Effects of heat transfer on the sorption kinetics of complex hydride reacting systems
-
Upload
independent -
Category
Documents
-
view
0 -
download
0
Transcript of Effects of heat transfer on the sorption kinetics of complex hydride reacting systems
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 3
Avai lab le at www.sc iencedi rect .com
journa l homepage : www.e lsev ie r . com/ loca te /he
Effects of heat transfer on the sorption kinetics of complexhydride reacting systems
Gustavo A. Lozano*, Nico Eigen1, Claude Keller2, Martin Dornheim, Rudiger Bormann
Institute of Materials Research, GKSS Research Centre, Max Planck Strasse 1, D-21502 Geesthacht, Schleswig-Holstein, Germany
a r t i c l e i n f o
Article history:
Received 5 November 2008
Received in revised form
5 December 2008
Accepted 6 December 2008
Available online 21 January 2009
Keywords:
Sodium alanate
Hydrogen storage
Thermal conductivity
Hydride tanks scale-up
* Corresponding author. Tel.: þ49 41 5287 26E-mail address: [email protected] (
1 Present address: MT Aerospace AG, D-862 Present address: Linde Engineering Divis
0360-3199/$ – see front matter ª 2008 Interndoi:10.1016/j.ijhydene.2008.12.028
a b s t r a c t
In this work, the effect of powder bed size on the absorption and desorption kinetics of
NaAlH4 catalyzed with TiCl4 was studied experimentally. For this purpose, volumetric
titration measurements were performed using cells of different diameters. The tempera-
ture was measured during the process at different positions inside the hydride bed,
providing detailed information about the influence of heat conduction. Experimental
results show that, under the applied conditions up to a critical size, larger diameters can
lead to faster kinetics for the first and second absorption reactions. At larger cell diameters,
however, temperatures up to 200 �C were measured during the first absorption step in the
hydride bed. This leads to a significant delay in the start of the second absorption step,
reducing the overall rate of the process. Reasons for the observed behaviour are discussed
and measures for optimization are proposed.
ª 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights
reserved.
1. Introduction in the order of 2 wt.%, while the 2010 US Department of Energy
Hydrogen is a very promising energy carrier for a compre-
hensive clean-energy concept in mobile applications.
Regarding its use as fuel for the zero-emission vehicle, one of
the main problems is the storage. Hydrogen storage systems
should fulfil the demands for automotive applications i.e. high
gravimetric and volumetric capacities, fast charging and dis-
charging rates at moderate operating conditions and high
safety levels. Metal hydrides offer a safe alternative to
hydrogen storage in compressed or liquid form. In addition,
metal hydrides have higher storage capacity by volume [1].
Room temperature hydrides are advantageous, because they
require low operating pressures and temperatures. The main
drawback is their poor gravimetric hydrogen storage density
43; fax: þ49 41 5287 2625.G.A. Lozano).153 Augsburg, Germany.ion, Linde AG, D-82049 Puational Association for H
(DOE) target is 6 wt.% for hydrogen storage systems. There-
fore, in recent years, research has focused on light weight
complex hydrides, which also have a high capacity by weight,
e.g. LiBH4 about 18 wt.%. However, LiBH4 only decomposes at
temperatures over 400 �C [2]. Sodium alanate, NaAlH4, offers
a suitable compromise with relatively large gravimetric
storage capacity at relatively moderate temperatures. Bogda-
novic and Schwickardi showed in 1996 that hydrogen can be
reversibly stored and released in sodium alanate, if doped
with a suitable catalyst [3]. NaAlH4 is reversibly formed in
a two-stage reaction within the technically favourable
temperature range of up to about 125 �C.
3NaHþAlþ 1.5H2 4 Na3AlH6 (DH¼�47 kJ/mol H2) (1)
llach, Germany.ydrogen Energy. Published by Elsevier Ltd. All rights reserved.
Fig. 1 – Cells of different inner diameters utilized for
sorption measurements in Sievert’s apparatus. Inner
diameters from left to right: 2 mm, 4.5 mm, 10 mm and
15.2 mm.
o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 3 1897
Na3AlH6þ 2Alþ 3H2 4 3NaAlH4 (DH¼�37 kJ/mol H2) (2)
Charging of the material is possible within a few minutes if
suitable catalysts and synthesis methods are used. Titanium-
based catalysts in particular have proven to be very effective,
e.g. TiCl3, TiCl4 [4,5]. Furthermore, sodium alanate can be
produced from cost-effective raw materials in kg amounts [6].
Up-scaling of the milling process is considered to be feasible
up to tonnage quantities at low costs [7].
Fundamental quantitative research on kinetics of the
absorption–desorption processes of metal hydrides is nor-
mally carried out with samples of some hundreds of mg in
standard apparatus e.g. Sievert’s apparatus. However, in order
to exploit the properties of sodium alanate in a suitable
hydrogen storage system, it is required to understand the
performance of the material in larger powder beds.
The heat transfer mechanisms and the correlation to the
hydriding behaviour of large material beds are relatively well
investigated on LaNi5 and FeTi-based transition-metal/rare-
earth metal alloys (room temperature hydrides). For some
reactor systems, the hydrogen sorption kinetics is limited by
heat transfer constraints of the reactor. In most cases,
however, the metal hydride is the principal heat transfer
resistance [8,9]. Many studies therefore focus on the deter-
mination of the effective thermal conductivity of the metal
hydride beds [9–12] and demonstrate heat transfer enhance-
ment by using extended areas for heat transfer and by binding
the hydrides into a solid matrix of high conductivity materials
[13–15] as well as the use of external fins [16]. Additionally,
there are numerous studies on dynamic metal hydride storage
systems and simulations [17–25].
On the other hand, the sorption behaviour of large beds of
complex hydrides is relatively unknown. Thermal conduc-
tivity values around 0.5 W m�1 K�1 for sodium alanate have
been reported based on static measurements in powder beds
at different pressures and degree of reaction [26,27], indicating
relatively poor thermal conductivity of light-metal hydrides
even compared to conventional transition-metal/rare-earth
metal hydrides (average values between 1 and 2 W m�1 K�1
under similar conditions are reported [9–11,14]). A brief study
about the bed temperature during the reactions in a scaled-up
test bed of NaAlH4 is reported in [28]. It was shown that poor
heat conductivity of the bed of hydride has a negative effect
on the absorption and desorption kinetics. However, the
determining mechanism for the sorption kinetics on NaAlH4
could not be clarified. Additionally, the effect of additions, e.g.
carbon, on the heat transfer in alanate-based systems is not
clear up to this point.
Therefore, the present work pursues an advanced under-
standing of the factors that influence charging and discharg-
ing kinetics in large beds of NaAlH4. In particular, we study the
effect of carbon additions, temperature and the powder bed
size on the reaction kinetics of NaAlH4 catalyzed with TiCl4.
For this purpose, titration measurements were performed
using cylindrical cells of different diameters. To understand
the underlying mechanisms, additionally, the temperature
was measured during the process at different positions inside
the bed of material, providing detailed information about the
influence of heat conduction and temperature on the reaction
kinetics.
i n t e r n a t i o n a l j o u r n a l o f h y d r
2. Experimental procedures
2.1. Materials’ production
The material used for the investigations was produced in
industrial milling equipment optimised towards fast kinetics
[6]. As raw materials, commercial NaH (95%, Sigma Aldrich
Chemie GmbH, Steinheim, Germany), aluminium (99.5%,
Johnson Matthey GmbH & Co. KG, Karlsruhe, Germany) and
2 mol.% aluminium reduced TiCl4 (3TiCl3þAlCl3) (Fluka
Chemie GmbH, Buchs, Switzerland) as catalyst and carbon
powder (Alpha Aesar) as milling agent were used. All handling
including milling was carried out in a glove box with purified
argon atmosphere. Prior to milling, NaH and aluminium were
mixed with a tumbling shaker for 0.5 h in a molar ratio of
1.08:1 according to the reaction
1.08NaHþAlþ 0.02(TiCl3þ 1/3AlCl3) /
NaHþ 1.0067Alþ 0.08NaClþ 0.02Tiþ 0.04H2 (3)
which is expected to occur during milling. Milling was
carried out in a modified vibratory tube mill (ESM 236, Siebt-
echnik, Muhlheim a. d. R., Germany), using 30 mm hardened
balls at a rotational speed of 1000 rpm and a ball-to-powder
ratio (BPR) of 140:1. In order to avoid agglomeration, 100 ml of
cyclohexane was added prior to milling. After milling for 2 h,
the mill was evacuated to dry the samples (batch 1). In order to
investigate the influence of graphite on the hydrogenation
behaviour, a second batch was prepared in the same way as
described before. After ball milling an additional 5 wt.% of
carbon (graphite powder, Alfa Aesar GmbH & Co. KG, Karls-
ruhe, Germany) was added to the second batch and the
composite milled for another 30 min (batch 2).
2.2. Sorption kinetics and cells
The sorption kinetics was characterized by a volumetric
method using a Sievert’s apparatus (HERA, Quebec, Canada),
utilizing cylindrical sample cells of different sizes (Fig. 1).
During all the experiments the external temperature on the
wall of the cells was measured. Cells with inner diameters of
2, 4.5, 10 and 15.2 mm were used. All material used during
these experiments came from the same batch 1 or 2. The
loading of the milled material in each of the cells was done
Fig. 2 – Thermocouples’ positions in the 15.2 mm diameter cell (thermocell) for heat flow analysis of the system during
absorptions and desorptions.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 31898
inside an argon glove box. Each cell was loaded with the
milled material homogeneously up to its top and immediately
weighted afterwards (mg precision scale). The weight of the
material in the cells was around 40 mg (2 mm cell), 150 mg
(4 mm cell), 1300 mg (10 mm cell) and 5000 mg (15.2 mm cell).
For the second part of the investigation, the 15.2 mm
diameter cell was additionally equipped with 5 thermocouples,
Fig. 2, what we call a thermocell. The thermocouples used are
type K, with outer diameter 1 mm and fast response grounded
junction. The cell was designed to enable measurements up to
temperatures of 400 �C and pressures up to 200 bar. The
temperature measurements throughout the bed of the mate-
rial with this cell make it possible to analyze the heat flow in the
system during the reactions and to estimate the effective
thermal conductivity of the bed.
Fig. 3 – Hydrogen content of material from batch 1 (a) and
batch 2 (b) during 3rd hydrogen absorption with different
cells at 100 bar in the Sievert’s apparatus. The initial
temperature was 125 8C.
3. Results and discussion
3.1. Absorption in cells of different sizes
The hydrogen absorption process, which corresponds to filling
an automotive tank, is demanded to last only a few minutes
and is therefore much more critical than the discharging
process. Consequently, it will predominantly determine the
design of the heat transfer system. In order to analyze the
effect of powder bed size on kinetics, the third absorption is
compared. In this absorption the material shows almost
steady kinetics with respect to cycling. In this subsection the
analysis is focused on the effect of the cell size on the
absorption profiles. The conclusions and discussion with
respect to the cell size are valid for the results obtained with
material from both batch 1 and batch 2. The absorption
profiles of the batches 1 and 2 are further compared and
analyzed in Section 3.3 (see Figs. 11 and 12).
Fig. 3 shows the absorption profiles of both materials
(batches 1 and 2) obtained with the cells of different dimen-
sions. Although the hydrogen capacities after 20 min are
similar, the slopes of the curves, and therefore the reaction
rates, are notably different. Intermediate diameters of 4.5 mm
and 10 mm lead to the fastest kinetics. This results from
different temperature developments in the beds of material
within the cells during the hydrogenation process. Fig. 4
shows the external wall temperature with respect to absorp-
tion time. The smaller cells exhibit almost isothermal
behaviour, while for the larger cells the outer temperature
raises by about 10–20 �C. This can be explained by ineffective
heat transfer with the surrounding air and a limited heat
capacity of the vessel. The temperature rise caused by limited
heat capacity of the vessel should decrease as the ratio mass
Fig. 4 – Outer wall temperature of material from batch 1
(a) and batch 2 (b) during 3rd hydrogen absorption using
different cells at 100 bar in the Sievert’s apparatus. The
initial temperature was 125 8C.
Fig. 5 – Cross-section of any cell used in the experiments.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 3 1899
of cell wall to mass of hydride, MCell wall/MHydride, increases,
because the heat capacity of the cell acts as sink for the heat
generated from the reaction. The calculated ratio MCell wall/
MHydride is presented below in Table 1 for the cells. The ratio
was calculated from the cross-section of cell wall and the
hydride bed (Fig. 5) and their respective densities, assuming
infinite cylindrical cells. By comparing results of Table 1 with
those of Fig. 4, it can be concluded that the calculated mass
ratios relate well with the observed rise of the outer wall
temperature. The temperature does not increase in the 2 mm
Table 1 – Ratio of mass of cell wall to mass of hydride forthe experimental cells. The density of the bed material is0.6 g/cm3 and the density of the walls of the cells is 7.85 g/cm3.
Cell Innerdiameter
(mm)
Outerdiameter
(mm)
MCell wall/Mmaterial
(g/g)
1 2 8 196.3
2 4.5 8 28.3
3 10 15.2 17.1
4 15.2 25 22.3
cell, the one with the highest ratio MCell wall/MHydride, whereas
the corresponding lowest ratio in cell 3 leads to the highest
temperature response.
Additionally, due to the low effective thermal conductivity,
temperature profiles inside the bed should develop. Large
beds would therefore have their maximum temperature at
their centre.
The above outlined thermal effects explain the variance of
the performances among the cells (Fig. 3). From the thermal
and kinetics profiles in Figs. 3 and 4 it is noted that a rise in
temperature improves the kinetics at the beginning of the
process, when the first step of hydrogenation proceeds.
Consequently, it improves the hydrogenation kinetics in the
larger cells compared to the smaller cells. Indeed in cell 3, with
the lowest ratio MCell wall/MHydride and the highest rise in the
outer wall temperature, the fastest kinetics for the first step is
observed. Respectively, cell 1 having the highest ratio MCell
wall/MHydride presents no rise in temperature and the slowest
kinetics for step one. Thus, the optimum temperature for
kinetics is substantially higher than the initial temperature of
125 �C but should be lower than the equilibrium temperature
at 100 bar for the first step of hydrogenation (Eq. (1)), which
amounts to about 250–260 �C [29].
At 100 bar, however, the equilibrium temperature for the
second step of hydrogenation (Eq. (2)) is around 170 �C [29]. If
the hydride reaches a temperature higher than 170 �C after the
first step of reaction, the second step will not proceed until the
temperature of the material has decreased significantly below
that temperature. This would cause a delay between the two
steps, and may explain the delays in Fig. 3 for the larger cell
sizes. This conclusion is supported firstly by the hydrogen
content of the material during the delay, which amounts to
around 1.6 wt.% and corresponds to the theoretical hydrogen
capacity of the first step of hydrogenation, and secondly, by
the observation that the time of the delay increases with the
diameter of the cells.
In addition, the time dependence of the outer wall
temperature of cell 3, in Fig. 4, reflects the delay between the
steps. The temperature rises during the first step, almost
20 �C, and decreases when the first reaction is completed. At
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 31900
the end of the delay, the temperature increases once again.
Both increments can be explained by the exothermic hydro-
genation reactions. However, the absolute value of tempera-
ture rise on the outer wall of the cells does not explain the
delay in the reaction, because the equilibrium temperature is
neither reached nor exceeded. This indicates significantly
higher temperatures in the powder bed. Therefore, experi-
ments with the thermocell (see Section 2.2) were performed in
which the temperature throughout the bed was monitored.
3.2. Absorptions in the thermocell
Sets of experiments were carried out using the thermocell
described in Section 2.2, with material from both batch 1 and
batch 2. Fig. 6 shows the 3rd absorption of carbon containing
sodium alanate, batch 2. The initial temperature was again
125 �C. Quite similar results are obtained with the material of
batch 1, and the description presented below is valid for both
batches.
As soon as the first step of absorption starts, there is
a small peak in temperature, which occurs quite shortly for
some seconds and had not been detected from outer wall
temperature measurements. It can be attributed to hydrogen
heating up upon expansion and subsequent fast flow of heat
into the hydride. At the same time, the exothermic hydroge-
nation of NaH and Al occurs with fast kinetics, leading to
a substantial increase of temperature in the powder bed. As
expected, due to the limited heat conduction in the material,
after about 1 min a temperature profile develops with
a maximum of 210 �C in the centre of the bed (position T1).
This is still lower than the equilibrium temperature of step 1
(Eq. (1)), which is 250–260 �C at the working pressure (100 bar),
but much higher than the equilibrium temperature of step 2
(Eq. (2)), 170 �C. Consequently, after absorbing 1.6 wt.% of
hydrogen, corresponding to the completion of step 1, hydro-
genation does not continue. Therefore, temperatures in the
bed decrease due to the heat flow through the cell wall and the
surrounding air. It is observed that the second step does not
start at the point when the temperature falls below the
Fig. 6 – Hydrogen content and temperature inside the bed
of hydride during 3rd hydrogen absorption at 100 bar,
measured in the Sievert’s apparatus. The initial
temperature was 125 8C. The positions of temperature
measurement are found in Fig. 2.
equilibrium temperature of 170 �C. The temperature
decreases to 140 �C and remains almost constant in the entire
bed. Only after 2 min does further hydrogenation proceed; the
temperature increases correspondingly again and a new
temperature gradient develops throughout the bed. According
to the slower kinetics of step 2 of hydrogenation, the
temperature gradient is not as high as in step 1.
The delay between step 1 and 2, labelled as section b in
Fig. 6, cannot be entirely explained by the constraint related to
the equilibrium temperature for the formation of NaAlH4,
since during the delay the temperature in the bed is at 140 �C.
This is 30 �C lower than the corresponding equilibrium
temperature, and even at this temperature and 100 bar the
material absorbs once again after its end, when section c in
Fig. 6 starts. A possible cause is that heating of the material to
high temperatures in the first hydrogenation step has resulted
in recovery processes of the material, which impede the
nucleation of the NaAlH4 phase. In order to support this
explanation, the totally absorbed material of batch 1 and
batch 2 was desorbed at 125 �C and 5 bar. This results in the
complete decomposition of NaAlH4 into Na3AlH6 and Al with
the respective release of hydrogen, because the equilibrium
pressure for the further dehydrogenation is 3 bar [29]. The
material was then hydrogenated at different initial tempera-
tures (100 �C, 125 �C, 140 �C and 150 �C) by applying 100 bar H2.
In order to start the absorption, the pressure increase in the
cells using the experimental Sievert’s apparatus is performed
within less than a second. Interestingly, the reaction starts
immediately, as shown in Fig. 7, accompanied by temperature
increase. Consequently, the delay between the two steps has
to be caused by the condition of Na3AlH6 or Al after the first
step of hydrogenation and can, for example, be unfavourable
for the NaAlH4 nucleation process.
These findings are important for modelling, design and
layout of a storage tank based on sodium alanate. It has to be
noted that the significant heating upon hydrogenation of the
cells observed in this investigation can be avoided by efficient
cooling using water or oil. Aspiring towards cost effectiveness,
simple constructions, and the consequently large dimensions
Fig. 7 – Absorptions of Na3AlH6/Al at 100 bar H2, previously
formed by desorption of NaAlH4 at 5 bar and 125 8C. Initial
temperatures of the absorptions are indicated for each
curve (100, 125, 140 and 150 8C). T1 is the temperature in
the centre of the hydride bed.
Fig. 8 – Desorption of NaAlH4 at 5 bar and 125 8C in the
thermocell. Initial temperature is 125 8C. For a reaction
time of 10 min, the temperature profile was fitted and the
desorption rate was calculated in order to estimate the
effective thermal conductivity of the bed of material.
Fig. 10 – Estimated effective thermal conductivity of
hydride bed for NaAlH4 with and without carbon addition
during desorption at 5 bar H2 in comparison with literature
data [26]. The temperatures reported are the mean
temperatures of the bed.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 3 1901
of the powder beds, however, will always result in strong
temperature changes in the hydride as in this work. A possible
approach to overcome the detrimental effect observed before
the reaction step 2 would be to increase the cooling intensity
for a short time after the reaction step 1 is completed.
3.3. Estimation of effective thermal conductivity
A further important factor necessary for the layout of
a storage system based on hydrides is the effective thermal
conductivity. The positions of the thermocouples in the ther-
mocell in Fig. 2 were chosen in order to estimate the effective
thermal conductivity of the material bed upon the hydroge-
nation/dehydrogenation reaction in-situ. The measuring
positions were arranged in such a way that a parabolic
temperature profile throughout the bed could be read with
uniform and sufficient temperature variation between the
positions: calling DT the total difference of temperature
Fig. 9 – Example of temperature of the hydride bed with
respect to the distance from the centre corresponding to
the indicated data in Fig. 8. The fitting was used to
estimate the effective thermal conductivity of the bed of
material.
between the material close to the wall of the cell and the
material in the centre (axis) of the cell, T4 reads 0% of DT,
T3 reads 33%, T2 reads 67% and T1 reads 100% of DT. T1 is
located exactly in the centre of the bed.
In order to evaluate the measurements, a homogeneous
cylindrical system is assumed in a stationary process with
only radial conduction, no axial conduction, and a constant
and homogeneous heat generation, as well as a constant
density and heat conduction of the hydride bed. In this case
a parabolic temperature profile is obtained by the standard
heat conduction equation:
T� TC ¼ �_qrr2
4Keff(4)
where T (K) is the temperature at a radial distance r (m) from
the centric axis of the cylinder, TC(K) the temperature in the
axis (centre) of the cylinder, _q the heat generation (W/kg), r the
density of the bed of material (kg/m3), and Keff the effective
thermal conductivity of the bed of material (W m�1 K�1).
Fig. 11 – Absorptions of the material at 100 bar with and
without carbon. Initial temperature is 125 8C. T1 is the
temperature in the centre of the bed.
Fig. 12 – Desorptions of the material with and without
carbon into vacuum. Initial temperature is 125 8C. T1 is the
temperature in the centre of the bed. As noted also in other
investigations, desorption kinetics at 125 8C for the second
step is relatively slow. Therefore on this time scale, only
desorption of the first reaction step is visible.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 31902
The effective thermal conductivity of the hydride during
in-situ measurements is estimated using Eq. (4). For this
estimation, a variable m is defined according to Eq. (5) below,
such that Eq. (4) turns into Eq. (6).
m ¼ �_qr
4Keff(5)
T� TC ¼ mr2 (6)
First, m is numerically adjusted to the experimental set of data
(T� TC) vs. r2 taken during an endothermic or exothermic
reaction in a moment with constant temperature profile
(vT/vt¼ 0 inside the bed). After adjusting m, Keff is calculated
by Eq. (5). A density of the bed in the desorbed state of about
0.6 g/cm3 was thereby taken, as experimentally determined._q is calculated from the measured reaction rate and the
enthalpy of reaction, derived from PcT-measurements [29]
_q ¼ �r0Dh (7)
where r0 is the hydrogen sorption rate (gH2 g�1 s�1) and Dh is
the enthalpy of reaction (J/gH2).
The condition vT/vt¼ 0 inside the bed was not valid during
most of the experiments, but it is clearly fulfilled after about
10 min during the desorptions of NaAlH4 at 5 bar and 125 �C,
as shown in Fig. 8. From the experimental temperature
profiles and the desorption rate r0, m and _q are calculated as
above explained. The experimental and fitted temperature
profile is presented in Fig. 9. The estimations of the effective
thermal conductivity were done for the material without and
with carbon, batch 1 and batch 2 respectively. In the moment
of the estimation, the material is a mixture of NaAlH4, Na3AlH6
and Al.
The estimated heat conductivities are summarized in
Fig. 10. The values obtained with our in-situ measurements
for the material without carbon and the values reported in
the literature from static measurements at 5 bar hydrogen
pressure [26] are in very good agreement, and range between
0.5 and 0.6 W m�1 K�1. The mean temperature of the bed
during the measurements does not seem to have a distinct
influence on the values obtained. The differences between
the various measurements are within the accuracy of the
estimation.
The addition of carbon leads to a roughly 50% higher
thermal conductivity, amounting to around 0.8 W m�1 K�1.
Nevertheless, the behaviour during the absorptions and
desorptions is qualitatively the same as without carbon, both
showing a significant delay between the two steps of reactions
as shown in Figs. 11 and 12. However, carbon improves the
kinetics of both absorption and desorption.
The improvement of hydrogenation kinetics and thermal
conductivity by carbon additions is still not sufficient for
technical applications. Therefore, further investigations on
the mechanisms effective in sodium alanate with carbon
additions are of high interest.
4. Conclusions
The size of the hydride bed as well as the container has
a significant effect on the hydrogenation kinetics of NaAlH4
system. This size effect is to be taken into consideration when
designing applications where large quantities of hydrogen are
to be stored. For cells of diameters between 2 mm and
15.2 mm, different absorption curves and temperature profiles
are obtained using a Sievert’s apparatus. For an initial
temperature of 125 �C, larger diameters up to a critical size
between 4.5 mm and 10 mm lead to faster absorption kinetics
due to optimised temperatures developing for the first step of
reaction. However, high peaks in temperature are detected in
the axis of the bed of the material during hydrogenation,
particularly for larger powder bed sizes. A delay between the
two steps of formation of the alanate is present. Although
Na3AlH6 is formed and thermodynamic conditions for the
formation of NaAlH4 are present, hydrogenation does not
proceed. This delay has to be caused by the condition of
Na3AlH6 or Al after the first step and can, for example, be
unfavourable for the NaAlH4 nucleation process.
Heat conductivities of NaAlH4 are measured to be around
0.5 W m�1 K�1, which agree well with values from reported
static measurements. The addition of carbon to the NaAlH4
system leads to higher effective heat conductivity of the
hydride bed. With the addition of 5 wt.% of carbon, the
effective heat conductivity was increased by about 50%.
Furthermore, the addition of carbon slightly improved sorp-
tion kinetics.
Further work will focus on the theoretical analysis of the
requirements of engineering properties of the system and
how they affect the performance of the system, focused on
hydrogen volumetric and gravimetric capacities. The analysis
will be supported by experiments with large beds of material.
5. Outlook
The experimental work presented and the subsequent anal-
ysis have given essential information concerning the
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 4 ( 2 0 0 9 ) 1 8 9 6 – 1 9 0 3 1903
influence of heat transfer during the hydrogenation and
dehydrogenation of sodium alanate. Nevertheless, not only is
further experimental work with possible modifications of the
material required, but also fundamental work focused on the
performance of the system as a function of the physical
properties of the material and the system (heat capacity, heat
conductivity, density) is necessary. In a first approach, and
supported by the performed experiments, calculations and
simulations of a large bed system will lead to the real
requirements of these properties and how they affect the
feasibility of the system, mainly in terms of hydrogen gravi-
metric and volumetric capacities. This theoretical analysis
will guide the modification of the material as well as the
design of the necessary heat transfer system. Experimental
work with a large bed of material will validate these calcula-
tions and will show the feasibility to fulfil the requirements in
a real application.
Acknowledgments
The authors thank Mr. C. Holvey for the experimental support.
Partial funding by the European Project ‘‘Hydrogen Storage
Systems for Automotive Application’’ is gratefully acknowl-
edged by the authors.
r e f e r e n c e s
[1] Schlapbach L, Zuttel A. Hydrogen-storage materials formobile applications. Nature 2001;414:353–8.
[2] Zuttel A, Wenger P, Rentsch S, Sudan P, Mauron Ph,Emmenegger Ch. LiBH4 a new hydrogen storage material. JPower Sources 2003;118:1–7.
[3] Bogdanovic B, Schwickardi M. Ti-doped alkali metalaluminium hydrides as potential novel reversible hydrogenstorage materials. J Alloy Compd 1997;253–254:1–9.
[4] Anton DL. Hydrogen desorption kinetics in transition metalmodified NaAlH4. J Alloy Compd 2003;356–357:400–4.
[5] Sandrock G, Gross K, Thomas G. Effect of Ti-catalyst contenton the reversible hydrogen storage properties of the sodiumalanates. J Alloy Compd 2002;339:299–308.
[6] Eigen N, Gosch F, Dornheim M, Klassen T, Bormann R.Improved hydrogen sorption of sodium alanate by optimizedprocessing. J Alloy Compd 2008;465:310–6.
[7] Eigen N, Keller C, Dornheim M, Klassen T, Bormann R.Industrial production of light metal hydrides for hydrogenstorage. Scr Mater 2007;56:847–51.
[8] Bernauer O, Topler J, Noreus D, Hempelmann R, Richter D.Fundamentals and properties of some Ti/Mn based lavesphase hydrides. Int J Hydrogen Energy 1989;14:187–200.
[9] Goodell PD. Thermal conductivity of hydriding alloy powdersand comparisons of reactor systems. J Less Common Met1980;74:175–84.
[10] Sun D, Deng S. Theoretical descriptions and experimentalmeasurements on the effective thermal conductivity inmetal hydride powder beds. J Less Common Met 1990;160:387–95.
[11] Suissa E, Jacob I, Hadari Z. Experimental measurements andgeneral conclusions on the effective thermal conductivity ofpowdered metal hydrides. J Less Common Met 1984;104:287–95.
[12] Pons M, Dantzer P. Determination of thermal conductivityand wall heat transfer coefficient of hydrogen storagematerials. Int J Hydrogen Energy 1994;19:611–6.
[13] Zhang J, Fischer TS, Ramachandran PV, Gore JP, Mudawar I. Areview of heat transfer issues in hydrogen storagetechnologies. J Heat Transfer 2005;127:1391–9.
[14] Suda S, Komazaki Y, Kobayashi N. Effective thermalconductivity of metal hydride beds. J Less Common Met 1983;89:317–24.
[15] Nagel M, Komazaki Y, Suda S. Effective thermal conductivityof a metal hydride bed augmented with a copper wire matrix.J Less Common Met 1986;120:35–43.
[16] MacDonald BD, Rowe AM. Impacts of external heat transferenhancements on metal hydride storage tanks. Int JHydrogen Energy 2006;31:1721–31.
[17] MacDonald BD, Rowe AM. Experimental and numericalanalysis of dynamic metal hydride hydrogen storagesystems. J Power Sources 2007;174:282–93.
[18] Mohan G, Prakash Maiya M, Srinivasa Murthy S. Performancesimulation of metal hydride hydrogen storage device withembedded filters and heat exchanger tubes. Int J HydrogenEnergy 2007;32:4978–87.
[19] Mellouli S, Askri F, Dhaou H, Jemni A, Ben Nasrallah S. Anovel design of a heat exchanger for a metal-hydrogenreactor. Int J Hydrogen Energy 2007;32:3501–7.
[20] Botzung M, Chaudourne S, Gillia O, Perret C, Latroche M,Percheron-Guegan A, et al. Simulation and experimentalvalidation of a hydrogen storage tank with metal hydrides.Int J Hydrogen Energy 2008;33:98–104.
[21] Kumar Phate A, Prakash Maiya M, Srinivasa Murthy S.Simulation of transient heat and mass transfer duringhydrogen sorption in cylindrical metal hydride beds. Int JHydrogen Energy 2007;32:1969–81.
[22] Nakagawa T, Inomata A, Aoki H, Miura T. Numerical analysisof heat and mass transfer characteristics in the metalhydride bed. Int J Hydrogen Energy 2000;25:339–50.
[23] Muthukumar P, Madhavakrishna U, Dewan Anupam.Parametric studies on a metal hydride based hydrogenstorage device. Int J Hydrogen Energy 2007;32:4988–97.
[24] Gambini M, Manno M, Vellini M. Numerical analysis andperformance assessment of metal hydride-based hydrogenstorage systems. Int J Hydrogen Energy 2008;33:6178–87.
[25] Brown TM, Brouwer J, Samuelsen GS, Holcomb FH, King J.Accurate simplified dynamic model of a metal hydride tank.Int J Hydrogen Energy 2008;33:5596–605.
[26] Dedrick DE, Kanouff MP, Replogle BC, Gross KJ. Thermalproperties characterization of sodium alanates. J AlloyCompd 2005;389:299–305.
[27] Mosher DA, Arsenault S, Tang X, Anton DL. Design,fabrication and testing of NaAlH4 based hydrogen storagesystems. J Alloy Compd 2007;446–447:707–12.
[28] Gross KJ, Majzoub E, Thomas GJ, Sandrock G. Hydridedevelopment for hydrogen storage. In: Proceedings of 2002 U.S. DOE hydrogen program review, Golden, CO. NREL/CP-610-32405.
[29] Bogdanovic B, Brand R, Marjanovic A, Schwikardi M,Tolle J. Metal-doped sodium aluminium hydrides aspotential new hydrogen storage materials. J Alloy Compd2000;302:36–58.