Mechano-chemical activation synthesis (MCAS) of nanocrystalline magnesium alanate hydride [Mg(AlH4)2] and its hydrogen desorption properties

Article abstract of DOI:10.1016/j.jallcom.2006.08.080

A successful synthesis of the Mg(AlH₄)₂ + 2NaCl mixture by the mechano-chemical activation synthesis (MCAS) for 5 and 10 h has been achieved. The nanocrystalline Mg(AlH₄)₂ hydride in the mixture has the grain size on the order of 18 nm. Even relatively short milling for just 10 h seems to result in a partial decomposition of the initially formed Mg(AlH₄)₂ into the nanocrystalline β-MgH₂, elemental Al (grain size ～26 nm) and hydrogen gas. A number of DSC tests at the scan rate of 4 °C/min of the Mg(AlH₄)₂ + 2NaCl mixture synthesized for 5 and 10 h show either endothermic or exothermic heat flow changes at the ～125-180 °C range due to the decomposition of Mg(AlH₄)₂. Endothermic peaks with the maxima at ～271 and 316 °C due to the decomposition of β-MgH₂ and the formation of Al₃Mg₂ intermetallic as well as Al(Mg) solid solution are present. The peak at ～452 °C is due to the eutectic melting of the Al₃Mg₂ and Al(Mg) mixture. Prolonged milling for 40 h results in a complete decomposition of the Mg(AlH₄)₂ into two nanocrystalline solid phases such as β-MgH₂ (grain size ～11 nm) and the elemental Al (grain size ～20 nm), and hydrogen gas. DSC tests at the scan rate of 4 °C/min up to 500 °C of the 40 h milled powder, which contains β-MgH₂, results in two endothermic peaks with the maxima at ～292 °C due to the decomposition of β-MgH₂ and ～452 °C due to the eutectic melting. TGA tests for the 5 and 10 h milled sample give the total weight loss of ～2.45 and 2.16 wt.%, respectively. The hydrogen desorption kinetics of the as-synthesized powders directly after milling were also evaluated using a Sieverts-type apparatus.

Full text of DOI:10.1016/j.jallcom.2006.08.080

Journal of Alloys and Compounds 439 (2007) 302–311
Mechano-chemical activation synthesis (MCAS) of nanocrystalline
magnesium alanate hydride [Mg(AlH ) ] and
4
2
its hydrogen desorption properties
a,∗
a
a
a,b
R.A. Varin , Ch. Chiu , T. Czujko , Z. Wronski
a
Department of Mechanical Engineering, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1
b
Materials Technology Laboratory, CANMET, Natural Resources Canada, 568 Booth Street, Ottawa, Ontario, Canada K1A 0G1
Received 5 July 2006; received in revised form 21 August 2006; accepted 21 August 2006
Available online 22 September 2006
Abstract
A successful synthesis of the Mg(AlH
4
)
2
+ 2NaCl mixture by the mechano-chemical activation synthesis (MCAS) for 5 and 10 h has been
achieved. The nanocrystalline Mg(AlH ) hydride in the mixture has the grain size on the order of 18 nm. Even relatively short milling for just 10 h
4 2
seems to result in a partial decomposition of the initially formed Mg(AlH
4
)
2
into the nanocrystalline ␤-MgH
2
, elemental Al (grain size ∼26 nm)
◦
and hydrogen gas. A number of DSC tests at the scan rate of 4 C/min of the Mg(AlH
4
)
2
+ 2NaCl mixture synthesized for 5 and 10 h show either
◦
endothermic or exothermic heat flow changes at the ∼125–180 C range due to the decomposition of Mg(AlH
4 2
) . Endothermic peaks with the
◦
maxima at ∼271 and 316 C due to the decomposition of ␤-MgH
2 3 2
and the formation of Al Mg intermetallic as well as Al(Mg) solid solution are
◦
present. The peak at ∼452 C is due to the eutectic melting of the Al
3
Mg
2
and Al(Mg) mixture. Prolonged milling for 40 h results in a complete
(grain size ∼11 nm) and the elemental Al (grain size
20 nm), and hydrogen gas. DSC tests at the scan rate of 4 C/min up to 500 C of the 40 h milled powder, which contains ␤-MgH , results in two
and ∼452 C due to the eutectic melting. TGA tests for the
and 10 h milled sample give the total weight loss of ∼2.45 and 2.16 wt.%, respectively. The hydrogen desorption kinetics of the as-synthesized
powders directly after milling were also evaluated using a Sieverts-type apparatus.
2006 Elsevier B.V. All rights reserved.
4 2 2
decomposition of the Mg(AlH ) into two nanocrystalline solid phases such as ␤-MgH
◦
◦
∼
2
◦
◦
endothermic peaks with the maxima at ∼292 C due to the decomposition of ␤-MgH
2
5
©
Keywords: Hydrogen storage; Nanostructured materials; Complex hydrides; Magnesium alanate; Mechano-chemical activation synthesis; Reactive milling; X-ray
diffraction; Differential scanning calorimetry; Hydrogen desorption kinetics
1
. Introduction
[4,5]. Among them, a potentially attractive complex hydride
is the magnesium tetrahydroaluminate (magnesium alanate)
Mg(AlH4)2. Its theoretical hydrogen capacity is 9.3 wt.% and
the hydride is based on two inexpensive light metals Mg and Al.
This hydride was first synthesized by chemical methods over 50
years ago and then thoroughly investigated by Ashby et al. [6]
and more recently investigated by Fichtner et al. [7–9], Schwarz
et al. [10] and Fossdal et al. [11,12] who found that it decom-
Hydrogen storage in solid hydrides is the most attractive
method of on-board hydrogen storage for fuel cell powered
passenger vehicles [1–3]. Hydrogen storage targets established
for 2010 and 2015 by the FreedomCAR program [1] clearly
show that only hydrides with the gravimetric hydrogen storage
capacitieshigherthan6 wt.%areviableforpotentialcommercial
automotive application. So far, the only group of hydrides which
have both high volumetric and gravimetric hydrogen densities
meeting the targets are complex metal hydrides having a mixed
ionic-covalent bonding between metal and hydrogen complex
◦
poses in two major steps with the first step around 160 C in
which three quarters of the bound hydrogen was released. This
relatively low dehydrogenation temperature of the first step is
attractive enough to warrant further studies.
Extensiveresearchinthelastdecadeledtoaratherstrongcon-
sensus that nanocrystalline solid hydrides, specifically obtained
by mechanical milling in various types of ball mills and char-
acterized by the existence of nanometric grains within the
∗
Corresponding author. Fax: +1 519 888 6197.
E-mail address: ravarin@mecheng1.uwaterloo.ca (R.A. Varin).
0
925-8388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2006.08.080

R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
303
hydride particles, exhibit substantially enhanced hydrogena-
tion/dehydrogenation properties [5,13,14]. Our recent attempt
of a direct mechano-chemical synthesis of nanocrystalline
Mg(AlH4)2 by controlled reactive mechanical alloying [15] has
been unsuccessful due to accelerated formation of the Al(Mg)
solidsolutionwhich, mostlikely, inhibitedthereactionofMgH2,
Al and H2 to form Mg(AlH4)2.
However, an alternative indirect method of the synthesis
of Mg(AlH4)2 is the mechano-chemically activated metathesis
reaction [16,17]. Mamatha et al. [18,19] subjected a mixture of
NaAlH4 and MgCl2 to ball milling for 3 h (Retsch ball mill)
and obtained a mixture of synthesized Mg(AlH4)2 and NaCl.
Subsequent DSC analysis of the mechano-chemically prepared
formation of Mg(AlH4) and NaCl mixture. Second, to clarify the
thermal events occurring during DSC analysis of the synthesized
mixture of Mg(AlH4) and NaCl with a special emphasis on the
◦
heat flow event in the range of 115–150 C.
2. Experimental
As-received NaAlH4 (90% purity; Sigma–Aldrich) and anhydrous MgCl2
(
(
99% purity; Alfa-Aesar) were mixed together in the 2:1 stoichiometric ratio
corresponding wt.% of NaAlH4 and MgCl2 in the mixture are 53.1 and
4
6.9 wt.%, respectively, so as a result the weight ratio is ∼1.13 g NaAlH4:1 g
MgCl2). Subsequently, the mixture was subjected to controlled mechanical
milling under 600 kPa pressure of high purity argon (99.999% purity) in the
magneto-mill Uni-Ball-Mill 5 (A.O.C. Scientific Engineering Pty Ltd., Aus-
tralia) [21–23]. Two Nd–Fe–B super-strong magnets at 6 and 8 o’clock positions
with WD = 10 and 2 mm, respectively, were used (Fig. 1) (WD, working distance
between the vial and the magnet). Four hardened steel balls in the milling vial
yielded the ball-to-powder weight ratio of 40:1. The samples were milled con-
tinuously for 5, 10 and 40 h at 225 rpm.
powders showed three endothermic peaks with maxima around
◦
1
40, 270 and 450 C, which were interpreted as correspond-
ing to the decomposition of Mg(AlH4) into MgH2, Al and H2,
decomposition of MgH2 into Mg and formation of Al3Mg2, and
eventually melting of Al3Mg2 and Al mixture, respectively. Kim
et al. [20] synthesized a mixture of Mg(AlH4) and NaCl after
Morphological examination of powders was conducted with a high-
resolution, field emission gun SEM (FEG SEM) LEO 1530 equipped with
integrated EDAX Pegasus 1200 EDS/OIM. The size measurement of the pow-
der particles for various samples was carried out by attaching loose powder to
the sticky carbon tape and taking pictures under secondary electron (SE) mode
in SEM. The images were analyzed by the Image Tool v.3.00 software. The
size of the powders were calculated as the particle equivalent circle diameter,
1
h in a Spex 8000 ball mill. DSC test was carried out only up
◦
to slightly over 300 C and showed the first exothermic peak at
the range of 115–150 C and the second endothermic peak at
the range of 240–290 C. They concluded that the first exother-
◦
◦
1
/2
ECD = (4A/π) , where A represents the projected particle area.
mic reaction corresponds to the decomposition of Mg(AlH4)
into MgH2, Al and H2 and the second endothermic peak to the
decomposition of MgH2. Apparently, there is a clear contro-
versy between the results of Mamatha et al. [18,19] who report
endothermic and those of Kim et al. [20] who report exothermic
reaction for the decomposition of Mg(AlH4).
The objective of this work is two-fold. First, to study the
effect of the milling time longer than 3 h on the microstructural
development of the initial mixture of NaAlH4 and MgCl2 and the
The crystalline structure of the as-milled powders was characterized by
Bruker D8 powder diffractometer using Cu K␣ (not monochromated) radia-
◦
◦
◦
tion at the scan range from 20 to 90 , step size of 0.05 /s, accelerating voltage
of 40 kV and a current of 30 mA. The nanograin (crystallite) size of phases resid-
ing in the milled powders was calculated 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 (nanograin/crystallite) and lattice 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
Fig. 1. Angular positions of Nd–Fe–B super-strong magnets at 6 and 8 o’clock for ball milling under high-energy impact mode (IMP68) in the Uni-Ball-Mill 5. The
angular positions of external magnets can be changed at each of the controlled modes of milling [21–23].

3
04
R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
Fig. 2. The XRD patterns of the initial mixture of NaAlH4 and MgCl2 powders after milling for 5, 10 and 40 h.
the following equation [24]:
3. Results and discussion
ꢀ
ꢁ
δ2(2θ)
Kλ
= L
δ(2θ)
2
3
.1. Microstructure of synthesized powders after milling
+ 16e
(1)
2
tan θ
tan θ sin θ
Mean ECD values for the particles of the powders after
where the term Kλ/L is the slope, the parameter L is the mean dimension
of the nanograin (crystallite) composing the powder particle, K is constant
(
milling for 5, 10 and 40 h are 2.7 ± 2.3, 2.4 ± 2.2 and
∼1); e the so-called “maximum” microstrain (calculated from the intercept); λ
2
.3 ± 2.2 ␮m, respectively. EDS analysis has shown that the
the wave length; θ is the position of the analyzed peak maximum. The term
distribution of Mg, Al, Na and Cl is homogeneous in all the
powders.
2
2
δ(2θ) = B[1 − (b /B )](rad) is the instrumental broadening-corrected “pure”
XRD peak profile breadth [24], where B and b are the breadths in radians of
The XRD pattern in Fig. 2 shows the presence of Mg(AlH4)2
(identified from JCPDS # 47-980) and NaCl (identified from
JCPDS # 05-0628) after milling for 5 and 10 h. It can be assumed
that after 5 h of MCAS the synthesis of Mg(AlH4)2 is completed
according to the following reaction:
the same Bragg peak from the XRD scans of the experimental and reference
TM
powder, respectively. They were calculated by the software Traces
v.6.5.1
as the full-widths at half maximum, FWHM, after K␣2 stripping. A compound
LaB6, the National Institute of Standards and Technology (NIST) standard refer-
ence 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 instru-
mental line profiles were obtained in this manner, the Bragg peaks for the LaB6
SRM were at different 2θ 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.
2NaAlH4 + MgCl2 → Mg(AlH4)2 + 2NaCl
(2)
as also reported by Mamatha et al. [18,19] and Kim et al. [20].
After milling for 5 h only a single, very weak XRD peak of
elemental Al is recognizable in the XRD pattern (Fig. 2; at
Differential scanning calorimetry (DSC) was performed on a Netzsch 404
instrument using ∼20 mg of a sample heated in an alumina crucible at the scan
◦
◦
rate of 4 and 20 C/min, and a 50 ml/min Ar flow rate. Thermogravimetric anal-
2θ ∼ 38 ). Increasing the milling duration to 10 h results in the
ysis (TGA) was performed on a TA INSTRUMENT Q600 instrument using
appearance of more Al peaks and one or two very weak peaks
of ␤-MgH2 (Fig. 2). After 40 h, the peaks of Mg(AlH4)2 disap-
pear while the well-discerned peaks of ␤-MgH2 and Al appear.
The pressure inside the milling cylinder increased by ∼465 kPa
◦
∼
10 mg of a sample heated at a heating rate of 4 C/min under a 100 ml/min
nitrogen gas flow rate. All samples were tested almost immediately after com-
pletion of milling to avoid aging phenomena due to hydrolysis of MgH2 [25].
The hydrogen desorption kinetics of the as-synthesized powders directly
after milling were also evaluated using a Sieverts-type apparatus custom-built
by A.O.C. Scientific Engineering Pty Ltd., Australia (see Fig. 2 in Ref. [26]).
The weight of the powder sample in the kinetics experiments was ∼100–150 mg.
The calibrated accuracy of desorbed hydrogen capacity is about ± 0.2 wt.% H2
from which one can estimate that ∼2.5 wt.% of H was desorbed
2
during milling for 40 h. It is clear that initially, the mixture of
Mg(AlH4)2 and 2NaCl is being formed but subsequently, during
further prolonged milling, there is a gradual decomposition of
Mg(AlH4)2 according to
◦
and that of temperature reading ± 5 C. Before starting the desorption test, the
inner tubing of the apparatus was also evacuated and purged four times with
argon and then two times with hydrogen. Subsequently, hydrogen under pres-
sure of 41 bars was admitted into the airtight cylindrical desorption chamber
containing the hydride powder and the temperature of the chamber was grad-
ually increased up to the desired desorption temperature. The 41 bar pressure
barrier was applied to prevent any desorption during the period of temperature
stabilization (15–20 min) within the desorption chamber. Once the temperature
was stabilized the hydrogen pressure was quickly reduced to 0.1 MPa and the
hydrogen desorption process was monitored. No elaborate activation procedure
was applied to the powders.
Mg(AlH4)2 (+2NaCl) → ␤-MgH2 + 2Al + 3H2 (+2NaCl)
(
3)
It is to be pointed out that in the present work the decompo-
sition of Mg(AlH4)2 upon milling occurs without the addition
of Ti metal catalyst, which was used by Mamatha et al. [19] to
induce similar reaction.

R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
305
Table 1
Grain (crystallite) size, L, and lattice strain, e, of the phases in the powders after milling for 5, 10 and 40 h as estimated from Eq. (1)
Milling time (h)
Phase
Nanograin (crystallite) size, L (nm)
Lattice strain, e
R2
No. of XRD peaks
5
Mg(AlH4)2
Mg(AlH4)2
Al
Al
␤-MgH2
18
18
26
20
11
0
0.973
0.975
0.995
0.945
0.983
11
9
4
4
3
−
4
1
1
4
4
0
0
0
0
6.6 × 10
0
0
0
As opposed to the reaction (2), which occurs during short
sition of Mg(AlH4)2 into ␤-MgH2 and the elemental Al. The
XRD peaks of decomposition product ␤-MgH2 in Fig. 4 are
rather weak since the amount of ␤-MgH2 is small (in the reac-
tion (2) there is only 1 mol of Mg(AlH4)2 per 2 mol of NaCl).
milling, the reaction (3) is not a metathesis reaction. Instead,
a long milling triggers a reduction–oxidation (redox) reaction.
Indeed, theHandAlaremutuallyoxidizedandreducedasshown
below:
◦
After a DSC run to the temperature of 295 C, higher than the
◦
temperature of the first DSC peak maximum at ∼271 C, the
Mg2 + 2(Al+3H4
+
−1 −1
)
→ Mg2+H2 + 3H2 + 2Al
−1
0
0
(4)
microstructure consists of residual NaCl, a negligible amount of
␤
-MgH2, small amount of Al3Mg2 intermetallic (very low peak
and the partial reactions are as follows:
intensities in Fig. 4) and large amount of Al(Mg) solid solu-
tion (high intensity XRD peaks in Fig. 4). Apparently, around
−
0
2
H − 2e → H2 |×3
(5)
(6)
◦
the 271 C peak the majority of ␤-MgH2 decomposes in the
Al3 + 3e → Al |×2
+
0
endothermic reaction into free Mg and H2. Simultaneously,
reactions of Mg and Al to form Al3Mg2 and Al(Mg) solid
solution must have occurred. The reactions can be written as
follows:
Nanocrystalline Mg(AlH4)2 with the grain size of ∼18 nm
(
Table 1) is produced after MCAS for 5 and 10 h in excellent
agreement with Mamatha et al. [19]. Grain sizes of Al (∼20 nm)
and ␤-MgH2 (∼11 nm) after decomposition of Mg(AlH4)2 dur-
ing a prolonged milling for 40 h are still in a nanometric range
Table 1). However, it must be noted that with cycling, the com-
plex hydrides generally may not maintain a small crystallite size.
␤
-MgH2 + 2Al (+2NaCl) → Mg + H2
◦
◦
+
2Al (+2NaCl) (225–340 C at 4 C/min)
(7a)
(7b)
(
Mg + 2Al → 0.5Al3Mg2
◦
3
.2. DSC and TGA thermal analysis up to 500 C
◦
◦
+
0.5Al(Mg) (225–340 C at 4 C/min)
◦
The NaCl by-product of the reaction (2) was not removed
After DSC run up to the temperature of 350 C, which is
higher than the temperature of the second DSC peak maxi-
from the powders milled for 5 and 10 h because it does not
react with Mg(AlH4)2 during heating and removing salt using
chemical methods might bring impurities, which may affect the
thermal behavior of Mg(AlH4)2 [20].
◦
mum at ∼316 C (Fig. 3c), the decomposition of the remnant
MgH2 is completed and the microstructure consists of residual
NaCl, large amount of Al3Mg2 intermetallic compound (high
intensity XRD peaks in Fig. 4) and large amount of Al(Mg)
solid solution (high intensity XRD peaks in Fig. 4). As such,
◦
DSC traces registered at the scan rate of 4 and 20 C/min
for the powders milled for 5 and 10 h exhibit some small heat
◦
◦
flow fluctuations at the low-temperature range of 125–180 C
two endothermic peaks at ∼271 and ∼316 C might arise either
(
Fig. 3a and b) while the DSC trace for the 40 h milled powder is
rather flat at this low-temperature range. These low-temperature
due to the decomposition of ␤-MgH2 and the formation of the
Al3Mg2 intermetallic compound or alternatively, due to the two-
step decomposition of ␤-MgH2 although the latter is not so
clear.
thermal events will be discussed in more detail in Section
◦
3
.3. DSC traces registered at the scan rate of 4 C/min clearly
show two endothermic peaks for the 5 and 10 h milled pow-
ders at the 225–340 C range and one endothermic event at a
It must be pointed out that the formation of Al(Mg) solid solu-
tion in due course and after complete decomposition of MgH2
observed by XRD in Fig. 4 was first identified by visible shifting
of Al peaks to lower angles, with respect to their position for a
pure Al in the XRD patterns of powders heated in DSC to 295,
◦
◦
4
50–460 C range (Fig. 3a). As expected, increase of the scan
◦
rate to 20 C/min shifts the medium-temperature range peak
◦
maxima to higher temperatures (∼250–375 C) but does not
◦
change the shape of the peaks (Fig. 3b). In order to analyze the
350 and 425 C (Fig. 4). Subsequently, the lattice parameters of
◦
TM
nature of the thermal events observed in Fig. 3a above 180 C the
Al(Mg) were calculated by the software Traces
v.6.5.1 and
powder milled for 10 h which contains Mg(AlH4)2 was analyzed
by running small samples of this powder in DSC at 4 C/min to
the temperatures marked by the vertical lines in Fig. 3c. The cor-
are listed in Table 2. It is clear that the lattice parameter of Al
in the powders milled for 10 h is much larger after annealing
◦
◦
in DSC at 295, 350 and 425 C than that of the as-milled and
◦
responding XRD patterns are shown in Fig. 4. After heating to
180 C annealed samples. The lattice expansion is consistent
◦
1
80 C the only phases existing in the microstructure are residual
with the magnitude of the atomic radius of Al and Mg, which
is 0.1432 and 0.1604 nm, respectively [27]. Similar effect of
NaCl, ␤-MgH2 and the elemental Al confirming the decompo-

3
06
R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
◦
◦
◦
Fig. 3. DSC curves of powders after milling for 5, 10 and 40 h upon heating to 500 C registered at the scan rate of: (a) 4 C/min; (b) 20 C/min; (c) the DSC curve
◦
of the powder after milling for 10 h showing a separation into the four segments corresponding to the thermal events registered on the curve (scan rate 4 C/min).
the Al(Mg) solid solution formation during reactive milling of
the elemental powders with Mg–2Al ratio, pre-alloyed Mg–2Al
ingot and AZ91 alloy was reported in [15]. Fossdal et al. [12]
claimed the formation of the Al(Mg) solid solution during low-
upon heating (Fig. 3a–c) is due to the eutectic melting of the
mixture of Al3Mg2 and Al(Mg) solid solution which essentially
occurs according to the equilibrium phase diagram of Al–Mg as
reported in [15].
◦
temperaturedesorptionofMg(AlH4)2 below180 Cwhichisnot
In the DSC curve of powder after milling for 40 h, only a
single endothermic peak due to the decomposition of ␤-MgH2
and formation of Al3Mg2 and Al(Mg) is observed regardless
of the scan rate (Fig. 3a and b). It is to be pointed out that the
confirmed in the present work (Fig. 4 and Table 2). Mamatha et
al. [18,19] never reported the formation of the Al(Mg) solid
solution although Kim et al. [20] mentioned about increase
in the lattice parameter of Al most probably due to forma-
◦
DSC run of the 40 h milled powder at the scan rate of 4 C/min
◦
◦
◦
tion of the Al(Mg) solid solution. After heating to 425 C the
up to 350 C which is above the peak maximum at ∼292 C
in Fig. 3a, resulted in the formation of the mixture of Al3Mg2
and Al(Mg) solid solution whose lattice parameter is also listed
◦
microstructure remains the same as that after the 350 C anneal-
ing. The nature of the endothermic reaction at ∼452–454 C
◦

R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
307
◦
Fig. 4. XRD patterns corresponding to the powders heated to 180, 295, 350 and 425 C in a DSC test shown in Fig. 3c.
in Table 2. This clearly confirms that the formation of Al3Mg2
and Al(Mg) in the present work occurs with the assistance of
temperature rather than during ball milling as reported in [15].
This can be related to different methods of synthesis used in the
present work (MCAS) and in [15] (reactive mechanical milling
of elemental powders or alloys). Mamatha et al. [18,19] and Kim
et al. [20] reported only a single endothermic peak with the max-
∼9.3 wt.% as in a pure Mg(AlH4)2. From TGA analysis, the
largest total weight loss ∼2.45 wt.%, which is observed for the
5 h milled powder, is much less than the theoretical value of
∼3.97 wt.%. A short of ∼1.5 wt.% might be a combined effect
of the partial decomposition of Mg(AlH4)2 during milling for
5 h and the underestimation of the weight loss in the second
◦
step (∼270–300 C) due to the oxidation of Mg during TGA
◦
imum around 280–290 C for the mixture Mg(AlH4) + 2NaCl.
run under flowing nitrogen. The total hydrogen desorption of
only ∼2.16 wt.% in TGA of the 10 h milled powder is most
probably due to the prior partial decomposition of Mg(AlH4)2
◦
The peak at ∼452 C is due to the eutectic melting of the Al3Mg2
and Al(Mg) mixture.
◦
◦
The temperature ranges of two weight losses at ∼120–170 C
during milling. No weight loss in the 100–180 C range for the
◦
range and ∼270–300 C, which were observed in TGA tests
40 h milled powder further confirms that Mg(AlH4)2 is fully
decomposed after a prolonged milling time.
for the powders containing Mg(AlH4)2 (5 and 10 h) (not shown
here), match well with those at DSC curves (∼120–170 and
◦
∼
225–340 C ranges; Fig. 3c). The TGA weight losses in the
3
.3. The nature of low-temperature heat flow events in DSC
first and second step decomposition for the 5 h milled powder are
.37 and 0.08 wt.%, respectively. Corresponding weight losses
2
The heat flow effects due to the decomposition of Mg(AlH4)2
for the 10 h milled powder are 2.14 and 0.02 wt.%. Assuming the
MCAS is completed (i.e., products have a 1 mol of Mg(AlH4)2
and 2 mol of NaCl) (Fig. 2), the theoretical hydrogen capacity
in the mixture of Mg(AlH4)2 + 2NaCl is ∼3.97 wt.% rather than
in DSC are difficult to monitor due to its very low enthalpy of
decomposition [19] and corresponding small amount of heat
flow (small DSC peaks). Fig. 5a shows the enlargement of
the low-temperature region up to 180 C observed on the full
DSC curve in Fig. 3a (scan rate 4 C/min) of the powder syn-
◦
◦
thesized for 10 h. Taking the flat DSC portion for the 40 h
sample as a baseline, one can invoke the existence of either
two exothermic or one endothermic peaks due to the decom-
position of Mg(AlH4)2. In order to shed more light on these
low-temperature heat flow events, we tested a few powder sam-
Table 2
Lattice parameters of Al and Al(Mg) solid solution in the as-milled and DSC
tested samples
Milling
time (h)
Temperature of
DSC test ( C)
Lattice parameter
of Al (FCC) (nm)
◦
◦
ples milled for 5 and 10 h at the scan rate of 20 C/min, which
10
10
10
10
40
10
As-milled
180
295
350
350
0.4037 ± 0.0015
0.4039 ± 0.0015
are plotted in Fig. 5b together with the flat trace for the 40 h
sample, which can be considered as a baseline. With respect
to this baseline, one 5 h and one 10 h milled samples show
the DSC humps, which could be considered as endothermic
events. In contrast, other 5 and 10 h samples show a diffuse
trough that could be considered as an exothermic reaction. In
a
0.4059 ± 0.0018
a
0.4062 ± 0.0016
a
0.4065 ± 0.0012
a
425
0.4063 ± 0.0013
a
Al(Mg) solid solution.

3
08
R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
◦
◦
Fig. 5. An enlargement of the low-temperature section of DSC curve of the powders milled for 10 and 40 h registered at the scan rate of: (a) 4 C/min; (b) 20 C/min;
◦
(
c) corresponding XRD patterns of the powders after heating to 125, 140, 150 and 180 C in a DSC test.
◦
◦
order to investigate the nature of microstructural changes up to
Mg(AlH4)2 starts occurring around 125 C (scan rate 4 C/min).
◦
◦
1
80 C a few samples of the 10 h milled powder which con-
After a DSC run up to 150 C, the microstructure contains only
tains Mg(AlH4)2 were run in a DSC test at the scan rate of
a very small amount of Mg(AlH4)2 as can be deduced from
a greatly reduced intensity of the strongest Mg(AlH4)2 peak
◦
4
C/min up to the temperatures marked by the vertical lines
◦
in Fig. 5a. Subsequently, XRD phase analysis was performed
for each sample after DSC test and the pertinent patterns are
at 2θ ∼ 25 (100% intensity) and almost complete disappear-
◦
ance of two other strong peaks at 2θ ∼ 34.5 (80% intensity)
◦
◦
shown in Fig. 5c. Samples annealed to 125 and 140 C still show
and 2θ ∼ 36.6 (75% intensity) (JCPDS# 47-980) in Fig. 5c
◦
◦
the presence of Mg(AlH4)2, residual NaCl, traces of ␤-MgH2
and the elemental Al rather than the Al(Mg) solid solution as
mentioned earlier. This in essence is almost the same microstruc-
ture as that after ball milling for 10 h (Fig. 2). The presence of
(150 C). The absence of Mg(AlH4)2 after DSC run to 180 C
is an indication of its complete decomposition.
◦
This analysis clearly shows that approximately up to 180 C,
Mg(AlH4)2 is already decomposed. However, the nature of this
transformation as being either endothermic [18,19] or exother-
␤
-MgH2 indicates that the beginning of the decomposition of

R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
309
◦
Fig. 6. (a) Hydrogen desorption kinetic curves obtained in a Sieverts-type apparatus at 150 and 350 C of the Mg(AlH4)2 + 2NaCl mixture synthesized for 5 h. (b)
XRD of the powders desorbed at 150 and 350 C.
◦
mic [20] cannot be unambiguously established. Most probably,
this situation arises owing to a very small enthalpy of decom-
position of Mg(AlH4)2, which is cited in the literature as being
equal to ∼1.7 kJ/molH [19]. It is quite possible that at such
small value of the enthalpy of decomposition, which in all
practical terms is almost close to zero, even small microstruc-
tural fluctuations always present in the powders synthesized by
mechano-chemical reactions may change the character of the
decomposition reaction from endothermic [18,19] to exother-
mic [20].
ever, the value 0.45 wt.% is much smaller than ∼2.37 wt.% H2
of the first step desorption estimated by TGA and much smaller
than the theoretical value of ∼2.98 wt.% H2 resulting from the
reaction (3) (the theoretical hydrogen capacity is ∼3.97 wt.%
in the mixture of Mg(AlH4)2 + 2NaCl, which is equal to the
amount stored within four H2 in Mg(AlH4)2; as a result, one
H2 should give us ∼0.99 wt.% (=3.97/4) in the mixture; taking
into account that three H2 are desorbed in the reaction (3) and
one H2 is released in the reaction (7a), the theoretical amount of
hydrogen desorbed in the reactions (3) and (7a) are ∼2.98 and
∼
0.99 wt.%, respectively). The phase make up after desorption
◦
3
.4. Kinetics of desorption
at 350 C (Fig. 6b) confirms that both the Al3Mg2 intermetallic
and Al(Mg) solid solution always form side by side when des-
◦
Hydrogen desorption kinetics of the powder mixture
orption is carried out above about 180 C as discussed earlier
Mg(AlH4)2 + 2NaCl synthesized for 5 h were studied in a
Sieverts-type apparatus at 150 and 350 C (Fig. 6a). A hydrogen
in the text. The observed microstructure also indicates that at
the temperature of 350 C the hydrogen desorption was com-
◦
◦
pressure barrier of 41 bar was supposed to prevent desorption
before the system temperature was stabilized at the desired test
pleted, i.e. both reactions (3) and (7) occurred. However, the
desorbed value of 0.75 wt.% H2 is still much smaller than the
total ∼2.45 wt.% H2 desorbed in TGA and ∼3.97 wt.% H2 of the
total amount of H2 stored in the Mg(AlH4)2 + 2NaCl mixture.
The question therefore arises as to the possible cause of
incomplete desorption in a Sieverts-type apparatus. Fichtner et
temperature. The desorbed amounts of hydrogen at 150 and
◦
3
50 C are 0.45 and 0.75 wt.%, respectively. XRD patterns of
◦
the powder after desorption at 150 C indicate a complete first
step decomposition (reaction (3)) of Mg(AlH4)2 (Fig. 6b). How-

3
10
R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
al. [7] reported the equilibrium pressure of pure Mg(AlH4)2 in
the 75–95 C range to be between 88 and 130 bar. As a result,
product requires some additional long-time annealing in vac-
uum [19]. Such a long production time may not be well suited
for industrial environment. There is also a problem with the
reversibility of the reaction (7). The reactions (7a) and (7b) are
reversible producing MgH2 and Al but still very sluggish [28].
However, our previous studies [15] suggest that the reaction (3)
between MgH2 and Al to form Mg(AlH4)2 is most likely irre-
versible.
◦
a hydrogen pressure barrier of 41 bar, which we used during
temperature stabilization period in a Sieverts-type apparatus in
the present work might have been insufficient to prevent the
◦
reaction (3) to occur during heating to 150 and 350 C in the
temperature stabilization period, giving rise to some desorption
according to the reaction (3) even before the temperature was
fully stabilized. This explanation is supported by the pressure
increase monitored during temperature stabilization period at
4. Conclusions
◦
1
50 C. Since the pressure increase is due to two factors such
as temperature and hydrogen release then its increment due to
just hydrogen release can be estimated by subtracting the pres-
sure increase due to temperature estimated from the heating
of empty crucible from the pressure registered while heating
crucible containing a powder sample synthesized for 5 h. This
estimate gives ∼1.79 wt.% H2 already released during temper-
A mixture of Mg(AlH4)2 + 2NaCl has been successfully syn-
thesized by the mechano-chemical activation synthesis (MCAS)
of the stoichiometric mixture of 2NaAlH4 and MgCl2 com-
pounds by ball milling for 5 and 10 h in a magneto-mill. The
particle size of the as-milled powder mixture is on the order of
2 ␮m after milling for 5, 10 and 40 h. The Mg(AlH4)2 hydride is
nanocrystalline having the grain size of around 18 nm. Even rela-
tively short milling for just 10 h results in a partial decomposition
of pre-synthesized Mg(AlH4)2 into ␤-MgH2 and the nanocrys-
talline elemental Al (grain size ∼26 nm). Prolonged milling for
40 h results in a complete decomposition of the Mg(AlH4)2 into
nanocrystalline ␤-MgH2 (grain size ∼11 nm) and the elemental
Al (grain size ∼20 nm).
◦
ature stabilization period at 150 C. Therefore, the total amount
of hydrogen desorbed will be the sum of the hydrogen desorbed
during temperature stabilization period (∼1.79 wt.%) and the
hydrogen desorbed in the isothermal desorption period of the
◦
test at 150 C, i.e. 0.45 wt.%. This summation gives the total of
2
◦
.24 wt.% H2 desorbed at 150 C, which is still slightly smaller
than the theoretical 2.98 wt.% H2.
Similar estimate taking into account the pressure increase
DSC test of the Mg(AlH4)2 + 2NaCl mixture synthesized for
5 and 10 h, results in the decomposition of Mg(AlH4)2 into ␤-
◦
calculated from the heating of empty crucible up to 350 C
◦
gives ∼2.5 wt.% H2 desorbed during temperature stabilization
MgH2 + 2Al + 3H2 (+2NaCl) within the range 125–180 C (at
◦
◦
period up to 350 C. Adding this up to the amount of ∼0.75 wt.%
the scan rate of 4 C/min) as confirmed by X-ray diffraction.
desorbed during the isothermal desorption period of the test at
However, thenatureofthistransformationbeingeitherendother-
mic [18,19] or exothermic [20] cannot be unambiguously estab-
lished because a number of samples tested in DSC at the scan
◦
3
50 C (as discussed above) gives us ∼3.25 wt.% of the total
hydrogen desorbed which is still smaller than the total theoret-
ical value of ∼3.97 wt.%H2 stored in the Mg(AlH4)2 + 2NaCl
mixture. We are not sure if this is a pure coincidence but in
◦
rate of 4 and 20 C/min showed both types of transformation,
most likely, owing to a very small enthalpy of decomposi-
tion of Mg(AlH4)2, which is cited in the literature as being
equal to ∼1.7 kJ/molH [19]. Desorption experiments of the
Mg(AlH4)2 + 2NaCl powder synthesized for 5 h, carried out in a
◦
both cases of Sieverts desorption at 150 and 350 C, the total
amount of hydrogen desorbed in the experiment is short of
∼
0.7 wt.% with respect to the theoretical hydrogen capacity val-
◦
ues. Sincethecalibratedaccuracyofdesorbedhydrogencapacity
is about ± 0.2 wt.% H2 in our Sieverts-type apparatus (calibrated
on a commercial MgH2) then the experimental deficiency of
Sieverts-type apparatus at 150 C, show that the total amount of
hydrogen desorbed during temperature stabilization period and
isothermal desorption period is ∼2.24 wt.%.
◦
∼
0.7 wt.% is not due to the experimental error and as such must
DSC test from 180 to 500 C of the Mg(AlH4)2 + 2NaCl mix-
be related to some other factor. One explanation could be that
the powder synthesized for 5 h already lost about ∼0.7 wt.% of
hydrogen due to partial decomposition of Mg(AlH4)2 during
milling. However, this needs further study.
ture synthesized for 5 and 10 h results in three endothermic
◦
reactions with the peak maxima at ∼271, ∼316 and ∼452 C
◦
◦
(at the scan rate of 4 C/min). In the first endo effect at ∼270 C
the ␤-MgH2 hydride decomposes and the free elemental Mg
reacts with the pre-existing free elemental Al forming an equi-
librium mixture of intermetallic compound Al3Mg2 and Al(Mg)
The major problem in the MCAS technology is the forma-
tion of a large amount of waste by-product NaCl. With 2 mol of
NaCl per only 1 mol of Mg(AlH4)2 (reaction (3)) the excellent
theoretical hydrogen capacity ∼9.3 wt.% of pure Mg(AlH4)2 is
reduced to the very inferior theoretical ∼3.97 wt.% H2 in the
mixture of Mg(AlH4)2 + 2NaCl. As reported by Mamatha et al.
◦
solid solution. The nature of the second peak at ∼315 C is not
clear and can be interpreted either as due to the formation of
the Al3Mg2 intermetallic compound or further decomposition
of the remnant ␤-MgH2. Heating of the powder synthesized for
40 h, which contains only a single ␤-MgH2 hydride results in
[
19] it is possible to separate Mg(AlH4)2 from NaCl by the
◦
Soxhlet extraction method which is based on the suspension
of the Mg(AlH4)2 + 2NaCl mixture in the diethyl ether (Et2O)
solvent. However, there are a couple of serious disadvantages of
this extraction method. First, the extracted Mg(AlH4)2 is usually
contaminated with the solvent adduct [20]. Second, the extrac-
tion process needs several days to be completed and then the
two endothermic peaks with the maxima at ∼290 and ∼451 C
◦
(at the scan rate of 4 C/min). The former peak is due to the
decomposition of ␤-MgH2. The endo peaks with the maximum
◦
at 451–452 C, observed in all the synthesized powders, are due
to the eutectic melting of the mixture of Al3Mg2 and Al(Mg)
as required by the binary phase diagram Al–Mg. Desorption

R.A. Varin et al. / Journal of Alloys and Compounds 439 (2007) 302–311
311
experiments on the powder synthesized for 5 h, carried out in
a Sieverts-type apparatus at 350 C (complete decomposition
[9] M. Fichtner, O. Fuhr, J. Alloys Compd. 345 (2002) 286–296.
10] M. Schwarz, A. Haiduc, H. Still, P. Paulus, H. Geerlings, J. Alloys Compd.
04–406 (2005) 762–765.
11] A. Fossdal, H.W. Brinks, M. Fichtner, B.C. Hauback, J. Alloys Compd.
87 (2005) 47–51.
12] A. Fossdal, H.W. Brinks, M. Fichtner, B.C. Hauback, J. Alloys Compd.
404–406 (2005) 752–756.
13] A.M. Seayad, D.M. Antonelli, Adv. Mater. 16 (2004) 765–777 (a review).
14] A. Zaluska, L. Zaluski, J.O. Str o¨ m-Olsen, J. Alloys Compd. 288 (1999)
◦
[
[
[
4
of ␤-MgH2), show that the total amount of hydrogen desorbed
during temperature stabilization period and isothermal desorp-
tion period is ∼3.25 wt.%. In both cases of Sieverts desorption
3
◦
at 150 and 350 C, the total amount of hydrogen desorbed in
[
[
the experiment is short by ∼0.7 wt.% with respect to the the-
oretical hydrogen capacity values which one could expect at
these respective temperatures. The factor causing such a discrep-
ancy may be related to the prior loss of hydrogen due to partial
decomposition of Mg(AlH4)2 during ball milling but the exact
explanation needs more studies. However, our recent results on
the attempts to directly synthesize Mg(AlH4)2 from the elemen-
tal metals by reactive milling [15] strongly suggests that the
reaction (7a) is most likely irreversible.
2
17.
15] R.A. Varin, Ch. Chiu, T. Czujko, Z. Wronski, Nanotechnology 16 (2005)
261–2274.
[
2
[16] T.N. Dymova, N.N. Mal’tseva, V.N. Konoplev, A.I. Golovanova,
D.P. Alexandrov, A.S. Sizareva, Russ. J. Coord. Chem. 29 (2003)
385–389.
[
[
[
[
17] T.N. Dymova, V.N. Konoplev, A.S. Sizareva, D.P. Alexandrov, Russ. J.
Coord. Chem. 25 (1999) 312–315.
18] M. Mamatha, C. Weidenthaler, A. Pommerin, M. Felderhoff, F. Sch u¨ th, J.
Alloys Compd. 416 (2006) 303–314.
19] M. Mamatha, B. Bogdanovi c´ , M. Felderhoff, A. Pommerin, W. Schmidt,
F. Sch u¨ th, C. Weidenthaler, J. Alloys Compd. 407 (2006) 78–86.
20] Y. Kim, E.-K. Lee, J.-H. Shim, Y.W. Cho, K.B. Yoon, J. Alloys Compd.
Acknowledgement
This work was financially supported by a grant from the Nat-
ural Sciences and Engineering Research Council of Canada,
which is gratefully acknowledged.
4
22 (2006) 283–287.
[21] Calka, A.P. Radlinski, Mater. Sci. Eng. A 134 (1991) 1350–1353.
[
[
22] Patents: WO 9,104,810; US 5,383,615; CA 2,066,740; EP 0,494,899; AU
43,949.
6
23] A. Calka, R.A. Varin, in: T.S. Srivatsan, R.A. Varin, M. Khor (Eds.), Pro-
ceedings of the International Symposium on Processing and Fabrication of
Advanced Materials IX (PFAM-IX), ASM International, Materials Park,
OH, 2001, pp. 263–287.
[24] H.P. Klug, L. Alexander, X-ray Diffraction Procedures for Polycrystalline
and Amorphous Materials, John Wiley & Sons, New York, 1974, pp.
618–708.
[25] R.A. Varin, S. Li, A. Calka, J. Alloys Compd. 376 (2004) 222–231.
[26] R.A. Varin, S. Li, Z. Wronski, O. Morozova, T. Khomenko, J. Alloys
Compd. 390 (2005) 282–296.
[27] D.R. Askeland, P.P. Phul e´ , The Science and Engineering of Materials,
Thomson Canada Ltd., Toronto, 2006, p. 842.
References
[
[
1] J.A. Ritter, A.D. Ebner, J. Wang, R. Zidan, Mater. Today (2003) 18–23.
2] A. Z u¨ ttel, P. Wenger, S. Rentsch, P. Sudan, Ph. Mauron, Ch. Emmenegger,
J. Power Sources 118 (2003) 1–7.
[
[
[
[
[
3] L. Schlapbach, A. Z u¨ ttel, Nature 414 (2001) 353–358.
4] B. Bogdanovi c´ , G. Sandrock, MRS Bull. 27 (2002) 712–716.
5] M. Fichtner, Adv. Eng. Mater. 7 (2005) 443–455 (a review).
6] E.C. Ashby, R.D. Schwartz, B.D. James, Inorg. Chem. 9 (1970) 325–332.
7] M. Fichtner, J. Engel, O. Fuhr, O. Kircher, O. Rubner, Mater. Sci. Eng. B
1
08 (2004) 42–47.
8] M. Fichtner, O. Fuhr, O. Kircher, J. Alloys Compd. 356/357 (2003)
18–422.
[
[28] M.H. Mintz, Z. Gavra, G. Kimmel, Z. Hadari, J. Less-common Met. 74
(1980) 263–270.
4