Reduction of hydrogen desorption temperature of ball-milled MgH2 by NbF5 addition

Article abstract of DOI:10.1016/j.jallcom.2007.09.130

Enhanced sorption properties of ball-milled MgH₂ are reported by adding NbF₅. Among various catalyst amounts, 2 mol% of NbF₅ reveals to be the optimum concentration leading to significant reduction of the desorption temper

Full text of DOI:10.1016/j.jallcom.2007.09.130

Journal of Alloys and Compounds 464 (2008) 377–382
Reduction of hydrogen desorption temperature of
ball-milled MgH2 by NbF5 addition
N. Recham, V.V. Bhat, M. Kandavel, L. Aymard, J.-M. Tarascon, A. Rougier∗
Laboratoire de Re´activite´ et Chimie des Solides, UMR 6007, Universite´ de Picardie Jules Verne, 33 rue St. Leu, 80039 Amiens, France
Received 3 June 2007; received in revised form 21 September 2007; accepted 29 September 2007
Available online 5 October 2007
Abstract
Enhanced sorption properties of ball-milled MgH2 are reported by adding NbF5. Among various catalyst amounts, 2 mol% of NbF5 reveals to
be the optimum concentration leading to significant reduction of the desorption temperature as well as faster kinetics of ball-milled MgH2. At
200 ◦C, temperature at which MgH2 does not show any activity, MgH2NbF /2 mol% composite desorbs 3.2 wt.% of H2 in 50 mins. Interestingly, the
5
addition of NbF5 is also associated with an increase in the desorption pressure. At 300 ◦C, MgH2NbF /2 mol% composite starts to desorb hydrogen at
5
600 mbar in comparison with 1 mbar for MgH2. Further improvements were successfully achieved by pre-grinding NbF5 prior to ball-milling the
catalyst with MgH2. Such pre-ground NbF5 catalyzed MgH2 composite desorbs 3 wt.% of H2 at 150 ◦C. Improved properties are associated with
smaller activation energies down to values close to the enthalpy of formation of MgH2. Finally, the mechanism at the origin of the enhancement is
discussed in terms of catalyst stability, MgF2 formation and electronic density localization.
© 2007 Elsevier B.V. All rights reserved.
Keywords: MgH2; Hydrogen storage; Catalyst; NbF5; Mechanism
1. Introduction
leads to a significant desorption of 4.5 wt.% at 200 ◦C in ball-
milled nanocrystalline MgH2, conditions for which commercial
Nb2O5 (SBET < 1 m2/g) do not show any catalytic activity
[17].
Mg-based alloys have been extensively investigated as
regards their interesting hydriding properties [1,2]. MgH2 is a
low cost and non-toxic material with a high, 7.6 wt.%, hydro-
genstoragecapacity. However, thehighthermodynamicstability
of MgH2 (−75 kJ/mol H2) limits the practical application of
MgH2 for PEM-fuel cells that operate around 80–100 ◦C. Sev-
eral ways have been reported to enhance the sorption kinetics
of magnesium-based hydrides including particle size reduction
by high-energy ball-milling [3] and carbon addition [4]. A suc-
cessful route remains the addition of catalysts [5] among which
pure transition metals [6–8], carbides nitrides [9] and oxides
[10,11] are often reported. Among the transition metal oxides,
in agreement with its remarkable catalytic effect, Nb2O5 has
received extensive interest [12–14]. Significant improvement
in the kinetics and further reduction of the desorption tem-
perature were observed by using nanometric catalyst [15,16].
The addition of high surface area Nb2O5 (SBET ≈ 200 m2/g)
Although, the various studies clearly demonstrate the Nb2O5
superior activity, the exact role played by the additive and the
importance of its oxide chemical nature are still being dis-
cussed. Indeed, in our group, faster kinetics were reported for
MgH2NbCl composite as compared with the MgH2Nb O one
5
2
5
[18]. Recently, the benefits of using halides catalysts have been
largely confirmed [19–24]. Through the examples of Niobium-
based additives, the present paper illustrates the advantage of
using fluoride catalyst as compared to oxide ones. Further
improvement in H-desorption kinetics and reduction in desorp-
tiontemperatureareachievedbyball-millingNbF5 catalystprior
to its addition to MgH2.
2. Experimental
Commercial MgH2 (Aldrich) and its mixtures with a calculated amount of
catalyst were milled under an argon atmosphere in a 50 ml hardened stainless
steel container (powder/ball ratio is 1/16) for 25 h using a Retsch planetary
ballmiller, and named as (MgH2)bm and (MgH2)A/Xmol%, where ‘A’ and ‘X’
represent the type of catalyst and its concentration, respectively. Niobium-based
∗
Corresponding author.
E-mail address: aline.rougier@u-picardie.fr (A. Rougier).
0925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2007.09.130

378
N. Recham et al. / Journal of Alloys and Compounds 464 (2008) 377–382
catalysts, namely, Nb2O5, NbCl5 and NbF5, from commercial origin, were
initially added in the 0.1–2 mol% range.
Samples were characterized by X-ray diffraction (XRD) using a Philips
˚
diffractometer (θ − 2θ configuration and λ (Cu K␣, 1.5418 A)). Scanning elec-
tron microscopy (SEM) was carried out on Philips XL 30 field emission gun FEG
microscope equipped with an EDX analysis detector. The differential scanning
calorimetric (DSC) and thermogravimetric measurements were performed, in
Ar using an alumina crucible, on a METTLER DSC 25 and METTLER TG 50
apparatus, respectively.
The sorption properties were investigated using a Hiden IGA thermobalance.
Hydrogen absorption was performed under continuously increased hydrogen
pressure up to 15 bar and desorption was carried out under primary vacuum
(10−1 mbar) and operating at different temperatures. As the samples were trans-
ferred in air, the following performances in terms of capacity and kinetics may
be underestimated.
3. Results
After milling, (MgH2)bm has the tetragonal structure of ␤-
MgH2 (S.G., P42/mnm) (Fig. 1). Traces of metastable ␥-MgH2
(S.G., Pbcn), together with a small percentage of Mg (S.G.,
P63/mmc) and MgO (S.G., Fm3m) impurities are also detected
Fig. 2. Hydrogen desorption curves of ball-milled (MgH2)catalyst composites at
300 ◦C in vacuum.
on the XRD patterns of both (MgH2)bm and (MgH2)catalyst
.
Regardless of the counter ions, at 300 ◦C in vacuum, faster
kinetics are recorded for catalyzed (MgH2) as compared to
(MgH2)bm (Fig. 2). Among the various commercial niobium-
based catalysts, remarkable improvement is shown by adding
NbF5 with a hydrogen desorption of 5.6 wt.% in 25 min
Table 1
Desorption rate of (MgH2)NbF /X mol% (X = 0.2, 1, 2) composites
5
NbF5 (mol%)
0.2
Desorption rate (g/s/100 kg)
Desorption limits (min)
2.8 (first cycle)
3.5 (third cycle)
3.7 (first cycle)
4.1 (first cycle)
5//30
5//30
5//20
5//20
for (MgH2)
[24] and a corresponding desorption
1
2
NbF5/0.2 mol%
rate, calculated between 5 and 30 min at the third cycle, of
3.5 g/s/100 kg (Table 1). As shown in Table 1, the desorption
rate increases during the very first cycles (2.8 g/s/100 kg on the
first cycle <3.5 g/s/100 kg on the third cycle). This activation
period lasting from two to three cycles is slightly shorter for
(MgH2)catalyst than for (MgH2)bm. An increase in niobium flu-
oride content up to 2 mol% leads to faster kinetics (i.e. higher
0.2 mol% of NbF5 is added. Similar 2% optimum amount of
NbF5 was also reported by Luo et al. [22].
Despite its importance for practical applications, the influ-
ence of the additive on the desorption pressure is scarcely
addressed. Interestingly, at300 ◦C, (MgH2)
compositestarts
desorption rates). At 300 ◦C in vacuum, (MgH2)NbF /2 mol% des-
NbF5
5
to desorb above 400 mbar as compared to 10 mbar for (MgH2)bm
(Fig. 3). This increase is rather promising. Similarly, we recently
orbs 5.4 wt.% in 12 mins as compared with 3.9 wt.% if only
Fig. 3. Hydrogen desorption pressure of (MgH2)bm and (MgH2)NbF /2 mol%
Fig. 1. X-ray diffraction patterns of (MgH2)bm and as-milled (MgH2)catalyst
composites.
5
composites at 300 ◦C.

N. Recham et al. / Journal of Alloys and Compounds 464 (2008) 377–382
379
gen amount, namely 0.2 wt.% in 45 min is also desorbed at
150 ◦C. This trend highlights that besides the faster desorption
of the sample, the desorption of MgH2 is feasible at much lower
temperatures than thermodynamics predict.
As largely illustrated in the literature, sorption properties are
generally highly improved by using nano and/or high surface
area catalyst [15–17,25]. Commercial NbF5 was milled for 1 h
using the same grinding conditions as above (argon atmosphere,
Retsch planetary ball-miller and 1/16 powder/ball ratio). After
1 h of grinding, ball-milled NbF5, named as NbF5 1 h b.m.,
exhibits a featureless XRD pattern. Longer grinding times result
in the formation of NbF3. 2 mol% of NbF5 1 h b.m. catalyst was
then mixed with commercial MgH2.
The comparison of the hydrogen desorption curves at
various temperatures (T < 250 ◦C) shows faster kinetics and
slightly higher capacity for (MgH2)NbF 1 h b.m./2 mol% than for the
5
(MgH2)NbF /2 mol% composites (Fig. 5). At 250, 200 and 150 ◦C
5
Fig. 4. Evolution of the hydrogen desorption curves of (MgH2)NbF /2 mol% com-
posites in vacuum with decreasing desorption temperature.
(MgH2)
desorbs nearly 4.4, 3.8 and 2.8 wt.%
5
NbF51 h b.m./2 mol%
in 40, 60, and 250 min, respectively. This result shows that at
150 ◦C the addition of NbF5 ball-milled 1 h is very promising,
even if the desorption kinetics need to be improved.
The activation energy of H-desorption for catalyzed MgH2
was calculated using the Kissinger plot (Fig. 6). For clarity rea-
sons, Fig. 6 does not show the ln(β/T2) = f(1000/T) curves, where
β corresponds to the heating rate used in the DSC measurements
for all (MgH2)catalyst. However, all plots confirm that any cata-
lyst addition leads to smaller activation energy than pure MgH2.
reported an increase in the desorption pressure up to 1 bar for
MgH2 catalyzed using high surface area Raney Ni [25].
Desorption properties were investigated down to 150 ◦C.
(MgH2)bm and (MgH2)catalyst/0.2 mol% composites did not show
any activity below 250 ◦C. On the contrary, (MgH2)
NbF5/2 mol%
composite desorbs nearly 3.8 and 3.2 wt.% in 35 and 50 min
at 250 and 200 ◦C, respectively (Fig. 4). A very small hydro-
Fig. 5. Hydrogen desorption curves of (MgH2)NbF /2 mol% composites in vacuum with decreasing desorption temperature 250, 200 and 150 ◦C.
5

380
N. Recham et al. / Journal of Alloys and Compounds 464 (2008) 377–382
Fig. 7. DSC profiles for (MgH2)bm and (MgH2)NbF /2 mol%. The heating rate
5
Fig. 6. The Kissinger plot for the hydrogen desorption reaction for (MgH2)bm
was 10 ◦C/min.
and (MgH2)NbF 1 h b.m./2 mol%. The heating rate ␤ used in DSC measurements
5
was 5, 10, 50 and 80 ◦C/min.
ites, however its amount cannot be properly quantified by this
technique.
The activation energy value of (MgH2)NbF /2 mol% composite of
5
88 kJ/mol is similar to the one of 90 kJ/mol reported by Luo et
al. [22]. Confirming, the more pronounced catalytic effect of
ball-milled NbF5, the activation energy decreases from 88 to
4.2. XRD structural and textural characterizations
To find out the eventual transformation of NbF5 for as-milled
and cycled (MgH2)catalyst composites, XRD measurements were
performed on cycling (Fig. 8). After the sorption (i.e. desorption
and absorption) cycles, MgH2 composites exhibit sharper ␤-
MgH2 X-ray peaks than as-milled one. The crystallite size of
sorbed composite has increased from 10 to about 50 nm while
the lattice strain has decreased from 10−2 down to 3 × 10−3. Due
to irreversibility in the absorption process, the XRD patterns of
cycled (MgH2)catalyst show the presence of Mg peaks. As the
NbF5 content increases, the sample becomes more reactive and
the Mg/MgH2 ratio increases. Interestingly after sorption cycle,
72 kJ/mol for (MgH2)
composite.
NbF51 h b.m./2 mol%
4. Discussion
In this work, comparing niobium-based catalysts, NbF5 is
presented as the most promising for improving MgH2 hydro-
gen desorption/absorption reaction. However, NbF5 melts at
81 ◦C and evaporates at 236 ◦C. Those temperatures are rather
low compared with the cycling temperature (300 ◦C), raising
therefore questions on the NbF5 stability upon cycling.
4.1. Thermal stability
To outline the effect of the NbF5 catalyst on the decompo-
sition temperature of MgH2, the composites were characterized
by DSC measurements. (MgH2)bm exhibits two endothermic
peaks at 410 and 435 ◦C (Fig. 7). The reason for the appear-
ance of two peaks above 400 ◦C is unclear and may be nested in
particle size sample inhomogeneity. MgH2 composite exhibits
a single peak shifted towards lower temperature (T ≈ 360 ◦C).
The slight exothermic peak at higher temperature corresponds
to the formation of MgO. The absence of peaks below 300 ◦C,
temperature below which NbF5 shows two peaks at 81 and
250 ◦C corresponding to the melting and vaporisation, sug-
gests that either in the as-milled MgH2 composite NbF5 has
been already transformed or it may be “protected”, respec-
tively. Such behavior was confirmed by thermogravimetric
analysis showing no weight loss below 300 ◦C for MgH2 com-
posite. Chemical analysis by EDS measurements indicate a
similar Nb/Mg ratio in the as-milled and after cycling com-
posites, corresponding to the starting mixture. The presence
of F is clearly detected in all as-milled and cycling compos-
Fig. 8. X-ray diffraction patterns of (MgH2)NbF /2 mol% composites as milled,
after one desorption and after three cycles.
5

N. Recham et al. / Journal of Alloys and Compounds 464 (2008) 377–382
381
the MgF2 phase is present with a proportion that increases both
on cycling and as the NbF5 content increases. This observation
raises the question of when the MgF2 phase is formed and is
it associated to a complete reduction of NbF5 into Nb or is it
a partial transformation? From our XRD data, the presence of
MgF2 in as-milled composite is hardly detectable, neither is the
one of NbF5. The formation of MgF2 is already visible after
a single dehydrogenation. Indeed, such behavior was already
reported and the formation of MgF2 was further supported by
XPS measurements [22].
Based on various observations, including a not as marked
enhancement in the sorption properties of ball-milled MgH2
when using Nb or MgF2 as catalysts, Luo et al. argue that Nbxt
containing compounds instead of Nb mainly contribute to the
overall sorption behavior [22]. Interestingly, from an evalua-
tion of the hydrogen sorption kinetics using an optical indicator
technique, Borgschulte et al. reported an increase in the cat-
alytic effect of NbOx oxides with an increase in x, reaching a
maximum of the activation energy for x = 2.5 (i.e. Nb2O5) [26].
The catalytic effect of Nb2O5 was discussed by Friedrichs et al.
claiming a more complex mechanism involving Nb2O5 decom-
position on the first heating leading to the local formation of
Nb, MgO and a mixed Mg–Nb oxide; therefore Nb2O5 cannot
act as a catalyst. The enhancement in the sorption properties
would therefore be linked to the formation of pathways of nio-
bium oxide species with lower oxidation state facilitating the
hydrogen transport into the sample [27]. The existence of an
intermediate Mg–Nb-oxide layer, helping in the splitting of H2
was also discussed by Gijis Schimmel et al. [28].
The reduction of Nb2O5 or NbF5 by metallic Mg being ther-
modynamically possible, we investigated the possibility of the
existence of Mg–Nb–F intermediate compound forming during
the desorption process when Mg is formed. Various equimo-
lar mixtures of Mg/NbF5 and MgH2/NbF5 were ball-milled and
heated at 300 ◦C for 2 h. At ambient temperature and 300 ◦C,
Mg reacts with NbF5 to form an MgNbF7 compound [29]. Mg
and NbF5 (at ambient temperature) or Mg (at 300 ◦C) are also
present in small proportions. Except for the presence of MgO,
the MgH2/NbF5 mixture remains as it is after heat treatment at
300 ◦C. AsimilarbehaviorwasobservedforNb2O5 catalystwith
the formation of Mg4Nb2O9 [30] for the Nb2O5/Mg mixture
whereas Nb2O5/MgH2 remains unchanged after heat treatment.
The catalytic property of MgNbF7 is under investigation.
Finally, it should be recalled that NbF5 is unstable in air and
reactstoformawhitecompound, NbO2F. Interestingly, weexpe-
Fig. 9. Hydrogen desorption curves of (MgH2)Nb O /0.1 mol%
,
2
5
(MgH2)NbF /0.2 mol% and MgH2 composites at 300 ◦C in vacuum.
5
ing of the surface Mg␦+ H␦− bonds. As the electronegativity
of fluorine is greater than that of oxygen, this electronic delo-
calization effect should be more pronounced for F than for O,
hence leading to a more pronounced effect for F-based catalyst,
in agreement with our data.
5. Conclusion
In summary, in this paper, the influence of the counter-
ion on niobium-based catalysts comparing oxides and halides
on the hydrogen desorption properties of ball-milled MgH2 is
investigated. We show faster kinetics and reduced hydrogen
desorption temperature for halides catalysts with remarkable
benefits by using fluorides. Working with an optimum compo-
sition of 2 mol% of NbF5, further improvement was achieved
by pre ball-milling the NbF5 prior to its addition to MgH2
with a reduction of 150 ◦C of the desorption temperature. At
150 ◦C, MgH2 composite desorbs 3 wt.% exhibiting a very
promising behavior. Preliminary understanding of the mecha-
nism based on various observations such as the formation of
MgF2, the existence of an intermediate phase or the larger elec-
tronic delocalization of the Nb X bond as the electronegativity
of X anion increases. In order to clarify the key role played by
NbF5, further investigations, such as XPS measurements, are in
progress.
rienced both (MgH2)
and (MgH2)
NbF5/0.2 mol%
NbO2F/0.2 mol%
show similar desorption properties (Fig. 9) indicating that the
presence of a single fluoride anion is associated with a strong
enhancement of the MgH2 performances. Although the cat-
alytic mechanism seems to govern the enhanced reactivity of
Mg towards hydrogen sorption/desorption a legitimate question
still resides on why fluorides or even oxy-fluorides performed
better? A likely assumption for the improvement in sorption
properties by NbF5 addition, could be nested on the formation of
an intermediate H␦− Nb␦+ bond favored by the large electronic
delocalization of the Nb F bond (i.e. high electronegativity of
the X anions in NbyX5 catalyst), leading to a further weaken-
References
[1] L. Schlapbach, A. Zu¨ttel, Nature 414 (2001) 353–358.
[2] A.M. Seayad, D.M. Antonelli, Adv. Mater. 16 (2004) 765–777.
[3] A. Zaluska, L. Zaluski, J.O. Strom-Olsen, J. Alloys Compd. 289 (1999)
197–206.
[4] R. Janot, L. Aymard, A. Rougier, G.A. Nazri, J.-M. Tarascon, J. Mater. Res.
8 (2003) 1749–1752.
[5] B. Sakintuna, F. Lamari-Darkrim, M. Hirscher, Int. J. Hydrogen Energy 32
(2007) 1121–1140.
[6] G. Liang, J. Huot, S. Boily, A.-V. Nestea, R. Schulz, J. Alloys Compd. 292
(1/2) (1999) 247–252.

382
N. Recham et al. / Journal of Alloys and Compounds 464 (2008) 377–382
[7] J. Charbonnier, P. De Rango, D. Fruchart, S. Miraglia, N. Skryabina, J.
[19] L.-P. Ma, P. Wang, H.-M. Cheng, J. Alloys Compd. 432 (2007) L1–L4.
[20] S. Deledda, A. Borissova, C. Poinsignon, W.J. Botta, M. Dornheim, T.
Klassen, J. Alloys Compd. 404–406 (2005) 409–412.
[21] J.F.R. de Castro, A.R. Yavari, A. LeMoulec, T.T. Ishikawa, W.J.F. Botta, J.
Alloys Compd. 389 (2005) 270–274.
Huot, B. Hauback, M. Pitt, S. Rivoirard, J. Alloys Compd. 404–406 (2005)
541–544.
[8] J.-L. Bobet, E. Akiba, B. Darriet, Int. J. Hydrogen Energy 26 (2001)
493–501.
[9] W. Oelerich, T. Klassen, T. Bormann, J. Alloys Compd. 322 (1/2) (2001)
[22] Y. Luo, P. Wang, L.-P. Ma, H.-M. Cheng, J. Alloys Compd. 453 (2008)
138–142.
L5–L9.
[10] W. Oelerich, T. Klassen, T. Bormann, J. Alloys Compd. 315 (1/2) (2001)
237–242.
[23] Y. Luo, P. Wang, L.-P. Ma, H.-M. Cheng, Scripta Materiala 56 (9) (2007)
765–768.
[11] C.X. Shang, M. Bououdina, Y. Song, Z.X. Guo, Int. J. Hydrogen Energy
29 (2004) 73–80.
[12] G. Barkhordarian, T. Klassen, R. Bormann, Scripta Mater. 49 (2003)
213–217.
[13] N. Hanada, T. Ichikawa, S. Hino, H. Fujii, J. Alloys Compd. 420 (1/2)
(2006) 46–49.
[14] G. Barkhordarian, T. Klassen, R. Bormann, J. Alloys Compd. 407 (2006)
249–255.
[15] O.O. Friedrichs, T. Klassen, J.C. Sanchez-Lopez, R. Bormann, A. Fernan-
dez, Scripta Mater. 54 (2006) 1293.
[16] N. Hanada, T. Ichikawa, H. Fujii, J. Alloys Compd. 404–406 (2005)
716–719.
[17] V.V. Bhat, A. Rougier, L. Aymard, G. Nazri, J.-M. Tarascon, J. Alloys
Compd. 460 (2008) 507–512.
[24] V. Bhat, A. Rougier, L. Aymard, G. Nazri, J.-M. Tarascon, Presented at:
Materials Research Society, Spring Meeting (Hydrogen Storage Materials
Symposium), San Francisco (Etats-Unis), April, 2006.
[25] V. Bhat, A. Rougier, L. Aymard, G. Nazri, J.-M. Tarascon, Int. J. Hydrogen
Energy 32 (2007) 4900–4906.
[26] A. Borgschulte, J.H. Rector, B. Dam, R. Griessen, A. Züttel, J. Catal. 235
(2005) 353–358.
[27] O. Friedrichs, J.C. Sanchez-Lopez, C. Lopez-Cartes, T. Klassen, R. Bor-
mann, A. Fernandez, J. Phys. Chem. B 110 (2006) 7845–7850.
[28] H. Gijis Schimmel, J. Hout, L.C. Chapon, F.D. Tichelaar, F.M. Mulder, J.
Am. Chem. Soc. 1127 (2005) 14348.
[29] C. Monteil, Chassaing, Rev. Chim. Miner. 16 (1979) 104.
[30] W. Wong-Ng, H. McMurdie, B. Paretzkin, C. Hubbard, A. Dragoo, NBS
(USA), ICDD Grant-in-Aid, 1986.
[18] V.V. Bhat, A. Rougier, L. Aymard, X. Darok, G. Nazri, J.-M. Tarascon, J.
Power. Sources 159 (2006) 107–110.