Hydrogen cycling of niobium and vanadium catalyzed nanostructured magnesium

Article abstract of DOI:10.1021/ja051508a

The reaction of hydrogen gas with magnesium metal, which is important for hydrogen storage purposes, is enhanced significantly by the addition of catalysts such as Nb and V and by using nanostructured powders. In situ neutron diffraction on MgNb_0.05 and MgV_0.05 powders give a detailed insight on the magnesium and catalyst phases that exist during the various stages of hydrogen cycling. During the early stage of hydriding (and deuteriding), a MgH_{1 phase is observed, which does not occur in bulk MgH2 and, thus, appears characteristic for the small particles. The abundant H vacancies will cause this phase to have a much larger hydrogen diffusion coefficient, partly explaining the enhanced kinetics of nanostructured magnesium. It is shown that under relevant experimental conditions, the niobium catalyst is present as NbH1. Second, a hitherto unknown Mg-Nb perovskite phase could be identified that has to result from mechanical alloying of Nb and the MgO layer of the particles. Vanadium is not visible in the diffraction patterns, but electron micrographs show that the V particle size becomes very small, 2-20 nm. Nanostructuring and catalyzing the Mg enhance the adsorption speed that much that now temperature variations effectively limit the absorption speed and not, as for bulk, the slow kinetics through bulk MgH2 layers.}

Full text of DOI:10.1021/ja051508a

Published on Web 09/27/2005
Hydrogen Cycling of Niobium and Vanadium Catalyzed
Nanostructured Magnesium
†
‡
§
|
H. Gijs Schimmel, Jacques Huot, Laurent C. Chapon, Frans D. Tichelaar, and
,
†
Fokko M. Mulder*
Contribution from the Department of Radiation, Radionuclides and Reactors, Faculty of Applied
Sciences, Delft UniVersity of Technology, Mekelweg 15, 2629JB Delft, The Netherlands, the
Institut de recherche sur l’hydrog e` ne, UniVersit e´ du Qu e´ bec a` Trois-RiVi e` res, (Qc)G9A 5H7
Canada, the ISIS facility, Rutherford Appleton Laboratory, Chilton, Didcot, United Kingdom,
and the National Centre for HREM, Delft UniVersity of Technology, The Netherlands
Received March 9, 2005; E-mail: f.m.mulder@tnw.tudelft.nl
Abstract: The reaction of hydrogen gas with magnesium metal, which is important for hydrogen storage
purposes, is enhanced significantly by the addition of catalysts such as Nb and V and by using
nanostructured powders. In situ neutron diffraction on MgNb0.05 and MgV0.05 powders give a detailed insight
on the magnesium and catalyst phases that exist during the various stages of hydrogen cycling. During
the early stage of hydriding (and deuteriding), a MgH1<x<2 phase is observed, which does not occur in bulk
MgH and, thus, appears characteristic for the small particles. The abundant H vacancies will cause this
2
phase to have a much larger hydrogen diffusion coefficient, partly explaining the enhanced kinetics of
nanostructured magnesium. It is shown that under relevant experimental conditions, the niobium catalyst
is present as NbH . Second, a hitherto unknown Mg-Nb perovskite phase could be identified that has to
1
result from mechanical alloying of Nb and the MgO layer of the particles. Vanadium is not visible in the
diffraction patterns, but electron micrographs show that the V particle size becomes very small, 2-20 nm.
Nanostructuring and catalyzing the Mg enhance the adsorption speed that much that now temperature
variations effectively limit the absorption speed and not, as for bulk, the slow kinetics through bulk MgH
layers.
2
1
. Introduction
Mg nor form intermetallic compounds, these will be hetero-
geneously catalyzed materials.
A conditio sine qua non for the development of sustainable
1
.1. Kinetic Factors in the Reaction of Hydrogen and
technologies for the hydrogen economy is the availability of
an economical, safe, and practical hydrogen storage mechanism.
Since hydrogen gas can be used as a clean fuel both in
conventional combustion engines and in fuel cells, it has a large
potential as an energy carrier. Hydrogen is a very light and
volatile gas, which forces the use of low temperatures, high
pressures, a sorption material, or a combination of these to store
a practical amount of hydrogen (6 wt % in the year 2010
according to the demand of the Department of Energy of the
USA). Magnesium metal stores approximately 7.6 wt % in the
form of MgH2, which is a transparent insulator. Its use is at
present hampered by the slow kinetics, which makes it practical
Magnesium. Magnesium metal transforms into magnesium
hydride in a gas-solid reaction. The driving force for this
process scales with ln(Papplied/Pequilibrium). The hydriding reaction
of metallic magnesium involves several steps: (1) gas perme-
ation through the particle bed, (2) surface adsorption and
hydrogen dissociation, (3) migration of hydrogen atoms from
the surface into the bulk, (4) diffusion through the particle and
finally (5) nucleation and growth of the hydride phase.4 In
principle each of these steps might be rate-limiting, and many
,5
6-14
reports are known that study one or more of these steps.
(
2) Zaluski, L.; Zaluska, A.; Str o¨ m-Olsen, J. O. J. Alloys Compd. 1997, 253-
254, 70.
1
only at temperatures above 300 °C. By means of a ball milling
(3) Huot, J.; Liang, G.; Boily, S.; Neste, A. V.; Schulz, R. J. Alloys Compd.
treatment the particle size can be reduced, which speeds the
sorption properties by a factor of 10. The addition of a catalyst
like V or Nb increases the kinetics by another order of
1999, 293-295, 495.
(4) G e´ rard, N.; Ono, S. In Hydrogen in Intermetallic Compounds II; Schlapbach,
L., Ed.; Springer: 1992.
(5) Sastri, M. C. V.; Viswanathan, B.; Murthy, S. S. Metal Hydrides; Springer-
1-3
Verlag: Berlin, 1998.
magnitude. Because niobium and vanadium do not alloy with
(
6) Liang, G.; Huot, J.; Boily, S.; Schulz, R. J. Alloys Compd. 2000, 305, 239.
7) Zaluska, A.; Zaluski, L.; Str o¨ m-Olsen, J. O. J. Alloys Compd. 1999, 289,
197.
(
†
Department of Radiation, Radionuclides and Reactors, Delft University
(
8) Bloch, J.; Mintz, M. H. J. Alloys Compd. 1997, 253-254, 529.
of Technology.
(
9) Martin, M.; Gommel, C.; Borkhart, C.; Fromm, E. J. Alloys Compd. 1996,
‡
Universit e´ du Qu e´ bec a` Trois-Rivi e` res.
Rutherford Appleton Laboratory.
National Centre for HREM, Delft University of Technology.
238, 193.
§
(
10) Rudman, P. S. J. Appl. Phys. 1979, 50, 7195.
(11) Luz, Z.; Genossar, J.; Rudman, P. S. J. Less-Common Met. 1980, 73, 113.
(12) Belkbir, L.; Joly, E.; Gerard, N. Int. J. Hydrogen Energy 1981, 6, 285.
|
(
1) Huot, J.; Liang, G.; Schulz, R. Appl. Phys. A 2001, 72, 187.
14348
9
J. AM. CHEM. SOC. 2005, 127, 14348-14354
10.1021/ja051508a CCC: $30.25 © 2005 American Chemical Society

H
2
Cycling of Nb and V Catalyzed Nanostructured Mg
A R T I C L E S
Regarding the first step, this is more of an engineering
2. Experimental Section
problem, and we will therefore not discuss it here. The second
step is rate-limiting in magnesium since the addition of catalysts
for this process increases the kinetics drastically. The kinetics
Magnesium hydride powder and niobium or vanadium metal powder
were ball-milled in a Spex 8000 ball milling apparatus for 20 h. Under
an inert atmosphere, respectively 2.15 g of the MgNb0.05H and 2.11 g
2
1
of the third step will, like the second step, scale with the surface
area of the particles. Previously we have shown using ab initio
computer calculations that the diffusion of hydrogen through
magnesium metal is not rate-limiting at practical temperatures
and particles sizes around 1 µm, even though the concentration
of hydrogen in the metal phase is very low.15 The nucleation
and growth process, finally, will lead to considerable interfacial
energy because magnesium metal has a completely different
crystal structure than magnesium hydride. To overcome such a
barrier, excess pressure for the creation of nucleation centers is
needed. In this respect it is not unlikely that some hydrogen
saturated catalyst particles in close contact with a Mg particle
act as nucleation centers as well.
of the MgV0.05H sample was loaded in quartz tubes that were connected
2
to a gas-handling system. After heating to 320 °C hydrogen gas was
first released from the cell while maintaining the equilibrium pressure,
with a controlled constant hydrogen flow, and diffraction spectra were
taken during desorption. Deuterium was then loaded as this isotope
provides a different contrast in neutron diffraction experiments.
Successive release of deuterium and finally the loading of hydrogen
completed the experiment. Diffraction patterns were collected during
all these stages using the GEM diffractometer at the ISIS facility. At
the temperatures of our experiment, no structural differences between
the hydride and deuteride form of the catalyst exist, at least in bulk
form. The hydrogen containing samples produce a huge background
due to the large incoherent cross section of hydrogen, but nevertheless
GEM provides a remarkably good signal-to-noise ratio sufficient for
detailed Rietveld analysis showing that these patterns yield the same
results as those of the deuterides.
The loading of deuterium or hydrogen gas was performed at a
temperature that was regulated to be 320 °C within 1 °C and mostly at
a pressure around 4-5 bar, finishing at 11 bar, much higher than the
equilibrium pressure of 2.5 bar at this temperature. A neutron diffraction
pattern was recorded for each of the loading steps (20 for the niobium
sample, 11 for the vanadium sample) in the loading process. From a
calibrated volume at room temperature and the pressure drop in it, the
amount of deuterium or hydrogen that was absorbed in the magnesium
sample was determined. During unloading, also at a constant temper-
ature of 320 °C, deuterium or hydrogen was released at a constant
desorption rate of 5 mL/min (STP conditions; 1 bar, rt), and pressures
were recorded. A diffraction pattern was acquired every 3 min, and a
total of 83 patterns for the niobium sample and 87 patterns for the
vanadium sample were taken.
GEM is a high intensity, good resolution neutron diffractometer at
the ISIS facility of the Rutherford Appleton Laboratory in the United
Kingdom.23 The 7290 neutron detectors are grouped into 7 banks, each
with a different range of accessible d-spacing. GEM combines an
unprecedented neutron count rate with a very large Q range and good
resolution, facilitating detailed in situ measurements. The data were
corrected for the contribution from the empty quartz sample container.
All patterns of the complete cycles were sequentially fitted using the
Rietveld refinement program GSAS.24 The phase fractions, lattice
parameters, hydrogen site occupancies, and line widths were allowed
to vary freely. The patterns of the different detector banks were fitted
simultaneously.
High-resolution electron microscopy photographs and selected area
diffraction patterns were obtained using a 300 kV Philips CM300UT-
FEG machine. Quantitative elemental composition was studied by
energy dispersive spectroscopy. Samples were prepared by ultrasonic
shaking of the powder in ethanol and drying it on a copper grid with
a holey carbon foil. Separate powder particles were examined with
diffraction, imaging, and EDS (Energy Dispersive Spectroscopy) for
elemental analysis.
On decreasing the particle size by ball milling magnesium
hydride, a high density of defects and distortions is introduced.
Previously, we have shown that the abundant defects and
distortions in the magnesium hydride structure play no decisive
role in the increase of the sorption speed as they are annealed
out when the hydrogen content is cycled, while the kinetics
1
6,17
remain fast.
We will therefore regard the ball milling
treatment mainly as a method to produce nanostructured and
catalyzed magnesium particles in large quantities.
The main contributions to the increase in speed when the
particles are nanostructured are therefore sought in the increase
of the specific surface area, the decrease of diffusion path
lengths, and factors concerning nucleation and growth in nano
particles. Next, Nb and V catalysts are known for their catalytic
activity in the reaction of hydrogen with Mg. Mg does not form
18
alloys or intermetallic compounds with Nb or V. Thus when
preparing Nb catalyzed Mg samples, one does not expect to
alloy Mg and Nb. The questions where the catalyst atoms are
and in which phase are important for a more detailed under-
standing of the mechanism behind the catalytic activity. In this
contribution a time-dependent, in situ neutron diffraction study
on the hydriding and dehydriding properties of ball milled and
catalyzed MgNb0.05H2 and MgV0.05H2 is presented. In previous
work the Nb phase has been observed using X-ray diffrac-
tion,1
9,20
and a solid solution gateway phase of H in Nb was
proposed. In contrast to the X-ray experiments, the use of
neutrons enabled us to locate hydrogen in the structures and
refine hydrogen site occupancies during hydrogen loading and
unloading of the samples. This makes an improved understand-
ing of the nanostructured Mg-Nb-H and Mg-V-H systems
possible.
(
13) B. Vigeholm, K.; Jensen, B. L.; Pedersen, A. S. J. Less-Common Met. 1987,
3
. Experimental Results and Analysis
1
31, 133.
(
(
14) Mintz, M. H.; Gavra, Z.; Hadari, Z. J. Inorg. Nucl. Chem. 1978, 40, 765.
15) Schimmel, H. G.; Kearley, G. J.; Huot, J.; Mulder, F. J. Alloys Compd.
The patterns obtained during dehydriding of the Mg-Nb
2
004, in press.
sample are shown as a color plot in Figure 1 together with the
(
(
(
(
(
16) Schimmel, H. G.; Johnson, M. R.; Kearley, G. J.; Ramirez-Cuesta, A. J.;
Huot, J.; Mulder, F. M. Mater. Sci. Eng. B 2004, 108, 38.
17) Schimmel, H. G.; Johnson, M. R.; Kearley, G. J.; Ramirez-Cuesta, A. J.;
Huot, J.; Mulder, F. J. Alloys Compd., accepted for publication.
18) Massalski, T. B. Binary alloy phase diagrams, 2nd ed.; ASM Interna-
tional: Metals Park, 1990; Vol. 3.
(21) Flanagan, T. B.; Oates, W. A. In Hydrogen in Intermetallic Compounds
II; Schlapbach, L., Ed.; Springer: 1992.
(22) F. D. Manchester, Ed. Phase diagrams of binary hydrogen alloys; ASM
International: Materials Park, 2000.
(23) Williams, W. G.; Ibberson, R. M.; Day, P.; Enderby, J. Physica B 1998,
241-243, 234.
(24) Larson, A. C.; Dreele, R. B. V. General Structure Analysis System (GSAS),
LAUR 86-748; Los Alamos National Laboratory: 1994.
19) Pelletier, J. F.; Huot, J.; Sutton, M.; Schulz, R.; Sandy, A. R.; Lurio, L.
B.; Mochrie, S. G. J. Phys. ReV. B 2001, 63, 052103.
20) Yavari, A. R.; de Castro, J. F. R.; Vaughan, G.; Heunen, G. J. Alloys Compd.
2
003, 353, 246.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005 14349

A R T I C L E S
Schimmel et al.
MgD2y<2 phase for loadings higher than Dtot ) 0.3 next to a
stoichiometric MgD2 phase, but this appeared not to be necessary
for a good fit. Nevertheless it appears likely that individual Mg
particles start with deuteriding (hydriding) with such substo-
ichiometric composition. The same hydrogen deficient MgH2y<2
phase is observed in the protonated sample. The incoherent cross
section of deuterium for neutrons contributes significantly to
the background level; therefore a decrease of the background
level during deuterium extraction is observed in Figure 1.
During the deuteration (or hydriding) process (triangles in
Figure 2) the volume of the unit cell of magnesium (hexagonal
3
space-group P63/mmc) starts to increase by about 0.05 Å per
unit cell. The deuterium entering the magnesium metal phase
will increase the volume of the unit cell due to the extra volume
of the hydrogen atom which is generally assumed to be 2-3
3
25
Å per hydrogen atom. From the assessment of the Mg-H
system by Zeng et al. we derive a concentration of hydrogen in
-3
magnesium metal of about 10 under our experimental condi-
1
5,26
Figure 1. Image of neutron diffraction patterns during the desorption of
deuterium, plotted against desorption time. Against the same time axis also
the fitted integral content Dtot of deuterium is given (×) as well as the gas
pressure (b). The horizontal axis to the right shows a fraction of the
d-spacing range measured. Reflections from the five different phases are
indicated: Mg (m), MgD2 (h), MgO (o), Nb (n), and the perovskite phase
tions.
However, the lattice expansion that we observe is
much larger than expected from the composition MgH0.001 and
therefore has to be attributed mainly to temperature effects
induced by the energy released during hydrogen absorption. This
occurs inside the bulky sample, notwithstanding the fact that
the temperature regulation was constant within (1 K. The linear
thermal expansion coefficient of magnesium metal at these
(p).
hydride fraction and the pressure as a function of time. Five
phases could be identified in the patterns: Mg metal, MgH2,
MgO, R-NbDx and a small, but clearly visible and identifiable
contribution from a perovskite phase that we discuss below.
The data of the Mg-V sample was analyzed in the same way
and produced equivalent results for the Mg, MgH2, and MgO
phases. However, no sign of vanadium in metallic, deuteride,
hydride, or other form was found. One of the reasons for this
may be that V has a very low coherent neutron cross section.
Below, we will discuss the electron microscopy measurements
that we performed to detect the vanadium. First we present the
findings for the niobium catalyzed sample from the neutron
scattering experiment.
-
6
temperatures is 30 × 10 /K. This means that the volumetric
-
6
expansion is about 90 × 10 /K, from which it is derived that
the temperature of the magnesium metal has to increase by about
1
2 K to show this effect. The peak in the unit cell volume at a
loading of 0.2 may indicate that initially the temperature rise is
quite high as the heater circuit has to adapt to the situation in
which heat is generated inside the sample. The other phases in
the system show a similar and consistent dependence of the
unit cell volume upon hydrogen cycling.
The diffraction peaks of the R-Nb phase can be fitted
consistently in all of the patterns. The fits indicate that Nb is
present in an effective composition of approximately NbD
during the loading and unloading of Mg. In the last stage, when
all MgD2 has been converted into Mg, the pressure becomes
lower than the equilibrium pressure of NbD (about 0.1 bar for
this phase at this temperature), which then converts to Nb metal
by lowering the D content gradually. The lattice of NbDx shrinks
when its deuterium content x is reduced, which is clearly visible
in the s-shapes around d ) 2.4 and 1.7 Å in the upper part of
the color plot. When NbDx releases its deuterium, the MgD2
has vanished already completely. This means that for the actual
catalytic hydrogen splitting into atoms one should consider NbD
to be the more relevant phase.
3
.1. Neutron Scattering Results. From the fits we obtained
the phase fractions, lattice parameters, site occupancies, tem-
perature factors, and line widths. The line widths are determined
by the effective particle size and the strain in the particles. In
these experiments, no significant changes in the line widths of
the phases were found. Detailed results were obtained for the
deuterides as well as the hydrides of the Nb and V catalyzed
materials. Because the samples show identical results concerning
the Mg, Mg(H/D)2, and MgO, for the sake of brevity only the
MgNb0.05D2 sample is described in detail, and where necessary
remarks are made concerning the other samples. For the Mg-
Nb compound, the phase fractions, hydrogen occupancies, and
lattice parameters of the different phases are plotted as a function
of the effective composition MgNb0.05DDtot in Figure 2.
At room temperature bulk niobium metal possesses an
R-phase which contains hydrogen up to a concentration of
NbH0.25. It is a metallic hydride which possesses the typical
R-R′ transition such as that present in other metallic hydrides
The results for the site occupation of the magnesium deuteride
phase when its concentration is low is remarkable. We see that
2
1
such as palladium. In analogy to a gas-to-liquid transition a
2
2
critical point is present at 171 °C. Under our experimental
conditions the niobium particles are above their critical point
and the R-phase may contain hydrogen in the composition NbHx,
with 0 < x < 1, depending on the pressure.
(
i) the site occupancy y of deuterium in MgD2y can be as low
as about y ) 0.5, much lower than its stoichiometric value y )
in the bulk, and (ii) this affects the lattice parameters of the
1
unit cell. Above an effective sample composition Dtot ) 0.3,
the structural parameters of the magnesium deuteride phase
converge to the usual values, which remain constant from then
on. In the analysis it was also considered to have a separate
(
(
25) Fukai, Y. Z. Phys. Chem. 1989, 164, 165.
26) Zeng, K.; Klassen, T.; Oelerich, W.; Bormann, R. Int. J. Hydrogen Energy
1999, 24, 989.
14350 J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005

H
2
Cycling of Nb and V Catalyzed Nanostructured Mg
A R T I C L E S
Figure 2. Structural parameters of Mg (blue), MgD2 (red), and Nb (green) during loading (4) and unloading (O) of hydrogen as a function of Dtot, the
hydrogen content: (a) phase fractions of Mg, MgD2, and Nb; (b) site occupancies y of D in MgD2y and z of D in NbD6z; (c) unit cell volume of Mg; and
(d) lengths of the a- and c-axes of the unit cell of MgD2y.
with lattice parameters around 4.05 Å at room temperature.
Other cubic phases such as cubic NbO and MgO all have lattice
parameters that are clearly larger (close to 4.21 Å) than what is
observed here. We therefore propose that during the ball-milling
process part of the Nb has been alloyed with the outer oxide
layers of the Mg particles, leading to materials with the
composition like Nb(Mg0.333Nb0.667)O3. Because the peaks are
best visible in the fully dehydrided patterns, we first analyzed
these patterns carefully to be able to characterize this new phase
in more detail. Subsequently all patterns were refined with the
amount of this perovskite phase as free variable. The refinements
improved when this phase was added: for a single spectrum
Chi-square reduced from 2.23 to 2.10. Notably, the fitted phase
fraction appeared constant for all patterns during (de-)sorption,
which shows the stability of the fitted contribution during large
changes of the overall spectrum. Setting the molar amount of
Mg to 100%, the sample contains 4% of Nb metal, 0.3% of
Nb(Mg0.333Nb0.667)O3, and 7% of MgO, which leads to a bulk
stoichiometry Mg-Nb0.04-O0.08. We present the data that were
obtained for this phase in Table 1.
Figure 3 also gives the subpatterns of the individual phases
that occur in the measurements. Note that different D site
occupations in NbDx and MgDx have a relatively large influence
on the peak intensities of the respective phases. Because of the
very high counting rate of GEM these intensities are determined
very accurately. The perovskite phase has only a small number
of peaks of which several are visible by eye.
The case of V-catalyzed Mg is analogous to the Nb-catalyzed
material with regards to the Mg, Mg(H/D)2, and MgO phase.
Though pure V metal is not visible in this particular experiment
because of the extremely low coherent cross section of
Figure 3. Rietveld refinement of the fully dehydrided sample. Three unique
peaks that belong to the perovskite phase are indicated by the arrows in
the insets. Below: simulated patterns of the subspectra identified in the
Nb catalyzed Mg-H system, with, from top to bottom, MgD2, MgD1.2,
Mg, Nb, NbD, Nb(Mg0.333Nb0.667)O3, and MgO using the parameters from
the fits.
A careful observation of the patterns shows that peaks are
present that do not belong to one of the expected phases. Three
peaks are identifiable by eye: at 2.035, 1.661, and 1.175 Å as
shown in Figure 3. These are reflections at 4.07 Å divided by
(
27) de Mattan, N.; Husson, E.; Gaucher, P.; Morell, A. Mater. Res. Bul. 1990,
25, 427.
(28) Bonneau, P.; Garnier, P.; Calvarin, G.; Husson, E.; Gavarri, J. R. J. Solid
State Chem. 1991, 92, 350.
(29) Turzhevsky, S. A.; Novokov, D. L.; Gubanov, V. A.; Freeman, A. J. Phys.
ReV. B 1994, 50, 3200.
2,
x
6 and
x
12, which points toward a cubic phase. We found
(30) Casais, M. T.; Alonso, J. A.; Rasines, I.; Hidalgo, M. A. Mater. Res. Bul.
1995, 30, 201.
27,28
that known perovskites such as Pb(Mg0.333Nb0.667)O3,
SrN-
all possess a cubic perovskite structure
(31) Grin, Y.; M u¨ ller-Buschbaum, H.; von Schnering, H. G. Z. Naturforsch.,
2
9
30,31
bO3, and BaNbO3
B: Chem. Sci. 1997, 52, 153.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005 14351

A R T I C L E S
Schimmel et al.
Table 1. Structural Parameters of the Phases Observed during
Deuterium Cycling of Nb Catalyzed and Ball Milled MgD2
also in an amount that is consistent with the given sample
composition MgV0.05. Below we discuss the results of our TEM
experiments on both the niobium and vanadium catalyzed
sample.
MgD2
lattice parameter
space group
atom site in Wyckoff notation
a,b ) 4.530 Å, c ) 3.025 Å
P42/mnm no. 136 (tetragonal)
Mg: 2a (0,0,0)
In the Philips CM30UT-FEG TEM, separate powder particles
were examined with diffraction, imaging, and EDS (Energy
Dispersive Spectroscopy) for elemental analysis. Diffraction
D: 4f (0.3044(1),0.3044(1),0)
Mg: 1.000(1)
occupancy
D: 1.000(1)
2
patterns from 1 µm areas showed the presence of MgO and
MgD1.2
Mg clearly. In the electron microscopy graphs, shown in Figure
lattice parameter
space group
atom site in Wyckoff notation
a,b ) 4.56 Å, c)3.00 Å
P42/mnm no. 136 (tetragonal)
Mg: 2a (0,0,0)
4
, a typical result of the Nb sample is presented, where dark
particles with sizes up to 50 nm can be identified, but particles
of several hundred nanometers were also found. EDS on such
particles confirmed the presence of Nb, and diffraction patterns
consistent with metallic Nb were recorded.
D: 4f (0.3053(3),0.3053(3),0)
Mg: 1.000(1)
occupancy
D: 0.60(5)
Mg
Diffraction from powder particles in the V sample showed
MgO lattice spacings clearly. In some powder particles, both
MgO and Mg reflections were found. In this sample also dark
particles were found but smaller (2-20 nm) than those in the
Nb sample. Separate imaging of these particles turned out to
be difficult since the cubic lattice structure and lattice parameter
of metallic V (the phase of V as shown below) coincide with
those of MgO within a few percent. Thus, lattice imaging does
not distinguish between V and MgO with similar crystal size,
neither does dark field imaging. Most successful was imaging
without objective aperture in order to see absorption contrast
of the heavier V particles in a powder particle. In this way about
lattice parameter
space group
atom site in Wyckoff notation
occupancy
a,b ) 3.235 Å, c ) 5.253 Å
P63/mmc no. 194 (hexagonal)
Mg: 2c (1/3,2/3,1/4)
Mg: 1.000(1)
Nb
lattice parameter
space group
atom site in Wyckoff notation
occupancy
a ) 3.324 Å
Im 3h m No. 229 (body centered cubic)
Nb: 2a (0,0,0)
Nb: 1.000(1)
NbD
a ) 3.44 Å
Im 3h m no. 229 (solid solution bcc)
Nb: 2a (0,0,0)
D: 12d (1/4,0,1/2)
Nb: 1.000(1)
lattice parameter
space group
atom site in Wyckoff notation
occupancy
1
0 particles were selected for local EDS, showing a clear V
D: 0.16(2)
signal compared to no signal in the matrix surrounding such a
particle. An example from the V sample is shown in Figure 5.
Local diffraction on a particle showed a metallic cubic V phase.
The neutron diffraction appears unable to observe the V and
VDx directly because of several reasons: the low coherent V
cross section, small crystal size of the metallic V phase which
causes line broadening in the measurements, and the position
of the diffraction lines with respect to MgO.
Nb(Mg0.333Nb0.667)O3
lattice parameter
space group
a ) 4.070 Å
Pm 3h m no. 221 (cubic perovskite)
Nb: 1a (0,0,0)
atom site in Wyckoff notation
Nb,Mg: 1b (1/2,1/2,1/2)
O: 3c (0,1/2,1/2)
Nb: 1.0(1)
Nb: 0.7(2), Mg: 0.3(2)
O: 1.0(1)
occupancy
MgO
a ) 4.227 Å
4. Effect of Nanostructuring
lattice parameter
space group
atom site in Wyckoff notation
Fm 3h m no. 225 (cubic)
Mg: 4a (0,0,0)
O: 4b (1/2,1/2,1/2)
Mg: 1.000(1)
O: 1.000(1)
The materials investigated here are nanostructured heteroge-
neous mixtures of Mg, Mg(H/D)2, Nb(H/D)x or V(H/D)x, and
Mg containing oxides. The magnesium material undergoes a
clear first-order phase transition upon (de-)intercalation with
hydrogen or deuterium. There appears to be a remarkable and
surprising influence of the nanostructuring on the phase transi-
tion and intercalation behavior, which to our knowledge was
not observed before in hydrogen storage or more generally in
intercalation compounds.
occupancy
vanadium, VD or a V-perovskite similar to the one observed
above could be visible because of the presence of good coherent
scatterers such as D and O. However, the VD/H phase appears
to be invisible, which may also be due to the fact that its
strongest peak would overlap with the strongest MgO peak. Bulk
vanadium hydride also possesses an R-R′ transition with a
critical point at about 200 °C. Therefore under the experimental
conditions the hydrogen content of vanadium hydride particles
is a continuous function of hydrogen pressure. Above 200 °C
no phase transitions occur on hydrogen loading.22 One may find
compositions of VHx in the range 0 < x < 2, again depending
on the hydrogen pressure.
It is known that, for bulk micron sized particles, a shell of
magnesium hydride is formed, which prevents the hydriding of
4,5
the remaining metal core. For the nanostructured materials
22
the diffraction line widths indicate particle size in a range of
50 nm for the Mg and Mg(H/D)2 phases. From the rather equal
1
line widths of the Mg metal phase and the Mg(H/D)2 phase,
one deduces that there cannot be formed a closed shell of
Mg(H/D)2 around a Mg metal nucleus that blocks further
hydriding of the Mg core; the observed deuteride/hydride phase
particle size is too large for that compared to the grain size.
This is unlike the situation in bulk Mg, and it gives a clear
explanation why nanostructuring is advantageous. Because in
stoichiometric MgH2 H diffusion is very slow, the absence of
such a closed shell removes an important kinetic bottleneck for
application of Mg as storage material.
3
.2. Electron Microscopy Results for Mg-Nb and Mg-
V. As stated above, we could not identify any vanadium in our
series of diffraction patterns of the vanadium catalyzed sample.
To clarify this situation, we performed additional measurements
using transmission electron microscopy (TEM). Elemental
analysis confirmed that V is indeed present in the sample and
14352 J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005

H
2
Cycling of Nb and V Catalyzed Nanostructured Mg
A R T I C L E S
Figure 4. Bright field electron diffraction image showing several Nb particles. The diffraction pattern (taken with a convergent electron beam) shows 002
reflections with mutual 90° angles and distances to the origin that are consistent with the [001] beam direction in cubic Nb. The EDS shows a Nb line
together with Mg and O signals from the surrounding material. The Cu signal is caused by the supporting grid.
Figure 5. Absorption contrast image of part of a powder particle from the Mg-V sample. EDS shows the presence of V in particle P2. The Mg and O
signals are caused by the surrounding matrix. The Cu signal is caused by the supporting grid. The diffraction pattern (taken with a convergent electron beam)
shows 011 and 121 reflections with mutual angles and distances to the origin consistent with a [113] beam direction in the body centered cubic V phase.
Some spurious reflections from the surrounding matrix are also visible.
Second, it appears that substoichiometric nanostructured
MgHx < < 2 exists, which is impossible in bulk MgH2. This is
best observed in the first diffraction patterns of the hydrogen
loading, where the hydride phase is becoming increasingly
observable, and in the last patterns during unloading. The lower
hydrogen occupancy is also reflected in altered unit cell
dimensions. Such a hydrogen deficient phase likely exhibits a
much faster diffusion of hydrogen due to the large concentration
of vacancies, which will be a second factor in the enhancement
of the hydrogen sorption speed for nanostructured Mg. After
hydrogen and deuterium absorption cycles, these Mg particles
remain smaller than a few hundered nanometers (on average
compositions may be found, in contrast with the macroscopic
phase diagram. Whether or not such intermediate composition
occurs clearly depends on the difference in energy between such
a phase and the stoichiometric phase, compared to the energy
of the phase boundary. Since the area of the boundary, and
therefore its energy, scales with the square of the particle size,
whereas the energy difference between intermediate and sto-
ichiometric phases scales with the cube of the particle size, there
exists a certain particle size below which the intermediate
composition phase appears energetically favorable. Altered
intercalation behavior likely occurs more generally in nano-
structured hydrogen and lithium intercalation materials and in
more phenomena involving structural phase transitions.
1
50 nm according to diffraction peak widths). The occurrence
of substoichiometric MgHx<2 may be considered a characteristic
of nanocrystallinity.
Role of Nb(H/D) and V(H/D) . The addition of Nb or V
x
x
catalyzes both the desorption and the absorption process. In
19
The rationale for finding a hydrogen depleted MgH2 phase
may be as follows. In bulk particles the two-phase equilibrium
leads to a limited amount of phase boundaries because the strains
and mismatches between the two different lattices will be
minimized. Nanostructured particles may not easily accom-
modate two phases with a (moving) boundary between them,
because of the relatively large energy penalty of the induced
strains. In such cases transition phases with intermediate
contrast to the X-ray experiments of Pelletier et al. and Yavari
2
0
et al., we did not find the NbD0.6 or NbH0.6 phase but rather
higher D and H contents. We think that their finding is merely
indicative that the pressure in the sample chamber is very low
compared to our case during their desorption with high
desorption rates. We find that during the complete Mg hydrogen
cycling stage, Nb(H/D)x≈1 is present and only when the pressure
is reduced below about 100 mbar the H or D content starts
J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005 14353

A R T I C L E S
Schimmel et al.
lowering. The catalytic properties of NbH1 rather than those of
Nb metal are important for the desorption and the absorption
process under real conditions, with pressures around 0.5-5 bar.
The Nb diffraction peak shape shows a particle size of about
generates prevents further nucleation especially for these small
particle sizes. This means that only one nucleus may be formed,
which cannot form a closed layer around a core.
The importance of heat effects for nucleation and growth may
be estimated as follows. A particle contains N mol of magnesium
atoms of which a fraction, f, has been converted to magnesium
hydride. The heat that is released by the conversion to the
hydride phase is Hf ) fHMgH2N, where HMgH2 is the formation
enthalpy of magnesium hydride with respect to magnesium
metal and hydrogen gas, 74.5 kJ/mol. The heat will increase
the temperature of the magnesium particle by ∆T ) Hf/
(MMgCMgN), where we used the molar mass of magnesium MMg
) 0.0243 kg/mol and the heat capacity CMg ) 1026 J/(kg K).
The equilibrium pressure becomes equal to the applied pressure
(which we assumed to be 2 times the equilibrium pressure
initially) when the temperature of the magnesium particle
increases by about ∆T ) 30 K (for temperatures around 320
°C, for lower temperatures ∆T is lower). Knowing the param-
42 nm. As for the role of V in catalyzing the reactions of
hydrogen and magnesium metal, we found that vanadium breaks
up in much smaller particles than niobium during the ball milling
treatment. This means that more catalytic centers are available
on the surface of the magnesium particles. This is thought to
enhance the activity of vanadium with respect to niobium, which
1
is supported by the finding that vanadium is the better catalyst.
Unfortunately, we could not extract any information about the
phase that vanadium has in the ball milled samples from the
neutron diffraction.
Nb(Nb,Mg)O3 Perovskite a Possible Proton Conductor?
A MgO phase is present on the surface of the Mg particles,
even when the starting materials are always handled in an inert
atmosphere. During (de-)sorption the MgO is a barrier for the
transport of hydrogen. In this respect the small amount of the
cubic Nb-Mg-O perovskite phase resulting from alloying of
Nb and the MgO is an interesting finding. It is apparently present
with particle sizes some two times larger than the average MgO
eters, f ) ∆TMMgCMg/HMgH ) 1% can be calculated. A fraction
2
of 1% of a Mg particle transforming to MgH2 generates
sufficient heat to make the driving force zero due to the
temperature rise of the particle. When this heat has been
removed, clearly more hydrogen may absorb. A difference
between a nanoscale particle and bulk will be that a relatively
large fraction of the Mg in the nanoparticle is involved in a
hydride nucleation center. Again this will make it more likely
that only a single nucleation site is present in a nanoscale
particle. In this way heat transfer is important not only for
engineering sized batches but also for nucleation behavior at
the particle level.
3
2,33
particle size. Such perovskites are known to transport
and
store34 hydrogen and in addition may be catalysts themselves
for splitting of H2. The perovskite phase, penetrating the MgO
layer, could therefore be the missing part of the hydrogen
transport pathway.
The finding of such transition metal and Mg mixed oxide
phases may be more general. Oxides such as Cr2O3, V2O5,
3
5
36
Fe3O4, and Nb2O5, as well as some pure metals such as V
3
7
and Nb, are known to catalyze the hydrogen sorption properties
of magnesium. Because all these transition metals T form
compounds of the form MgxTyOz (perovskite, spinel and other
phases), it is interesting to investigate the catalytic activity of
these compounds too. For instance Mg3VxO8 are well-known
catalysts used for hydrogenation reactions in the petrochemical
industry.3 Simulation of such a V containing material shows
that the diffraction peaks strongly overlap with MgO.
Heat Effects. The observation of an increase in the unit cell
volume of Mg during loading corresponds to a temperature rise
of ∼12 K. The direct impact of this rise will be a higher
equilibrium pressure, i.e., a reduction of the driving force for
the growth of the hydride. Also the forming of new hydride
nucleation centers inside the same particle becomes less
probable. Both factors may play a role in the finding of a
hydrogen depleted MgHx<2, as well as in the absence of a closed
MgH2 shell around a Mg metal particle core (this should be
formed by hydride nucleation centers on all sides of a Mg
particle). We think that the heat that one growing nucleus
Conclusions
In situ neutron diffraction experiments reveal that nanostruc-
tured and Nb or V catalyzed Mg has characteristic properties
explaining the enhanced sorption speeds. The occurrence of
hydrogen depleted MgHx<2 and the absence of a closed MgH2
shell induced by a single MgHxe2 center in each particle may
be considered as characteristic of the nanocrystallinity of the
material. Unlike in bulk, heating of the material has now become
the remaining limiting factor. Under conditions relevant for
absorption and desorption, NbH was shown to be the relevant
Nb hydride phase for the catalytic activity. In addition a new
cubic perovskite phase with an estimated composition of Nb-
(Mg0.333Nb0.667)O3 was indicated that may also be catalytically
active as well as proton transporting. The V catalyzed material
shows smaller V particle sizes than the Nb catalyzed material,
but otherwise similar effects on the Mg phases are observed.
8,39
Acknowledgment. This work is a contribution from the Delft
Center of Sustainable Energy (DCSE). Financial support was
received from the Dutch Science Foundation (NWO) for access
to ISIS. Special sample containers were provided by J. Dreyer
and his team at ISIS.
(
(
(
32) Goodenough, J. B. Rep. Prog. Phys. 2004, 67, 1915.
33) Kreuer, K. D. Annu. ReV. Mater. Res. 2003, 33, 333.
34) Mandal, T. K.; Sebastian, L.; Gopalakrishnan, J.; Abrams, L.; Goodenough,
J. B. Mater. Res. Bull. 2004, 39, 2257.
(
35) Oelerich, W.; Klassen, T.; Bormann, R. J. Alloys Compd. 2001, 315, 237.
36) Barkhordarian, G.; Klassen, T.; Bormann, R. J. Alloys Compd. 2004, 364,
(
Supporting Information Available: Crystallographic infor-
mation files (CIF) are available for the Mg, MgD2, MgD1.2,
MgO, Nb, NbD, and Nb(Mg0.333Nb0.667)O3 phases. This material
is available free of charge via the Internet at http://pubs.acs.org.
2
42.
(
(
(
37) Liang, G.; Huot, J.; Boily, S.; Neste, A. V.; Schulz, R. J. Alloys Compd.
1
999, 291, 295.
38) Balderas-Tapia, L.; Wang, J. A.; Hernandez-Perez, I.; Aguilar-Rios, G. G.;
Schacht, P. Catal. Lett. 2004, 58, 3034.
39) Carraz a´ n, S.; Peres, C.; Bernard, J.; Ruwet, M.; Ruiz, P.; Delmon, B. J.
Catal. 1996, 158, 452.
JA051508A
14354 J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005