Influence of particle size on electrochemical and gas-phase hydrogen storage in nanocrystalline Mg

Article abstract of DOI:10.1016/j.jallcom.2007.09.085

Nanocrystalline Mg powders of different particle size were obtained by inert gas evaporation and studied during electrochemical and gas-phase hydrogen cycling processes. The samples were compared to dehydrided samples obtained by mechanical milling of MgH₂ with and without 2 mol% Nb₂O₅ as catalyst. The hydrogen overpotential of the pure Mg, which is a measure of the hydrogen evolution at the electrode surface, was observed to be reduced with smaller particle sizes reaching values comparable to samples with Nb₂O₅ additive. On the other hand gas-phase charging experiments showed the capacity loss with smaller particle sizes due to oxidation effects. These oxidation effects are different depending on the synthesis method used and showed a major influence on the hydrogen sorption kinetics.

Full text of DOI:10.1016/j.jallcom.2007.09.085

Journal of Alloys and Compounds 463 (2008) 539–545
Influence of particle size on electrochemical and gas-phase
hydrogen storage in nanocrystalline Mg
O. Friedrichsa,b,∗, L. Kolodziejczyk , J.C. Sanchez-Lopez , A. Fernandez ,
a
a
a
´
´
´
c
c
c
d
¨
L. Lyubenova , D. Zander , U. Koster , K.F. Aguey-Zinsou ,
T. Klassend, R. Bormannd
a Instituto de Ciencia de Materiales de Sevilla (CSIC-US), Avda. Ame´rico Vespucio 49, 41092 Seville, Spain
b EMPA Materials Sciences and Technology, Department of Environment, Energy
¨
and Mobility, Div. “Hydrogen and Energy”, Uberlandstrasse 129, CH-8600 Du¨bendorf, Switzerland
c Department of Biochemical and Chemical Engineering, University of Dortmund, D-44221 Dortmund, Germany
d Institute for Materials Research, GKSS Research Centre, Max-Planck-Street 1, 21502 Geesthacht, Germany
Received 3 June 2007; received in revised form 21 September 2007; accepted 21 September 2007
Available online 29 September 2007
Abstract
Nanocrystalline Mg powders of different particle size were obtained by inert gas evaporation and studied during electrochemical and gas-phase
hydrogen cycling processes. The samples were compared to dehydrided samples obtained by mechanical milling of MgH2 with and without 2 mol%
Nb2O5 as catalyst. The hydrogen overpotential of the pure Mg, which is a measure of the hydrogen evolution at the electrode surface, was observed
to be reduced with smaller particle sizes reaching values comparable to samples with Nb2O5 additive. On the other hand gas-phase charging
experiments showed the capacity loss with smaller particle sizes due to oxidation effects. These oxidation effects are different depending on the
synthesis method used and showed a major influence on the hydrogen sorption kinetics.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Hydrogen storage; Metal hydrides; Nanocrystalline magnesium hydride; Electrochemical; Gas-phase; Kinetics
1. Introduction
By using nanocrystalline MgH2 obtained by mechanical
milling [4–6] with transition metal oxides as milling additives
[7–11] big improvements of the sorption kinetics have been
achieved for the sorption kinetics. Nb2O5 has been shown to
behave as one of the best additives [12,13] for hydrogenation
in the gas-phase at the present time, and has been demonstrated
to catalyze significantly also the electrochemical charging pro-
cess as reported in [14]. The mechanism however is not yet
understood. One of many possible explanations for the sorp-
tion improvement is the particle size reduction and formation
of MgH2 nanoparticles due to the Nb2O5 [15,16]. The milling
additivehelpstobreaktheMgH2 particlesandavoidsitsagglom-
eration leading to cold welding during the milling process.
The aim of the present work was to investigate the influence
of the Mg particle size on the hydrogen sorption mechanisms by
electrochemical and gas-phase methods. Therefore samples of
variable size prepared by inert gas evaporation technique (IGE),
which consist of ultrafine Mg nanoparticles, were compared to
samples synthesized by high energy ball milling (BM) with and
Magnesium as a light, abundant and cheap metal with high
hydrogen capacity of 7.6 wt.% is a very attractive material
for hydrogen storage. Hydrogen charging can hereby be per-
formed from the gas-phase [1] or in solution by electrochemical
methods [2,3]. The disadvantage of using Mg in the gas-phase
however lies in the slow sorption kinetics and the high thermody-
namic stability of the MgH2, which demand high temperatures
(>300 ◦C). In the case of the electrochemical hydrogenation the
corrosion behavior of Mg in electrolytes reduces the hydrogen
capacity and kinetics as well as the cycle stability.
∗
Correspondingauthorat:EMPAMaterialsSciencesandTechnology, Depart-
ment of Environment, Energy and Mobility, Div. “Hydrogen and Energy”,
¨
Uberlandstrasse 129, CH-8600 Du¨bendorf, Switzerland. Tel.: +41 823 4153;
fax: +41 823 4022.
E-mail address: oliver.friedrichs@empa.ch (O. Friedrichs).
0925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2007.09.085

540
O. Friedrichs et al. / Journal of Alloys and Compounds 463 (2008) 539–545
Table 1
Overview of samples analyzed in the presented work
Sample type
Label
Preparation method
–
Analysis
PM
Commercial (Mg rod)
Mg bulk
BM-Mg
BM-Mg/Nb2O5
BM-MgH2
BM of MgH2 + dehydrogenation
BM of MgH2/Nb2O5 + dehydrogenation
BM of MgH2
PM
PM
XRD
BM
IGE
IGE700-Mg
IGE900-Mg
IGE700-MgH2
IGE900-MgH2
IGE of Mg at 700 ◦C
TEM, HS, PM
TEM, HS, PM
XRD
IGE of Mg at 900 ◦C
IGE of Mg at 700 ◦C + hydridation
IGE of Mg at 900 ◦C + hydridation
XRD
BM: ball milling; IGE: inert gas evaporation; PM: polarization measurements; HS: hydrogen sorption measurements; XRD: X-ray diffraction; TEM: transmission
electron microscopy.
without Nb2O5 as milling additive with respect to their electro-
chemical properties and their gas-phase sorption behavior.
carried out in a de-aerated, not stirred, 6 M KOH electrolyte at 25 ◦C. The
samples were embedded in a thermoplastic resin to provide electrical isola-
tion. Sample surfaces were ground with a 1200 grid SiC-emery paper and
ethanol.
The kinetic sorption measurements from the gas-phase were carried out by
a volumetric Sievert apparatus designed by HERA Hydrogen Storage Systems
[20]. Themeasurementsconsistingofhydrogenabsorptionanddesorptioncycles
were performed at 300 ◦C and 1 MPa of hydrogen pressure for absorption and
0.1 kPa of hydrogen pressure for desorption. Therefore the samples of about
100 mg were loaded into a reactor under argon atmosphere and transferred to
the Sievert apparatus without air exposure.
2. Experimental
2.1. Sample preparation
The samples studied in this paper were prepared by two different methods,
the inert gas evaporation and the high energy ball milling. An overview of the
samples and the corresponding analysis methods is presented in Table 1.
The IGE method was performed by thermal evaporation of Mg (supplied by
Aldrich with purity >99.98%) in a He gas atmosphere of 3 Torr and by collecting
the powder (IGE-Mg) on a nitrogen-cooled container and subsequent transfer
of the sample in He to the glove box as described in [17].
The milling was done under argon atmosphere (<10 ppm water and oxygen)
with a Fritsch P5 planetary mill applying 100 g Cr-hardened balls as milling
tools within a 300 ml stainless steel vial. MgH2 (purchased from Goldschmidt
AGwithpurity>95%(restbeingMg))wasmilledfor20 h(BM-MgH2)andaddi-
tional milled for 100 h with Nb2O5 additive (BM-Mg/Nb2O5, 2 mol% Nb2O5,
supplied by Sigma–Aldrich with purity >99.99%). The ball to powder ratio
was 10/1.
For the electrochemical experiments the BM-MgH2 samples were first dehy-
drogenated at 300 ◦C. The obtained BM-Mg or the as-produced IGE-Mg is
then compacted with 5 wt.% graphite and 5 wt.% PTFE to a pressed pellet
electrode suitable for electrochemical hydrogenation (compaction pressure of
6.2 N/mm2). Thereby the polymer (PTFE) was chosen to increase the void frac-
tion for hydrogen transport and the graphite to enhance conductivity, which
results in higher capacity and improved hydrogen kinetics as shown in [18]. The
electrode preparation was performed inside the glove box under pure Ar atmo-
sphere. Since the preparation by IGE is very complex and the yield of sample is
very low only the electrochemical characterization of the IGE700-Mg sample by
potentiodynamic polarization was done up to now. For comparison the obtained
results are compared to a bulk Mg rod.
3. Results and discussion
3.1. Microstructural characterization of IGE samples by
TEM
Compared to samples prepared by high energy milling, with
particle sizes ranging from about 0.1 ␮m to 1 ␮m [19], the mag-
nesium synthesized by IGE displays a different microstructure
that consists of isolated particles of variable size and shapes
depending on the evaporation temperature. At 700 ◦C (IGE700-
Mg), the particles are almost spherical, single-crystal and well
dispersed with an average diameter of about 35 8 nm (see
Fig. 1a).
Fig. 1b shows the analysis of Mg powder prepared by inert
gas evaporation at a higher synthesis temperature of 900 ◦C
(IGE900-Mg). Here the spherical particles (from the synthesis
at 700 ◦C) are replaced by platelets or particles with isomet-
ric shapes of hexagonal symmetry as schematically illustrated
in Fig. 1c. The particle size distribution of Fig. 1b shows the
variation of hexagonal diameter and width of these particles.
The comparison of the particle size histograms for both samples
illustrates the increase in particle size and dispersion with the
increment in synthesis temperature. In other words, the parti-
cle size distribution is less homogeneous and the mean value
increases substantially with the increment of evaporation tem-
perature during the gas-phase condensation. A characteristic
selected area electron diffraction pattern (SAED) is represented
in Fig. 1d. The rings observed in the SAED can be assigned
to polycrystalline magnesium metal together with a magnesium
oxide phase. This latter phase is constituted by smaller nanocrys-
tallinegrainsthanthepuremagnesiummetalascanbeconcluded
by the presence of diffuse diffraction rings of MgO appearing
in the diffraction pattern.
2.2. Sample characterization
The microstructural characterization was carried out by X-ray diffraction
(XRD) with a Bruker D8 advance X-ray diffractometer using Cu K␣ radiation
in conventional Bragg–Brentano configuration and by transmission electron
microscopy (TEM) and electron diffraction (ED) in a Philips CM200 with
200 kV electron acceleration voltage and a point-to-point resolution of 0.24 nm.
For the TEM analysis the nanocrystalline Mg powder was dispersed in toluene
as protective agent against oxidation before introduction in the pre-vacuum
chamber of the microscope [19].
The potentiodynamic polarization curves were obtained using a FAS1
potentiostat (Gamry Instruments) with a voltage scan rate of 1 mV/s. For all
measurements a three-electrode electrochemical cell was used with an Hg/HgO
reference electrode and a platinum counter electrode. All the potential values
are reported with respect to the Hg/HgO reference electrode. The electro-
chemical tests using the DC (potentiodynamic polarization) technique were

O. Friedrichs et al. / Journal of Alloys and Compounds 463 (2008) 539–545
541
Fig. 1. TEM micrographs with particle size distributions of Mg nanoparticles produced by gas-phase condensation in He atmosphere (3 Torr) at evaporation
temperatures of (a) 700 ◦C and (b) 900 ◦C. A schematic picture of the hexagonal Mg particles is depicted in (c) and a representative ED for the samples is presented
in (d).
A representative HRTEM image of the microstructure for
these IGE-Mg samples is presented in Fig. 2. It is important to
note that the particles exhibit a core–shell structure showing a
crystalline core of metallic magnesium and an oxide passivation
layer, which shows some degree of crystallinity and also some
amorphous regions (insets A and B in Fig. 2). Thereby the thick-
ness of this oxide layer is in the range of 3–4 nm, which is of
similar thickness to that determined for a milled and dehydrided
MgH2 sample [19]. The inset in the top left side of the image
corresponds to the digital diffraction pattern (DDP) obtained
solutions, can be described as
M + H2O + e− ↔ MHads + OH− (Volmer reaction)
followed by solid state transfer of hydrogen:
MHads ↔ MH␣
MH␣ ↔ MH␤
The hydrogen atoms adsorb at the electrode surface by the
reduction of water (MHads), diffuse into the electrode mate-
rial and form a solid solution in the host lattice (MH␣). The
formation of a metal hydride (MH␤) proceeds after exceeding
solubility limit of hydrogen. If the hydrogen diffusion into the
material proceeds slower than the charge transfer step additional
competing reactions can be observed. The adsorbed hydrogen
participates in chemical and/or electrochemical recombination
steps resulting in the release of molecular hydrogen
◦
˚
from the bulk of the particle. The measurement of 2.8 A at 60
on this DDP allows us the identification of the [0 0 1] orientation
of the Mg h c p crystal structure. The same geometry is repeated
when the DDP is calculated over different areas in the image,
pointing out the single-crystalline nature of the particle. This
single-crystalline character is a general feature of this type of
sample.
2MHads ↔ H2 + 2M (Tafel reaction)
MHads + H2O + e−↔ H2 + OH− + M (Heyrowsky reaction)
3.2. Characterization of the electrochemical hydrogen
charging behavior
The main reactions occurring during electrochemical charg-
ing and discharging of metal hydride electrodes in, e.g. alkaline
The evolution of molecular hydrogen reduces the efficiency
of the hydriding/dehydriding cycles of the metal hydride elec-

542
O. Friedrichs et al. / Journal of Alloys and Compounds 463 (2008) 539–545
Fig. 2. HRTEM micrographs of the nanocrystalline Mg sample obtained by IGE showing the core–shell microstructure. The selected zones A and B show zooms on
the oxide surface layer whereas in the upper left corner of the graph the digital diffraction pattern (DDP) of the bulk of this particle is presented.
trode. Therefore it is necessary to develop electrodes which
allow a preferred Volmer reaction with low hydrogen overpo-
tential, which is a measure of the hydrogen evolution at the
electrode surface, relative to the Tafel as well as Heyrowsky
reaction and an increased diffusion into the metal.
Potentiodynamic polarization experiments (see Fig. 3) give
information regarding the surface reactions at the electrodes,
especially regarding the hydrogen overvoltage and the corrosion
and passivation behavior of the investigated IGE-Mg electrodes
in comparison to the ball milled with and without Nb2O5 addi-
tives (BM-Mg and BM-Mg/Nb2O5). In general it was observed
that Nb2O5 decreases the hydrogen overvoltage (η20, current
density of 20 mA/g), but reduces the corrosion resistance of the
milled and dehydrided MgH2 (Table 2). As reported earlier [21]
no significant influence of the addition of graphite as well as
PTFE on the hydrogen overvoltage was observed. The reduced
Fig. 3. Influence of the particle size on potentiodynamic polarization of Mg
(bulk, IGE-Mg700 and BM-Mg) and of a BM-Mg/Nb2O5-electrode compacted
with a pressure of 6.2 N/mm2. Particle size is indicated in the figure.
hydrogen overvoltage due to the catalyst is assumed to accelerate
the hydrogen adsorption at the electrode surface. At lower cur-
rent densities it is assumed that the Volmer reaction will become
the speed limiting reaction and the competing hydrogen recom-
bination reactions will be irrelevant for the BM-Mg/Nb2O5
electrodes. It is also assumed that at very low current densities
the electrochemical equilibrium is moved to the anodic reac-
tion, which are influenced by the corrosion current density and
the passive current density and leads to a lower storage capac-
ity of the BM-Mg/Nb2O5 electrodes because of an increased
oxidation in the 6 M KOH electrolyte.
In addition the potentiodynamic measurements revealed a
significant influence of the particle size on the hydrogen over-
potential as well as on the active dissolution and passivation
Table 2
Potentiodynamic polarization parameters of bulk Mg, IGE700-Mg, BM-Mg and Mg/Nb2O5-electrodes with different particle sizes compacted with a pressure of
6.2 N/mm2
Material
Corrosion current density (A/cm2)
Passive current density (A/cm2)
Hydrogen overpotential η20 at 20 mA/g (mV)
Mg bulk (polycrystalline Mg rod)
BM-Mg (␮m particle size)
IGE-Mg700 (35 nm particle size)
BM-Mg/Nb2O5 (␮m particle size)
1 × 10−5
8 × 10−5
1 × 10−3
4 × 10−3
4 × 10−5
2 × 10−4
4 × 10−3
4 × 10−4
226
130
58
61

O. Friedrichs et al. / Journal of Alloys and Compounds 463 (2008) 539–545
543
behavior of nanocrystalline Mg (Table 2). Decreasing the par-
ticle size down to 35 nm (IGE700-Mg sample) leads to an
improved cathodic reaction with a low hydrogen overpoten-
tial but also to strong anodic reactions with high corrosion
rates and passive current densities. Further, Mg particles of
35 nm revealed similar hydrogen overpotentials in comparison
to nanocrystalline Mg electrodes with Nb2O5 and a particle size
up to 1 ␮m but still the highest affinity for oxidation processes.
Thus indicates that smaller Mg particle sizes might improve
as well as catalysts the electrochemical hydrogen absorption
behavior whereas the electrochemical hydrogen desorption will
be strongly influenced by the anodic reaction.
3.3. Hydrogen desorption behavior in gas-phase
Fig. 5. Hydrogen capacity and desorption rate for the milled MgH2 and the
hydrided Mg powder obtained by IGE at 700 ◦C and 900 ◦C.
Despite the promising electrochemical behavior of the mag-
nesium particles synthesized by the inert gas evaporation
method, which resulted in small and little agglomerated par-
ticles as shown in Fig. 1, a relatively slow hydrogen sorption
behavior at only high sorption temperatures was obtained during
H2 sorption experiments from the gas-phase. Hydrogen desorp-
tion at 300 ◦C, as it is possible for milled MgH2, could not be
achieved for the material from IGE method, so that a temperature
of 350 ◦C was used during the experiments.
to the ball milled ones, which seems to be correlated to the
synthesis temperature, i.e. the particle size, of the IGE powders.
Also it can be observed for all samples a decrease of the capacity
for successive cycles, which occurs for the IGE samples together
with a decrease of desorption rate and for the milled sample with
anincreaseofdesorptionrate. Inthefollowingadetailedanalysis
of these effects with consequences on the kinetic behavior are
presented.
Fig. 4 shows the desorption kinetics at 350 ◦C for the MgH2
powder synthesized by hydridation of Mg particles prepared by
IGE at 700 ◦C and 900 ◦C in comparison to the nanocrystalline
MgH2 powder obtained by high energy ball milling. The des-
orption kinetics of the IGE powder is at 350 ◦C more or less
comparable to the one for milled MgH2. This is however not
very fast considering that the kinetics of the milled MgH2 is
about 10 times slower than for MgH2 milled with Nb2O5 addi-
tive, where the sample can de- and absorb within a few minutes
[12,13].
3.3.1. Particle size effect
The general decrease of capacity for the IGE samples with
decreasing synthesis temperature can be explained considering
the amount of oxide phase by means of some simple geometrical
models for spherical (IGE700-Mg) and hexagonal plate particles
(IGE900-Mg) as they have been observed by TEM analysis (see
Fig. 1). These geometrical models are used to calculate a theo-
retical chargeable metallic Mg content in the sample based on
the following assumption and calculations.
The characteristic parameters of desorption are the hydrogen
capacity and desorption rate, which are shown quantitatively in
Fig. 5. With respect to the hydrogen capacity it can be seen a
general decrease for the smaller IGE particles in comparison
For the IGE700-Mg sample spherical magnesium particles
with a surface oxide layer of 3.6 nm (as determined by XPS anal-
ysis in [19] and TEM observation) were assumed. Based on this
assumption volume ratios between the metallic magnesium and
the oxide phase could be calculated with respect to variations in
theparticlesize(“sphericalmodel”). FortheIGE900-Mgsample
this model had to be modified due to the different morphology
of the particles. Here the particles are assumed to be formed of
boxes with hexagonal base as illustrated in Fig. 1c “hexagonal
model” and they are also surrounded by a magnesium oxide sur-
face layer as observed for the spherical model. Assuming the
same oxide layer thickness as for the spherical model (3.6 nm)
and a fixed ratio of 0.3 between the width and the hexagonal
diameter of the particles as determined by TEM analysis, the
volume ratios between the metallic magnesium and the oxide
phase was calculated again with respect to the variations in the
particle size. For both models we assumed that the particles con-
sist of metallic Mg and MgO with densities of 1.738 g/cm3 and
3.58 g/cm3, respectively. By these densities the chargeable Mg
content as mass ratio was calculated.
Fig. 4. Desorption kinetics of BM-MgH2 and MgH2-IGE produced by hydri-
dation of IGE-Mg powder prepared at 700 ◦C and 900 ◦C.

544
O. Friedrichs et al. / Journal of Alloys and Compounds 463 (2008) 539–545
Both models together with selected points for the different
IGE-Mg samples (IGE700-Mg and IGE900-Mg), where the par-
ticle size values were determined from TEM observations (see
Fig. 1) are presented in Fig. 6. Fig. 6 shows the calculated mass
ratio of chargeable metallic magnesium with respect to particle
size in the comparison to the mass ratio obtained experimen-
tally by the hydrogen weight capacity (see Fig. 5) of the IGE-Mg
samples as measured during the first and second hydrogen cycle.
Here it can be seen clearly the effect of particle size on the
hydrogen capacity. With smaller particles the surface to bulk
ratio increases drastically and leads to a high capacity loss in the
sample(i.e. decreaseofchargeableMg), althoughtheoxidelayer
thickness is identical to that of bigger particles. The hydrogen
capacity, obtained by titration measurements, follows the trend
of lower metallic Mg content with smaller particle size, but is in
general higher than the calculated chargeable Mg content based
on the TEM analysis. This may be explained by the uncertainty
in the density of the magnesium oxide layer that has been used
for the calculations of the gravimetric hydrogen content in the
spherical and hexagonal model. Surely the oxide layer on the
IGE-Mg particles is not as crystalline and stoichiometric as the
crystalline MgO phase, which has been assumed for the conver-
sion. In the real samples we most probably deal with oxide layers
of lower densities, which may explain the underestimation of
the chargeable Mg content measured by capacity measurements
during cycling.
Fig. 7. XRD diffraction patterns of nanocrystalline MgH2 prepared by IGE at
(a) 700 ◦C and (b) 900 ◦C or synthesized by (c) high energy ball milling as-
obtained for the samples after two cycles of H2 sorption. Symbols correspond
to the following indexed phases: MgH2 (ꢀ), MgO (ꢀ) and Mg (+).
pure metal. The fraction of this metallic Mg is the lowest for the
BM-sample and the highest for the IGE700-MgH2.
This may be explained by assuming the formation of a MgH2
layer on the surface of the Mg particles during the hydrogen
absorption. The superficial hydride layer grows into the Mg crys-
tals transforming the Mg into MgH2 (contracting volume model
[22]). Since the MgH2 presents a barrier for hydrogen diffu-
sion, the hydrogen diffusion to the Mg core is more and more
hindered with growing MgH2 phase, so that finally part of the
Mg remains inaccessible for the hydrogen. The BM-MgH2 con-
tains bigger particles but very small crystalline domains in the
range of 10 nm [19] that are suitable for an almost full charging
by hydrogen diffusion. Taking into account the contracting vol-
ume model the opposite effect for the two IGE samples would
be expected. On the smaller particles as it is the case for the
IGE700-Mg sample less hydrogen barrier would be produced
by a full charging. However, this sample shows experimen-
tally the highest remaining metallic Mg fraction after cycling
of all three samples. To explain these results the microstruc-
tural properties of the oxide layer should be considered. Since
the thickness of the oxide layer is more or less constant for
all samples, the composition and crystallinity can play a major
role in hindering the hydrogen diffusion into the material. A
more ordered and dense oxide layer can hinder the hydrogen
diffusion into the particle [23]. This fact may explain why the
hydrided Mg produced by IGE shows slower hydrogen sorp-
tion kinetics than the BM-MgH2 produced by milling despite of
its smaller particle size. While during the gas-phase condensa-
tion the oxide layer can grow unaffected from any mechanical
influences, the oxide layer on the BM-MgH2 is influenced by
mechanical deformation and high compressive stresses [24,25],
induced by the milling process, which may generate a less dense,
crystalline and easier to penetrate oxidation layer. This results
also in a different behavior with repeated cycling for the milled
sample. In contrast to the IGE samples here the hydrogen sorp-
tion kinetics improves with repeated cycling. It can be assumed
that micro/nano-cracks in the oxide layer are formed during the
3.3.2. Capacity decrease with cycling (oxide layer barrier
effect)
The decrease of capacity with cycling (seen in Figs. 4 and 5)
occurs for all samples and is mainly due to an incomplete hydri-
dation of the Mg in the samples. In Fig. 7 the XRD for the
BM-MgH2 and the IGE-MgH2 samples after two cycles are pre-
sented. By comparison of the XRD patterns of the three samples
it is seen that a certain amount of Mg remains unhydrided as
Fig. 6. Influence of particle size on chargeable metallic Mg content: the curves
present semi-theoretical calculations (based on TEM and XPS) assuming hexag-
onal and spherical particles consisting of Mg and MgO with densities of
1.738 g/cm3 and 3.58 g/cm3, respectively. The expected and experimental hydro-
gen capacities are shown for IGE-Mg samples prepared at 700 ◦C (spherical)
and 900 ◦C (hexagonal).

O. Friedrichs et al. / Journal of Alloys and Compounds 463 (2008) 539–545
545
cycling [26] which facilitates the hydrogen uptake through the
surface.
Apart from this, the crystal growth during hydridation has
also to be taken into account. In contrast to the electrochemical
experiment, in a gas-phase experiment the Mg sample is heated
during cycling, which can lead to a considerable crystal and
particle growth. This effect does not occur in electrochemical
charging, where pure particle size effects can be observed.
References
[1] T. Klassen, W. Oelerich, R. Bormann, Proceedings of Ismanam-2000 on
Metastable Mechanically Alloyed and Nanocrystalline Materials, Trans
Tech Publications Ltd., 2001, pp. 603–608.
[2] T. Sakai, I. Uehara, H. Ishikawa, J. Alloy Compd. 293–295 (1999) 762–
769.
[3] J. Kleperis, G. Wojcik, A. Czerwinski, J. Skowronski, M. Kopczyk,
M. Beltowska-Brzezinska, J. Solid State Electrochem. 5 (2001) 229–
249.
[4] A. Zaluska, L. Zaluski, J.O. Strom-Olsen, J. Alloy Compd. 289 (1999)
197–206.
4. Conclusions
[5] A. Zaluska, L. Zaluski, J.O. Strom-Olsen, J. Alloy Compd. 288 (1999)
217–225.
[6] K.J. Gross, P. Spatz, A. Zuettel, L. Schlapbach, J. Alloy Compd. 240 (1996)
206–213.
[7] W. Oelerich, T. Klassen, R. Bormann, J. AlloyCompd. 315(2001)237–242.
[8] G. Liang, J. Huot, S. Boily, R. Schulz, J. Alloy Compd. 305 (2000) 239–
245.
[9] J. Huot, J.F. Pelletier, L.B. Lurio, M. Sutton, R. Schulz, J. Alloy Compd.
348 (2003) 319–324.
[10] F.J. Castro, J.-L. Bobet, J. Alloy Compd. 366 (2004) 303–308.
[11] J.-L. Bobet, S. Desmoulins-Krawiec, E. Grigorova, F. Cansell, B. Chevalier,
J. Alloy Compd. 351 (2003) 217–221.
[12] G. Barkhordarian, T. Klassen, R. Bormann, Scripta Mater. 49 (2003)
213–217.
[13] G. Barkhordarian, T. Klassen, R. Bormann, J. Alloy Compd. 364 (2004)
242–246.
[14] D. Zander, L. Lyubenova, U. Koster, M. Dornheim, F. Aguey-Zinsou, T.
Klassen, J. Alloy Compd. 413 (2006) 298–301.
[15] I.-H. Kwon, J.-L. Bobet, J.-S. Bae, M.-Y. Song, J. Alloy Compd. 396 (2005)
264–268.
[16] K.-F. Aguey-Zinsou, J.R. Ares Fernandez, T. Klassen, R. Bormann, Mater.
Res. Bull. 41 (2006) 1118–1126.
[17] O. Friedrichs, L. Kolodziejczyk, J.C. Sa´nchez-Lo´pez, C. Lo´pez-Carte´s, A.
Ferna´ndez, J. Alloy Compd. 434–435 (2007) 721–724.
[18] D. Zander, L. Lyubenova, U. Ko¨ster, T. Klassen, M. Dornheim, in: Proceed-
ings of the Ismanam-2005, Paris, France, J. Alloy Compd. 234/5 (2007)
753–755.
[19] O. Friedrichs, J.C. Sa´nchez-Lo´pez, C. Lo´pez-Cartes, M. Dornheim,
T. Klassen, R. Bormann, A. Ferna´ndez, Appl. Surf. Sci. 252 (2006)
2334–2345.
[20] R. Schulz, S. Boily, J. Huot, Patent Canada 2207149 (1999).
[21] D. Zander, L. Lyubenova, U. Ko¨ster, T. Klassen, M. Dornheim, in: M.
Nazri, G-A. Nazri, R.C. Young, C. Ping (Eds.), Mat. Res. Soc. Symp.
Proc., Warrendale, Hydrogen Storage Mater. 801 (2004) BB 3.1.
[22] G. Friedlmeier, M. Groll, J. Alloy Compd. 253 (1997) 550–555.
[23] V.E. Henrich, Rep. Prog. Phys. 48 (1985) 1481–1541.
[24] J. Huot, G. Liang, S. Boily, A. Van Neste, R. Schulz, J. Alloy Compd. 295
(1999) 495–500.
[25] R. Schulz, J. Huot, G. Liang, S. Boily, G. Lalande, M.C. Denis, J.P. Dodelet,
Mater. Sci. Eng. A 267 (1999) 240–245.
[26] V. Bhat, A. Rougier, L. Aymard, X. Darok, G.A. Nazri, J.-M. Tarascon, J.
Power Sources 159 (2006) 107–110.
The studies of nanocrystalline Mg powder, which was
obtained by inert gas evaporation technique, in comparison
to that obtained by dehydridation of nanocrystalline MgH2
prepared by mechanical milling with and without 2 mol%
Nb2O5 revealed an influence of the Mg particle size on the
electrochemical and gas-phase hydrogen sorption. In the case
of electrochemical charging, a reduced particle size (35 nm)
yielded a performance similar to that obtained using Nb2O5
as additive. The hydrogen overpotential of nanocrystalline Mg
with a particle size up to 1 ␮m is much higher than for a smaller
particle size of 35 nm as well as for Mg with Nb2O5 catalyst.
However, the electrochemical oxidation behavior became more
favorable with the decrease in particle size which might influ-
ence the hydrogenation negatively at very low current densities.
The promising electrochemical behavior of the Mg samples pro-
duced by inert gas evaporation could however not be observed
for the gas-phase sorption. The gas-phase sorption experiments
have shown that the sorption properties of the IGE-Mg samples
do not only depend on the particle size, but are influenced very
much by the composition and crystallinity of a thin magnesium
oxide layer (3–4 nm) covering the magnesium metal underneath.
The use of smaller particles leads inevitably to a higher contri-
bution of the oxide component in the sample and consequently
to a loss of capacity. The formation in the case of a more ordered
and dense oxide layer hinders the hydrogen diffusion into the
particle. These factors explained that IGE-Mg particles did not
demonstrate better kinetics in spite of their small size and were
not suitable to prove any particle size influence on the gas-phase
sorption kinetics.
Acknowledgements
Authors thank the financial support (project nos. HPRN-CT-
2002-00208 and MRTN-CT-2006-035366) from the European
Commission.