A kinetics model of hydrogen absorption and desorption in Ti-doped NaAlH4

Article abstract of DOI:10.1016/j.jallcom.2004.05.004

The kinetics of the hydrogen sorption of Ti-doped direct-synthesized NaAlH₄ has been studied by measuring sorption rates at various temperatures and applied pressures. Formation and decomposition rate equations for NaAlH₄ and Na₃AlH₆ are proposed, and pre-exponential factors and activation energies have been calculated for these reactions. These equations were used to calculate alanate decomposition curves. The results fit the experimental data very well. The predictive capabilities of this empirical approach provide a very useful modeling tool for optimizing the performance of the alanates over a range of hydrogen absorption temperatures and pressures. Under a typical hydrogen-loading condition (constant pressure), the optimum temperature for fast fill can be determined.

Full text of DOI:10.1016/j.jallcom.2004.05.004

Journal of Alloys and Compounds 385 (2004) 224–231
A kinetics model of hydrogen absorption
and desorption in Ti-doped NaAlH4
Weifang Luo∗, Karl J. Gross
Analytical Material Science Department, Sandia National Laboratories, P.O. Box 969, Livermore, CA 94550, USA
Received 3 May 2004; received in revised form 13 May 2004; accepted 13 May 2004
Abstract
The kinetics of the hydrogen sorption of Ti-doped direct-synthesized NaAlH4 has been studied by measuring sorption rates at various
temperatures and applied pressures. Formation and decomposition rate equations for NaAlH4 and Na3AlH6 are proposed, and pre-exponential
factors and activation energies have been calculated for these reactions. These equations were used to calculate alanate decomposition curves.
The results fit the experimental data very well. The predictive capabilities of this empirical approach provide a very useful modeling tool for
optimizing the performance of the alanates over a range of hydrogen absorption temperatures and pressures. Under a typical hydrogen-loading
condition (constant pressure), the optimum temperature for “fast fill” can be determined.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Ti-doped; Kinetics; NaAlH4
1. Introduction
drogen diffusion on the surface, H2 dissociation on the
surface, hydrogen absorption into the solid, hydrogen dif-
fusion and hydride formation [5–8]. The dependence of the
hydrogen sorption rate on temperature, hydrogen pressure
and geometry of solids has been deduced using different
models for the various rate-limiting steps [5–8].
The extensive work on the kinetics of interstitial hydrides
provides a foundation for the study of hydriding kinetics
in the alanates system. It is important, however, to recog-
nize that the mechanism of hydrogen uptake and release is
somewhat different for alanates than interstitial hydrides. In
this case, sorption occurs through complex solid–solid and
solid–gas phase formation and decomposition reactions. The
formation and decomposition of the alanates Na3AlH6 and
NaAlH4 are described by the following reactions
The discovery that hydrogen can be easily absorbed and
desorbed from Ti-doped sodium–aluminum hydrides (the
alanates) has created an entirely new prospect for lightweight
hydrogen storage [1]. These materials have nearly ideal equi-
librium thermodynamics [1,2]. Doping the alanates with Ti
and other transition metals enables the reverse (hydriding)
reactions to take place under moderate conditions (<100 ◦C
and <100 bar H2) [1,2]. However, sorption rates for this sys-
tem remain a critical issue for vehicular hydrogen storage
applications. Initial studies of alanate kinetics have focused
on the dependence of the reaction rates on temperature [3].
The effect of Ti-doping on the reaction rates have also been
measured [3]; however, the mechanism of enhanced kinet-
ics is still not fully understood. The effect of hydrogen gas
pressures and phase composition of reaction rates have not
been evaluated until recently by Anton et al. [4].
NaH + 1/3Al + H2 ⇔ 1/3Na3AlH6
1/3Na3AlH6 + 2/3Al + H2 ⇔ NaAlH4
(1)
(2)
Hydriding/dehydriding reaction kinetics has been
well studied for interstitial hydrides [5–8]. These are
multiple-step reactions, which include hydrogen transport
in the gas-phase, hydrogen adsorption on the surface, hy-
Unlike interstitial hydrides, these are stoichiometric com-
pounds with no hydrogen solid solutions of either phase.
The two reactions exhibit very different plateau pressures
observed as discrete jump in pressure in the single Na3AlH6
phase region [1].
The formation reactions involve more than one solid
phase, which implies the existence of long-range diffusion
of solid species aluminum or sodium [9,10].
∗
Corresponding author. Tel.: +1-925-294-3729;
fax: +1-925-294-3410.
E-mail address: wluo@sandia.gov (W. Luo).
0925-8388/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2004.05.004

W. Luo, K.J. Gross / Journal of Alloys and Compounds 385 (2004) 224–231
225
aluminum powder (∼20 micron, 99+% Sigma–Aldrich),
sodium hydride (Sigma–Aldrich) and solid titanium
tri-chloride (Sigma–Aldrich) were weighed in molar ratio
of NaH:Al:TiCl3 = 112:100:4 in an argon-filled glove box.
The excess NaH was added to compensate for the forma-
tion of NaCl [3]. The powders were mixed by ball milling
under argon in a high-energy SPEX mill for 30 min. After
mixing, a sample of about 3 g was transferred in the argon
atmosphere glove box to a stainless steel reaction vessel.
The sample holder had a thermocouple located in the center
of the sample to monitor temperature in the reaction zone.
Nomenclature
dC(H)/dt rate of concentration change of H in alanate
formed (+) or decomposed (−), as H wt.%.
Eaa1
Ead1
Eaa2
Ead2
activation energy of NaAlH4 formation
activation energy of NaAlH4 decomposition
activation energy of Na3AlH6 formation
activation energy of Na3AlH6
decomposition
H wt.% weight percent of H in sample
Ko
K
Pappl
Peq1
pre-exponential factor of rate constant
rate constant, K = Koe(−E
a1/RT)
2.2. Experimental details
applied pressure
plateau pressure of NaAlH4 at a given
temperature
Sorption rates and hydrogen capacities were obtained
volumetrically using a carefully calibrated Sieverts’ ap-
paratus. During sorption, the sample temperature, dosing
volume temperature and applied pressure were monitored
and recorded. Hydrogen pressures for absorptions were un-
der 120 bar, and monitored using a transducer of Teledyne
Taber model 206 piezoelectric, 0–200 bar with a resolution
of 0.1 bar. Desorption pressures were monitored using a
0–3 bar Baratron capacitance manometer with resolution of
0.001 bar.
Peq2
plateau pressure of Na3AlH6 at a given
temperature
The mechanism of catalyst of Ti-containing compounds
is not fully understood [11,12]. In previous studies, it was
found that rates of hydrogen desorption are strongly depen-
dent on temperature and the level of Ti-doping (between 1
and 6 mol.%) [3]. A detailed evaluation of desorption rates
in terms of an exponential temperature dependence using
the rate Eq. (3) provided a measure of the activation energy
(Ea) and a pre-exponential term (K) for the decomposition
of both alanate phases [3].
The sample was heated to 125 ◦C under vacuum for
20 min prior to being exposed to hydrogen. High purity
hydrogen (Matheson Trigas research purity, 99.999%) was
then introduced into the reaction vessel. The first absorption
cycle was slow; however, the reaction rates increased with
subsequent cycles. By the third cycle, the sample was acti-
vated and ready for sorption rate tests. During sorption rate
tests, the sample temperature was set at several tempera-
tures in steps between 60 ◦C and 180 ◦C for desorption, and
60 ◦C and 160 ◦C for absorption. All pressure and temper-
ature data were acquired and recorded by computer using
LabView-based software.
Before desorption measurements, the sample was heated
to 125 ◦C and exposed to about 100 bar of hydrogen in or-
der to completely hydride the sample. When the pressure in
the chamber was stable (generally about 2 h), the hydriding
process was completed and the sample was, then, cooled
down to room temperature. Desorption rate measurement
began at room temperature by opening the reaction vessel
to a large calibrated volume (approximately 1.2 l) that had
been evacuated. The temperature was increased gradually
in steps, about 15–40 min for each step. The amount of hy-
drogen desorbed from the sample was determined from the
pressure rise in the calibrated volume. Desorption rates were
calculated in weight percent from the change in hydrogen
pressure over a given time period. When a desorption-rate
test was completed, the sample was evacuated at 160 ◦C for
2 h to completely desorb the sample; the sample was then
ready for next absorption rate test.
ꢀ
ꢁ
Ea
Rate = K exp −
(3)
RT
It was shown that only minor doping levels (1 mol.%
TiCl3) were necessary to significantly reduce the activation
energy of decomposition for both NaAlH4 and Na3AH6.
However, the rates of decomposition did continue to increase
as the level of TiCl3 increased [3]. This increase was as-
cribed to the pre-exponential coefficient K.
In the present study, we measured the forward (absorption)
and reverse (desorption) rates of both Eqs. (1) and (2) under
varying temperature and pressure conditions. We used the
data to evaluate a number of different empirical rate equa-
tions, including a pressure-dependent term. The rate equa-
tions that gave the best overall results were then used to
determine the best fit to pre-exponential coefficients and ac-
tivation energies. These values were used to predict kinetic
behavior over a broad range of temperatures and pressures.
Such a model should prove useful for optimizing operational
conditions for real world hydrogen storage applications.
2. Experiment
2.1. Sample preparation
The absorption rate measurements were performed in
a similar manner to desorption rate measurements using
a smaller calibrated volume of approximately 0.1 l. The
Samples were prepared using the direct synthesis
method as described elsewhere [13]. Initial materials,

226
W. Luo, K.J. Gross / Journal of Alloys and Compounds 385 (2004) 224–231
amount of hydrogen absorbed was measured at different
temperatures, and the absorption rates were calculated in
weight percent from the pressure drop over a given period
of time.
the Na3AlH6 started to release hydrogen. The desorption
rate increased as the temperature was raised in steps until
the Na3AlH6 decomposition was complete.
In desorbing, the reaction rate drops off in time. This
is due to the combined action of the concentration of
the desorbing phase approaching zero and the hydrogen
over-pressure rising towards the equilibrium pressure. As
the hydrogen pressure approaches the equilibrium pressure,
the driving force for the reactions is lowered and the de-
composition rate drops to zero at equilibrium. If the applied
hydrogen pressure was higher than the equilibrium pres-
sure, such as when temperature is suddenly lowered, instead
of desorbing, the sample would begin to absorb hydrogen.
Therefore, an analytical model of the rate behavior must
necessarily include the relative concentration of reactants
and the driving force. This is the motivation behind the
empirical model we present here.
3. Results and discussion
3.1. Desorption measurements
A typical desorption measurement is shown in Fig. 1.
The amount of desorbed hydrogen is given in terms of
weight percentage of the total sample weight. The tem-
perature was raised from room temperature and gradually
increased to 180 ◦C. Hydrogen release became detectable
starting at 60 ◦C. When the amount of hydrogen released
from the sample approached 2.3 wt.%, the desorption rate
dropped off. This indicates that 2.3 wt.% was the end of
NaAlH4 decomposition. The cause analysis for incomplete
hydride formation will be discussed later in this paper.
After the decomposition of NaAlH4, the temperature was
reduced to 100 ◦C and kept at this temperature for 20 min.
No hydrogen release was observed at 100 ◦C even after the
applied hydrogen pressure was reduced to zero (by evacu-
ation). Then the temperature was increased to 120 ◦C and
3.2. Hydrogen capacity
Theoretical hydrogen capacity for alanate is 5.6 wt.%, i.e.,
3.7 wt.% from Eq. (2) and 1.9 wt.% from Eq. (1). However,
in this study the maximum H wt.% was about 3.9 wt.%;
2.23 wt.% comes from Eq. (2) and 1.67 wt.% from Eq. (1).
This is due the following reasons:
• Hydrogen weight percentage was calculated based on to-
tal weight of the sample, including active and inactive
components.
• The molar ratio of the sample was NaH:Al:TiCl3
= 1.12:1:0.04. Here, the NaH molar ratio used was 1.12,
instead of 1, because of the need to compensate for the
loss of Na resulting from formation of NaCl [3]. NaCl
formation, together with the weight of TiCl3, results in
total inactive components as 15 wt.%.
• Chemically prepared and purified NaAlH4 will release
the full 5.6 wt.% hydrogen [1]. However, previous work
by others showed that the NaAlH4 formation as shown
in Eq. (2) does not reach 100% completion; instead, it
stops at about 75% completion in the temperature range
of 150–200 ◦C [1]. This is probably due to the presence of
“isolated islands” of reactant, the Na3AlH6 and Al, after
about 75% completion of NaAlH4 formation [10]. XRD
results for samples hydrided at 100 bar typically show that
it is a mixture of NaAlH4 and Na3AlH6 with a molar ratio
of approximately 2:1.
With these considerations, the maximum hydrogen
weight percentage in the sample is 3.9 wt.%, the slightly
less-than-theoretical hydrogen content for the Na3AlH6
phase (1.67%) can be primarily accounted for by the weight
of the inactive component TiCl3. Thus, the reactions consist
of the full formation and decomposition of Na3AlH6 for
hydrogen weight percentage between 0 < H wt.% < 1.67%,
and the partial formation and decomposition of NaAlH4 for
hydrogen concentrations between 1.67% < H wt.% < 3.9%.
Fig. 1. NaAlH4 and Na3AlH6 decomposition profile at various tempera-
tures (A) and pressures (B). Thick lines are for hydrogen wt.%, and thin
lines for temperature (A) and hydrogen pressure (B), respectively.

W. Luo, K.J. Gross / Journal of Alloys and Compounds 385 (2004) 224–231
227
3.3. Dependence of reaction rates on applied hydrogen
pressure and reactant concentration at various temperatures
NaH content is considered. Therefore, the rate of Na3AlH6
formation is taken as first order of the NaH content (thus
first order in H content). For NaAlH4 formation, there is no
excess amount of Al in the system. The concentration of
both Na3AlH6 and Al should be considered for this reac-
tion. Therefore, the rate of NaAlH4 formation is taken to be
second order of the reactant contents (thus second order in
H concentration). By contrast, decomposition rates for both
reactions are taken as first order because only one solid re-
actant (NaAlH4 or Na3AlH6) is involved in each reaction.
Based on the above principles, the rate equations for for-
mation/decomposition for NaAlH4 and Na3AlH6 can be de-
scribed as given in Eqs. (4) – (7). Note that reactant contents
in these equations are now expressed as H wt.% in order to
simplify the calculation, and the pressure considered here is
within the range for hydrogen storage – about 0.1–150 bar.
Ultimately, one would like to study kinetics in terms of
the fundamental processes occurring during hydrogen ab-
sorption and desorption. Many models have been developed
for hydrides to examine these processes and to aid in de-
termining the rate-limiting mechanism [5–8]. However, to
make mechanistic determinations for a particular hydride
system based on such a kinetic analysis requires extremely
well-defined experimental conditions, such as near-perfect
isothermal conditions, and well-defined particle sizes and
shapes. At this point, very little is actually known about how
the alanate phase transitions even take place on an atomic
scale. Thus, the purpose of this study was not to attempt
to perform reaction rate analysis at a mechanistic level, but
rather to find a set of empirical rate equations that accurately
describe the hydrogen sorption behavior of the alanates on
a performance level. The goal was to achieve a predictive
model to aid in developing hydrogen storage systems and
better materials.
• For NaAlH4 formation:
ꢀ
ꢁ
ꢀ
ꢁ
Pappl
Peq1
Eaa1
d(H wt.%)/dt = Koa1exp −
ln
RT
× (3.9−H wt.%)2; (1.67<H wt.%<3.9)
To begin, the general sorption rate at a given temperature
is a function of reactant concentration and applied pressure
(4)
ꢀ
ꢁ
• For NaAlH4 decomposition:
Ea
d(H wt.%)/dt = Koexp −
∗ F(p)
ꢀ
ꢁ
ꢀ
ꢁ
RT
Ead1
Peq1
−d(H wt.%)/dt = Kod1exp −
ln
∗ F(reactant concentrations)
(3)
RT
Pappl
×(H wt.%−1.67); (1.67<H wt.%<3.9)
The term Ko exp(−Ea/RT) is the rate constant, which is
a function of temperature. Ko is a pre-exponential factor
and Ea is the activation energy. We use the difference of
logarithms of applied pressure and plateau pressure for the
functional dependence of reaction rates on pressure. Thus,
for both reactions 1 and 2, the rate dependence on hy-
drogen pressure is ln(Pappl/Peq) for hydride formation and
ln(Peq/Pappl) for hydride decomposition.
(5)
• For Na3AlH6 formation:
ꢀ
ꢁ
ꢀ
ꢁ
Pappl
Peq2
Eaa2
d(H wt.%)/dt = Koa2exp −
ln
RT
× (1.67 − H wt.%); (H wt.% < 1.67)
The reason for using the logarithm of pressure will be
discussed later in this paper.
(6)
(7)
As a first approximation, reactions (1) and (2) are con-
sidered to be independent reactions that take place one af-
ter the other. This is justified for alanates because far from
equilibrium, the reaction rates are at least an order of mag-
nitude slower for reaction 1 than reaction 2 for desorption
and faster for absorption at a given temperature and pres-
sure. We reported earlier on the serial behavior of the re-
actions for decomposition [9]. Recently, Gao et al. reported
that they have found that these two reactions also take place
sequentially on hydrogen absorption [13]. In order to sim-
plify the calculation, the concentrations of reactants, such
as Al, NaH, NaAlH4 and Na3AlH6, can be replaced by H
wt.%. This is because hydrogen is evolved or consumed in
both reactions, and so by material balance in this system,
the concentration of the solid phases are functions of the
hydrogen weight-percent.
• For Na3AlH6 decomposition:
ꢀ
ꢁ
ꢀ
ꢁ
Peq2
Ead2
−d(H wt.%)/dt = Kod2 exp −
ln
RT
Pappl
× (H wt.%); (H wt.% < 1.67)
The equilibrium plateau pressure (Peq) values used here
for reactions 1 and 2 are taken from literature [9,14,15].
There are some indications that materials prepared by direct
synthesis may have a more sloping character, and this of
course would affect the analysis to at least a small degree
[16], but this is not included in this model.
According to Eqs. (4)–(7), there should be straight lines
on the plots of (−d(H wt.%)/dt) versus (ln(Peq/Pappl)*(H
wt.%)) for formation and (ln(Pappl/Peg) (H wt.%)) for de-
composition at constant temperature. It can be seen from
the linear behavior at each temperature that this model fits
well. The slopes of these lines are the rate constants Ko
For Na3AlH6 formation, there are two solid reactants,
NaH and Al. Since there is an excess amount of Al, only the

228
W. Luo, K.J. Gross / Journal of Alloys and Compounds 385 (2004) 224–231
Fig. 2. Dependence of Na3AlH6 decomposition rates on applied pressure and content of Na3AlH4 at various temperatures.
exp(−Ea/RT) at a given temperature. Fig. 2 is an example
showing this dependence of sorption rates at given temper-
atures on applied/equilibrium pressures and reactant con-
centration for Na3AlH6 decomposition. During a sorption
measurement, the applied pressures and hydrogen wt.% are
changing, so do the reaction rates. For example, the 120 ◦C
line in Fig. 2 shows the process of Na3AlH6 decomposition
with the applied pressure increasing and H wt.% in solid de-
creasing due to desorption. The slope of this line is the rate
constant for Na3AlH6 decomposition at 120 ◦C. The for-
mation/decomposition rate constants for both tetra-hydride
and hex-hydride can be obtained in the same method at
various temperatures. These values will be used to calcu-
late activation energies by using Arrhenius plots in the next
section.
The same data were used to evaluate other potential func-
tional dependencies of rates on pressure and concentrations.
These included
Fig. 3. Arrhenius plot for NaAlH4 formation and decomposition.
• F (p) = (Pappl − Peq)/Peq
• First order of F (H wt.%) for NaAlH4 formation, and
• The rate equations for hydrides proposed by Wang and
Suda [8].
These functions did not provide nearly as good a represen-
tation or fit to the data as the chosen set of relationships.
3.4. Arrhenius plots
The rate constants calculated from the slopes of the
plots of (−d(H wt.%)/dt) versus (ln(Peq/Pappl) ∗ (H wt.%))
at given temperatures discussed above were used to gen-
erate the Arrhenius plots for NaAlH4 and Na3AlH6 for-
mation/decomposition are shown in Figs. 3 and 4. The
pre-exponential factors Ko and the activation energies Ea
for formation/decomposition of tetra- and hex-hydrides can
Fig. 4. Arrhenius plot for Na3AlH6 formation and decomposition.

W. Luo, K.J. Gross / Journal of Alloys and Compounds 385 (2004) 224–231
229
Table 1
Summary of formation and decomposition activation energy of NaAlH4 and Na3AlH6
Ko (abs)
Eaa (abs) (kJ/mol)
Ko (des)
Ead (des) (kJ/mol)
Ead (des) [3] (kJ/mol)
NaAlH4
Na3AlH6
6.25 × 108
1.02 × 108
61.6
56.2
1.9 × 1011
2.9 × 1010
85.6
88.3
80.0
97.5
be determined from these two figures, and the results are
listed in Table 1.
The activation energy for decomposition of NaAlH4 and
Na3AlH6 reported by Sandrock et al. [3] are listed for com-
parison.
decreases with increasing temperature for NaAlH4 forma-
tion due to the higher Peq at higher temperature. The net
effect on rate due to increasing temperature can be posi-
tive or negative. For example, using Ko = 6.25 × 108; Ea
= 61.6 kJ/mol and assuming reactant concentration is a con-
stant as H wt.% = 2.9%, the rate equation for NaAlH4 for-
mation, Eq. (4), becomes
3.5. Desorption simulations
ꢀ
ꢁ
Pappl
Peq
d(H wt.%)/dt = 6.25 × 108 e(−7.4/T)ln
(8)
To test our model, the rate Eqs. (4)–(7) were used to cal-
culate the decomposition curves of NaAlH4 and Na3AlH6
from temperature and pressure data. The results of these
simulations are presented in Fig. 1, along with the actual
experimental data for comparison. It can be seen that the set
of relationships that we determined to best represent these
complex reactions (Eqs. (4)–(7)) provide a very reasonable
means for modeling desorption behavior of the alanates over
a broad range of conditions. This provides confidence that
these relationships can be used in a predictive manner to op-
timize the alanate sorption processes for applications under
various operating conditions.
Using these values, the calculated rates from 60 to 160 ◦C
at applied pressure between 10 and 120 bar for NaAlH4 for-
mation are shown in Fig. 5. As an example of the counter-
acting effects of pressure and temperature, it can be seen
from Fig. 6 that the absorption rate at 140 ◦C is higher than
at 160 ◦C, when applied pressure is below 110 bar.
3.6.2. Calculated formation rates for Na3AlH6
The formation rates for Na3AlH6 can be calculated us-
ing the same approach as for NaAlH4. Unlike the NaAlH4
case, however, the crossover of the absorption rates with
temperature occurs at pressures below those commonly ap-
plied to these samples. For example, the plateau pressure
for Na3AlH6 at temperatures below 160 ◦C is below 7 bar,
which is lower than applied pressure generally used for the
charging process. Thus, for Na3AlH6, absorption rates typ-
ically increase with increasing temperature. Fig. 6 shows
the calculated formation rates at various temperatures and
3.6. Applications of the rate equations
Expanding on these simulations, we now present three
examples of the predictive capabilities of these empirical re-
lationships. This provides some insight into how this model
can be applied as a powerful tool for optimizing performance
in alanate hydrogen storage applications.
3.6.1. Optimizing temperature and pressure for hydrogen
absorption in forming NaAlH4
Temperature increase has two effects on a reaction rate: it
increases both rate constants and plateau pressures. For hy-
dride decomposition at a given applied pressure, both these
effects lead to rate increase and, therefore, these effects re-
inforce each other so that desorption rates always increase
by raising the temperature. For hydride formation, however,
it is not the case. Temperature increase will increase rate
constant, but also increases plateau pressure; that leads to
rate reduction according to Eqs. (4) and (6). Thus, the net
effect of temperature increase on absorption rate at a given
applied pressure is not necessarily positive. This counteract-
ing phenomenon has been reported for the alanates by the
observation of higher absorption rate at 50 ◦C than at 70 ◦C
for a given applied pressure [17].
This pressure versus temperature effect can be evaluated
using the proposed absorption rate Eqs. (4) and (6). Rate con-
stant K = Koe(−E
increases monotonically with tem-
a1/RT)
Fig. 5. Calculated formation rates of NaAlH4 at several temperatures and
applied pressures.
perature. However, the pressure driving force, ln(Pappl/Peq),

230
W. Luo, K.J. Gross / Journal of Alloys and Compounds 385 (2004) 224–231
Fig. 6. Calculated formation rates for Na3AlH6 at selected temperatures
and applied pressures.
Fig. 7. Calculated decomposition rates of NaAlH4 and Na3AlH6 over a
typical range of applied pressures and temperatures for H wt.% = 3.2
(NaAlH4) and H wt.% = 1 (Na3AlH6).
applied pressures. It can be seen that at applied pressure of
40 bar, the formation rate at hydrogen content of 0.67 wt.%
is 11 wt.%/h at 120 ◦C and 20 wt.%/h at 140 ◦C.
• at 120 ◦C and an applied pressure of 1 bar, the Na3AlH6
decomposition rate is very low, about 0.07 H wt.%/h,
compared to the decomposition rate of NaAlH4 which is
1.5 wt.%/h at the same temperature;
• at applied pressure of 1 bar, the decomposition rate for
Na3AlH6 becomes more significant when temperature is
increased to 160 ◦C.
3.6.3. Calculated decomposition rates for Na3AlH6 and
NaAlH4
Unlike hydride formation, the decomposition rates for
both Na3AlH6 and NaAlH4 increase with temperature re-
gardless of the pressure, because both the rate constant K
and decomposition driving force (Peq/Pappl) increase with
temperature. To evaluate the influence of temperature on the
hydrogen desorption rates, Eqs. (5) and (7) can be rewritten
using the values of activation energy and pre-exponential
factor listed in Table 1, and assuming values of H wt.%
= 1% for Na3AlH6 and H wt.% = 3.2% for NaAlH4, which
are close to the median hydrogen content in each plateau
region.
4. Conclusion
The kinetics of hydrogen absorption and desorption of
Ti-doped direct-synthesized NaAlH4 was studied by mea-
suring sorption rates over a wide range of temperatures and
applied pressures. From these measurements, a series of for-
mation and decomposition rate equations for NaAlH4 and
Na3AlH6 were determined. Using these rate equations, the
pre-exponential rate constants and activation energies were
calculated for the decomposition and reformation of these
two alanate compounds. The calculated alanate decomposi-
tion curves using these empirical rate equations were tested
and fit the experimentally measured results quite well. The
kinetic performance of the alanate reactions can be predicted
using these equations for a wide range of pressures and tem-
peratures. As a result, it is possible to optimize alanate for-
mation rates for any given pressure by selecting a proper
temperature. For the NaAlH4 formation reaction at a fixed
pressure, it is not always advantageous to increase the tem-
perature to achieve the highest rate. This model provides a
method for evaluating and optimizing the performance of
alanates for a variety of hydrogen storage applications.
For NaAlH4:
ꢀ
ꢁ
Peq1
−d(H wt.%)/dt = 1.9 × 1011
For Na3AlH6:
e
(−10.29/T)ln
(9)
Pappl
ꢀ
ꢁ
Peq2
−d(H wt.%)/dt = 2.9 × 1010
e
(−10.62/T)ln
(10)
Pappl
We used Peq values calculated from the van’t Hoff plot
[14,15] for Eqs. (9) and (10) to calculate decomposition rates
for Na3AlH6 and NaAlH4. The results are plotted in Fig. 7.
The vertical line at pressure of 1 bar is shown in this figure
to indicate the decomposition rates at 1 bar. This provides
a gauge for evaluating desorption rates for applications that
typically require hydrogen supplied at near ambient condi-
tions, such as fuel cells.
It can be seen from this figure that:
Acknowledgements
• it is possible to estimate discharge rates for the desorption
reaction at over a broad range of pressures and tempera-
tures;
The authors thank S. Spangler and K. Stewart for prepar-
ing samples and test equipment. This work is financially

W. Luo, K.J. Gross / Journal of Alloys and Compounds 385 (2004) 224–231
231
[8] X.L. Wang, S. Suda, Int. Hydrogen Energy 17 (2) (1992) 139–
147.
[9] K.J. Gross, S. Guthrie, G.J. Thomas, J. Alloys Comp. 297 (1999)
270–281.
supported by U.S. DOE under DE-AC04-94 AL 8500 and
Sandia’s Laboratory Directed Research and Development
Project Office.
[10] B. Bogdanovic, M. Felderhoff, M. Germann, M. Hartel, A. Pom-
merin, F. Schuth, C. Weidenthaler, B. Zibroius, J. Alloys Comp. 350
(2003) 246–255.
References
[11] D. Sun, T. Kiyobayashi, H.T. Takishita, N. Kutiyama, C.M. Jensen,
J. Alloys Comp. 337 (2002) L8–L11.
[12] E.H. Majzoub, K.J. Gross, J. Alloys Comp. 356–357 (2003) 363–367.
[13] K.J. Gross, E. Majzoub, G.J. Thomas, G. Sandrock, DOE Annual
Review, 2002.
[14] Y. Gao, J. Rijssenbeek, W. Heward, V. Smentkowski, Presenta-
tion at American Physics Society, Montreal, Canada, March 22–24,
2004.
[15] K. Gross, G. Thomas, C. Jensen, J. Alloys Comp. 330–332 (2002)
683–690.
[1] B. Bogdanovic, R.A. Brand, A. Marjanovic, M. Schwickardi, J. Tolle,
J. Alloys Comp. 302 (2000) 36–58.
[2] C.M. Jensen, K.J. Gross, Appl. Phys. A 72 (2001) 213–219.
[3] G. Sandrock, K. Gross, G. Thomas, J. Alloys Comp. 339 (2002)
299–308.
[4] D.L. Anton, D.A. Mosher, S.M. Opalka, Hydrogen, Fuel Cells and
Infrastructure Technologies Program Annual Merit Review Meeting,
Berkeley, CA, May 19–22, 2003.
[5] M.H. Mintz, J. Bloch, Prog. Solid State Chem. 16 (1985) 163–194.
[6] T.B. Flanagan, in: A.F. Anderson, A.J. Maeland (Eds.), Hydrides for
Energy Storage, Pergamon, Oxford, 1978, pp. 135–150.
[7] P.D. Goodell, P.S. Rudman, J. Less-Common Metals 89 (1983) 117–
125.
[16] K.J. Gross, DOE Hydrogen and Fuel Cell Program Annual Review,
2003.
[17] K.J. Gross, G. Sandrock, G.J. Thomas, J. Alloys Comp. 330–332
(2002) 691–695.