Hydrogen storage properties of Na-Li-Mg-Al-H complex hydrides

Article abstract of DOI:10.1016/j.jallcom.2006.12.089

Lightweight complex hydrides have attracted attention for their high storage hydrogen capacity. NaAlH₄ has been widely studied as a hydrogen storage material for its favorable reversible operating temperature and pressure range for automotive fuel cell applications. The increased understanding of NaAlH₄ has led to an expanded search for high capacity materials in mixed alkali and akali/alkaline earth alanates. In this study, promising candidates in the Na-Li-Mg-Al-H system were evaluated using a combination of experimental chemistry, atomic modeling, and thermodynamic modeling. New materials were synthesized using solid state and solution based processing methods. Their hydrogen storage properties were measured experimentally, and the test results were compared with theoretical modeling assessments.

Full text of DOI:10.1016/j.jallcom.2006.12.089

Journal of Alloys and Compounds 446–447 (2007) 228–231
Hydrogen storage properties of Na–Li–Mg–Al–H
complex hydrides
Xia Tanga,∗, Susanne M. Opalkaa, Bruce L. Laubea,
Feng-Jung Wub, Jamie R. Stricklerb, Donald L. Antonc
a United Technologies Research Center, 411 Silver Lane, East Hartford, CT 06108, United States
b Albemarle Corporation, Gulf States Road, Baton Rouge, LA 70805, United States
c Savannah River National Laboratory, 227 Gateway Dr., Aiken, SC 29808, United States
Received 25 September 2006; received in revised form 4 December 2006; accepted 14 December 2006
Available online 28 December 2006
Abstract
Lightweight complex hydrides have attracted attention for their high storage hydrogen capacity. NaAlH4 has been widely studied as a hydro-
gen storage material for its favorable reversible operating temperature and pressure range for automotive fuel cell applications. The increased
understanding of NaAlH4 has led to an expanded search for high capacity materials in mixed alkali and akali/alkaline earth alanates. In this
study, promising candidates in the Na–Li–Mg–Al–H system were evaluated using a combination of experimental chemistry, atomic modeling,
and thermodynamic modeling. New materials were synthesized using solid state and solution based processing methods. Their hydrogen storage
properties were measured experimentally, and the test results were compared with theoretical modeling assessments.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Metal hydrides; Chemical synthesis; Solid state reactions; X-ray diffraction
1. Introduction
2. Experimental
The starting materials, NaH (95%, with NaOH and NaNO3 as the major
impurities by X-ray diffraction (XRD)), LiH (95%, with Li2O as the major impu-
rity), MgH2 (90%, with Mg and MgO as major impurities), MgCl2 (99.9%) and
LiAlH4 (95%, with Al and Li3AlH6 as the major impurities), were purchased
from the Aldrich Co. AlH3 was obtained using a proprietary method. Detailed
characterization of this aluminum hydride will be described in a future pub-
lication [4]. The XRD pattern of this material shows solely ␣-AlH3, with no
detectable impurities. Purified NaAlH4 (98.7% by ICP analysis), was obtained
from Albemarle Co. All solid materials were used in the as-received condition
withoutfurtherpurification. Anhydroustolueneanddiethyletherwerepurchased
from Aldrich and dried over aluminum and Na/K, respectively. The chemicals
were stored, measured, and transferred under high purity N2 inside a glove box
with oxygen content less than 25 ppm. High pressure hydrogen gas (240 bar,
99.95% purity), was used for hydriding reactions.
The Na–Li–Mg–Al–H system materials were prepared by mixing a 1:1:1:1
mole ratio of NaH, LiH, MgH2, and AlH3 in a mortar and pestle for 15 min.
Approximately 10 g of the mixture were ball milled in a high energy SPEX
mill in N2 for 3 h. Immediately after ball milling, approximately 1 g samples
were hydrided under 190–195 bar hydrogen for 20 h each at 60, 80, 100, and
120 ◦C, respectively, inamodifiedSievert’sapparatus(AdvancedMaterialsCo.).
The Li–Mg–Al–H complex materials were prepared by a similar method using
LiH, MgH2, and AlH3 at a 1:1:1 mole ratio. LiMg(AlH4)3 was prepared by the
solution method developed by Bulychev, et al. [5], using 30–50% of the solvent
volume used in the original method. Approximately 0.6 g of sample was tested
Automotive hydrogen fuel cells require hydrogen storage
materials with high volumetric and gravimetric capacity. The
discovery of the reversible Ti-doped sodium aluminum hydride,
NaAlH4, system by Bogdanovic [1] inspired extensive investi-
gations of this light metal complex hydride for hydrogen storage
applications [2–3]. Although NaAlH4 has a 5.5 wt% theoretical
hydrogen storage capacity with reasonable desorption kinetics at
intermediate temperatures, it cannot meet the capacity require-
ments for on-board hydrogen storage. A new material with
higher capacity and more rapid desorption kinetics is needed.
Mixed alkali and alkaline earth alanates have theoretical hydro-
gen contents as high as 10.4 wt%. These compounds have the
potential to achieve the required reversible storage capacity. In
this paper, our results within the Na–Li–Mg–Al–H system are
reported.
∗
Corresponding author. Tel.: +1 860 610 7432.
E-mail address: tangx@utrc.utc.com (X. Tang).
0925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2006.12.089

X. Tang et al. / Journal of Alloys and Compounds 446–447 (2007) 228–231
229
in a modified Sievert’s apparatus for desorption under ca. 1 bar and absorption
under ca. 190 bar of hydrogen pressure.
milling process:
2AlH3 → 2Al + 3H2
(1)
(2)
(3)
(4)
XRD analyses were performed on a Rigaku D/Max-b diffractometer, Model
RU-200B, equipped with a rotating Cu anode, or a Scintag X1 Advanced Diffrac-
tion System equipped with a Cu anode housed in a glove box. Samples for the
Rigaku were mounted on a glass sample holder and covered with a layer of
DuraSealTM, sealed with a ring of silicone grease in a glove box. DSC measure-
ments were conducted on TA Instruments 2910 or 2920. The sample pans were
made of gold plated stainless steel, and the sample-heating rate was 10 ◦C/min.
The ground state atomic structures of known and hypothetical phases in
the Na–Li–Mg–Al–H system were predicted with the density functional the-
ory Vienna Ab-initio Simulation Package (VASP) code [6–7] using projector
augmented wave potentials [8] with generalized gradient PW91 exchange-
correlation corrections [9]. The thermodynamic properties of the most stable
phase structures were predicted with direct method lattice dynamics using the
Materials Design MedeA Phonon module [10–11].
NaH + AlH3 → NaAlH4
2NaH + LiH + AlH3 → Na2LiAlH6
NaH + MgH2 → NaMgH3
Upon heat-treatment under 195 bar of hydrogen, the sample
composition remained unchanged below 100 ◦C. At temper-
atures greater than 100 ◦C, the signal of NaAlH4 decreased
substantially. Signals from the Na2LiAlH6, NaMgH3, and Al
phases increased, indicating that Eqs. (5) and (6) may occur at
this temperature.
2NaAlH4 + LiH → Na2LiAlH6 + Al + 3/2H2
NaAlH4 + MgH2 → NaMgH3 + Al + 3/2H2
(5)
(6)
3. Results and discussion
3.1. Na–Li–Mg–Al–H
In this system, NaH appears to be the most reactive among
alkali and alkaline earth hydrides. It can form binary complex
hydrides with AlH3 and MgH2, respectively, and ternary com-
plex with LiH plus AlH3. Among the complex hydrides formed,
Na2LiAlH6 and NaMgH3 are most stable. The existence of
Na2LiAlH6 and NaMgH3 [1,12–13] has been reported in the
literature. No LiAlH4, Li3AlH6, Na3AlH6 or mixed Li–Mg–Al
hydride phases were observed, indicating they are less stable
than the Na2LiAlH6 and NaMgH3 under this reaction environ-
ment. It could also be due to the fact that their formation is
kinetically impeded.
In order to further understand thermodynamic stability with-
out the interference of the kinetic factors, first principles
thermodynamic property predictions were utilized to generate
data for equilibrium thermochemical calculations. Below 87 ◦C,
the LiH destabilization of NaAlH4 to form the Na2LiAlH6 inter-
mediate phase (Eq. (5)), with the release of only 2.6 wt% H2, is
predicted to predominate. At higher temperatures, the NaAlH4
decomposition to the intermediate phase, Na3AlH6, with the
release of 3.7 wt% H2 is predicted to predominate, even in
the presence of LiH. In contrast, the MgH2 destabilization of
NaAlH4 to form NaMgH3 (Eq. (6)), is predicted to predomi-
nate above 52 ◦C, leading to the release of 4.8 wt% H2. These
predictions confirm the above experimental observations that
Na2LiAlH6 and NaMgH3 are the predominant phases at higher
temperature.
The syntheses of mixed alkali and alkaline earth salts of alu-
minum hydride from simple hydrides were attempted by the
solid synthesis method described in the experimental section.
XRD examination of the hand mixed sample of NaH, LiH,
MgH2, and AlH3 at 1:1:1:1 molar ratio did not show evidence
of chemical reactions, although the LiH signal was absent in
mixture. It is suspected that the LiH signal, which should not
be masked by any of the other constituents, is absorbed by one
or more of the other compounds present. After SPEX milling,
several reaction products were detected by XRD, including
NaAlH4, Na2LiAlH6, NaMgH3, and Al (Fig. 1). This obser-
vation suggests that the following reactions occurred during
3.2. Li–Mg–Al–H
The initial attempted synthesis of Li–Mg–Al–H compounds
focused on solid state methods. A reaction mixture contain-
ing LiH, MgH2, and AlH3 (1:1:1) was hand mixed and SPEX
milled for 3 h. The SPEX milling resulted in the formation
of LiAlH4. Upon heat-treatment to 60 ◦C under 190–195 bar
hydrogen, LiAlH4 dissociated to Al and Li3AlH6. Further heat-
ing above 100 ◦C in hydrogen dissociated the Li3AlH6 to LiH
and Al. The MgH2 did not react. No mixed Li–Mg–Al hydride
phase formed using the solid-state method with simple hydride
precursors.
Fig. 1. XRD of the reaction mixture of LiH, NaH, MgH2, and AlH3 at 1:1:1:1
molar ratio: (a) SPEX milled for 3 h, (b) 80 ◦C hydriding at 195 bar hydrogen
pressure for 20 h, and (c) 120 ◦C hydriding at 195 bar hydrogen pressure for
20 h.

230
X. Tang et al. / Journal of Alloys and Compounds 446–447 (2007) 228–231
Fig. 2. XRD of LiMg(AlH4)3 synthesized by a solution based chemical synthe-
sis method.
3.3. LiMg(AlH4)3
It has been reported by Bulychev, et al. [5] that LiMg(AlH4)3
can be synthesized from MgCl2, LiAlH4, and NaAlH4 in ether.
Recent work by Mamatha, et al. [14] have shown the formation
of LiMg(AlH4)3 by ball milling of LiAlH4 and MgCl2. This
compound has a theoretical hydrogen capacity of 7.8–8.2 wt%,
depending on the decomposition products formed. In this study,
LiMg(AlH4)3 was prepared using the solution method based
on the procedures provided in Ref. [5]. XRD analysis (Fig. 2)
shows that the synthesized compound is reasonably pure based
on the XRD data reported in ICDD PDF #00-034-0926 with a
small amount of LiCl present as impurity. Preliminary index-
ing of the XRD data yielded a monoclinic structure with the
preliminary lattice parameters of: a = 8.40, b = 8.83, c = 14.59
Fig. 4. XRD of decomposition products of LiMg(AlH4)3 at 100, 150, and
340 ◦C, respectively.
The as-synthesized compound was tested for isothermal
hydrogen desorption at 100 and 165 ◦C (Fig. 3). At 100 ◦C,
4.0 wt% H2 was released at a slow rate. The desorption rate
increased significantly at 165 ◦C, with 6.9–7.0 wt% H2 released.
An additional 1.8 wt% H2 was released at 340 ◦C, yielding a total
capacity of 8.7–8.8 wt% H2.
The XRD characterization of the dehydrogenated LiMg
(AlH4)3 indicated that the phase composition varied with des-
orption temperature (Fig. 4) and desorption time. After 5 h of
desorption at 100 ◦C, XRD showed the presence of Al and an
unknown compound, proposed to be an intermediate LiMgAlH6
phase based on the H2 wt% released (Eq. (7)). This hypothesis
was corroborated by the identification of a VASP-minimized
LiMgAlH6 trigonal P321 candidate structure [16] whose simu-
lated powder pattern very closely matched the unknown XRD
peaks [17]. Desorption at 150 ◦C for 24 h resulted in the for-
mation of MgH2 and additional Al (Eq. (8)). Note that a small
amount of LiH cannot be determined by XRD in the presence of
Al. Increasing the temperature to 340 ◦C caused the appearance
of Al3Mg2 via Eq. (9).
˚
(in A). Numerous VASP–minimized LiMg(AlH4)3 candidate
models were very similar in energy, but the most favorable had
monoclinic symmetries. The refinement of this compound struc-
ture has been conducted in collaboration with the Institute for
Energy in Norway [15]. Elemental analysis by ICP determined
the compound composition to be: Li, 5.5 wt%; Mg, 19.1 wt%,
and Al, 63.9 wt% (theoretical: Li, 5.6 wt%; Mg, 19.6 wt%, and
Al, 65.1 wt%).
100 ◦C : LiMg(AlH4)3 → LiMgAlH6 + 2Al + 3H2
150 ◦C : LiMgAlH6 → LiH + MgH2 + Al + 3/2H2
340 ◦C : 2MgH2 + 3Al → Al3Mg2 + 2H2
(7)
(8)
(9)
Thermal analyses and first principle thermodynamic pre-
dictions confirmed the likelihood of these reactions. When
LiMg(AlH4)3 was heated at 10 ◦C/min in a DSC, an exothermic
reaction began at 140 ◦C with a latent heat of −5.5 kJ/mol-H2
(Eq. (7)), immediately followed by an endothermic reaction
with a latent heat of 13.1 kJ/mol-H2 (Eq. (8)). Results from the
first principle thermodynamic modeling confirm their exother-
mic (Eq. (7)), and endothermic (Eq. (8)), nature. The heat of
Fig. 3. Hydrogen desorption of LiMg(AlH4)3 at 100, 165, and 340 ◦C, respec-
tively.

X. Tang et al. / Journal of Alloys and Compounds 446–447 (2007) 228–231
231
reaction for Eq. (8) agrees with the value reported by Mamatha
[14].
H.W. Brinks and B.C. Hauback of the Institute for Energy,
Kjeller, Norway.
The reverse hydrogen absorption of Eq. (8) was exper-
imentally tested using both undoped and 2% TiCl3 doped
LiMg(AlH4)3. The doping was achieved by 15 min of SPEX
milling. After discharging at 150 ◦C for more than 5 h, the sam-
ples were charged under 190 bar of hydrogen at 80 and 100 ◦C
for a total of 24 h. No absorption was observed. The material is
not rechargeable under these conditions.
References
[1] B. Bogdanovic, M. Schwickardi, J. Alloys Compd. 253–254 (1997) 1.
[2] C.M. Jensen, R. Zidan, N. Mariels, A. Hee, C. Hagen, Int. J. Hydrogen
Energy 24 (5) (1999) 461.
[3] G. Sandrock, K. Gross, G. Thomas, J. Alloys Compd. 339 (2002) 299.
[4] X. Tang, B.L. Laube, D.L. Anton, R.B. Bowman, S.-J. Hwang, 2007. Amer-
ican Physics Society March Meeting, March 5–9, 2007, Denver, Colorado.
[5] B.M. Bulychev, K.N. Semenenko, K.B. Bitcoev, Koord. Khim. 4 (1978)
374.
4. Conclusion
[6] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 14251.
[7] G. Kresse, J. Furthmu¨ller, Comput. Mater. Sci. 6 (1996) 15.
[8] G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758.
[9] J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson, M.R. Pederson, D.J.
Singh, C. Fiolhais, Phys. Rev. B 46 (1992) 6671.
[10] K. Parlinski, Z.Q. Li, Y. Kawazoe, Phys. Rev. Lett. 78 (1997) 4063.
[11] MedeA-Phonon Version 1.0 using Phonon Software 4.24, Copyright Prof.
K. Parlinski.
[12] P. Claudy, B. Bonnetot, J.P. Bastide, J.M. Letoffe, Mater. Res. Bull. 17
(1982) 1499.
[13] E. Ronnebro, D. Noreus, K. Kadir, A. Reiser, B. Bogdanovic, J. Alloys
Compd. 299 (2000) 101.
[14] M. Mamatha, B. Bogdanovic, M. Felderhoff, A. Pommerin, W. Schmidt,
F. Schueth, C. Weidenthaler, J. Alloys Compd. 407 (2006) 78.
[15] H. Grove, H.W. Brinks, R.H. Heyn, X. Tang, S.M. Opalka, B.L. Laube,
F.-J. Wu, B.C. Hauback, J. Alloys Comp., submitted for publication.
[16] S.M. Opalka, X. Tang, D.A. Mosher, B.L. Laube, S. Arsenault, F.-J. Wu, J.
Strickler, D.L. Anton, R. Zidan, H.W. Brinks, O.M. Løvvik, B.C. Hauback,
C. Qiu, G.B. Olson, 2006 US DOE Annual Merit Review Proceedings,
http://www.hydrogen.energy.gov/pdfs/review06/st 10 opalka.pdf.
[17] The identity of these peaks was also corroborated by a private communi-
cation with H. Grove.
In this study, complex hydrides of the Na–Li–Mg–Al–H
system were synthesized by a solid-state method. Under the
testing conditions, NaAlH4 and Na2LiAlH6 are the most stable
compounds at temperatures below 100 ◦C, and NaMgH3 and
Na2LiAlH6 are the most stable compounds at 120 ◦C.
Reasonably pure LiMg(AlH4)3 was synthesized by a solu-
tion based method. Below 150 ◦C, this compound thermally
dissociated in two steps (Eqs. (7) and (8)), via LiMgAlH6
as an intermediate, releasing ca. 6.9–7.0 wt% hydrogen. Both
undoped and 2% TiCl3 doped LiMg(AlH4)3 were not recharge-
able at 80–100 ◦C under 190 bar hydrogen pressure after being
discharged at 150 ◦C.
Acknowledgements
The authors gratefully acknowledge support by the U.S.
Department of Energy Contract DE-FC04-02AL67610 and
important relevant discussions with our collaborators: H. Grove,