to investigate a long-term stability of ball
milled LiAlH4 in more detail. A powder was ball milled under IMP68
mode with R40 for 15 min and about 50 mg of ball milled powder
was loaded into a glass vial in a glove box purged with ultra-high
purity argon. The loaded vial was stored in a glove box and samples
for volumetric desorption tests in a Sieverts-type apparatus were
taken after 37 and 51 days of storage. Desorption was carried out at
140 ◦C for such a duration of time until desorption curves achieved
saturation as shown in Fig. 18a for desorption curves after 37 and
51 days compared to desorption curve tested shortly after milling
(0 days). Fig. 18b shows a plot of desorbed capacity of H2 as a func-
tion of storage time at room temperature. It is quite clear that up
to 51 days of storage there is no measurable loss of H2 capacity
due to room temperature storage which is close to 7.5 wt.%, i.e. the
same as after milling. Only rate of desorption after 51 days of stor-
age seems to be slightly slower than that after milling (Fig. 18a).
Slightly lower capacity of 6.9 wt.%H2 in Fig. 18b was obtained after
37 days simply because the test was stopped too early before the
3.6. Flammability assessment
When the milling cylinder with the ball milled LiAlH4 is opened
in a glove box under Ar atmosphere, there is no evidence of self-
ignition. Furthermore, flammability studies also show that the ball
milled LiAlH4 does not ignite spontaneously after opening the
milling cylinder in air. However, within 15 min from the opening
of the milling cylinder the powder can be ignited by scraping the
cylinder walls with a metal tool and then the powder burns with
an open flame. This behavior indicates that even ball milled LiAlH4
Fig. 20. (a) Desorption curve at 120 ◦C under 0.1 MPa H2 pressure for LiAlH4 ball milled for 15 min under high energy impact IMP68 mode. (b) Desorption curve at 120 ◦
C
under 0.1 MPa H2 pressure for LiAlH4 ball milled for 15 min under high energy impact IMP68 mode and exposed to 47% humid air for 1 h which resulted in water absorption
of 11.7%.

R.A. Varin, L. Zbroniec / Journal of Alloys and Compounds 504 (2010) 89–101
101
with greatly reduced particle size is a relatively benign hydride not
really prone to self-ignition on contact with air always containing
some amount of moisture.
to exposure to air for 1 h does not change in any drastic way
the desorption rate of ball milled LiAlH4 in Stage I, i.e. reaction
LiAlH4(s) → (1/3)Li3AlH6(s) + (2/3)Al(s) + H2.
(9) The ball milled LiAlH4 powder does not ignite spontaneously
by simply opening a milling cylinder in air. However, within
approximately 15 min from the opening of the milling cylinder
the ball milled LiAlH4 powder can be ignited by scraping the
cylinder walls with a metal tool and then the powder burns
with an open flame.
4. Conclusions
The more important conclusions are as follows:
(1) From the comparison of microstructural investigations of the
present LiAlH4 with those reported by Ares et al. [9] it is clear
that the morphology, microstructure and chemistry of LiAlH4
can be very dissimilar depending on the supplier from which
LiAlH4 powder was purchased. These differences may influence
thermal desorption behavior.
(2) The average ECD particle size with standard deviation of LiAlH4
ball milled for 2 h under high energy mode IMP68 mode with
R132 in this work is 2.8 2.3 ␮m which is much smaller than
the average particle size reported by Ares et al. [9] who inves-
tigated Sigma–Aldrich LiAlH4.
Acknowledgments
This research was supported by the NSERC Hydrogen Canada
(H2CAN) Strategic Research Network and NSERC Discovery grants
which are gratefully acknowledged. The authors are grateful to
Prof. Linda Nazar from the Department of Chemistry, University of
Waterloo, for the usage of XRD equipment. The authors thank Prof.
A. Calka from the University of Wollongong for XRD calculations.
(3) We do not observe a partial decomposition of LiAlH4 during
milling up to 5 h under high energy impact mode.
References
(4) The melting of LiAlH4 in a DSC test is a very volatile event.
Immediately after the beginning of melting the liquid LiAlH4
starts foaming and flowing out of the alumina crucible raising
the alumina lid. After completion of desorption and solidifica-
tion the powder resembles a lava rock.
[1] R.A. Varin, T. Czujko, Z.S. Wronski, Nanomaterials for Solid State Hydrogen
Storage, Springer Science+Business Media, New York, NY, 2009.
[2] J. Graetz, J.J. Reilly, Scr. Mater. 56 (2007) 835–839.
[3] J. Block, A.P. Gray, Inorg. Chem. 4 (1965) 304–305.
[4] M. McCarty Jr., J.N. Maycock, V.R. Pai Verneker, J. Phys. Chem. 72 (1968)
4009–4014.
[5] J.A. Dilts, E.C. Ashby, Inorg. Chem. 11 (1972) 1230–1236.
[6] V.P. Balema, V.K. Pecharsky, K.W. Dennis, J. Alloys Compd. 313 (2000) 69–74.
[7] H.W. Brinks, B.C. Hauback, P. Norby, H. Fjellvåg, J. Alloys Compd. 351 (2003)
222–227.
[8] A. Andreasen, T. Vegge, A.S. Pedersen, J. Solid State Chem. 178 (2005)
3672–3678.
(5) Thermal sectioning in DSC combined with a phase anal-
ysis using XRD shows that LiAlH4 starts decomposing
into Li3AlH6 immediately after transforming to
a melt
and half way at temperature between the temper-
a
ature peak of the reaction LiAlH4(s) → LiAlH4(l) and
LiAlH4(s) → (1/3)Li3AlH6(s) + (2/3)Al(s) + H2, LiAlH4 is already
fully decomposed. This also implies that Li3AlH6 formed is
already solidified when it starts decomposing.
[9] J.R. Ares, K.-F. Aguey-Zinsou, M. Porcu, J.M. Sykes, M. Dornheim, T. Klassen, R.
Bormann, Mater. Res. Bull. 43 (2008) 1263–1275.
[10] A. Calka, A.P. Radlinski, Mater. Sci. Eng. A 134 (1991) 1350–1353.
[11] Patents: WO9,104,810, US5,383,615, CA2,066,740, EP0,494,899, AU,643,949.
[12] A. Calka, R.A. Varin, in: T.S. Srivatsan, R.A. Varin, M. Khor (Eds.), International
Symposium on Processing and Fabrication of Advanced Materials IX (PFAM IX),
ASM International, Materials Park, OH, 2001, pp. 263–287.
[13] H.P. Klug, L. Alexander, X-ray Diffraction Procedures for Polycrystalline and
Amorphous Materials, John Wiley & Sons, New York, 1974, pp. 618–708.
[14] A. Calka, A.P. Radlinski, Mater. Sci. Eng. 97 (1988) 241–246.
(6) All volumetric desorption curves clearly exhibit a two-stage
desorption process, Stage I and II. The most striking discov-
ery is that as received LiAlH4 is able, in a fully solid state,
to desorb at 120 ◦C under pressure of 0.1 MPa H2 (atmo-
spheric) as much as 7.1 wt.%H2 within ∼259,000 s (∼72 h),
i.e. 93% of the purity-corrected H2 content from reaction
in Stage I (LiAlH4(s) → (1/3)Li3AlH6(s) + (2/3)Al(s) + H2) and II
((1/3)Li3AlH6 → LiH + (1/3)Al + 0.5H2). The apparent activation
energy for Stage I and II for unmilled LiAlH4 is ∼111 and
∼100 kJ/mol, respectively. For ball milled LiAlH4 the apparent
activation energy for Stage I and II is ∼92.5 and ∼92 kJ/mol,
respectively.
(7) The microstructural investigations presented in this work do
not support a notion that thermal decomposition of LiAlH4 in a
solid state may follow a direct decomposition path into LiH and
Al as suggested by Wiench et al. [22].
(8) It seems that the ball milled LiAlH4 is reasonably immune to
the exposure to most air. The water absorption up to 11.7% due
[15] NIST/SEMATECH
e-Handbook
of
Statistical
Methods,
http://www.itl.nist.gov/div898/handbook/eda/section3/eda366g.htm.
[16] D. Blanchard, H.W. Brinks, B.C. Hauback, P. Norby, Mater. Sci. Eng. B 108 (2004)
54–59.
[17] J.-P. Bastide, B. Bonnetot, J.-M. Lettofé, P. Claudy, Mater. Res. Bull. 20 (1985)
999–1007.
[18] S.-I. Orimo, Y. Nakamori, J.R. Eliseo, A. Züttel, C.M. Jensen, Chem. Rev. 107 (2007)
4111–4132.
[19] O.M. Løvvik, S.M. Opalka, H.W. Brinks, B.C. Hauback, Phys. Rev. B 69 (2004)
134117-1–134117-9.
[20] A. Andreasen, J. Alloys Compd. 419 (2006) 40–44.
[21] D. Blanchard, H.W. Brinks, B.C. Hauback, P. Norby, J. Muller, J. Alloys Compd.
404–406 (2005) 743–747.
[22] J.W. Wiench, V.P. Balema, V.K. Pecharsky, M. Pruski, J. Solid State Chem. 177
(2004) 648–653.
[23] D. Blanchard, A.I. Lem, S. Øvergaard, H.W. Brinks, B.C. Hauback, J. Alloys Compd.
458 (2008) 467–473.
[24] T.J. Francombe, G.-J. Kroes, Chem. Phys. Lett. 423 (2006) 102–105.