Solid-state 27Al NMR investigation of thermal decomposition of LiAlH4

Article abstract of DOI:10.1016/j.jssc.2003.08.006

Solid-state nuclear magnetic resonance is used to study the thermal decomposition of lithium tetrahydroaluminate into metallic aluminum, hydrogen and trilithium hexahydroaluminate. Aluminum sites in LiAlH₄ and Li₃AlH₆ were characterized using static, magic angle spinning (MAS) and multiple-quantum MAS NMR. By applying the in situ NMR method, it has been demonstrated that melting is not a prerequisite for the decomposition of LiAlH₄. Based on the observed data, a decomposition path has been established that is consistent with the concentrations of observed Al species at various stages of the thermally induced reaction.

Full text of DOI:10.1016/j.jssc.2003.08.006

ARTICLE IN PRESS
Journal of Solid State Chemistry 177 (2004) 648–653
2
7
Solid-state Al NMR investigation of thermal decomposition
of LiAlH4
a
a,1
a,b
J.W. Wiench, V.P. Balema, V.K. Pecharsky, and M. Pruski
a,ꢀ
a
Ames Laboratory-USDOE, Iowa State University, Ames, IA 50011-3020, USA
Department of Materials Science and Engineering, Iowa State University, Ames IA 50011, USA
b
Received 24 June 2003; received in revised form 31 July 2003; accepted 1 August 2003
Abstract
Solid-state nuclear magnetic resonance is used to study the thermal decomposition of lithium tetrahydroaluminate into metallic
aluminum, hydrogen and trilithium hexahydroaluminate. Aluminum sites in LiAlH and Li AlH were characterized using static,
magic angle spinning (MAS) and multiple-quantum MAS NMR. By applying the in situ NMR method, it has been demonstrated
4
3
6
that melting is not a prerequisite for the decomposition of LiAlH . Based on the observed data, a decomposition path has been
4
established that is consistent with the concentrations of observed Al species at various stages of the thermally induced reaction.
Published by Elsevier Inc.
27
4
Keywords: Hydrogen storage; Lithium tetrahydroaluminate; Trilithium hexahydroaluminate; Thermal decomposition of LiAlH ; In situ NMR; Al
27
MAS NMR; Al MQMAS NMR
1
. Introduction
uptake is very important for achieving further progress
in applications of these materials for high capacity
hydrogen storage. For several decades, the thermal
decomposition of pure complex alkali metal tetrahy-
During the last decade, an increasing demand for high
capacity sources of hydrogen compatible with modern
fuel cell applications shifted the major emphasis in
hydrogen storage research towards complex hydrides of
light elements such as aluminum and boron [1,2].
Among the known aluminohydrides, two alkali metal
derivatives containing more than 7% of hydrogen by
weight, LiAlH and NaAlH , have attracted particular
attention [2]. Unlike other high-capacity hydrogen
storage solids, such as magnesium hydride, pure lithium
and sodium aluminohydrides release significant
amounts of hydrogen at relatively low temperatures
doaluminates, MAlH where M=Li, Na, K or Cs, has
4
been interpreted as a series of events, which begins with
the melting of the hydride [5–7] and proceeds through
the formation of an intermediate trialkali metal hex-
ahydroaluminate M AlH :
3
6
4
4
MAlH ðsÞ-MAlH ðlÞ;
ð1Þ
ð2Þ
ð3Þ
4
4
1
3
2
3
MAlH4ðlÞ- M3AlH6 þ Al þ H2;
1
3
1
3
1
2
M AlH -MH þ Al þ H :
ꢁ
3
6
2
not exceeding 200 C. The use of an appropriate metal
catalyst can lead to further reduction of their decom-
position temperature all the way to room temperature
However, in the presence of an appropriate transition
metal catalyst, both lithium and sodium tetrahydoalu-
minohydrides release hydrogen and transform into
metallic aluminum and corresponding hexahydroalumi-
nates at temperatures well below their melting points
[
2–4].
The fundamental understanding of the processes
taking place in solids during hydrogen release and
[
2,4]. Hence, it is reasonable to assume that the
decomposition of pure alkali metal aluminohydrides
may follow several alternative routes, including ones
that do not require the formation of a liquid phase. In
this case, the role of the metal catalyst is to promote the
low temperature decomposition path.
ꢀ
Corresponding author. 230 Spedding Hall, Ames Laboratory,
Ames, IA, 50011-3020, USA. Fax: +1-515-294-5233.
E-mail address: mpruski@iastate.edu (M. Pruski).
Current address: Sigma-Aldrich Co. 5485, County Road V,
Sheboygan Falls, WI 53085, USA.
1
0022-4596/$ - see front matter Published by Elsevier Inc.
doi:10.1016/j.jssc.2003.08.006

ARTICLE IN PRESS
J.W. Wiench et al. / Journal of Solid State Chemistry 177 (2004) 648–653
649
Recently, nuclear magnetic resonance (NMR) has
been successfully employed for monitoring chemical
transformations in solid complex aluminohydrides [3,8],
where chemical reactions resulted in complete or partial
amorphization of the samples and other analytical
techniques, such as X-ray powder diffraction, could
not provide sufficient structural information. In pow-
rotor (20 kHz). Shearing and scaling of the MQMAS
spectra were performed as described previously [16,17].
In situ VT NMR experiments on LiAlH4 were
performed using static (non-spinning) 5-mm glass tubes,
which were loaded under helium in a glove box and
sealed using a Teflon tape. The tape allowed for
hydrogen gas to escape during heating. The measure-
ments were carried out in a nitrogen flow to further
minimize the possibility of oxygen contamination. The
VT unit was calibrated using Pb(NO ) as a ‘chemical
2
7
dered solids containing quadrupolar nuclei such as Al
(
I ¼ 5=2), the NMR spectra are usually broadened by
chemical shift anisotropy (CSA), dipolar, and quad-
rupolar interactions [9,10]. Broadening due to CSAand
dipolar interactions can be eliminated by using magic
angle spinning (MAS) combined with various decou-
pling strategies [9–11]. The quadrupolar effect can be
overcome by employing a two-dimensional multiple
quantum method combined with MAS (MQMAS)
3
2
shift thermometer’, which allowed to establish the
temperature of a sample under static and MAS
ꢁ
conditions with the accuracy of 71 C [18]. The
observed Pb line width revealed that the temperature
2
07
ꢁ
gradient across the sample did not exceed 1.5 C. The
2
Al spectra reported in this work use the d scale, with
positive values downfield, and are referenced to 1 M
7
[
12,13]. In the present study, we combine these high-
3
+
resolution techniques with variable temperature (VT) in
situ NMR of static samples to detail the mechanism
responsible for the thermal decomposition of pure
LiAlH4.
Al(H O) aqueous solution.
2 6
3. Results and discussion
2
7
The MAS and static Al NMR spectra of pure
LiAlH are shown in Figs. 1a and b, respectively. In
addition to the dominant resonance at around 100 ppm
representing four-coordinated Al in LiAlH , the spec-
trum in Fig. 1a revealed minor amount of metallic
aluminum (Al ), whose chemical shift of 1640 ppm
agrees with the literature data [19]. The corresponding
NMR spectra of as-synthesized Li AlH are shown in
2
. Experimental
4
2
.1. Sample preparation
4
LiAlH4 (97% purity) was purchased from Fluka.
Li AlH was prepared from LiAlH and LiH (Aldrich,
9
M
3
6
4
5% purity) as described in Ref. [14]. All operations on
3
6
metal hydrides were carried out in a glove box under
helium.
Figs. 1c and d. As expected, they feature a single
resonance at approximately ꢂ30 ppm representing six-
coordinated Al.
2
.2. NMR measurements
LiAlH
4
2
7
The Al NMR spectra were collected at 9.4 T on a
Varian/Chemagnetics Infinity 400 spectrometer, operat-
2
7
1
Al and H,
ing at 104.24 and 400.02 MHz for
respectively. The spectrometer was equipped with a
VT unit and doubly tuned MAS probes that use 5-mm
8 kHz MAS rate) and 3.2-mm (20 kHz MAS rate)
zirconia rotors. Since the spin–spin relaxation times
T1), as determined by saturation method, were 1.7, 1.3,
and 1.0 s for LiAlH , Li AlH and metallic aluminum,
Al
M
a) 8 kHz MAS
b) static
(
(
4
3
6
3
Li AlH6
respectively, a pulse delay of 10 s was used in all
2
experiments in order to detect fully equilibrated Al
7
c) 8 kHz MAS
d) static
1
magnetization. H decoupling was accomplished by
using the two-pulse phase modulation (TPPM) scheme
[11,13], which performed better than the cw method
under MAS conditions.
The MQMAS experiment utilized the Z-filter method
1
800 1600 1400 1200 1000
800
ppm
600
400
200
0
-200
[15] with the RF fields of B170 and B17 kHz during the
excitation/conversion and selective Z-filter pulses, re-
27
Fig. 1. Al NMR spectra of LiAlH
4
3 6
(a,b) and Li AlH (c,d) recorded
spectively. The t step of 50 ms for data acquisition in F1
domain was synchronized with the spinning rate of the
1
using 8 kHz MAS (a,c) and static (b,d). All spectra were acquired at
1
room temperature, using H decoupling.

ARTICLE IN PRESS
J.W. Wiench et al. / Journal of Solid State Chemistry 177 (2004) 648–653
650
To facilitate the ensuing discussion of the mechanism
of thermal decomposition, a separate set of experiments
allowed for observation of well-defined powder pattern
representing the central (m ¼ þ1=22m ¼ ꢂ1=2) transi-
2
was performed to examine the width of observed Al
7
27
tion of Al nuclei which is broadened by the second-
lines in LiAlH under various conditions. The full-width
4
order quadrapolar interaction. The highest resolution in
solid LiAlH4 has been achieved by the method of
MQMAS (Fig. 2e), which will be discussed in more
detail below. Finally, Fig. 2f demonstrates the single-
2
at half-magnitude (FWHM) of the Al signal of a static
7
sample acquired under single pulse excitation is
1
approximately 9.5 kHz (see Fig. 2a). The use of
TPPM decoupling reduced the FWHM by a factor of
H
2
7
pulse Al spectrum of molten LiAlH4 acquired at
ꢁ
1
almost two (Fig. 2b). The combination of MAS and H
decoupling (Fig. 2d) yielded further improvement and
180 C. Due to increased mobility, the observed signal
narrowed to 400 Hz, i.e., by a factor of almost 20 when
compared with the spectrum of Fig. 2a.
2
The Al MAS and MQMAS spectra of LiAlH
acquired with the assistance of H decoupling via TPPM
13,15] are shown in Fig. 3. The sheared MQMAS
7
4
1
[
spectrum is presented as a two-dimensional correlation
plot between single and triple quantum coherences. The
single quantum dimension of the MQMAS spectrum
a) FWHM = 9.5 kHz
static
1
H coupled
(horizontal, d2) has the same resolution as the standard
MAS spectrum. Note that MAS and MQMAS yielded
somewhat different line shapes of the central transition
powder pattern. This is due to the dependence of the
MQ excitation and conversion on the crystallite
orientation in a powdered sample, which skews the
spectral intensities in the single quantum dimension
b) FWHM = 5.0 kHz
static
1
H coupled
[
17]. The multiple quantum dimension (vertical, dISO)
c) FWHM = 3.2 kHz
8
kHz MAS
1
H coupled
(a)
d) FWHM = 1.4 kHz
8
kHz MAS
1
H coupled
ppm
SSB
5
0
e) FWHM = 1.3 kHz
MQMAS projection
1
H coupled
1
00
QIS
f) FWHM = 0.4 kHz
150
liquid (melted at 180°C)
CS
1
H coupled
SSB
(b)
150
100
50
ppm
3
0
20
10
0
kHz
-10
-20
-30
27
Fig. 3. (a) Al spectrum of LiAlH
27
4
obtained under MAS at 8 kHz; (b)
1
4
Al MQMAS spectrum of LiAlH obtained under MAS at 20 kHz. H
2
7
Fig. 2. Al NMR lineshapes observed in LiAlH . Spectra (a)–(e) were
acquired at room temperature, other experimental conditions are
shown in the figure.
4
TPPM decoupling was used in both experiments. Chemical shift (CS)
and quadrupole induced shift (QIS) axes are represented by dashed
lines [17]. ‘SSB’ denotes the spinning sidebands.

ARTICLE IN PRESS
J.W. Wiench et al. / Journal of Solid State Chemistry 177 (2004) 648–653
651
Table 1
27
Al NMR parameters obtained for pure hydrides by QUASAR simulation of the MAS spectra
diso (ppm)
CQ (MHz)
ZQ
DdCSA (ppm)
ZCSA
LiAlH
Li AlH
4
103.870.8
ꢂ33.770.8
3.970.1
1.470.1
0.2470.05
0.0270.05
30710
25710
0.070.1
0.070.1
3
6
The following definitions were used: isotropic shift diso is a sum of chemical shift (dCS) and quadrupole induced shift (dQIS), quadrupole coupling
2
constant CQ ¼ e qQ=h; anisotropy of quadrupolar tensor Z ¼ ðVyy ꢂ VxxÞ=Vxx; chemical shift anisotropy DdCSA ¼ 1:5ðd33 ꢂ dCSÞ; anisotropy of
Q
CSA Z
¼ ðd22 ꢂ d11Þ=ðd33 ꢂ dCSÞ; see Refs. [9,10,17] for details.
CSA
represents the isotropic resonance that is free from the
second-order effects [12,17]. The numerical simulation
of the MAS spectrum of Fig. 3a by QUASAR [20] and
the standard analysis of MQMAS spectrum of Fig. 3b
a) 22°C
LiAlH
4
Al
M
[
tetrahedral site in LiAlH , which are given in Table 1.
17] yielded the same set of line shape parameters for the
4
b) 100°C
c) 150°C
The measured quadrupolar coupling constant CQ of
3
.9 MHz agrees with the previous NMR study [21]. The
Al MAS and sheared MQMAS NMR spectra of
Li AlH (not shown) exhibit a single symmetric
2
7
Al
O
3 6
Li AlH
3
6
resonance whose line parameters are consistent with
the six-coordinated aluminum (see Table 1). The
MQMAS spectra of LiAlH and Li AlH disclosed the
12 min.
4
3
6
presence of weak spinning sidebands due to the CSA
interaction, which is magnified by a factor of p (in our
approach p ¼ 3) along the MQ dimension [17,22]. Our
simulation yielded the CSAvalues of 30 and 25 ppm for
LiAlH and Li AlH , respectively.
3
2min.
4
3
6
The thermal decomposition of LiAlH was studied in
4
ꢁ
ꢁ
situ between 20 C and 150 C under static conditions, as
described in the Experimental section (Fig. 4). After
ꢁ
30 min of exposure to 100 C, only trace amounts of
metallic aluminum (Al ) and Li AlH were detected. At
150 C, however, the decomposition process accelerated
significantly. Resonances representing Al , Li AlH ,
and a new peak at 60 ppm, which was assigned as six-
III
coordinated oxidized Al derivative (abbreviated as
4
8 min.
M
3
6
ꢁ
M
3
6
Al ), intensified at the expense of LiAlH , which
4
61 min.
30 min.
O
disappeared almost completely within 2 h. The amount
of AlO increased further as a function of time, most
likely due to a reaction of gaseous oxygen impurity with
highly active metallic aluminum.
The estimated relative concentrations of various Al
ꢁ
1
species during the thermal decomposition at 150 C are
shown in Fig. 5. We note that the quantification of
different aluminum species in the spectra of Fig. 5 is not
straightforward [10,19]. First, the resonance assigned to
LiAlH is dominated by the central transition, and thus
4
2
7
2000
1500
1000
500
0
represents only 9/35 of the Al resonance signal in this
hydride [10]. The satellite transitions for this site appear
as a wide manifold of spinning sidebands in the MAS
spectrum of Fig. 1a. Similar effect is expected for
Li AlH , although non-negligible contribution of the
ppm
2
7
1
measured under H decoupling as
Fig. 4. Al static spectra of LiAlH
a function of heating conditions.
4
3
6
satellite transition intensity may be expected due to
more symmetric coordination of Al in this compound.
In contrast, the metallic aluminum has a cubic crystal
structure, which leads to degeneracy of all the transi-
tions [19]. Consequently, the entire intensity of the metal
peak is accounted for in the resonance at 1640 ppm.

ARTICLE IN PRESS
J.W. Wiench et al. / Journal of Solid State Chemistry 177 (2004) 648–653
652
were then used to calibrate the intensities of the spectra
in Fig. 4.
Our NMR studies offer new insights into the
1
00
LiAlH4
Li AlH
Al
Al
3
6
×
M
90
80
70
60
50
40
30
20
10
0
O
mechanism of thermal decomposition, which deserve
ꢁ
further discussion. First, while heating LiAlH to 150 C
4
caused disintegration of the hydride and the formation
of Li AlH and Al/Al , no line narrowing correspond-
ing to the melting of LiAlH (see Fig. 2e) could be
3
6
O
4
2
7
detected. Instead, the observed Al line widths are
consistent with those measured in solid LiAlH , Li AlH
6
4
3
and AlM (see Figs. 1, 2 and 6), which clearly
demonstrates that undoped LiAlH can decompose in
4
the solid state, and its melting is not a prerequisite for
this process. Apossibility that thermal decomposition of
LiAlD may not require melting of the sample was also
4
mentioned by Brinks et al. [23], who studied thermal
transformations of LiAlD4 using neutron and X-ray
powder diffraction techniques. Second, the analysis of
2
7
integral intensities of the signals in the Al NMR
spectrum of the decomposed sample (Fig. 5) reveals an
apparent deviation of the sample composition from that
predicted by Eq. (2). Indeed, the concentration of
ꢁ
Li AlH in the hydride sample after exposure to 150 C
6
0
20
40
60
80
100
120
140
3
time [min]
for 130 min was found to be B18 mol% instead of the
expected value of 33.3 mol% if the reaction proceeded
according to Eq. (2). At the same time, it is well known
that Li AlH decomposes thermally at the temperatures
Fig. 5. Relative concentrations of the observed Al species versus
ꢁ
exposure time at 150 C.
3
6
ꢁ
far above 150 C [6,7,14]. Note that very little, if any,
change in the concentration of Li AlH (and other
3
6
Al
M
aluminum species) is observed in Fig. 5 after B80 min at
ꢁ
1
50 C, i.e., when the decomposition of LiAlH is
4
3 6
Li AlH
practically completed. Therefore, the deviation from
the expected stoichiometry as a result of the decom-
position of trilithium hexahydroaluminate (Eq. (3)) is
rather unlikely. Remarkably, gas-volumetric data on
LiAlH
4
a) 20 kHz MAS
b) static
thermal decomposition of LiAlH published by Dymova
4
et al. [7] clearly support the results of our NMR studies.
Based on the amount of hydrogen released by LiAlH
1
800 1600 1400 1200 1000
800
ppm
600
400
200
0
-200
4
ꢁ
during its exposure to 150–200 C [7], the concentration
of Li AlH in the sample cannot exceed B22 mol%,
27
Fig. 6. Al NMR spectra of 1:1:1 molar mixture of LiAlH
4
, Li
obtained under H decoupling using 20 kHz MAS (a) and
static (b) conditions.
3
AlH
6
3
6
1
and Al
M
which disagrees with the stoichiometry represented by
the Eq. (2) but it is in a fairly good agreement with our
data.
Second, the RF excitation and the probe tuning
were optimized for the central transition of LiAlH4,
which had the opposite effect of reducing the intensity
of the peak corresponding to metallic aluminum, as it
was measured with a large resonance offset. In order
to properly account for these effects, we took a static
4. Conclusion
Our studies on the thermal decomposition of lithium
aluminohydride using in situ high-temperature solid-
2
state Al NMR spectroscopy revealed that undoped
7
2
7
ꢁ
Al NMR spectrum of the 1:1:1 molar mixture of
LiAlH , Li AlH and Al (see Fig. 6b). (We note that
commercial LiAlH can decompose at 150 C without
4
melting, and the chemical composition of the sample
kept at this temperature for 130 min does not agree with
that expected for a two-step process suggested pre-
viously. Therefore, it seems feasible that thermal
decomposition of lithium tetrahydroaluminate follows
4
3
6
the MAS spectrum of the same mixture shown in
Fig. 6a yielded the same relative intensities, provided
that the spinning sidebands associated with Al were
included in the calculation.) The obtained intensities

ARTICLE IN PRESS
J.W. Wiench et al. / Journal of Solid State Chemistry 177 (2004) 648–653
653
several different reaction paths including that repre-
sented by
[8] B. Bogdanovic, M. Felderhoff, M. Germann, M. Hartel, A.
Pommerin, F. Schuth, C. Weidenthaler, B. Zibrowius, J. Alloys
Compd. 350 (2003) 246.
3
2
LiAlH -LiH þ Al þ H :
ð4Þ
Note that the transformation of LiAlH directly into
4
2
¨
[9] A.-R. Grimmer, B. Blumich, NMR Basic Principles Progr. 30
(
10] D. Freude, J. Haase, NMR Basic Principles Progr. 29
1994) 1.
4
[
LiH, Al and H is also feasible from the thermodynamic
2
point of view [7].
(1993) 1.
[11] A.E. Bennett, C.M. Rienstra, M. Auger, K.V. Lakshmi, R.G.
Griffin, J. Chem. Phys. 103 (1995) 6951.
[
[
[
[
[
12] L. Frydman, J.S. Harwood, J. Am. Chem. Soc. 117 (1995)
5367.
Acknowledgments
13] V. Lacassagne, P. Florian, V. Montouillout, C. Gervais, F.
Babonneau, D. Massiot, Magn. Reson. Chem. 36 (1998) 956.
14] V.P. Balema, V.K. Pecharsky, K.W. Dennis, J. Alloys Compd.
Different aspects of this research were supported at
Ames Laboratory by the US Department of Energy,
Office of Basic Energy Sciences, Division of Chemical
Sciences, and Division of Materials Sciences under
Contract W-7405-Eng-82.
3
13 (2000) 69.
15] J.-P. Amoureux, C. Fernandez, S. Steuernageel, J. Magn. Reson.
A123 (1996) 116.
16] D. Massiot, B. Touzo, D. Trumeau, J.P. Coutures, J. Virlet, P.
Florian, P.J. Grandinetti, Solid State Nucl. Magn. Reson. 6
(
1996) 73.
[
17] J.-P. Amoureux, M. Pruski, in: D.M. Grant, R.K. Harris (Eds.),
Encyclopedia of NMR, Vol. 9, John Wiley & Sons Ltd.,
Chicester, 2002, p. 226.
References
[
1] M. Gardiner, Presented at the 224th ACS National Meeting,
Boston, MA, United States, August 18–22, 2002.
[18] A. Bielecki, D.P. Burum, J. Magn. Reson. A 116 (1995) 215.
[19] L. Kellberg, H. Bildsoe, H.J. Jakobsen, J. Chem. Soc. Chem.
Commun. 19 (1990) 1294.
[
[
2] B. Bogdanovic, G. Sandrock, MRS Bull. 27 (2002) 712.
3] V.P. Balema, J.W. Wiench, K.W. Dennis, M. Pruski, V.K.
Pecharsky, J. Alloys Compd. 329 (2001) 108.
[20] J.-P. Amoureux, C. Fernandez, Y. Dumazy, Presented at the 37th
Rocky Mountain Conference, Denver, CO, USA, July–August
1995.
[
4] V.P. Balema, K.W. Dennis, V.K. Pecharsky, Chem. Commun.
1
7 (2000) 1665.
[21] V.P. Tarasov, S.I. Bakum, S.F. Kuznetsova, Russ. J. Inorg.
Chem. 42 (1997) 694.
[
[
[
5] E.C. Ashby, Adv. Inorg. Chem. Radiochem. 8 (1966) 283.
6] J.A. Dilts, E.C. Ashby, Inorg. Chem. 11 (1972) 1230.
7] T.N. Dymova, V.N. Konoplev, D.P. Aleksandrov, A.S. Sizareva,
T.A. Silina, Russ. J. Coord. Chem. 21 (1995) 165.
[22] L. Marinelli, L. Frydman, Chem. Phys. Lett. 275 (1997) 188.
[23] H.W. Brinks, B.C. Hauback, P. Norby, H. Fjellvag, J. Alloys
Compd. 351 (2003) 222.