Hydrogenation properties of Mg2AlNi2 and mechanical alloying in the Mg-Al-Ni system

Article abstract of DOI:10.1016/j.jallcom.2008.10.023

Samples with Mg_mAlNi_n composition (m, n ≤ 3) were synthesized by ball milling in form of crystalline nanoparticles, and were found to be a single phase with partially disordered CsCl-type cubic structure. For compositions with large

Full text of DOI:10.1016/j.jallcom.2008.10.023

Journal of Alloys and Compounds 477 (2009) 420–424
Contents lists available at ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Hydrogenation properties of Mg2AlNi2 and mechanical alloying
in the Mg–Al–Ni system
A. Parentea, A. Nalea, M. Cattia,∗, E. Kopninb, P. Caracinob
a Dipartimento di Scienza dei Materiali, Università di Milano Bicocca, via Cozzi 53, 20125 Milano, Italy
b Pirelli Labs, viale Sarca 336, 20126 Milano, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Samples with MgmAlNin composition (m, n ≤ 3) were synthesized by ball milling in form of crystalline
nanoparticles, and were found to be a single phase with partially disordered CsCl-type cubic structure. For
compositions with large m and small n, minor quantities of Mg and Mg2Ni are segregated. XRD Rietveld
Received 24 June 2008
Received in revised form
30 September 2008
Accepted 2 October 2008
Available online 21 November 2008
¯
refinements indicate that the two Pm3m sublattices basically have (Mg, Al) and (Mg, Al, Ni) occupa-
tions, respectively. Hydrogenation experiments on the Mg2AlNi2 sample by the PCI technique showed a
reversible absorption/desorption of 1.4 wt% H2 with formation/decomposition of the MgH2 and Mg2NiH4
hydrides. The van’t Hoff plot gives ꢀrH = 67.7 kJ mol−1 (enthalpy of dehydrogenation), and T0 = 524 K (tem-
perature of dehydrogenation at p = 0.1 MPa). A kinetic study of the gas release at 535 K showed a satisfactory
behaviour, with a first fast step followed by a slower one in both the absorption and desorption processes.
© 2008 Elsevier B.V. All rights reserved.
Keywords:
Hydrogen storage material
Mechanical alloys
Nanocrystalline phases
1. Introduction
Al–H bonding in aluminium hydride [5], a mixed Mg–Al hydride
with intermediate bond strength between MgH2 and AlH3 could
Despite its well known weak points, magnesium hydride is still
often considered to be the best hydrogen storage material which
is presently available. Many efforts were made in the last years to
improve its performance by lowering the decomposition temper-
ature and/or improving the corresponding kinetics. In most cases,
doping or alloying of elemental Mg with very minor quantities of
other metals, oxides or different compounds was attempted; pos-
itive results were often obtained as far as the kinetics, but not
thermodynamics, of the MgH2 → Mg + H2 reaction is concerned
[1–3].
The way leading to true Mg-containing inter-metallic com-
pounds or alloys was comparatively less explored. Although the
storage capacity may be expected to decrease drastically on reduc-
ing the Mg contents of the system, significant improvements
could be achieved with a lower dehydrogenation temperature.
Within this line of research, the Mg–Ni and Mg–Al binary sys-
tems were investigated. The first one, which provides the important
hydrogenation reaction Mg2Ni + 2H2 → Mg2NiH4, is also attractive
because of the well known catalytic properties of Ni [3]. Nanocrys-
talline Mg + Mg2Ni composites were thus claimed to improve the
hydrogenation kinetics of both components [4]. The Mg–Al sys-
tem aroused interest in the hope that, owing to the quite unstable
be formed. This was not found to occur; yet the Mg17 Al12 binary
compound proved to hydride reversibly with formation of MgH2
and Al [6–8].
Concerning the Mg–Al–Ni ternary composition, some studies
were performed by doping Mg2Ni with very minor Al quanti-
ties [9–11], and an improvement of the desorption kinetics was
generally observed. The Mg1-xAlxNi [12] and Mg2-xAlxNi [13–16]
systems were explored within the x ≤ 0.5 compositional range, but
in most cases the reaction with hydrogen was characterized only
partially or was not considered at all. In particular, the Mg3AlNi2
compound turned out to present different cubic phases according
¯
to the preparation method: a Fd3m NiTi2-type crystal structure,
when synthesized by conventional heat treatments [13,15,16], and a
¯
Pm3m CsCl-type structure when prepared by ball milling [14]. Also
the ball milled MgAlNi2 compound showed the CsCl-type phase
[12]. The hydrogenation capacity proved to decrease substantially
with the increase of the Al contents, but the equilibrium tempera-
ture of dehydrogenation at/or above room pressure was not studied.
We thus started an investigation of a broader compositional area
within the Mg–Al–Ni ternary system, along the MgmAlNin lines
with m, n = 1, 2, 3. The first aim was to synthesize samples with
several compositions by mechanical alloying, in order to obtain
nanocrystalline materials with best performance for the hydro-
genation reaction. Secondly, we wanted to understand in what
conditions single-phase or poly-phase materials are obtained, and
to characterize their crystallographic features. The last goal was to
∗
Corresponding author. Fax: +39 02 64485400.
E-mail address: catti@mater.unimib.it (M. Catti).
0925-8388/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2008.10.023

A. Parente et al. / Journal of Alloys and Compounds 477 (2009) 420–424
421
Fig. 2. Powder XRD patterns (CuK␣ radiation) of the MgmAlNi2 samples with
m = 1–3.
Fig. 1. Powder XRD patterns (CuK␣ radiation) of the MgmAlNi samples with m = 1–3.
The main phase is a pseudo-body-centered-cubic (Pm3m space group) AlNi-type
¯
solid solution.
tribution in view of the close scattering powers of such atoms, we
can conclude that some disordering of (Mg, Al) vs. Ni should occur
in at least one of the two sites of the CsCl-like structure, at variance
with the full order observed in the AlNi alloy. A progressive sepa-
ration of Mg out of the pseudo-BCC solid solution appears clearly
to occur as m increases along each MgmAlNin series (Figs. 1–3). In
addition to elemental Mg, also Mg2Ni and Ni may separate for the
compositions richest in Mg (Table 1). We also prepared samples
with Mg/Al ratio of 4:1; in this case larger quantities of the same
phases were observed in addition to the p-BCC solid solution, and
results are not reported here.
Rietveld profile refinements of the collected XRD powder pat-
terns were performed by the FULLPROF code [17]. Taking into
account the poor quality of the data (small scattered intensity and
few very broad Bragg peaks), due to low crystallinity of the nanopar-
ticles, it was not possible to distinguish between Mg and Al atoms.
The starting structure was then modelled with a full Mg atom in the
0, 0, 0 site and a mixture of Mg and Ni consistent with the sample
nominal composition in the 1/2, 1/2, 1/2 site, considering Al equiv-
alent to Mg. The Mg/Ni fraction was refined on each site, but the
quantity of Ni in 0, 0, 0 never attained 10%, so that in most cases the
starting composition was finally kept fixed. A pseudo-Voigt func-
tion was used for the Bragg peak profile, and the background was
treated by interpolation through a number of selected points in
the pattern. For the samples of the MgmAlNi series, with larger
amounts of impurities (Fig. 1), Mg was included as a second phase
in the refinement. Convergence was attained with average Rp and
Rwp values of 0.13 and 0.18. The unit-cell parameters are reported
in Table 1. The uncertainty is estimated to be of the order of 0.01 Å
from results on different samples with the same composition. A
assess the thermodynamic and kinetic behaviour of some of these
samples with respect to hydrogen absorption/desorption. In par-
ticular, we intended to clearly identify the active chemical species
in the hydride formation, and to elucidate their effects on the tem-
perature of dehydrogenation of the material at pressures not lower
than 0.1 MPa.
2. Experimental
The materials were prepared by mechanical alloying in Retsch PM100 planetary
and a Spex 8000 high-energy ball mills (Ar atmosphere), with milling times varying
from 30 to 100 h, and ball-to-powder-mass-ratios of 10:1. Reagent-grade pure metals
were used as starting materials, and all samples were always handled in a glove-box
filled with purified argon. The progress of the mechanochemical reaction was peri-
odically checked by X-ray powder diffraction (XRD) in the course of the milling. The
reaction was considered to be ended when no appreciable XRD change appeared
on further milling. A Bruker D8 Advance powder diffractometer, with CuK␣ radi-
ation (ꢁ = 1.5418 Å) and secondary-beam monochromator, was used for the X-ray
characterization. Microstructural analyses were performed by means of a Scanning
Electron Microscope (SEM: Tescan Vega Ts 5136) with an energy-dispersive X-ray
spectrometer (EDX: EDAX Genesis 4000).
Thermodynamic and kinetic measurements of the hydrogenation/dehydro-
genation reactions were done by the method of pressure-composition-isotherm
(PCI), employing an automatic Sievert-type apparatus of the Advanced Materials
Corporation. The sample of 0.5 g was loaded into the 5 cc reactor located in a fur-
nace with temperature control. After evacuation, hydrogen gas of high purity was
introduced at increasing pressures.
3. Results and discussion
3.1. Synthesis and XRD characterization of the Mg–Al–Ni alloys
Samples were synthesized (Spex 8000, 30 h milling time) with
the 1:1, 1:2, and 1:3 Al/Ni molar ratios; for each of them, the
Mg/Al ratio was increased from 1:1 to 3:1. In case of the Mg2AlNi2
composition, for which detailed hydrogenation measurements are
reported below, other samples were also prepared by 100 h milling
(Retsch equipment). As main feature, the final XRD pattern of all
¯
samples showed four broad peaks of the CsCl-type Pm3m cubic
structure in the 2ꢂ ≤ 70◦ range (Figs. 1–3): (1 0 0), (1 1 0), (1 1 1),
and (2 0 0) at increasing angles. The same structure is observed for
the AlNi alloy, with a = 2.882 Å. The (1 0 0) and (1 1 1) reflexions,
¯
¯
distinguishing the Pm3m from the Im3m space group (BCC lattice),
appear to be comparatively weak in all diffractograms. Thus, the
main phase present in all samples looks like a partially ordered
solid solution with AlNi-type structure; if the two atomic sites at
x = 0, y = 0, z = 0 and 1/2, 1/2, 1/2 had the same average chemical
composition, a true BCC lattice would be attained. In particular, as
the intensity of Bragg peaks can be hardly affected by the Mg/Al dis-
Fig. 3. Powder XRD patterns (CuK␣ radiation) of the MgmAlNi3 samples with
m = 2–3.

422
A. Parente et al. / Journal of Alloys and Compounds 477 (2009) 420–424
Table 1
Phase compositions of the samples with nominal MgmAlNin stoichiometry, from powder XRD data (cf. Figs. 1–3). The cell edge a of the pseudo-body-centred-cubic AlNi-type
¯
main phase (Pm3m space group) is given in parentheses (the estimated uncertainty is 0.010 Å). AlNi itself has a = 2.882 Å.
n = 1
n = 2
n = 3
m = 1
m = 2
m = 3
MgAlNi
p-BCC (2.880 Å) + traces of Mg
MgAlNi2
p-BCC (2.943 Å)
Mg2AlNi
p-BCC (2.882 Å) + Mg
Mg2AlNi2
p-BCC (2.950 Å) + traces of Mg2Ni
Mg2AlNi3
p-BCC (2.955 Å)
Mg3AlNi
Mg3AlNi2
Mg3AlNi3
p-BCC (2.884 Å) + Mg + Ni
p-BCC (2.925 Å) + traces of Mg2Ni
p-BCC (2.946 Å) + traces of Mg2Ni
clear trend of increasing a with the Ni content n is observed, con-
sistent with its slightly larger metallic radius; in particular, the a
values in the MgmAlNi series are comparable with the a of AlNi.
The effect of Mg content on the unit-cell dimension can not be
assessed clearly, taking also into account the tendency of Mg3AlNin
compositions to segregate Mg-rich phases.
maximum intensity, was applied to the (1 1 0) peak of the pseudo-
BCC phase in order to estimate the L size of the particles obtained by
ball milling. Values in the 6–12 nm range are obtained, depending
on sample composition and milling time.
3.2. Hydrogenation reactions
In summary, the p-BCC solid solution is obtained by ball milling
as a pure phase in a wide Mg–Al–Ni compositional range. The
molar ratio Ni/(Mg + Al + Ni) seems to have a lower threshold of
about 1/3, below which the solid solution starts to decompose seg-
regating Mg-rich phases; similarly, a higher threshold of 1/2 can
be observed for the Mg/(Mg + Al + Ni) ratio (Table 1). It should be
remarked that this p-BCC phase is not reported in the phase dia-
grams of the Mg–Al–Ni ternary system [18], so that it is probably
thermodynamically metastable, and the driving force for its forma-
tion may be related to the large surface area and small nanosize of
the crystal particles.
The prepared compounds were examined by SEM with micro-
analysis attachment, which confirmed the nominal compositions
by several spot samplings. No chemical heterogeneities could be
detected, so that the size of separated Mg-rich particles should be
below the space resolution of the instrument. A typical SEM image
is shown in Fig. 4 for the Mg2AlNi2 composition. The Scherrer equa-
tion L = 0.94ꢁ/(Bcos ꢂ), where B(2ꢂ) is the full-width (radians) at half
Some of the prepared samples were tested for hydrogenation
by the PCI technique. In terms of wt% H2 absorption, Mg2AlNi2
gave the best results, which are hereafter reported in detail. A
short activation treatment at 3 MPa and 473 K was sufficient to
achieve the hydrogenation reaction. The H2 absorption and des-
orption isotherms in the 485–583 K range (Fig. 5) show that about
1.4 wt% of gas reacts reversibly with the alloy. At the lowest tem-
perature of 485 K hydrogenation occurs at about 0.1 MPa, but the
release of H2 has to be forced below room pressure (p = 25 kPa), yet
with a limited ꢀp hysteresis. In order to dehydrogenate the sam-
ple at 0.12 MPa, T has to be raised to 531 K. In those conditions,
the plateau of the hydrogenation step (average pressure 0.4 MPa)
is not as stable as at lower temperature, and a larger hysteresis
is observed. On rising temperature the pressure hysteresis is con-
firmed to increase, and the H2 absorption process is kinetically
hindered; however, the dehydrogenation isotherms are still very
regular with flat plateaux, indicating good equilibrium conditions
for the desorption process.
Fig. 5. Pressure-composition-isotherms of the Mg2AlNi2/H2 system at four different
temperatures. Full and open symbols refer to hydrogen desorption and absorption,
respectively.
Fig. 4. Scanning electron microscope picture of the sample of Mg2AlNi2 prepared
by ball milling.

A. Parente et al. / Journal of Alloys and Compounds 477 (2009) 420–424
423
Fig. 7. Kinetics of hydrogen absorption and desorption (full and open symbols) of
Mg2AlNi2 at 535 K.
Fig. 6. van’t Hoff plot of the dehydrogenation reaction of hydrogenated Mg2AlNi2,
from data of Fig. 5.
depleted of Mg and Ni with respect to the initial alloy, correspond-
ing to a Mg0.5AlNi1.5 composition in order to match the required
stoichiometry. A slight decrease of the full width at half-maximum
of the (1 0 0) Bragg peak is observed, with respect to the original
alloy, indicating a particle size of about 15 nm.
On repeated cycling of hydrogenation/dehydrogenation, the
original p-BCC solid solution with Mg2AlNi2 composition is not
formed any more, but the MgH2 and Mg2NiH4 hydrides only are
involved in the process. This is proved by the fact that XRD patterns
of the dehydrogenated samples always show significant amounts
of Mg2Ni and Mg, consistent with the following reaction:
From the equilibrium pressures of the four desorption curves
of Fig. 5 (0.03, 0.12, 0.22 and 0.52 MPa), a van’t Hoff plot of
type ln(p/p0) = ꢀrH/(RT0) − (ꢀrH/R)(1/T) was derived and displayed
in Fig. 6. A good linear trend is shown by the data, and the
following values of thermodynamic parameters are obtained by
least-squares fitting: ꢀrH = 67.7 kJ mol−1 (enthalpy of dehydro-
genation), and T0 = 524 K (temperature of dehydrogenation at
p = 0.1 MPa). The decomposition enthalpy is similar to the value
reported for Mg2NiH4 (68 kJ mol−1 [19]), and significantly lower
than that of MgH2 (76 kJ mol−1 [20]).
Kinetic measurements were also performed for both the absorp-
tion and desorption reactions at T = 535 K. The degree of reaction
advancement in each case is plotted vs. time in Fig. 7; the absorp-
tion and desorption pressures were 0.4 and 0.1 MPa, as at T = 531 K.
Both processes are characterized by a first fast step, followed by a
quite slower step with quasi-linear behaviours. In the case of hydro-
genation, the fast step lasts about 1500 s for 85% completion of the
reaction, with an average rate of 5.7 × 10−4 s−1. Much shorter and
faster is the first stage of dehydrogenation, with 90% of the process
completed in 800 s (average rate of 1.1 × 10−3 s−1). The second steps
have rates of 1.1 × 10−5 and of 7.0 × 10−6 s−1 for the absorption and
desorption processes, respectively.
Mg2NiH4 + MgH2 + 2Mg0.5AlNi1.5 ↔ Mg2Ni + Mg
+ 2Mg0.5AlNi1.5 + 3H2.
(2)
The very fine dispersion of the MgH2 and Mg2NiH4 particles in
the nanosized p-BCC matrix with presumed Mg0.5AlNi1.5 composi-
tion is likely to be responsible for the low temperature of hydrogen
release.
An XRD analysis of the hydrogenated sample gave the pat-
tern shown in Fig. 8. Both the Mg2NiH4 and (tetragonal) ␣-MgH2
hydrides are present, and they appear to be the only phases
responsible for the hydrogen absorption and desorption processes.
Further, a phase with p-BCC structure and a lattice constant of
a = 2.944 Å is left in considerable amount in the reacted material.
A few very weak other peaks of uncertain identification are also
observed in the XRD pattern. Thus, the Mg2AlNi2 alloy turns out to
be decomposed by hydrogen, according to a reaction which may be
plausibly guessed as follows:
2Mg2AlNi2 + 3H2 → Mg2NiH4 + MgH2 + 2Mg0.5AlNi1.5
.
(1)
This scheme corresponds to a theoretical absorption of 1.5 wt%
of H2, not far from the observed value, and it is consistent with the
semi-quantitative phase composition shown by the diffractogram.
In particular, the p-BCC phase in the reacted material must be
Fig. 8. Powder XRD pattern (CuK␣ radiation) of the Mg2AlNi2 hydrogenated sample.

424
A. Parente et al. / Journal of Alloys and Compounds 477 (2009) 420–424
4. Conclusions
Albizzati and Vincenzo Boffa (Pirelli Labs) for their continuous sup-
port to this research project. Bruno Vodopivec helped us with the
SEM characterization.
Mechanical alloying by ball milling in a wide compositional
range of the Mg–Al–Ni system was shown to produce a nanocrys-
talline solid solution with Pm3m cubic symmetry. The solid solution
¯
References
is partially ordered over the two independent sites of the CsCl-type
[1] K.S. Jung, E.Y. Lee, K.S. Lee, J. Alloys Compd. 421 (2006) 179–184.
[2] L.P. Ma, P. Wang, H.M. Cheng, J. Alloys Compd. 432 (2007) L1–L4.
[3] R.A. Varin, T. Czujko, E.B. Wasmund, Z.S. Wronski, J. Alloys Compd. 432 (2007)
217–231.
[4] G. Liang, S. Boily, J. Hulot, A. Van Neste, R. Schulz, J. Alloys Compd. 267 (1998)
302–306.
[5] J. Graetz, J.J. Reilly, J. Alloys Compd. 424 (2006) 262–265.
[6] S. Bouaricha, J.P. Dodelet, D. Guay, J. Huot, S. Boily, R. Schulz, J. Alloys Compd.
297 (2000) 282–293.
[7] A. Andreasen, M.B. Sørensen, R. Burkarl, B. Møller, A.M. Molenbroek, A.S. Peder-
sen, J.W. Andreasen, M.M. Nielsen, T.R. Jensen, J. Alloys Compd. 404–406 (2005)
323–326.
structure, according to the presence of weak hkl reflections with
¯
odd h + k + l values violating the Im3m symmetry (BCC). The two
sites have (Mg, Al) and (Mg, Al, Ni) atomic occupations, respec-
tively. These results confirm and generalize the previously reported
ball milling synthesis of CsCl-type phases for the MgAlNi2 [12] and
Mg3AlNi2 [14] particular compositions, belonging to the presently
explored range. By production of new metastable phases with
nanosized particles in metallic systems, access may be given to
novel alloys with interesting properties which can not be obtained
by equilibrium thermodynamics methods. This holds particularly
for the hydrogen storage application.
[8] H. Yabe, T. Kuji, J. Alloys Compd. 433 (2007) 241–245.
[9] T. Sato, H. Blomqvist, D. Noréus, J. Alloys Compd. 356–357 (2003) 494–
496.
Within the pseudo-BCC synthesized ternary alloys, the sample
with Mg2AlNi2 composition proved to react with hydrogen forming
a mixture of the MgH2 and Mg2NiH4 hydrides with a Mg-poorer
p-BCC solid solution. About 1.4 wt% of H2 is absorbed/desorbed
reversibly for several cycles; hydrogen can be extracted at 531 K
at a pressure not smaller than 0.1 MPa. Although the quantity of
stored hydrogen is limited, the temperature of release at p ≥ 0.1 MPa
is significantly lower than for pure MgH2, and also favourable
with respect to Mg2NiH4. This indicates that appreciable improve-
ments in the dehydrogenation process of Mg-alloy hydrides can
be achieved, by producing the hydride in a convenient chemical–
physical environment. The contribution of interface free energy
may be responsible for the modified hydrogenation response of
nanosized systems dispersed in a heterogeneous matrix.
[10] S. Doppiu, P. Solsona, T. Spassov, G. Barkhordarian, M. Dornheim, T. Klassen, S.
Surinach, M.D. Baró, J. Alloys Compd. 404–406 (2005) 27–30.
[11] Q. Li, K.C. Chou, Q. Lin, L.J. Jiang, F. Zhan, Int. J. Hydrogen Energy 29 (2004)
1383–1388.
[12] S. Orimo, A. Züttel, K. Ikeda, S. Saruki, T. Fukunaga, H. Fujii, L. Schlapbach, J.
Alloys Compd. 293–295 (1999) 437–442.
[13] H.T. Yuan, L.B. Wang, R. Cao, Y.J. Wang, Y.S. Zhang, D.Y. Yan, W.H. Zhang, W.L.
Gong, J. Alloys Compd. 309 (2000) 208–211.
[14] A. Gasiorowski, W. Iwasieczko, D. Skoryna, H. Drulis, M. Jurczyk, J. Alloys Compd.
364 (2004) 283–288.
[15] G.L. Lü, L.S. Chen, L.B. Wang, H.T. Yuan, J. Alloys Compd. 321 (2001)
L1–L4.
[16] L.B. Wang, J.B. Wang, H.T. Yuan, Y.J. Wang, Q.D. Li, J. Alloys Compd. 385 (2004)
304–308.
[17] J. Rodriguez-Carvajal, FULLPROF: a program for Rietveld Refinement and Pat-
tern Matching Analysis, http://www-llb.cea.fr/fullweb/powder.htm.
[18] S.B. Prima, in: P. Villars, H. Okamoto, K. Cenzual (Eds.), ASM Alloy
Phase Diagrams Center, ASM International, Materials Park, OH. http://www.
asminternational.org/AsmEnterprise/APD, 2006.
[19] M.L. Post, J.J. Murray, J. Less Common Met. 134 (1987) 15–26.
[20] M. Yamaguchi, E. Akiba, in: R.W. Cahn, P. Haasen, E.J. Kramer (Eds.), Materials
Science and Technology, vol. 3B, VCH, New York, 1994, pp. 333–345.
Acknowledgements
We thank Claudia Parola, Cristina De Piccoli and Cristina
Carpanese for their help with the sample preparation, and Enrico