How carbon affects hydrogen desorption in NaAlH4 and Ti-doped NaAlH4

Article abstract of DOI:10.1016/j.jallcom.2006.09.025

The hydrogen storage properties of doped and undoped NaAlH₄ samples are studied after mixing them with different percentages of high surface carbon. Manually mixed samples are compared with ball milled ones; it was found that manual mixing was a simple and effective way to dope NaAlH₄. A morphological and micro-structural analysis has been carried out in order to understand the effect of carbon. Carbon added samples show a marked enhancement of hydrogen desorption rate. The desorption temperature and the total hydrogen content remain almost unchanged for undoped sample. The desorption temperatures of Ti-doped samples increase with carbon content.

Full text of DOI:10.1016/j.jallcom.2006.09.025

Journal of Alloys and Compounds 437 (2007) 360–366
How carbon affects hydrogen desorption in
NaAlH4 and Ti-doped NaAlH4
C. Cento, P. Gislon∗, M. Bilgili, A. Masci, Q. Zheng, P.P. Prosini
ENEA CR Casaccia, Via Anguillarese 301, 00060 S. Maria di Galeria, Rome, Italy
Received 17 May 2006; received in revised form 25 August 2006; accepted 1 September 2006
Available online 19 October 2006
Abstract
The hydrogen storage properties of doped and undoped NaAlH4 samples are studied after mixing them with different percentages of high surface
carbon. Manually mixed samples are compared with ball milled ones; it was found that manual mixing was a simple and effective way to dope
NaAlH4. A morphological and micro-structural analysis has been carried out in order to understand the effect of carbon. Carbon added samples
show a marked enhancement of hydrogen desorption rate. The desorption temperature and the total hydrogen content remain almost unchanged
for undoped sample. The desorption temperatures of Ti-doped samples increase with carbon content.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Metal hydrides; Hydrogen storage materials
1. Introduction
Different techniques and precursors have been used to pre-
pare Ti-doped NaAlH4. A solution chemistry was followed by
Bogdanovic and Schwickardi [3] and by Jensen and co-workers
[4] to prepare the Ti-doped material. A solution of TiCl3 or
Ti(OBun)4 in THF was added to a solution or a suspension of
thehydrideinthesamesolvent. Thedopedmaterialwasobtained
by removing the solvent under vacuum.
Hydrogen storage for vehicle application is still an open
challenge. Different routes have been followed but a satisfy-
ing solution has not achieved yet. Liquid hydrogen shows good
volumetric and gravimetric density but its production cost is
relatively high. Compressed hydrogen suffers for low volumet-
ric hydrogen density whilst the use of traditional metal hydride
is detrimental in terms of gravimetric hydrogen density. High
amount of hydrogen (about 7 wt.%) can be absorbed at very
high temperatures (350–450 ◦C) in light weight metals or metal
nitrides (i.e. Mg or LiN3) [1,2]. The desorption enthalpy (about
80 kcal/mol) can be supplied burning part of the stored hydro-
gen. In such a way the available hydrogen content is reduced to
about 5wt.%.
The theoretical hydrogen content of NaAlH4 is 5.6 and
5.1 wt.% hydrogen was desorbed from NaAlH4 at 225 ◦C [5].
Material modifications induced by ball milling strongly decrease
the desorption temperature. The rate of decomposition of 2 h
milled sampled was so fast to release about 3.0 wt.% hydrogen
in 2 h at 160 ◦C [6]. Impressive results were obtained with Ti-
doped NaAlH4. Three weight percent hydrogen was desorbed
in 4 h from the 0.9 mol% doped sample at temperature as low
as 125 ◦C and still faster desorption rates have been observed
increasing the doping level [5]. Since part of the hydride reacts
irreversibly with the dopant, high dopant levels are detrimental
from gravimetrical point of view. Thus, the very positive ben-
efits on kinetics of Ti-catalyst are seriously offset by losses in
reversible H-capacity [5,7]. It has been shown that many types
of dopant precursors including non-titanium based compounds
result in enhancing sorption kinetics, among them carbon and
graphite. Zaluska et al. [6] modified NaAlH4 and Na3AlH6 by
mechanical grinding with or without carbon. Milled materials
Complex metal hydrides represent a good compromise
between hydrogen content, desorption temperature and reaction
enthalpy. Ti-doped NaAlH4 is emerging as one of the most inter-
esting candidate to efficiently store hydrogen since the discovery
of its reversibility in hydrogen desorption.
∗
Corresponding author. Tel.: +39 06 30483797; fax: +39 06 30486357.
E-mail address: paola.gislon@casaccia.enea.it (P. Gislon).
0925-8388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2006.09.025

C. Cento et al. / Journal of Alloys and Compounds 437 (2007) 360–366
361
showed enhanced desorption kinetics, and the addition of carbon
in the milling process remarkably improved their performance.
The amount of added carbon varied from 7 to 18 wt.%. It was
found that in the case of activated carbon or powdered graphite
the results did not significantly depend on the initial form of car-
bon, whilst carbon nanotubes (CNT) admixed NaAlH4 shows
higher performances [8]. In a recent study, the positive syner-
gistic effect of 10 wt.% graphite powder as co-dopant of 2 mol%
Ti-doped NaAlH4 was reported [9].
The observed phenomena were interpreted in mechanical
and electronic terms. Graphite can serve as lubricant or micro-
grinding agent through the formation of titanium carbide. The
electronic interaction of graphite ␲-electrons with Ti could
also facilitate Ti–hydrogen bond formation and consequently
enhancing Al–hydrogen bond cleavage. In the same work it was
shown that graphite in absence of titanium is essentially inactive
as a catalyst [9].
In this work the effect of high surface carbon on de-
hydrogenation and re-hydrogenation of NaAlH4 was investi-
gated. Hydrogen desorption properties of manually mixed and
high energy milled materials were compared. The effect of car-
bon as a co-dopant on the de-hydrogenation kinetics of Ti-doped
NaAlH4 was also studied.
Fig. 1. Apparatus for desorption rate measurements.
The reactor was heated by an external oven. To monitor the temperature
a thermocouple was connected to the reactor. When the thermal equilibrium
was reached hydrogen was injected into the sample holder at constant flow and
pressure changes have been quantified by a calibrated 8 MPa range pressure
transducer. An additional 0.08 MPa pressure transducer was used for a good
resolution at low pressure fields. The flow was accurately monitored by a flow
controller. The signal from the flow-meter was recorded and the volume of
flowed hydrogen obtained by numerical integration.
2. Experimental
Sodium aluminum hydride (NaAlH4, Aldrich, tech. 90%) and hydrogen
titanium ortobutylate (Ti(OBu)4, Aldrich, purity ≥97%) were used as received.
High surface area carbon (Carbon Ketjen Black, Akzo Nobel, specific surface
1250 m2/g) was dried at 120 ◦C for 12 h. The amount of added carbon varied
from 2 to 20 wt.%. Carbon was added following two methods, namely ball
milling or manual mixing. In a typical ball-milled sample NaAlH4 and carbon
were weighed to form 0.5g-batches which were loaded in a stainless steel vial
together with two stainless steel balls. The vial was evacuated and filled with
hydrogen. It was mounted in a shaker mill (SPEX 8000, CertPrep, Metuchen,
NJ) and milled for 30 min. Manually mixed samples were prepared by grinding
the powders in a mortar for 5 min. Doped samples were obtained by gently
mixing the mixture with titanium ortobutylate in its liquid form for additional
5 min. The amount of added titanium varied from 0.5 to 4 mol%.
The materials were characterized by X-ray diffraction (XRD) analysis (Mini-
flex Rigaku diffractometer) using Cu K␣ radiation.
Powders morphology was studied by scanning electron microscopy (SEM,
JEOL JSM-5510LV). Energy dispersive X-ray (EDS) analysis was performed
with an IXRF EDS-2000 System. The conditions were: accelerating voltage
25 kV, spot size 21 and working distance 21 mm.
3. Results
Ball-milled Ti-doped samples showed a very high reactivity
with air: the powder burned immediately after the evacuated vial
was opened, even if this operation was carried out in a glove
box. The reactivity of ball-milled samples induced us to find
an alternative method to prepare Ti-doped alanate. It was found
that manually grinding of the NaAlH4 powder together with the
Ti precursor was an effective method to prepare air-stable Ti-
doped alanate. There are several practical merits related to the
dry approach since it is very easy to produce large amounts of
doped material and it avoids the handling of solvents.
3.1. SEM analysis results
Thermal decomposition was monitored by the home-made apparatus illus-
trated in Fig. 1. The sample (about 200 mg-batches) was introduced in a
cylindrical glass holder. Before each measurement the sample holder was evacu-
ated and filled with nitrogen for three times and finally connected to the measure
apparatus. This latter consists in a BronkHorst F-100 flow-meter, 30 nl/h range.
The volume of desorbed hydrogen was calculated by numerical integration. A
thermocouple positioned on the sample holder was used to record the tempera-
ture. When the set temperature was reached, the sample was introduced into a
B-585 Buchi Glass Oven. Both the signals from thermocouple and flow-meter
were recorded.
Hydrogenation/de-hydrogenation cycles at various temperatures were car-
ried on in a specifically designed pressure composition isotherm apparatus (PCI)
from DeltaE, University of Calabria, Italy (www.deltae.it). About 1.0 g of sam-
ple was loaded in a cylindrical SS reactor (1.0 cm inner diameter). In order to
achieve identical starting condition, the reactor was heated at the operative tem-
perature and evacuated at low pressure (300 Pa) using a turbomolecular pump
before each hydrogenation process.
A morphological investigation has been carried out to detect
modifications in the alanate structure due to addition of carbon or
toballmilling, beforeandafterde-hydrogenation. InFig. 2, SEM
images of the NaAlH–0.5mol% Ti sample are shown before
(Fig. 2a) and after (Fig. 2b) de-hydrogenation. The very small
amount of Ti has not been revealed by micro-analysis.
In Fig. 3a and b, the mortar grinded 10% carbon sample is
shown, before and after de-hydrogenation, respectively. The car-
bon clusters can easily be recognized in Fig. 3a since their grain
size is smaller, as compared to the smoother NaAlH4 grains.
The sample with the same C content as Fig. 3 but prepared by
ball milling is shown in Fig. 4. The structures of both the NaAlH4
and C grains are no more distinguishable, the ball milling had

362
C. Cento et al. / Journal of Alloys and Compounds 437 (2007) 360–366
Fig. 2. SEM images of Ti-doped samples (a) before and (b) after de-hydrogenation.
Fig. 3. SEM images of 10% C added samples (a) before and (b) after de-hydrogenation.
the effect of “flattening” the hydride; in reality, as the XRD anal-
ysis showed (Section 3.3), the macroscopic loss of granularity
corresponds to a loss of microscopic cristallinity. The same pic-
ture at a larger view shows a columnar stratification of planes,
Fig. 4b.
The “flat aspect” disappears after the first de-hydrogenation,
and the de-hydrogenated product appears similar to the non-ball
milled one (Fig. 5).
different carbon percentage, is shown in Fig. 6. The tempera-
ture was kept constant at 220 ◦C for several hours. The effect
of the carbon content on desorption kinetic is evident: passing
from 2 to 20% carbon, the hydrogen flow increases by a factor
of 4.
In Fig. 7, the plot of the hydrogen yield versus time at 220 ◦C
for various carbon content samples is shown. Again it is evident
that increasing carbon content has the effect of speeding up the
hydrogen dissociation: after 150 min the 20% carbon samples
attained the maximum hydrogen yield, 5.1 wt.%, whilst lower
percentage carbon samples are still desorbing.
3.2. Hydrogen desorption in carbon added samples
A plot of the hydrogen flow, normalized to the sample weight,
versus time, released by un-doped, carbon added samples with
The weight percentage refers to the alanate weight only; if we
calculate the hydrogen yield with respect to the sample weight,
Fig. 4. SEM images of 10% C added ball-milled samples before de-hydrogenation at two magnifications.

C. Cento et al. / Journal of Alloys and Compounds 437 (2007) 360–366
363
Fig. 5. SEM images of 10% C added ball-milled samples and after de-hydrogenation at two different magnification.
Fig. 6. Desorbed hydrogen flow of different C content NaAlH4 samples at
220 ◦C; the flow is normalized to the alanate weight.
Fig. 8. Comparison between hydrogen flow released from a ball milled and a
just grinded 10% C added samples.
the 10% C sample shows better results in terms of gravimetric
hydrogen density than the 20% C still at a reasonable desorption
rate.
of carbon is to increase the pre-exponential factor, and conse-
quently to increase the kinetics of the reaction, with respect to
the undoped NaAlH4 [5].
The behavior of the manually grinded samples compared to
the ball milled one with the same carbon content is shown in
Fig. 8. The hydrogen flow is reduced in the just grinded sample,
although the hydrogen amount and the desorption temperature
of the two samples are very similar.
The Arrhenius plot of the flow versus temperature for the
non-doped 20% carbon added sample (Fig. 9) shows a single
decomposition step characterized by an activation energy of
158 kJ/mol/K and a pre-exponential factor of 5.3E18. The effect
It is well known that the decomposition of NaAlH4 is a two-
step reaction described by the following formula:
3NaAlH4 → Na3AlH6 + 2Al + 3H2
(1)
(2)
3
2
Na3AlH6 → 3NaH + Al + H2
Fig. 7. Desorbed hydrogen of various C content NaAlH4 samples at 220 ◦C; the
hydrogen weight percentage is referred to the alanate weight.
Fig. 9. Arrhenius plot of an undoped 20% C added sample.

364
C. Cento et al. / Journal of Alloys and Compounds 437 (2007) 360–366
Fig. 10. XRD spectrum of a 10% C added sample after decomposition at 195 ◦C.
Fig. 11. XRD spectrum of a 10% C added sample after decomposition at 220 ◦C.
To detect only the first decomposition step of NaAlH4 we set the
temperature at 195 ◦C and analyzed the decomposition reaction
products by XRD. The hydrogen released by the first step of
reaction was 3.7 wt.%.
3.4. X-ray diffraction in carbon-added Ti doped samples
X-ray diffraction spectra for a mixture of NaAlH4 containing
5 wt.% carbon and doped with 0.5 mol% Ti shows exclusively
crystalline NaAlH4. The absence of the peaks related to titanium
metal does not allow to recognize whether Ti is in its crystalline
state or not. The absence of peaks attributed to carbon is not
surprising since the carbon we used is amorphous.
3.3. X-ray diffraction of carbon added samples
The progressive turning from NaAlH4 into two different
Na3AlH6 phases, the ␣-Na3AlH6 and ␤-Na3AlH6, and succes-
sively into Al and NaH, was clearly detected by X-ray analysis.
In Figs. 10 and 11, the X-ray diffraction of the desorption
products of the 10% carbon ball-milled sample after decompo-
sition, respectively at 195 and 220 ◦C, are shown. In the low-
temperature reaction (Fig. 10) the NaAlH4 completely decom-
posed into Al and Na3AlH6 (␣ and ␤ phases) following Eq. (1);
after a further increase in temperature (Fig. 11), together with a
small amount of Na3AlH6, the peaks of Al and NaH appear, as
predicted by Eq. (2).
The spectra for the ball-milled sample, before any de-
hydrogenation process, evidences that the macroscopic amor-
phous structure seen by SEM corresponds effectively to an
absence of microscopic structure: the peak intensity related to
NaAlH4 was strongly decreased and Al was found in the spec-
trum. The presence of Al evidences that some decomposition
reaction occurred during the ball milling.
3.5. Hydrogen desorption in Ti-doped carbon added
samples
As stated by different authors [4,5], Ti doping is a very effec-
tive way to improve performance and de-hydrogenation kinetic.
We tested therefore if the beneficial effect found by adding car-
bon to the alanates is effective also for Ti-doped samples.
In Fig. 12a, the desorbing curves of four 0.5 mol% Ti samples
added, with0, 5, 10and20 wt%carbonarecompared. According
to Sandrock et al. [5] the Ti-doped sample without carbon shows
two well defined decomposition step. The addition of carbon
results in an increase of flow that again reaches its maximum
for the sample containing 10 wt.% carbon. At the same time the
second decomposition step disappeared and it seems that the
reaction follows a single decomposition pathway.
Fig. 12. Ti-doped, C added NaAlH4 samples in: (a) the desorbed hydrogen flow is shown for different C percentages; (b) the corresponding hydrogen content is
plotted. Temperature of measurement: 200 ◦C.

C. Cento et al. / Journal of Alloys and Compounds 437 (2007) 360–366
365
Fig. 13. Arrhenius plot of 0.5 mol% Ti-doped samples for different C added
percentages; the 0% carbon curve shows the typical behavior with two different
activation energies corresponding to the two decomposition steps.
Fig. 15. Hydrogen absorbed by a 10% C sample during a PCT cycle, with a
constant pressure ramp up to 80 bar at various temperatures.
The total amount of hydrogen evolved versus time is plotted
in Fig. 12b. The temperature was raised up to 200 ◦C, so only
the first reaction step is likely to occur; the hydrogen yield is
therefore only 3.8%.
the sample was completely de-hydrogenated by pumping it at
230 ◦C till a 0.1 mbar pressure was attained.
The Arrhenius plots for three 0.5 mol% Ti-doped samples at
various carbon content, 0, 5 and 20 wt%, are plotted in Fig. 13.
By increasing the carbon content both the activation energy
and pre-exponential factor are increased with respect to the no-
carbon, Ti-doped samples.
Differently from the un-doped sample, the desorption tem-
peratures are strongly affected by carbon content. The effect of
the carbon content on the doped sample desorption temperature
can be argued from Fig. 14, where the flow curves of Fig. 12a
are plotted versus temperature: increasing the carbon percent-
age from 0 to 20% has the effect to move the desorption onset
at higher temperatures, from 140 to 160 ◦C.
The absorption and desorption curves of the 10% C added
sample at various temperatures are shown in Figs. 15 and 16,
respectively, the sample absorbed reversibly about 5 wt.%
hydrogen. It is worth to note that the pure alanate do not exhibit
any re-hydrogenation property at those conditions. For the 2.5 h
charging time, the rate of hydrogenation as well as the total
amount of H2 absorbed does not seem to be affected by tem-
perature. A little change in the absorption rate is observed at
the end of process and it causes a little variation in the amount
of absorbed hydrogen with temperature. The desorption curves
show a more pronounced dependence on temperature: the des-
3.6. Hydrogenation and cyclability
The PCT apparatus has been used with P = 80 bar maxi-
mum pressure, F = 40 ncc/min hydrogen flow, temperatures in
the 200–230 ◦C range. Before any subsequent re-hydrogenation,
Fig. 16. Hydrogen desorbed by a 10% C sample during a PCT cycle, with a
constant pressure ramp down from 80 bar at various temperatures.
Fig. 14. Desorption temperatures for doped C added NaAlH4.

366
C. Cento et al. / Journal of Alloys and Compounds 437 (2007) 360–366
orbed hydrogen amount is larger at higher temperature, and
larger is also the desorption rate.
increase of the activation energy of the process. It could be due
to some interaction of Ti with carbon that decreases the amount
of “active” titanium or reduces Ti mobility. As stated for non-
dopedsamples, thepresenceofcarbonondopedsamplesinhibits
the grain aggregation of NaAlH4.
4. Discussion
Beneficial effects in both hydrogen desorption rate and des-
orption temperature due to carbon presence have been detected
by different authors [6,8]. How the carbon could enhance the
alanate performance is anyway still an open question. Mor-
phological analysis shows that the aggregate structure of the
de-hydrogenated samples has been modified by carbon. The
presence of carbon produces rough particles. By examining
every single particle it is possible to observe that it is an aggre-
gation of particles of smaller diameter (few micron); anyway
X-ray spectra after de-hydrogenation show that no modification
occurred in the crystalline structure of alanate. The role of car-
bon could be more “physical” than “chemical” in the sense that it
modifies the grain surface of the de-hydrogenated phases prob-
ably due to the inhibition of grain aggregation. The increase of
the surface/volume ratio upon the first de-hydrogenation could
enhance the second de-hydrogenation step. This is in accordance
with the obtained results, which show that only the desorption
rate is improved by carbon addition, being the desorption tem-
perature and the hydrogen yield only slightly affected.
The effect of carbon to increase the hydrogen desorption rate
is noticeable for carbon content as high as 10 wt.%. By dou-
bling this amount only a slight change in the hydrogen rate
was observed. The amounts of hydrogen desorbed from 10 to
20 wt.% carbon added samples was almost the same. It is easy
to recognize a “threshold” behavior appearing when the car-
bon content lies between 5 and 10%. By considering that large
amount of dopant is detrimental in term of gravimetrical hydro-
gen density, the 10 wt.% sample can be considered the best
compromise between the de-hydrogenation rate and the hydro-
gen yield.
With respect to what previously found [9] our results pointed
out that the effect of carbon, independently from its allotropic
state, has beneficial effect on both Ti-doped or undoped NaAlH4
decomposition. The main differences between what has been
found by Wang et al. by adding graphite instead of high surface
carbon [9] and our results are in the lower decomposition tem-
perature; a very light effect on it should be ascribed to the ball
milling of the sample.
5. Conclusions
In this paper, the positive effect of carbon on hydrogen
absorbtion/desorption process has been shown. Carbon acts with
a threshold behavior and 10 wt.% carbon was found the best
carbon content. The 10% carbon sample, obtained by manual
mixing carbon and NaAlH4, showed to reversibly desorption of
hydrogen to 4 wt.% in the temperature range between 200 and
300 ◦C. The smaller grain size noted for carbon-doped materi-
als could enhance the rate of the second de-hydrogenation and
the following hydrogenation/de-hydrogenation steps. Syner-
gismwasfoundbetweencarbonandTiindouble-dopedsamples.
It follows that the two dopants act with different mechanisms.
The higher desorption temperature showed from Ti-doped sam-
ples mixed with carbon was related to a decrease of the amount
of available titanium.
References
[1] H. Imamura, K. Masanari, M. Kusuhara, H. Katsumoto, T. Sumi, Y. Sakata,
J. Alloys Compd. 386 (2005) 211–216.
[2] P. Chen, Z. Xiong, J. Luo, J.L. Lin, K.L. Tan, Nature 420 (2002) 302–304.
[3] B. Bogdanovic, M. Schwickardi, J. Alloys Compd. 253–254 (1997) 1.
[4] S.S. Srinivasan, H.W. Brinks, B.C. Hauback, D. Sun, C.M. Jensen, J. Alloys
Compd. 377 (2004) 283–289.
[5] G. Sandrock, K. Gross, G. Thomas, J. Alloys Compd. 339 (2002) 299–308.
[6] A. Zaluska, L. Zaluski, J.O. Strom-Olsen, J. Alloys Compd. 298 (2000)
125–134.
[7] J.M. Bellosta von Colbe, B. Bogdanovic, M. Felderhoff, A. Pommerin, F.
Schuth, J. Alloys Compd. 370 (2004) 104–109.
[8] D. Pukazhelvan, J. Alloys Compd. 403 (2005) 312–317.
[9] J. Wang, A.D. Ebner, T. Prozorov, R. Zidan, J.A. Ritter, J. Alloys Compd.
395 (2005) 252–262.
Also in Ti-doped materials the rate of the second de-
hydrogenation step is increased after carbon was added. As a
result the two desorption steps clearly visible in the Ti-doped
material, disappeared in presence of carbon (Figs. 12 and 13)
and a single desorption process was observed in carbon added
titanium-doped NaAlH4 samples. The temperature of the first
desorption step is slightly increased by increasing the amount
of carbon. This could be explained by considering that Ti-doped
samples are very sensitive to the amount of Ti used as cata-
lyst [5]. Carbon inhibits the effect of the titanium through an