Magnesium alanate - A material for reversible hydrogen storage?

Article abstract of DOI:10.1016/S0925-8388(02)01236-7

Magnesium alanate was synthesized in a metathesis reaction of magnesium chloride and sodium alanate followed by purification. The material obtained was sufficiently pure and it was investigated by X-ray diffraction (XRD) and by thermogravimetry (TG) and mass spectrometry (MS) of the evolved gas, respectively. Thermal analysis showed a decomposition with a release of hydrogen proceeding in two major steps. Measured in vacuum, the peak decomposition temperature of the first step was found to be 163°C and the residue at 200°C consisted of MgH₂ and Al which continues to release hydrogen and transforms into an Al₃Mg₂/Al mixture at higher temperatures. In the first decomposition step 6.6 wt.% of hydrogen was released. To enhance the kinetics of the decomposition, magnesium alanate was doped with a titanium based promoter and ball milled for up to 100 min, resulting in a significantly reduced peak decomposition temperature. First results on determining thermodynamic properties indicate equilibrium desorption pressures in the range of AB₅ compounds. In total, the promising properties determined so far have inspired further investigations on the thermodynamics, kinetics and reabsorption properties of the compound.

Full text of DOI:10.1016/S0925-8388(02)01236-7

Journal of Alloys and Compounds 356–357 (2003) 418–422
L
www.elsevier.com/locate/jallcom
Magnesium alanate—a material for reversible hydrogen storage?
*
Maximilian Fichtner , Olaf Fuhr, Oliver Kircher
Karlsruhe Research Center, Institute of Nanotechnology, POB 3640, D-76021 Karlsruhe, Germany
Received 1 June 2002; accepted 25 October 2002
Abstract
Magnesium alanate was synthesized in a metathesis reaction of magnesium chloride and sodium alanate followed by purification. The
material obtained was sufficiently pure and it was investigated by X-ray diffraction (XRD) and by thermogravimetry (TG) and mass
spectrometry (MS) of the evolved gas, respectively. Thermal analysis showed a decomposition with a release of hydrogen proceeding in
two major steps. Measured in vacuum, the peak decomposition temperature of the first step was found to be 163 8C and the residue at
200 8C consisted of MgH2 and Al which continues to release hydrogen and transforms into an Al3Mg2 /Al mixture at higher temperatures.
In the first decomposition step 6.6 wt.% of hydrogen was released. To enhance the kinetics of the decomposition, magnesium alanate was
doped with a titanium based promoter and ball milled for up to 100 min, resulting in a significantly reduced peak decomposition
temperature. First results on determining thermodynamic properties indicate equilibrium desorption pressures in the range of AB5
compounds. In total, the promising properties determined so far have inspired further investigations on the thermodynamics, kinetics and
reabsorption properties of the compound.
2002 Elsevier B.V. All rights reserved.
Keywords: Metal hydrides; Hydrogen absorbing materials; Earth alkali aluminium hydrides; X-ray diffraction; Thermal analysis
1. Introduction
erties and could meet at least some of the requirements
listed above.
Mobile applications in combination with hydrogen fuel
cell systems require sustainable storage materials which
contain large amounts of hydrogen. Furthermore, low
decomposition temperatures and fast kinetics for absorp-
tion and desorption of hydrogen are required. Stimulated
by the work of Bogdanovic et al. [1], researchers investi-
gated the reversible properties of sodium and lithium
alanates (e.g. [2–6]). While the hydrogen content of these
compounds is high and desorption kinetics seems to be
promising, the absorption kinetics still has to be improved.
Consequently, the search continues for new materials with
a high hydrogen content and sufficiently fast reaction
kinetics of absorption and desorption at moderate tempera-
tures.
2. Experimental
The solids were handled in an argon-filled glove box
equipped with a recirculation system to keep the water and
oxygen concentrations below 1 ppm during operation. The
synthesis was carried out on the bench under a vacuum or
in inert gas (N2) using Schlenk tube techniques.
The solvents were distilled over sodium before use.
Sodium alanate (purum, Sigma-Aldrich) was used in a 1 M
THF solution as received. Solvent-free NaAlH4 was
obtained by drawing the THF off under a vacuum and
heating the dry white solid to 50 8C until a residual
pressure of 231023 mbar was reached. Magnesium chlo-
ride (98%, Sigma-Aldrich) was used as received. Addition-
ally promoted samples were prepared by doping
Mg(AlH4)2 with 2 mol% TiCl3 (99.999%, Sigma-Aldrich)
and milled in a SPEX 8000 ball mill with a ball to
powder weight ratio of about 10:1.
In this contribution we present some of our results on
the synthesis, structure, and thermal properties of mag-
nesium alanate, Mg(AlH4)2, which contains 9.3 wt.% of
hydrogen. The compound has promising desorption prop-
*
Corresponding author. Tel.: 149-7247-82-5340.
E-mail address: fichtner@int.fzk.de (M. Fichtner).
The powder X-ray diffraction measurements were per-
0925-8388/02/$ – see front matter
doi:10.1016/S0925-8388(02)01236-7
2002 Elsevier B.V. All rights reserved.

M. Fichtner et al. / Journal of Alloys and Compounds 356–357 (2003) 418–422
419
formed with a PHILIPS X’PERT diffractometer (Cu Ka
radiation, 2 kW, with X’Celerator RTMS detector, auto-
matic divergence slit). The powder was wetted with
mineral oil, and the slurry was spread and measured on a
Si single crystal. For data acquisition and evaluation, the
X’PERT 1.3e and ProFit 1.0c software were used.
Thermal decomposition of magnesium alanate at am-
bient pressure was performed in a N2 (5.0) stream in a
JUWE IR 05 combustion/reduction furnace equipped with
a radiant heating system.
Thermogravimetric analysis was conducted under a
vacuum between room temperature and 400 8C at a heating
rate of 2 K/min using a NETZSCH STA 409C analyser
equipped with a BALZERS quadrupole mass spectrometer
for analysis of the evolved gas. During the measurement,
the m/z intensities between 1 and 80 were recorded as a
function of time. Typical sample quantities were 5–10 mg.
Elemental analysis of carbon and hydrogen was per-
formed by using a CE Instruments Flash EA 1112 Series
analyser with helium as purge gas. The sample quantity
typically was 4–6 mg. Benzoic acid and a standardized
LiAlH4 were used as standards for calibration.
Fig. 1. X-ray powder diffractogram of Mg(AlH4)2.
signal pattern of magnesium alanate in Fig. 1 exhibits a
structure of relatively high symmetry which is produced
from the less symmetric ether adduct [8]. We simulated a
congruent XRD pattern by utilizing calculated spatial
positions of the atoms. The positions were obtained from
an ab initio DFT calculation of the structure in the solid
state. According to the calculations, which were also
3. Results
supported by FTIR measurements, magnesium alanate
crystallizes in a trigonal structure, space group P3m1,
]
3.1. Synthesis of Mg(AlH4 )2
consisting of corner sharing MgH6 octahedrons and AlH4
tetrahedrons. A detailed description of the method and the
structure of the magnesium alanate is in preparation [9].
A peak shape analysis of the X-ray reflections of the
final product indicated that the magnesium alanate pro-
duced in this way is a nanocrystalline material with a
volume-weighted mean grain size of 30 nm. The structure
probably breaks during the drying procedure, as it is likely
that high internal stresses are induced into the structure
during the thermal release of the diethylether formerly
bound.
The synthesis of magnesium alanate was performed via
a metathesis reaction between sodium alanate and mag-
nesium chloride as proposed earlier in the literature by
Wiberg and Bauer [7]:
2NaAlH4 1 MgCl2 → Mg(AlH4)2 ? Et2O 1 2NaCl/in Et2O
(1)
D
Mg(AlH4)2 ? Et2O →Mg(AlH4)2 1 Et2O↑
(2)
During the process, the solvent adduct precipitates first. It
must be separated from the coprecipitated NaCl by an
extraction procedure and dried in order for the pure
substance to be obtained. The synthesis and structures of
two solvent adducts as well as the solvent-free magnesium
alanate are described in detail elsewhere [8]. The yield of
the end product was 81%. According to the elemental
analysis, it contained 9.0260.08 wt.% H and 2.360.2
wt.% C (theoretical: 9.34 wt.% H). The residual content of
NaCl, determined by the Mohr method, was 1.7 wt.%.
Taking into account the impurities of sodium chloride and
the residual ether, the content of Mg(AlH4)2 in the final
product turned out to be 95%.
4.1. Thermal decomposition
For evaluation of the thermogravimetric measurements,
the sample mass at T5100 8C was taken as the 100%
starting value. As can be seen from the TG/MS data in
Fig. 2, decomposition started approximately at 110 8C as
indicated by a rise in the signal at m/z52 (H12 ). However,
detailed analysis of the H12 intensity revealed an exponen-
tial rise of its slope with an increase in intensity by a factor
of 10 between 50 8C and 120 8C. It was also observed that
the alanate lost 0.5 wt.% hydrogen when it was kept in the
glove box at room temperature for 4 months.
Fig. 2 indicates that there were two contributions to the
signal at 132 8C and 163 8C in the first major decomposi-
tion step. Repeated measurements with different samples
obtained by different preparation methods showed that the
relative amounts of the two components of the signal
4. Structure
The end product of the reaction described in Section 3.1
was investigated by X-ray powder diffractometry. The

420
M. Fichtner et al. / Journal of Alloys and Compounds 356–357 (2003) 418–422
Fig. 2. TG/MS data of the thermal decomposition of Mg(AlH4)2.
Fig. 3. X-ray powder diffractogram of Mg(AlH4)2 treated in N2 at 200 8C
for 30 min.
varied while their positions remained stable within a few
degrees centigrade. The apparent rise in the TG curve at
around 160 8C was caused by the repulsion of the sample
holder, which is induced by the high hydrogen production
rate at these temperatures. Hence, the highest intensity of
the MS signal was found together with the maximum
repulsion at the same temperature. The first event at 132 8C
appeared as a small bump in the TG curve. After peaking
at 163 8C, the TG curve declined to a value of 93.4 mass
%, which was reached at approximately 200 8C. This
completed the first major decomposition step. A third
component of the mass signal at 180 8C was not always
reproducible. In the first step, the sample lost 6.6 wt.%,
which was attributable almost exclusively to hydrogen. As
the sample contained 9.0 wt.% hydrogen before the
measurement, 73.3%, i.e., about three quarters of the total
stored hydrogen within the experimental error, was re-
leased.
These observations support a decomposition mechanism
similar to the one observed for alkali alanates, which is in
agreement with the observations made by Claudy et al.
[10]. Accordingly, the following transition occurs in the
first step of the decomposition (110–200 8C):
Mg(AlH4)2 → MgH2 1 2Al 1 3H2
(3)
The second step (240–380 8C, decomposition of mag-
nesium hydride) proceeds as follows:
MgH2 → Mg 1 H2
(4)
At temperatures around 400 8C an intermetallic compound
is formed consisting of the precipitated aluminum from
step 1 and the magnesium from step 2:
1
1
]
]
2Al 1 Mg → 2 Al3Mg2 1 2 Al
(5)
Although there are apparent similarities in the decomposi-
An X-ray powder diffractogram of magnesium alanate
treated for 30 min in an inert N2 stream at 200 8C is shown
in Fig. 3. The signals in the pattern could be assigned to
MgH2 and Al, the products of the first decomposition step.
The small peak at 2u531.68 was attributed to the most
intense reflection of NaCl, which was present as an
impurity.
Further heating lead to the decomposition of MgH2 and
the production of hydrogen gas in two steps at 287 8C and
345 8C. At 400 8C, the hydrogen was released completely
and the sample decomposed. According to the TG curve,
the loss of mass between 200 8C and 400 8C was 2.4 wt.%,
which is about one quarter of the total amount of the stored
hydrogen.
Fig. 4 shows an X-ray diffraction pattern of a sample
which has been heated at a temperature of 400 8C for 30
min. The signals in the pattern can be attributed to a
mixture of Al and Al3Mg2 (and NaCl). A diffractogram of
another sample, which has been treated in N2 at 300 8C,
includes additional MgH2 signals.
Fig. 4. X-ray powder diffractogram of Mg(AlH4)2 treated in N2 at 400 8C
for 30 min.

M. Fichtner et al. / Journal of Alloys and Compounds 356–357 (2003) 418–422
421
tion behaviour of magnesium alanate and the alkali ala-
nates, there are differences as well. According to the well
reproduced data obtained so far by TG, DSC, XRD and
volumetric measurements, the first decomposition step (3)
seems to be a transition from the alanate directly to the
metal hydride MgH2. This is different compared to the
decomposition mechanism of LiAlH4 and NaAlH4 [11,12],
where the analogous transition is known to be a two-step
procedure: The alkali alanates decompose first to M3AlH6
(M5Li, Na), a relatively stable intermediate with a
cryolithe structure, and then to the alkali hydride and
aluminum. Nevertheless, it might be that there are one or
more intermediates in the transition of Mg(AlH4)2 →MgH2
which were not resolved by our methods. It is therefore
intended to make more detailed investigations in the region
of decomposition, e.g. by temperature programmed diffrac-
tion measurements.
The reduced hydrogen release with increasing milling
times is based on the fact that TiCl3 may be reduced by the
alanate under consumption of hydrogen and that ball
milling provides sufficient mechanical and thermal energy
to release a significant part of the stored hydrogen during
the milling process.
5. Conclusion and outlook
The results show that Mg(AlH4)2 exhibits some promis-
ing features as a hydrogen storage material. Due to its high
hydrogen content of 9.3 wt.%, a maximum reversible
content of 7 wt.% may be expected if reaction (3) can be
made reversible. Another interesting aspect is the
nanocrystallinity of the material when it is synthesized as
described above. This seems to be a good starting point, as
it has already been observed that the kinetic barriers in
nanocrystalline hydrides should be low compared to
coarse-grained materials [2]. The peak decomposition
temperature of the unmilled magnesium alanate in vacuum
is 163 8C, which is in the range of the values for alkali
alanates. By adding a TiCl3 promoter and ball milling the
samples the peak temperature could be reduced by up to
45 8C, however accompanied by a reduced hydrogen
release. Finally, it may be of interest that the substance has
a sustainable composition of elements, which is an im-
portant issue only for mass applications.
Although the properties investigated so far are promis-
ing for mobile applications, some important issues still
have to be explored. These include the thermodynamic
properties of the substance as well as the kinetics of
hydrogen absorption and release. The formation of stable
MgH2 as an intermediate or Al3Mg2 as an end product
might negatively influence the reabsorption of hydrogen
when starting from the respective dehydrogenated state of
the material.
Another difference in the decomposition mechanisms is
the formation of a stable intermetallic compound (Al3Mg2)
in the last step (5) which obviously does not occur with the
alkali alanates because of the positive formation enthalpies
of the respective compound.
To enhance the kinetics of the dehydrogenation of
Mg(AlH4)2 we doped the sample with 2 mol% of TiCl3
and reduced the grain size of the compound by high energy
ball milling for up to 100 min. Fig. 5 shows the peak
temperature of the MS signal and the mass loss of the first
decomposition step in magnesium alanate for the unmilled,
pure sample and for three different milling times of 10, 30
and 100 min, respectively. While the starting point of
hydrogen release could be clearly shifted to lower tempera-
tures by milling the doped magnesium alanate samples, the
completion of the first decomposition step still was ob-
served to be at around 200 8C, i.e., not all of the hydrogen
was released at lower temperatures, which, however, may
not necessarily be a mechanistic effect. Moreover, it is also
conceivable that the broad signal was due to inhomo-
geneities in the sample, caused by the short milling times.
The high desorption pressures in the range of AB5
compounds make the material on one hand promising as a
low temperature hydride. On the other hand, to avoid very
high loading pressures, this would require reabsorption at
low temperatures which might lead to an unfavorable
kinetic behaviour.
One focus of our future work will therefore be the
investigation of the reabsorption properties of the material,
which can be accomplished when equilibrium desorption
pressures for different temperatures have been determined.
References
[1] B. Bogdanovic, M. Schwickardi, J. Alloys Comp. 253/254 (1997)
1–9.
Fig. 5. Peak decomposition temperature and mass loss by H2 of the first
decomposition step of unmilled Mg(AlH4)2 and Mg(AlH4)2 doped with 2
mol% TiCl3 for 3 different milling times.
¨
[2] A. Zaluska, L. Zaluski, J.O. Strom-Olsen, J. Alloys Comp. 298
(2000) 125–134.

422
M. Fichtner et al. / Journal of Alloys and Compounds 356–357 (2003) 418–422
[3] R.A. Zidan, S. Takara, A.G. Hee, C.M. Jensen, J. Alloys Comp. 285
(1999) 119–122.
[4] C.M. Jensen, K.J. Gross, Appl. Phys. A 72 (2) (2001) 213–219.
[5] K.J. Gross, G.J. Thomas, G. Sandrock, Proceedings of the U.S. DOE
2000 Hydrogen Program Review, NREL/CP-570-28890, Liver-
more, CA.
[8] M. Fichtner, O. Fuhr, J. Alloys Comp. 345 (2002) 286–296.
[9] M. Fichtner, O. Fuhr, A. Gloess, O. Rubner, R. Ahlrichs, The
structure of magnesium alanate, Inorg. Chem., submitted.
[10] P. Claudy, B. Bonnetot, J.M. Letoffe, J. Therm. Anal. 15 (1979)
119–128.
[11] J.A. Dilts, E.C. Ashby, Inorg. Chem. 11 (1972) 1230.
[12] P. Claudy, B. Bonnetot, G. Chahine, J.M. Letoffe, Thermochim.
Acta 38 (1980) 75–88.
[6] J. Chen, N. Kuriyama, Q. Xu, H.T. Takeshita, T. Sakai, J. Phys.
Chem. B 105 (2001) 11214–11220.
[7] E. Wiberg, R. Bauer, Z. Naturforschung 5b (1950) 397.