Factors affecting hydrogen release from lithium alanate (LiAlH4)

Article abstract of DOI:10.1016/j.jallcom.2005.02.017

The release of hydrogen from two batches of milled LiAlH₄-2 mol % TiCl₃ was studied by temperature-programmed desorption and in-situ X-ray diffraction. Milling at room temperature can cause decomposition into Li₃AlH_6<

Full text of DOI:10.1016/j.jallcom.2005.02.017

Journal of Alloys and Compounds 398 (2005) 245–248
Factors affecting hydrogen release from lithium alanate (LiAlH4)
D.S. Easton, J.H. Schneibel∗, S.A. Speakman
Metals and Ceramics Division, Oak Ridge National Laboratory, PO Box 2008, 1 Bethel Valley Road, Oak Ridge, TN 37831-6115, USA
Received 20 January 2005; received in revised form 4 February 2005; accepted 9 February 2005
Available online 23 March 2005
Abstract
The release of hydrogen from two batches of milled LiAlH4–2 mol % TiCl3 was studied by temperature-programmed desorption and in-situ
X-ray diffraction. Milling at room temperature can cause decomposition into Li3AlH6. Our results suggest that the decomposition occurs
because the TiCl3 addition reduces the temperature of the LiAlH4 to Li3AlH6 decomposition to values near room temperature.
© 2005 Elsevier B.V. All rights reserved.
Keywords: Hydrogen storage materials; High-energy ball milling; X-ray diffraction
1. Introduction
the possibility that pure LiAlH4 is thermodynamically sta-
ble at room temperature, and that decomposition of LiAlH4
samples at room temperature is due to destabilizing contam-
inants or dopants [3]. Dopants can indeed have a distinct
effect on the decomposition of LiAlH4: Blanchard et al. [2]
ball-milled mixtures of LiAlD4 and a few mol % of VCl3 or
TiCl3·1/3AlCl3. The additives reduced the thermal decom-
position temperatures for Eq. (1) by 50–60 ◦C. Even more
dramatic effects were observed by Balema et al. [4]. Balema
et al. mixed LiAlH4 with 3 mol % of TiCl4, subjected it to
high-energy ball-milling for 5 min at room temperature, and
observed complete decomposition into Li3AlH6. Balema et
al. did not determine to what extent the decomposition tem-
perature of LiAlH4 was reduced due to the milling with TiCl4.
If that decomposition temperature was well above room tem-
perature, the high-energy milling was directly responsible
for driving Eq. (1) to the right, i.e., the high-energy impacts
occurring during the milling caused the release of hydrogen
through a mechanochemical decomposition mechanism. If
the TiCl4 additions reduced the decomposition temperature
to room temperature or below, then purely thermal decompo-
sition would have occurred during the milling. The purpose
of this paper is to show that mechanical milling of LiAlH4
with TiCl3 (which acts very similar to TiCl4 [5]) does reduce
the decomposition temperature for reaction (1) to values near
ambient, and that batch-to-batch variations of the LiAlH4 are
also likely to affect the decomposition temperature.
Lithium alanate, LiAlH4, is of interest as a potential hy-
drogen storage material because of its high hydrogen content.
Hydrogen release occurs in three stages [1,2]:
3LiAlH4 = Li3AlH6 + 2Al + 3H2 ↑
[150–175 ◦C, 5.3 wt.%]
(1)
(2)
(3)
Li3AlH6 + 2Al = 3LiH + 3Al + 3/2 H2 ↑
[180–220 ◦C, 2.6 wt.%]
3LiH + 3Al = 3LiAl + 3/2 H2 ↑ [400 ◦C, 2.6 wt.%]
At the decomposition temperature indicated in Eq. (1),
LiAlH4 decomposes in time intervals on the order of hours.
LiAlH4 is thought to be metastable at room temperature –
it is said to decompose by 50% over a period of 10 years
[1]. Thermodynamic property calculations aided by lattice
phonon vibrational simulations, on the other hand, leave open
∗
Corresponding author. Tel.: +1 865 576 4644; fax: +1 865 574 7659.
E-mail address: schneibeljh@ornl.gov (J.H. Schneibel).
0925-8388/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2005.02.017

246
D.S. Easton et al. / Journal of Alloys and Compounds 398 (2005) 245–248
2. Experimental procedure
starting crystal structure for Li3AlH6 was taken from Brinks
and Hauback [6]. Atomic parameters such as position, occu-
pancy, and thermal parameter were not refined, but the lattice
parameters, overall thermal parameters, specimen displace-
ment, and peak profiles were refined. The diffraction pattern
of LiH highly overlaps that of Al. Therefore the amount of
LiH was not refined, and any content of LiH was convoluted
into the refinement of Al. The hydrogen release during the
ramping up of the temperature was calculated from the mea-
sured decomposition of the LiAlH4 and the Li3AlH6.
Quantities of 25 g (“batch 1”) and 100 g (“batch 2”) of
LiAlH4 with nominal purities of 95% were purchased from
the same supplier at different times. However, batch 1 was a
white powder, whereas batch 2 consisted of pressed tablets,
which were either white or gray. All handling was car-
ried out in an argon glove box in which the argon was
continuously purified by flowing through oxygen and wa-
ter vapor getters. LiAlH4 (typically 1–2 g) was mixed with
2 mol % TiCl3 and ball milled in a Spex mill. A hardened
tool steel vial with two 1/2ꢀꢀ diameter and four 1/4ꢀꢀ diameter
steel balls was used. After the first 15 min, the milling was
stopped and the vial was cooled for a few minutes. Milling
was then continued for another 15 min, resulting in a total
milling time of 30 min. Temperature-programmed desorp-
tion (TPD) experiments with furnace ramp rates ranging from
0.015 to 1 ◦C/min were carried out in a pressure-composition-
isotherm (PCI) apparatus (Advanced Materials Corporation,
Pittsburgh). This equipment was a Sieverts type apparatus,
and its operation was based on pressure measurements in
known volumes, and the gas equation for hydrogen. Approx-
imately0.7 gofpowderwasplacedina2 ccsamplecupwhich
was inserted into a stainless steel container connected to the
PCI equipment. After an initial evacuation step the hydrogen
release from the powder was determined by monitoring the
pressure rise due to expansion into a known volume. During
the hydrogen release, the pressure increased up to a value
of 0.7 atm. The TPD runs were typically started within 1 h
from the end of the milling. As a check, TPD runs were also
made without powder. As anticipated, these runs did not show
release within an error of 0.1 wt.%. In some cases, the spec-
imen weights before and after desorption were determined in
order to verify the accuracy of the hydrogen desorption data.
In-situ XRD analysis was carried out using a PAN-
alytical X-ray diffractometer equipped with a PANalyt-
ical X’celerator high-speed detector and an Anton Paar
XRK900 reaction chamber. In a glove box, a freshly milled
LiAlH4–2 mol % TiCl3 specimen was placed in a 0.5 mm
deep specimen holder and sealed by gluing a 6 ␮m mylar
film onto the circular edge of the holder. The sample was
transported from the glove box to the diffractometer, the my-
lar film was removed (which briefly exposed the specimen to
air), and the specimen holder was immediately inserted into
the reaction chamber. N2–4%H2 gas was flowing through the
reaction chamber at a rate of ∼5 cc/min. No evidence of oxide
formation or other impurities was observed in the diffraction
patterns before and after removing the mylar film. The reac-
tion chamber furnace was heated at a rate of 0.1 ◦C/min and
successive XRD scans, each lasting approximately 10 min,
were carried out. Approximately half an hour elapsed be-
tween the end of the milling and the start of the data collec-
tion. The data were analyzed using PANalytical HighScore
Plus software to perform quantitative Rietveld analysis. Start-
ing crystal structures for LiAlH4, Al, and LiCl were taken
from the inorganic crystal structure database (ICSD). The
3. Experimental results
3.1. LiAlH4, batch 1
Fig. 1 shows TPD results for LiAlH4, batch 1, in the
as-received condition, and after milling without and with
2 mol % TiCl3. The as-received and received + milled pow-
ders showed the two-step hydrogen release corresponding to
Eqs. (1) and (2). Consistent with Balema et al. [4], milling
with TiCl3 reduced the amount of hydrogen released during
the TPD because reaction (1) took place during the milling;
only the hydrogen release corresponding to reaction (2) was
observed. While the amount of hydrogen released during the
milling was not measured directly, it can be estimated from
the fact that the combined release due to the milling and the
TPD is on the order of 7.5 wt.%. This is why the endpoint of
the curve annotated “milled with TiCl3” in Fig. 1 was fixed
at −7.5 wt.% H2. The decomposition temperatures were de-
fined by determining the temperatures at which reactions (1)
or (2) were 50% complete, and are listed in Table 1.
3.2. LiAlH4, batch 2
In contrast to batch 1, TPD of LiAlH4, batch 2, milled
with TiCl3 showed decomposition according to Eqs. (1) and
(2) (Fig. 2). This means that the LiAlH4 did not decompose
into Li3AlH6 during the milling. Decreasing ramp rates re-
duced the decomposition temperatures as seen in Fig. 2 and
Fig. 1. Thermal programmed desorption (ramp rate 1.3 ◦C/min) of LiAlH4,
batch 1, in the as-received, milled, and milled with 2 mol % TiCl3 conditions.

D.S. Easton et al. / Journal of Alloys and Compounds 398 (2005) 245–248
247
Table 1
Decomposition temperatures T1 and T2, corresponding to reactions (1) and (2), respectively, as a function of processing history and ramp rate
Batch of LiAlH4
Measurement technique
Composition
Condition
Ramp rate (◦C/min)
T1 (◦C)
T2 (◦C)
1
1
1
2
2
2
2
Sieverts
Sieverts
Sieverts
Sieverts
Sieverts
Sieverts
In-situ XRD
LiAlH4
LiAlH4
As-received
Milled
Milled
Milled
Milled
1.3
1.3
1.3
1.3
0.11
0.015
0.1
158
140
Unknown
67
37
29
187
180
94
97
77
LiAlH4–2 mol % TiCl3
LiAlH4–2 mol % TiCl3
LiAlH4–2 mol % TiCl3
LiAlH4–2 mol % TiCl3
LiAlH4–2 mol % TiCl3
Milled
Milled
54
152
52
Fig. 4. Quantitative analysis of decomposition of LiAlH4–2 mol % TiCl3
(batch 2) during TPD at 0.1 ◦C/min.
Fig. 2. Thermal programmed desorption of LiAlH4, batch 2, at different
ramp rates.
Table 1, and at the lowest ramp rate the LiAlH4 decomposi-
tion temperature was on the order of room temperature. Con-
sistent with this, storing of milled LiAlH4–2 mol % TiCl3 at
room temperature in the glove box for a day or longer re-
duced the amount of hydrogen released to that expected for
the decomposition of Li3AlH6, indicating that the LiAlH4
had decomposed.
The in-situ XRD of LiAlH4, batch 2, demonstrated
conclusively that the decomposition occurred according to
Eqs. (1) and (2). Fig. 3 depicts the changes in the diffraction
peaks associated with the various compounds forming or
disappearing during the TPD. The change of the phase
fractions as a function of temperature is shown in Fig. 4.
The decomposition started at 20 ◦C. In Fig. 5, the hydrogen
release calculated from the data in Fig. 4 is compared to the
PCI measurement. While the decomposition temperatures
for reaction (1) are similar for both measurement techniques,
the decomposition temperatures for reaction (2) are quite
different, see also Table 1. This difference is not thought to be
due to inaccuracies in the temperature measurements; it prob-
ably depends upon experimental factors such as differences
in the gas pressure and composition during release.
Fig. 5. Comparison between the hydrogen release determined from the PCI
apparatus, at a ramp rate of 0.11 ◦C/min, and from in-situ XRD analysis, at
a ramp rate of 0.1 ◦C/min.
Fig. 3. In-situ XRD data for the decomposition of milled LiAlH4–2 mol %
TiCl3 (batch 2) during TPD at 0.1 ◦C/min.

248
D.S. Easton et al. / Journal of Alloys and Compounds 398 (2005) 245–248
4. Discussion
differential spectroscopy (TDS) measurements employed to
quantify the amount of released hydrogen. Unless the TDS
experiments were performed immediately after the milling,
it is possible that significant thermal decomposition (at room
temperature) of the LiAlH4 milled with TiCl3·1/3AlCl3
took place prior to the TDS experiments. In that case one
might incorrectly conclude that the TiCl3·1/3AlCl3 caused
significant mechanochemical decomposition, while in fact
thermal decomposition was the main factor.
According to Table 1, the decomposition temperature for
Li3AlH6 was similar for both batches. The decomposition
temperature for Li3AlH6, therefore, did not appear to be sen-
sitive to differences in the composition or contamination of
the two batches. It should be noted that the decomposition
temperatures determined here are not thermodynamic equi-
librium values – they depend on the ramp rate. Also, if the
decomposition temperature would be defined as the onset of
decomposition, lower values would be obtained. However,
the values in Table 1 are suitable for comparative purposes.
Batch 2 of the LiAlH4 did not decompose during milling
with TiCl3; however, TPD and XRD both showed that de-
composition began near room temperature. The milling with
TiCl3 did reduce the decomposition temperature for reaction
(1) to values near room temperature. Therefore, the decom-
position during ball milling observed by Balema et al. [4]
is likely to have been partially or exclusively a thermal one.
However, the behavior observed for the two batches in this
study prevent a complete demonstration. In principle it is
possible that, for batch 1, the thermal decomposition tem-
perature for reaction (1) was well above room temperature,
and that mechanochemical desorption was the reason why
LiAlH4 converted into Li3AlH6 during the milling. To verify
this possibility would require milling at temperatures well
below room temperature (in the hope of preventing any ther-
mal desorption), followed by TPD measurements. However,
it is also possible that the decomposition temperature for re-
action (1) in batch 1 is lower than that for batch 2. If that was
the case, then conventional thermal decomposition occurring
during the milling of batch 1 was the reason for the formation
of Li3AlH6.
5. Summary and conclusions
Mechanical milling of LiAlH4 with TiCl3 reduces the
temperature for the thermal decomposition of LiAlH4 and
Li3AlH6. While the decomposition temperature for Li3AlH6
does not appear to be sensitive to contamination levels, that
for the decomposition of LiAlH4 may very well be. It is
likely that the decomposition of LiAlH4 into Li3AlH6 during
millingwithTiCl3 occursnotbymechanochemicaldecompo-
sition, but rather because of a reduction of the decomposition
temperature to room temperature or below.
Acknowledgements
This research was sponsored by the Laboratory Directed
Research and Development Program of Oak Ridge National
Laboratory (ORNL), managed by UT-Battelle, LLC for the
U.S. Department of Energy under Contract No. DE-AC05-
00OR22725.
References
Our findings demonstrate that care must be taken when
comparing the influence of a doping agent on the promotion
of mechanochemical decomposition versus purely thermal
decomposition. Blanchard et al. [2] noted that while VCl3
additions were more effective than TiCl3·1/3AlCl3 additions
in reducing the decomposition temperature for LiAlH4
according to reaction (1), TiCl3·1/3AlCl3 additions were
more effective for mechanochemical decomposition. How-
ever, Blanchard et al. did not specify the time that elapsed
between the milling of their doped LiAlH4 and the thermal
[1] H.W. Brinks, B.C. Hauback, P. Norby, H. Fjellvag, J. Alloys Comp.
351 (2003) 222.
[2] D. Blanchard, H.W. Brinks, B.C. Hauback, P. Norby, Mater. Sci. Eng.
B108 (2004) 54.
[3] O.M. Løvvik, S.M. Opalka, H.W. Brinks, B.C. Hauback, Phys. Rev.
B 69 (2004) 134117.
[4] V.P. Balema, J.W. Wiench, K.W. Dennis, M. Pruski, V.K. Pecharsky,
J. Alloys Comp. 329 (2001) 108.
[5] D.D.L. Anton, J. Alloys Comp. 356–357 (2003) 400.
[6] H.W. Brinks, B.C. Hauback, J. Alloys Comp. 354 (2003) 143–147.