Sodium alanates for reversible hydrogen storage

Article abstract of DOI:10.1016/S0925-8388(99)00666-0

In this paper we show that sodium alanates may be used for reversible hydrogen storage, with the advantage of having high storage capacity combined with low cost. Both NaAlH₄ and Na₃AlH₆ have been investigated for this application, and two complementary techniques have been used: improvement of the reaction kinetics by mechanical grinding, and chemical modification of the alloys. By these methods remarkable desorption/absorption kinetics are obtained. Sodium alanates so modified are capable of reversible hydrogen storage at the relatively low temperatures of around 80-140 °C, with a capacity of between 2.5 and 3.0 wt.%. The hydrides have an even higher reversible capacity of about 4.5-5 wt.% when operated at temperatures around 150-180 °C.

Full text of DOI:10.1016/S0925-8388(99)00666-0

Journal of Alloys and Compounds 298 (2000) 125–134
L
www.elsevier.com/locate/jallcom
Sodium alanates for reversible hydrogen storage
*
A. Zaluska , L. Zaluski, J.O. Str o¨ m-Olsen
Centre for the Physics of Materials and Department of Physics, McGill University, Ernest Rutherford Building, 3600 University Street, Montr e´ al,
Qu e´ bec H3A 2T8, Canada
Received 15 June 1999; received in revised form 5 October 1999; accepted 9 November 1999
Abstract
In this paper we show that sodium alanates may be used for reversible hydrogen storage, with the advantage of having high storage
capacity combined with low cost. Both NaAlH4 and Na AlH have been investigated for this application, and two complementary
3
6
techniques have been used: improvement of the reaction kinetics by mechanical grinding, and chemical modification of the alloys. By
these methods remarkable desorption/absorption kinetics are obtained. Sodium alanates so modified are capable of reversible hydrogen
storage at the relatively low temperatures of around 80–1408C, with a capacity of between 2.5 and 3.0 wt.%. The hydrides have an even
higher reversible capacity of about 4.5–5 wt.% when operated at temperatures around 150–1808C.
 2000 Elsevier Science S.A. All
rights reserved.
Keywords: Reversible hydrogen storage materials; Alkali metal aluminium hydrides; Metal hydrides; Ball milling
1
. Introduction
at a temperature between 1858C (where melting of NaAlH4
is also suggested [4,5]) and 2308C [3]. The second reaction
occurs at about 2608C, with a further desorption of 1.85
wt.% of hydrogen (relative to the starting material,
Sodium alanates (i.e. complex hydrides of aluminum
and sodium: sodium tetrahydroaluminate–NaAlH4 and
trisodium hexahydroaluminate–Na AlH ) have been
NaAlH ). Finally, the decomposition of NaH occurs at a
3
6
4
known for decades as reducing agents in organic chemis-
try. NaAlH4 was first synthesized in the late 1940s by
Finholt et al. [1], and Na AlH about 15 years later by
Zakharin and Gavrilenko [2]. Although a good source of
hydrogen, sodium alanates were not considered appropriate
for reversible hydrogen storage, because hydrogen release
was extremely slow, and more importantly, because the
rehydrogenation after decomposition was in practice im-
possible.
much higher temperature, with the total hydrogen release
of 7.4 wt.%.
For hydrogen storage, only the first two reactions (with
total hydrogen release of 5.55 wt.%) need to be consid-
ered, because the decomposition of NaH occurs at too high
a temperature for practical storage systems. (The Merck
Index gives the temperature of decomposition of 4258C for
commercial NaH). As mentioned above, the reversibility
of these two reactions is a critical factor for the usability of
sodium alanates in hydrogen storage devices. Reversibility
means the ability to form NaAlH and Na AlH directly
3
6
NaAlH and Na AlH contain large amounts of hydro-
4
3
6
gen (7.4 and 5.9 wt.%, respectively). However the release
of hydrogen does not occur in a single-step reaction.
According to Ref. [3] (and references within), the de-
4
3
6
from NaH and Al (i.e. from the products of the second
stage of decomposition), under hydrogen pressure. How-
ever, so far methods for the direct formation of NaAlH4
and Na AlH use sodium in elemental form rather than in
composition of NaAlH proceeds as follows:
4
3
6
3
NaAlH → Na AlH 1 2 Al 1 3H ↑
the form of sodium hydride, NaH. It has been reported that
4
3
6
2
NaAlH can be formed by the following reaction [4,6,7]:
Na AlH → 3 NaH 1 Al 1 3/2 H ↑
4
3
6
2
NaH → Na 1 1/2 H2
Na 1 Al 1 2H → NaAlH
2
4
The first reaction releases 3.7 wt.% of hydrogen and starts
According to Ref. [6], the reaction should be performed for
several hours in a solvent, tetrahydrofuran (THF), at a
temperature of about 140–1608C. It requires high pressure
*
Corresponding author.
E-mail address: alicja@physics.mcgill.ca (A. Zaluska)
0
925-8388/00/$ – see front matter
 2000 Elsevier Science S.A. All rights reserved.
PII: S0925-8388(99)00666-0

1
26
A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
of hydrogen (140–350 atm.) and the use of activated
aluminum. The activation of the aluminum consists of
treating it in powder form with triethylaluminum. In
another method developed by Dymova et al. [4], hydrogen
pressure of 200–400 atm. is used to hydrogenate Na in the
presence of aluminum. The reaction is performed at higher
temperature (240–2608C) allowing sodium to melt, which
is the key to the process. The liquid form of the reagent is
advantageous because the reaction can be performed
without organic solvents. The above methods are, how-
ever, not very effective when using NaH instead of Na. As
methods) leading to viable and inexpensive materials for
hydrogen storage.
2. Experimental
The starting material, NaAlH4 (sodium aluminum hy-
dride), was purchased from Sigma-Aldrich Canada, with a
nominal purity of 90%. Na AlH was fabricated by
3
6
mechano-chemical synthesis through solid-state reaction,
as described in Ref. [10]. It involved direct reaction of
reported in Ref. [4], only traces of NaAlH were formed
4
from NaH and Al after hydrogenation at temperatures
between 180 and 2408C and under hydrogen pressures
between 100 and 400 atm., although the same conditions
NaAlH and NaH in a vibratory ball mill. The process was
performed in a laboratory apparatus from SCP Sience
(Model SP-2100). Appropriate amounts of the reagents
4
were sufficient for the formation of NaAlH from Na and
(NaAlH and 2NaH, in the form of powders or granules)
4
4
Al. In the work of Ashby et al. [6], the reaction with NaH
was indicated to occur under hydrogen pressure of 350
atm. in a ‘quantitative yield’, but no further data were
given. The above direct methods of the formation of
were ball milled in the stainless steel vials, with stainless
steel balls.
Ball milling was performed under an inert atmosphere of
argon [10]. Sodium hydride, NaH, was purchased from
Sigma-Aldrich Canada, having a nominal purity of 95%.
Ball milling was also used to improve kinetics of the
sorption properties of the hydrides. In this case the milling
time was shorter: between 0.25 and 2 h, performed in
argon, in stainless steel vials. The mass ball-to-powder
ratio (j ) was between 1.7 and 15.
NaAlH do not therefore solve the problem of reversibility
4
of the hydriding/dehydriding reactions. Still, from the
thermodynamic point of view, formation of sodium ala-
nates from Al and NaH under gaseous hydrogen is
possible, but perhaps requires either extreme pressures of
hydrogen or the presence of special solvents or catalysts.
Recent work of Bogdanovi c´ and Schwickardi [8,9] showed
for the first time that reversibility is indeed possible in
sodium alanates. It was confirmed that sodium alanates are
extremely difficult to rehydrogenate, but a new method of
catalysis (by doping with a special Ti-based catalyst)
improved kinetics of hydrogenation/dehydrogenation in
such a way that cycling was feasible. Although the reaction
still required the use of high hydrogen pressure, at least
Chemical modification of sodium alanates was per-
formed by ball milling NaAlH and Na AlH with other
4
3
6
elements or compounds. Various additives have been
studied, as will be presented in upcoming papers. In this
work we describe results obtained with carbon as an
additive and chemical modificator. The amount of carbon
added to NaAlH4 was varied from 1:3 to 1:1 in molar
ratio. This was equivalent to a content of 7–18 wt.% of
carbon in the sample. It has been found that the results did
not depend significantly on the initial form of carbon. For
example, similar results were obtained when using either
activated carbon or powdered graphite as a starting materi-
al.
1
50 atm., and took several hours at 1708C, the results
showed that formation of NaAlH from NaH and Al under
4
gaseous hydrogen was at least possible. The work of
Bogdanovi c´ and Schwickardi [8] initiated new interest in
sodium alanates as prospective hydrogen-storage media
(
e.g. [10–12]). An obvious advantage of sodium alanates is
The materials were characterized by an automated
Nicolet-Stoe powder diffractometer with CuKa radiation.
Hydrogenation experiments from the gas phase were
performed by using a volumetric method. A computerized
gas titration system was used to determine the absorption/
desorption kinetics and to measure the pressure–concen-
tration isotherms. The system operates over the tempera-
ture range from room temperature to 5008C.
Alkali metal hydrides are very sensitive to exposure to
air, and any contact with water results in their hydration.
Therefore exposure to air and humidity was carefully
avoided. All the material handling (including weighing and
loading) was performed in a glove box with a carefully
controlled atmosphere, with low oxygen and water vapour
content. Also, X-ray diffraction (XRD) samples were
specially prepared. The samples were covered with a
their high hydrogen capacity, combined with the low cost
of sodium and aluminum. Our efforts have been focused
recently on enabling sodium alanates to operate reversibly,
with improved kinetics and in the range of moderate
temperatures. We have investigated both NaAlH4 and
Na AlH , and have adopted two complementary ap-
proaches: firstly, improvement of the reaction kinetics by
mechanical grinding, and secondly, chemical modification
of the alloys. These methods result in a great enhancement
of the hydrogenation/dehydrogenation performance of
sodium alanates. In many aspects the results surpass the
catalytically enhanced alloys from Ref. [8]. More im-
portantly however, this work indicates possible new direc-
tions for the development of sodium alanates (to be
exploited independently of, or concurrently with catalytic
3
6

A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
127
plastic tape, which had a negligible or easily deductible
contribution to the diffraction pattern.
effects of ball milling and chemical additives on the
sorption properties of sodium alanates, without the inter-
ference with organic media being used for purification.
Commercially available NaAlH4 and NaH have rela-
tively low purity (90% and 95%, respectively). The use of
such materials obviously results in lower hydrogen capaci-
ties than the nominal values. Moreover, as will be ex-
plained in the following section, NaAlH4 is in fact
thermodynamically unstable when stored at room tempera-
ture and under hydrogen pressure lower than several tens
3. Results and discussion
3.1. Characteristics of the Na–Al–H system
of atm. Although NaAlH decomposes extremely slowly at
low temperatures (which is a serious disadvantage for
practical cycling) it can be expected that after prolonged
As indicated above, the Na–Al–H system exhibits
multistep dehydrogenation, with the following hydride
phases involved: NaAlH , Na AlH and NaH. In order to
4
4
3
6
storage the actual content of NaAlH in the sample may be
determine the usability of the system for practical hydro-
gen storage, temperature–pressure conditions for each of
these reactions should be determined. There are, to date,
two reports providing thermodynamic data for these re-
actions, namely the early paper of Dymova et al. [5] and
the more recent work of Bogdanovi c` and Schwickardi [8].
In the work of Dymova et al. [5] thermodynamic properties
of NaAlH4 and Na AlH were determined from the
4
even lower than that claimed by the manufacturer. In
particular, it is difficult to determine how long the material
has been stored by the supplier after the fabrication, and
whether it deteriorated because of the residual humidity in
the storage atmosphere, etc. For this reason, hydrogen
capacities in the purchased alanates depend not only on
their initial purity but also on their ‘freshness’. One way to
overcome this problem is to purify sodium alanates before
the hydrogenation experiments. There are methods allow-
ing effective purification of NaAlH4 (see for example
3
6
dissociation pressures, which have been measured at
temperatures between 190 and 2408C. In the work of
Bogdanovi c` et al. [8] several P–C isotherms for both
NaAlH4 and Na AlH have been also reported. These
[7,8]), which consist of solvation in THF, filtration of the
3
6
solid impurities and then precipitation in ether, toluene or
pentane. Although purification allows us to regain hydro-
gen capacities in the hydride, it is possible that this
treatment may by itself change the hydriding/dehydriding
properties of NaAlH4.
literature results are plotted in Fig. 1 and allow us to draw
the following conclusions about the Na–Al–H system:
1. Over a very wide range of temperatures (even below
room temperature), NaAlH is thermodynamically un-
4
Traces of organic media (if not carefully removed) may
decompose during heating (as indicated for example in
stable and should spontaneously decompose (although
[13]) and contribute to the measured weight-loss of the
sample, or to an apparent pressure increase when measured
by volumetric methods. Moreover, THF (even in small
amounts) can strongly affect hydriding/dehydriding kinet-
ics, as shown by Imamura et al. [14,15] for Mg–Pd–C
composites. It was suggested that in this case THF was
involved in the formation of a charge-transfer complex,
having a high ability of catalytic activation [14,15].
Moreover, for some alanates, cleavage of THF occurs at
low temperatures, and opening of the THF ring was
observed even at room temperature for example in a
THF–AlH complex [3]. These effects may cause the first
3
desorption after purification to produce apparently more
hydrogen than can be obtained in the subsequent, revers-
ible cycles.
For the above reasons, the effects of synthesis and
purification of NaAlH4 perhaps require a more detailed
study. We avoided any contribution of these effects, in
particular the eventual catalytic action of the purifying
media, by performing experiments on a batch of commer-
cial NaAlH , untreated and unpurified. As a result, the
4
Fig. 1. Equilibrium pressures for the decomposition of NaAlH4 and
hydrogen capacities were reduced (in some cases by as
much as 20 wt.%), but they can be regained if purification
is applied. In consequence, the results demonstrate the
Na AlH in dependence on temperature: (+) from Ref. [5] and (ȣ) from
3
6
Ref. [8]. (Blank and filled symbols are used to differentiate the decompo-
sition of NaAlH4 and Na AlH ).
3
6

1
28
A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
in practice very slowly), if hydrogen pressure is lower
Dymova et al. [5] and Bogdanovi c´ and Schwickardi [8]
can be extrapolated beyond the measured temperature
ranges. Also, we do not know the effect of melting of
NaAlH4 (suggested to occur at 1858C) on its thermo-
dynamics of dehydrogenation/hydrogenation. (After sub-
mission of this paper we learned that recent results indeed
confirm that melting changes the slope of the van’t Hoff
plot [16,17]). Although the above basic characteristics can
be derived from Fig. 1 alone, evidence for these properties
has also been found in the course of the present work, as
described below. The most characteristic feature of the
Na–Al–H system is that the kinetic curves for both
decomposition and synthesis of NaAlH4 always exhibit
two steps, reflecting the two stages of the reaction. This is
illustrated in Fig. 2, which shows the kinetics of the
than about several tens of atmospheres. If simply
extrapolated from the literature data [5,8], the equilib-
rium pressure at room temperature should be equal to
about 30–40 atm. In consequence gaseous synthesis of
NaAlH is thermodynamically possible only if hydro-
4
gen pressure exceeds these pressures, for example 130
atm. at 1808C, or 155 atm. at 2108C, as measured in
Refs. [5] and [8], respectively. (It should be noted that
in practice even higher pressure has to be applied in
order to produce a meaningful driving force for the
reaction and to give usable kinetics).
2
3
. At room temperature, Na AlH is very stable and in
3 6
practice cannot decompose. As estimated from Fig. 1,
this hydride has equilibrium pressure of 1 atm. at about
1
108C [8] or 1358C [5]. Therefore in practice one
decomposition of NaAlH (ball milled for 15 min). The
4
cannot expect decomposition of Na AlH below these
first reaction, 3NaAlH →Na AlH 12Al13H ↑, occurred
3
6
4
3
6
2
temperatures, unless the kinetics of desorption are
excellent.
. In the range of intermediate temperatures (above |110–
when NaAlH was heated from room temperature up to
4
1608C, as shown in Fig. 2a. Further hydrogen desorption
ceased after about 16 h because the temperature was too
low for the decomposition of Na AlH . Subsequent heat-
1
358C and below |3008C), the decomposition of
3
6
NaAlH4 proceeds in two steps and involves firstly
decomposition into Na AlH , and subsequently into
ing of the sample up to the temperature of 2208C resulted
in the decomposition of Na AlH . Two steps are also seen
3
6
3
6
NaH, consistent with the reactions cited above. At
lower temperatures NaAlH4 cannot be fully decom-
posed because Na AlH remains stable at these tem-
in Fig. 2b, which shows the decomposition of NaAlH at
4
1808C, with both reactions occurring, but at different rates.
Fig. 3 shows the XRD patterns taken at each stage of the
reaction: before the decomposition, after the first reaction,
and after the second reaction. In the first pattern, the
structure of NaAlH4 is shown, with some aluminum
impurity. The second XRD pattern shows Na AlH formed
3
6
peratures.
4
. The reaction: NaH1Al → NaAlH (proposed in the
4
literature as a possible path for direct formation of
NaAlH e.g. [6]) is in fact not substantiated thermo-
4
3
6
dynamically. Under gaseous hydrogen, formation of
as an intermediate stage of the decomposition, with an
NaAlH has to be also a stepwise process, involving
4
formation of Na AlH as an intermediate phase. Direct
3
6
synthesis of NaAlH without the formation of Na AlH
4
3
6
may eventually occur, but only at high temperature and
under very high hydrogen pressure. As can be deduced
from Fig. 1, this can be probably effected at tempera-
tures above 3008C and under hydrogen pressure exceed-
ing about 250 atm., where possibly Na AlH becomes
3
6
less stable than NaAlH . In this context, one comment
4
should be made regarding the formation of NaAlH . In
4
previous papers [4,6] it has been established that direct
synthesis of NaAlH from Na and Al always required a
4
substantial excess of aluminum (up to as much as 100
at.%). The excess of aluminum ‘‘has been used to
exclude formation of Na AlH ’’ [4], being interpreted
3
6
as an unwanted by-product of the reaction which
perhaps indicated insufficient amount of aluminum in
the process. It turns out that Na AlH is actually a
3
4
necessary, intermediate stage of the reaction. Therefore,
in the case of an incomplete synthesis, Na AlH can
3
4
indeed be seen among the reaction products, and its
amount simply depends on the stage of the reaction.
Fig. 2. Decomposition of NaAlH4 (ball milled for 15 min., j51.7): (a)
during heating from RT to 1608C (first decomposition stage: NaAlH4
→
Na AlH ) and subsequent heating to 2208C (second decomposition stage:
3
6
It should be stressed that the above interpretation of Fig.
is based on an arbitrary assumption that the data of
Na AlH → NaH); (b) during heating from RT to 1808C. Heating rate:
3
6
1
|58C/min, first desorption of the unpurified material.

A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
129
Fig. 4. Rates of decomposition (first hydrogen desorption) for NaAlH4
ball-milled for various periods of time: (a) 15 min, j51.7, during heating
from RT to 1608C (shown in full in Fig. 2); (b) 1 h, j515, desorption at
1
608C; (c) 2 h, j515, desorption at 1608C.
Fig. 3. X-ray diffraction patterns illustrating decomposition of NaAlH4:
a) before decomposition: NaAlH4 with Al impurity; (b) after the first
reaction: Na AlH 1Al; (c) after the second reaction: NaH1Al.
(
3
6
Na AlH (fabricated by solid state reaction [10]) also
3 6
occurs at a higher rate than the same hydride prepared by
conventional methods, described in [2,18,8]. The rate of
hydrogen desorption shown in Fig. 5 is for example almost
ten times faster at 2008C than that presented by Bog-
danovi c´ and Schwickardi [8] for the Na AlH prepared in
increased amount of aluminum, resulting from the first
reaction. At the final stage of the decomposition, the
material consists only of sodium hydride, NaH, and
aluminum, as presented in the third XRD pattern in Fig. 3.
3
6
a solvent, and is similar to that for the catalyzed hydride.
Thus, even with no catalysts and activators, ball-milled
3
.2. Effects of ball milling on the kinetics of
decomposition
Commercial NaAlH is usually provided in the form of
4
granules. As-purchased, the hydride does not exhibit any
significant desorption of hydrogen up to about 1808C.
According to Ref. [8], after purification NaAlH decom-
4
poses very slowly at 1608C. It was shown that a desorption
of about 2 wt.% required more than 50 h at this tempera-
ture, and for the first 15–20 h almost no desorption was
observed [8]. It has been found in the present work that
NaAlH decomposes much more easily after ball milling.
4
The rate of decomposition depends on the milling intensity
(milling energy) and on the milling time, as illustrated in
Fig. 4. However, even samples milled for a very short
period of time (15 min.) and at a low energy (i.e. a mass
ratio of ball-to-powder j51.7) exhibited a much faster
decomposition at 1608C than that shown by Bogdanovic´
and Schwickardi [8] for the purified (but non-catalysed)
sample. Fig. 4a presents a portion of the plot, which is
shown in full in Fig. 2a. Prolonged milling with higher
energy (i.e. j515) resulted in even faster desorption. In
this case the first reaction can be completed within 1–2 h,
which is comparable with the rates reported in Ref. [8] for
Fig. 5. Rates of hydrogen desorption and absorption for Na AlH
3
6
synthesized by the solid-state process; (temperatures of desorption: 200
and 2408C; temperature of absorption: 2408C; absorption pressure: 58
atm.).
the catalyzed NaAlH . The decomposition of ball-milled
4

1
30
A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
Na AlH is able to absorb and desorb hydrogen relatively
quickly, e.g. within |1 h at a temperature of 2408C (Fig.
subsequent decomposition rate. The first desorption is the
slowest one, and in the subsequent cycles the rates are
always faster and reproducible. It should be noted that the
difference between the samples milled with and without
carbon is not limited to the rate of the first reaction, i.e.
NaAlH →Na AlH . In a more significant way, the second
3
6
5
).
3
.3. Carbon as activator
4
3
6
The hydrogen sorption properties of hydrides are in
reaction (i.e. the decomposition of Na AlH ) is also
3 6
general changed by alloying with other elements or
compounds. For the present hydrides, the most effective
method of such alloying is by using mechanical grinding
affected. The sample without carbon does not exhibit at
1608C any noticeable hydrogen desorption within several
hours after completion the first reaction. In contrast, the
sample milled with carbon exhibited a decomposition of
Na AlH at 1608C. This reaction was however signifi-
(
ball milling) of the hydride in the presence of additives.
Various mechanisms may be involved in this process, such
as thermodynamic modification of the hydride (by chang-
ing hydrogen bonding through substitution by atoms of the
additive) or kinetic modification (for example by catalysis).
Sodium alanates can be effectively modified by such
processes and one of the most efficient additives is carbon.
In most samples the amount of added carbon was 10 wt.%,
but samples with 7 and 18 wt.% C were also studied. The
optimum (i.e. the lowest) content of carbon still needs to
be determined. Too much carbon reduces the hydrogen
capacity by increasing the total weight of the material; on
the other hand, too little carbon produces insufficient
enhancement. All samples with carbon were ball milled
with a mass ball-to-powder ratio j of 15.
3
6
cantly slower than the first reaction, and was completed
only within about 12 h in the first dehydrogenation, and
within about 2–3 h in the subsequent cycles. The total
amount of hydrogen released reached |4.5 wt.% (which is
beyond the time scale in Fig. 6). A similar effect of
enabling the decomposition of Na AlH at lower tempera-
3
6
tures has been also observed in the case of the catalyzed
NaAlH [8,11].
4
After the decomposition of NaAlH4 (modified with
carbon), absorption experiments have been performed.
According to Fig. 1, the first stage of hydrogen absorption
(i.e. formation of Na AlH ) should not require high
3
6
hydrogen pressures at moderately elevated temperatures,
for example about 1008C. In practice however, formation
of Na AlH can occur only at higher temperatures because
Fig. 6 compares the rates of decomposition at 1608C for
NaAlH , milled with and without carbon (10 wt.%). With
4
3
6
the addition of carbon, NaAlH decomposes much faster
than any previous sample: either ball milled without
carbon or catalyzed (as reported in the literature [8]). Fig.
of the kinetic limitations. For example, Bogdanovi c´ and
Schwickardi [8] used a temperature of 2008C and a
hydrogen pressure of about 60 bar for the rehydrogenation
of the catalyzed Na AlH . The formation of NaAlH (i.e.
4
6
b and c shows also the difference between the first and
3
6
4
the second stage of the reaction) requires much higher
hydrogen pressures. It was indicated by Dymova et al. [4]
that a pressure of about 175 atm. represents the ‘threshold
pressure’, below which NaAlH cannot be synthesized, but
4
only Na AlH . However, in the work of Bogdanovi c´ and
3
6
Schwickardi [8], the absorption of NaAlH4 was accom-
plished, through the catalytic action of the chemical
dopant. In the case of the catalyzed NaAlH , the absorp-
4
tion was performed under hydrogen pressure of 152 bar, at
1
708C. Without catalysis only a negligible absorption was
observed.
Sodium alanates activated by our technique (i.e by ball
milling with carbon) cannot only be rehydrogenated with-
out the use of catalysts, but may also be rehydrogenated
under more practical conditions. For example, the forma-
tion of NaAlH4 has been effected under a hydrogen
pressure of about 75–90 atm., instead of the 150 atm. used
in Refs. [8] and [11]. As a result, the synthesis of NaAlH4
can be performed using regular cylinders with compressed
hydrogen, with no need for its additional compression. In
addition, as shown in Fig. 7, hydrogen absorption can be
completed under this pressure within 3–5 h at an even
lower temperature of 1308C. Because of the good kinetics
Fig. 6. Decomposition of NaAlH4 at 1608C: (a) ball milled for 2 h
without carbon, first decomposition (b) ball milled for 2 h with 10 wt.%
of carbon, first decomposition (c) ball milled with 10 wt.% of carbon,
second decomposition.
resulting from our method of synthesis of NaAlH , the
4
process can be performed even at temperatures below

A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
131
Fig. 7. Rehydrogenation of the decomposed NaAlH4 at various tempera-
tures. Initial hydrogen pressure: 88 atm.
Fig. 8. X-ray diffraction patterns for NaAlH ball milled with 10 wt.% of
4
1
008C (Fig. 7), although at these temperatures the reaction
carbon: (a) after ball milling (b) after full decomposition into NaH and Al
(
c) after the first stage of absorption: Na AlH formed (d) intermediate
takes more than 20 h to complete.
3 6
stage of absorption: both Na AlH and NaAlH are present (e) completed
3
6
4
All kinetic plots for the hydrogen absorption exhibit a
two-step reaction: firstly the relatively fast formation of
Na AlH and then the slower synthesis of NaAlH . This
absorption: NaAlH , with Al impurity only. Phases are marked with the
4
same symbols as in Fig. 3.
3
6
4
reaction route is confirmed by XRD measurements, as
shown in Fig. 8. The first XRD pattern represents the
kinetic behaviour, with a higher rate occurring at a higher
temperature. For the second reaction however (i.e. for the
structure of NaAlH , ball milled for 2 h with the addition
4
of 10 wt.% of carbon. The pattern is substantially similar
to that presented in Fig. 3, which was obtained for NaAlH4
milled for a very short time (15 min) and without carbon.
However, the reflections are broader, as is seen more
clearly when patterns with similar intensities are com-
pared. The broadening indicates a refinement of the
structure (perhaps even a nano-structure), and a possible
contribution from strain and defects. Another observation
is that carbon cannot be seen in the diffraction pattern,
either in the form of crystalline or amorphous contribution.
In Fig. 8, the second XRD pattern shows the same material
after complete decomposition into NaH and Al, and is
similar to that shown in Fig. 3c. The next three patterns
represent different stages of the rehydrogenation: the
formation of Na AlH (Fig. 8c; the pattern taken after the
formation of NaAlH ) the driving force is more critical.
4
The applied hydrogen pressure of above 80 atm. (which
decreases during the absorption by several atm.) represents
a much greater driving force for the reaction at 1308C than
at 1508C, so that under these conditions absorption is
actually slower at 1508C than at 1308C. We conclude that
the optimum temperature for the formation of NaAlH in
4
the pressure range of 80–90 atm. is between 120 and
1408C.
In parallel with the excellent kinetics of hydrogen
absorption, alanates modified with carbon also desorb
hydrogen at much lower temperatures. Figs. 9 and 10 show
decomposition rates of the modified NaAlH4 at various
temperatures. The most important result is again the ability
of NaAlH4 to decompose even at temperatures below
1008C, and with kinetics still better than these for cata-
lyzed NaAlH at 1408C [8]. As shown in Fig. 11, NaAlH
3
6
‘fast’ portion of the absorption curve), an intermediate
stage (both Na AlH and NaAlH are present, Fig. 8d) and
3
6
4
4
4
the final formation of NaAlH (taken after the completion
of the slower process; Fig. 8e).
can also desorb hydrogen in two steps: first a low-tempera-
ture desorption, then, as a result of heating the sample to
higher temperature, a second desorption corresponding to
the decomposition of Na AlH . In this two-step process,
high hydrogen capacities can be obtained without exces-
sive heating of the material.
4
It is interesting to note that the kinetics of absorption
reflect a competition between different kinetic factors at
different temperatures. For the formation of Na AlH , the
3
6
3
6
driving force of the reaction (i.e. the difference between
the actual hydrogen pressure and the equilibrium pressure)
is large enough for the reaction to exhibit the usual, simple
As shown above, ball milling alone has a significant
effect on the improvement of the kinetics of absorption and

1
32
A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
Fig. 9. Hydrogen desorption from NaAlH , ball milled with 10 wt.% of
carbon, at medium temperatures.
4
Fig. 11. Two-step decomposition of NaAlH4 at 130 and 1608C.
volumetric system provides good accuracy only for the
lower plateau. At pressures much over 60 atm. the system
is less accurate and the measured values of the desorption
pressure could be as much as 10–15% lower than the true
equilibrium values. This must be borne in mind when we
compare our measured desorption pressures with previous
results in the literature, as is shown in Fig. 13. For the first
reaction (i.e for the higher plateau) our measured desorp-
tion pressures are significantly lower than indicated by the
extrapolation of the results of both [8] and [5]. Absorption
experiments provide further support that the first plateau
really is occurring at a lower pressure. The absorption
desorption, but the addition of carbon improves it even
more remarkably. How carbon modifies sodium alanates is
an interesting question. To study possible thermodynamic
contribution to these modifications, pressure–composition
isotherms of NaAlH were measured at various tempera-
4
tures between 130 and 2308C. Fig. 12 shows a P–C
isotherm for hydrogen desorption taken at 1508C. The
isotherm has two-plateaus, the higher plateau corre-
sponding to the decomposition of NaAlH into Na AlH
6
4
3
and Al, and the lower plateau corresponding to the
decomposition of Na AlH . It should be noted that our
3
6
Fig. 10. Hydrogen desorption from NaAlH , ball milled with 10 wt.% of
Fig. 12. Pressure–composition isotherm for NaAlH , taken for hydrogen
4
4
carbon, at low temperatures.
desorption at 1508C.

A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
133
and Schwickardi [8] were chemically purified and had all
the excessive aluminum removed from the samples. It is
not clear in what way this could affect the equilibrium
pressures. However, if the higher plateau pressure is
indeed a result of purification of the alloys, then it could
explain the easier decomposition of Na AlH at lower
3
6
temperatures in the purified alloys (as in [8,11]) and also
points to an easy way to improve the decomposition of
Na AlH .
3
6
4
. Conclusions
Sodium alanates have been modified by mechanical
grinding without and with an addition — carbon. Milled
NaAlH or Na AlH exhibited great enhancement of the
4
3
6
kinetics of absorption and desorption, and the addition of
carbon in the milling process improved their performance
even more remarkably. Mixtures of NaAlH with carbon
4
Fig. 13. Van’t Hoff plot (temperature dependence of equilibrium pressure
are capable of reversible dehydrogenation/hydrogenation
at much lower pressures and with much faster kinetics than
conventional compounds. The kinetics of dehydrogena-
tion/hydrogenation cycling are many times faster even
than those seen in catalysed NaAlH and Na AlH [8,9].
for hydrogen desorption) for the decomposition of NaAlH ball milled for
4
2
h with 10 wt.% of carbon. Dashed lines represent literature results as in
Fig. 1.
4
3
6
The desorption behaviour is remarkable in that it can be
performed even at the relatively low temperature of 808C.
Rehydrogenation can be effected at 75–90 atm. instead of
experiments proved a complete rehydrogenation (as shown
both by kinetic measurements and by XRD) under the
initial pressure of 80–90 atm. at 1508C. One possible
explanation of this apparent discrepancy could be that
simple linear extrapolation of the data of Refs. [8] and [5]
to lower temperatures is in fact not justified and that the
thermodynamic behaviour of the Na–Al–H system is more
complex (involving the effect of melting). Extrapolation of
the present results gives much more usable equilibrium
pressures at room temperature for the decomposition of
1
50 atm., as is reported necessary for the rehydrogenation
of catalysed NaAlH [8,9,11,12]. Also, hydrogen absorp-
4
tion can be affected at much lower temperatures, even
below 1008C. In conclusion, these hydrides show a good
potential for hydrogen storage applications, being capable
of operation in the temperature range required for PEM
fuel cells, and having high hydrogen capacity at low price.
NaAlH , i.e. several atm. only.
4
Unfortunately we are not at present able to resolve the
question of the role of carbon directly (i.e. by comparing
samples milled with and without carbon). At low tempera-
tures the hydrogen absorption in samples without carbon
occurs too slowly for the equilibrium pressure to be
reliably measured, and at higher temperatures (for example
above 1808C), too high pressures are required, for which
our system is not best suited. We shall be addressing this
problem in more detail in the future.
Acknowledgements
This research has been supported by Le Minist e` re des
Ressources Naturelles (Qu e´ bec), Natural Resources
Canada, the Natural Science and Engineering Research
Council of Canada and Les Fonds pour la Formation des
Chercheurs et a` l’Aide a` Recherche (Qu e´ bec).
In Fig. 13, the equilibrium pressures for the second
plateau (reflecting the decomposition of Na AlH ) are also
lower than the results from Ref. [8], and overlap with the
results from Ref. [5]. The discrepancies between the results
obtained in [5] and [8] have been already pointed out in
3
6
References
[
1] A.E. Finholt, A.C. Bond, H. Schlesinger, J. Am. Chem. Soc. 69
1947) 1199.
[2] L.I. Zakharin, V .V . Gavrilenko, Dokl. Akad. Nauk SSSR 1 (145)
1962) 793.
(
[8]. However, there is no indication whether these differ-
(
ences reflect changes in the hydrides, or instrumental
differences. In this context, it is worth mentioning that the
hydrides of Dymova et al. [5] (and also our materials) were
not purified before the hydrogenation experiments and
contained aluminum, whereas the hydrides of Bogdanovic´
[
[
3] J.A. Dilts, E.C. Ashby, Inorg. Chem. 11 (6) (1972) 1230.
4] T.N. Dymova, N.G. Eliseeva, S.I. Bakum, Y.M. Dergachey, Dokl.
Akad. Nauk SSSR 215 (1974) 1369.
[5] T.N. Dymova, Y.M. Dergachev, V.A. Sokolov, N.A. Grechanaya,
Dokl. Akad. Nauk SSSR 224 (3) (1975) 591.

1
34
A. Zaluska et al. / Journal of Alloys and Compounds 298 (2000) 125–134
[6] E.C. Ashby, G.J. Brendel, H.E. Redman, Inorg. Chem. 2 (1963)
[12] C.M. Jensen, R. Zidan, N. Mariels, A. Hee, C. Hagen, Int. J.
4
99.
Hydrogen Energy 24 (1999) 461.
[
[
[
7] H. Clasen, Angew. Chem. 73 (1961) 322.
8] B. Bogdanovi c´ , M. Schwickardi, J. Alloys Compd. 253 (1997) 1.
9] B. Bogdanovi c´ , German Pat. 19526434, 1995.
[13] J.C. Bureau, J.P. Bastide, B. Bonnetot, H. Eddaoudi, Mat. Res. Bull.
20 (1985) 93.
[14] H. Imamura, N. Sakasai, J. Alloys Compd. 321 (1995) 810.
[15] H. Imamura, N. Sakasai, Y. Kajii, J. Alloys Compd. 232 (1996) 218.
[16] K.J. Gross, private communication.
[
10] L. Zaluski, A. Zaluska, J.O. Str o¨ m-Olsen, J Alloys Compd. 290
1999) 71.
11] R.A. Zidan, S.A. Takara, G. Hee, C.M. Jensen, J Alloys Compd.
85 (1999) 119.
(
[
[17] B. Bogdanovi c´ , private communication.
[18] E.C. Ashby, P. Kobetz, Inorg. Chem. 5 (9) (1996) 1616.
2