International Journal of Hydrogen Energy 31 (2006) 2052 – 2062

www.elsevier.com/locate/ijhydene

Dehydrogenation kinetics of pure and nickel-doped magnesium

hydride investigated by in situ time-resolved powder
X-ray diffraction
Torben R. Jensena,∗ , Anders Andreasenb, c,1 , Tejs Veggeb, d , Jens W. Andreasenb, e, f ,
Kenny Ståhle , Allan S. Pedersenb , Martin M. Nielsenf ,
Alfons M. Molenbroekg , Flemming Besenbacherh
a Department of Chemistry, Interdisciplinary Nanoscience Center (iNANO), University of Aarhus, DK-8000
Aarhus C, Denmark
b Materials Research Department, RisZ National Laboratory, DK-4000 Roskilde, Denmark
c Department of Chemical Engineering, Interdisciplinary Research Center for Catalysis, Technical University of Denmark,
DK-2800 Lyngby, Denmark
d Department of Physics and Center for Atomic Scale Materials Physics, Technical University of Denmark, DK-2800 Lyngby, Denmark
e Department of Chemistry, Technical University of Denmark, DK-2800 Lyngby, Denmark
f Danish Polymer Centre, RisZ National Laboratory, DK-4000 Roskilde, Denmark
g Haldor TopsZe A/S, DK-2800 Kgs. Lyngby, Denmark
h Department of Physics and Astronomy, Interdisciplinary Nanoscience Center (iNANO), University of Aarhus,
DK-8000 Aarhus C, Denmark

Received 18 May 2005; received in revised form 20 January 2006; accepted 2 February 2006
Available online 17 April 2006

Abstract
The dehydrogenation kinetics of pure and nickel (Ni)-doped (2 w/w%) magnesium hydride (MgH2 ) have been investigated
by in situ time-resolved powder X-ray diffraction (PXD). Deactivated samples, i.e. air exposed, are investigated in order to
focus on the effect of magnesium oxide (MgO) surface layers, which might be unavoidable for magnesium (Mg)-based storage
media for mobile applications. A curved position-sensitive detector covering 120◦ in 2
and a rotating anode X-ray source
provide a time resolution of 45 s and up to 90 powder patterns collected during an experiment under isothermal conditions.
A quartz capillary cell allowed the in situ study of gas/solid reactions. Three phases were identified: Mg, MgH2 and MgO
and their phase fractions were extracted by Rietveld refinement or integration of selected reflections from each phase. Dehy-
drogenation curves were constructed and analysed by the Johnson–Mehl–Avrami formalism in order to derive rate constants
at different temperatures. The apparent activation energies for dehydrogenation of pure and Ni-doped magnesium hydride
were EA ≈ 300 and 250 kJ/mol, respectively. Differential scanning calorimetry gave, EA = 270 kJ/mol for dehydrogenation
of the Ni-doped sample. The relatively high activation energies are due to MgO surface layers, retarding the diffusion of

* Corresponding author. Tel.: +45 8942 3894; fax: +45 8619 6199.
E-mail address: trj@chem.au.dk (T.R. Jensen).
1 Present address: MAN B&W A/S, Basic Research & Emission, Teglholmsgade 41, DK-2450 Copenhagen, Denmark.

0360-3199/$30.00 䉷 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2006.02.004
 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062
hydrogen (H2 ) out of MgH2 /Mg. The observed difference in EA of ca. 50 kJ/mol is likely due to the catalytic effect of Ni on
the recombination of H atoms to H2 molecules verified by theoretical considerations.
䉷 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
Keywords: Magnesium hydride; Nickel-doped magnesium hydride; Hydrogen storage materials; Gas–solid reactions; In situ powder X-ray
diffraction; Dehydrogenation kinetic; Differential scanning calorimetry
1. Introduction
Future utilisation of renewable energy can be realised
by implementing a ‘hydrogen society’ with hydrogen
(H2 ) as the energy carrier. One of the remaining prob-
lems is the development of a safe and economic H2
storage system [1,2]. Magnesium (Mg) and Mg-based
alloys have been explored intensively due to the high
theoretical gravimetric H2 density in Magnesium hy-
droxide (MgH2 ) m = 7.6 w/w% and the high volu-
metric H2 density, v = 110 g/L (i.e. 153% relative
to the density of liquid H2 , H2 (l) = 71 g/L). The H2
storage capacity in MgH2 exceeds the requirements
stated by the International Energy Association (IEA),
m > 5.0 w/w% [3]. In addition, the light metal Mg is
cheap and virtually limitless, i.e. occur to an extent of
0.13% in sea water and 2.76% by weight in the earth
crust. One of the problems however, is that the Mg–H
system has unfavourable thermodynamics and slow ki-
netics for H2 uptake and release. Thermodynamics dic-
tate that relatively high temperatures (T > 280 ◦ C) must
be applied to desorb H2 from pure MgH2 due to a forma-
tion enthalpy of MgH2 of,
Hf = −75 kJ/mol [4]. Fur-
thermore, investigations have shown that kinetics can be
improved remarkably either by reduction of the particle
size by ball milling, by mechanical alloying or sputter
deposition of other elements or by doping with carbon
[5–19]. In this way, the kinetics can be improved by re-
ducing the diffusion path of hydrogen atoms in Mg and
MgH2 , facilitating the diffusion process or by adding
catalysts on the metal surface for the dissociation or
formation of H2 molecules.
The detailed kinetics of both hydrogenation and de-
hydrogenation of Mg and its alloys have been inves-
tigated by a variety of different techniques and this is
possibly the group of H2 storage materials that has at-
tracted most interest during the past decades. A wide
range of apparent activation energies have been reported
suggesting a complex sorption mechanism sensitive to
the pretreatment and experimental conditions. This has
prompted us to investigate dehydrogenation of MgH2
and Nickel doped MgH2 using in situ time-resolved
powder X-ray diffraction (PXD) and determine phase
fractions of the involved phases, Mg, MgO and MgH2 .
2053
In situ PXD has proved to be a powerful tool for the
investigation of metal hydrides, for example Mg2 Ni
and LaNi5 [20,21]. Phase separation during hydrogena-
tion was observed for the alloys LaMg12 and La2 Mg17 ,
which decompose to LaH3 and MgH2 upon heating in
a H2 atmosphere [22]. H2 desorption from MgH2 /Nb
nanocomposite revealed an intermediate niobium hy-
dride phase suggested to act as a gateway for H2 flowing
out of the Mg reservoir, and the diffusion coefficient for
H2 was found to reach a maximum at an H/Nb ratio cor-
responding to the intermediate phase, NbH0.6 [23–27].
In situ PXD is an invaluable technique for studies of
complex hydrides such as NaAlH4 , which currently are
considered as very promising candidates for H atoms
storage materials and interact with H2 via a complex
reaction mechanism with several steps [28–30].
In this study, we will explore the use of in situ PXD
as a tool for kinetic investigations of new potential H2
storage materials. We will demonstrate that the low
X-ray scattering power of hydrogen is compensated by
a significant symmetry or volume change during hydro-
genation and dehydrogenation of most materials. There-
fore, H2 –metal interactions are often clearly detectable
by X-ray scattering. A kinetic study of the dehydrogena-
tion of MgH2 and Ni-doped MgH2 is presented, and the
results are discussed in relation to previous experimen-
tal work and theoretical calculations. This work is the
first part of our on-going investigation of deactivated,
i.e. air exposed, Mg and Mg-based alloys. The aim is to
gain further insight into the mechanism for hydrogena-
tion and dehydrogenation. For possible applications of
Mg-based storage media surface oxidation, i.e. creation
of a few nm thick MgO surface layer, can be unavoid-
able either due to oxygen impurities in the H2 gas or
due to air exposure.
2. Experimental
2.1. Materials and initial characterisation
The materials used in this investigation are Mg
(99%) and Ni-doped (2 w/w%) Mg commercially
available from Sigma-Aldrich and used as received.
For both materials, ex situ PXD analysis showed
2054 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062
diffraction peaks from Mg only. Elemental analysis for
the Ni-doped sample, denoted Mg(Ni) in the follow-
ing, was performed with Energy-Dispersive X-ray Spec-
troscopy (EDX) giving 87 ± 4, 11 ± 5 and 2 ± 2 w/w%
for Mg, O and Ni, respectively. The particle size dis-
tributions of the starting materials were determined by
laser diffraction using a Bechman Coulter LS 13 320
multi-wavelength particle size analyser. The Ni-doped
powder has a slightly higher mean particle diameter,
76±28 m compared to 60 ±22 m for the Mg sample.
Pure and Ni-doped MgH2 were prepared from the
starting materials in a Sartorius high-pressure balance
unit [31] at p(H2 ) = 30 bar and T = 400 ◦ C, using
99.9997% H2 from Air Liquide. The H2 uptake was
measured from the weight gain of the sample and veri-
fied by powder X-ray diffraction showing a nearly 100%
transformation of Mg into MgH2 . The materials were in-
vestigated with a scanning electron microscope (SEM)
JEOL JSM-5310LV operated at 15 kV. The morphology
of the hydrogenated samples appear more homogeneous
with a reduced number of deeper cracks possibly due
to the structural expansion of metal to hydride.
2.2. Calorimetry
Differential scanning calorimetry (DSC) data were
measured for the dehydrogenation of Mg(Ni)H2 from
RT to 490 ◦ C at heating rates, , of 2, 3.5, 5 and
10 ◦ C/ min in a flow of He (200 mL/min) using a Seiko
DSC 120 U. The hydride (ca. 10 mg) and the reference
(-Al2 O3 , ca. 10 mg) were loaded in open aluminium
crucibles. The heat flux is proportional to the rate of
reaction and goes through a maximum, TP . Apparent
activation energy can be extracted from these data
using the Kissinger equation [32].
d(ln(/TP2 )/d(TP−1 ) = −EA /R.
2.3. In situ powder X-ray diffraction
The X-ray instrument was built with a Rigaku rotating
anode X-ray source operated at 50 kV and 300 mA with
a curved graphite monochromator producing Cu K
radiation,  = 1.542 Å (∼ 107 photons/s). The inten-
sity of the diffracted beam was recorded with a curved
position-sensitive detector, INEL CPS 120, covering
120◦ in 2
with a resolution of ca. 0.03◦ . The samples
were exposed to X-rays 30–120 s corresponding to ac-
quisition time per powder pattern between 48 and 192 s
including read-out and overhead time.
The reactor for studies of gas–solid reactions at ele-
vated temperature and pressure was connected to a gas
Argon
Stainless steel tube
and fittings
Capillary
tube Diffracted beam
Oven
CPS 120 Detector
Xrays
Glass wool plugs Hydride bed
Fig. 1. Schematic illustration of the experimental set-up illustrating
the gas/solid reactor for in situ investigations allowing a controlled
gas flow, analysis of the exhaust gas, simultaneous heating and
exposure to X-ray radiation.
control system [33]. The powder sample was loaded be-
tween plugs of quartz glass wool in a quartz capillary
tube (0.7 mm OD) in order to fix the hydride bed and
allow a gas flow (Ar, 10 mL/min) through the sample
during data acquisition as shown in Fig. 1. The MgH2
was exposed to air during this procedure. The sam-
ples were heated by an external stream of hot nitro-
gen gas (20 ◦ C/ min) to a constant temperature in the
(1) range 350–400 ◦ C and the dehydrogenation was studied
at isothermal conditions. The experimental details for
the PXD experiments are described in Table 1. In each
experiment a new load of sample was dehydrogenated.
It was not possible to re-hydrogenate the samples under
the present experimental conditions.
Full profile Rietveld refinement of the in situ PXD
data was performed in order to determine the phase
fractions as a function of time during dehydrogenation
of MgH2 and Mg(Ni)H2 [34,35]. Three phases are ob-
served in the powder patterns, Mg, MgH2 and MgO
and the crystallographic information for these phases
are listed in Table 2. The Rietveld refinements are per-
formed with the winpow software [36] using pseudo-
Voigt profile functions, a shifted Chebyshev polynomial
(15 coefficients) for the background and the constraint
 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062 2055

Table 1
Experimental details for the in situ time-resolved powder X-ray diffraction investigations

No. MgH2 No. Mg(Ni)H2

Acquisition Exposure Acquisition Exposure

T /◦ C t/s t/s T /◦ C t/s t/s

1 360 182 120 5 350 182 120

2 380 182 120 6 370 182 120
3 390 48 30 7 380 182 120
4 400 48 30 8 390 182 120

Table 2
Unit cell parameters and crystallographic information for magnesium, magnesium hydride and magnesium oxide, the three phases included in
the Rietveld refinement

Property Substance

Mg MgH2 MgO

a/Å 3.2093 4.5168 4.2198

c/Å 5.2103 3.0205 —
V /Å3 46.47 61.62 75.14
Space group P63 /mmc P42 /mnm Fm3m
Crystal system Hexagonal Tetragonal Cubic
Ref. [39] [40] [41]

that the sum of phase fractions of Mg, MgH2 and MgO
is equal to one. The X-ray data is also analysed by nu-
merical peak integration of the MgH2 (1 1 0) reflection
and the (1 0 0) and (0 0 2) from Mg. MgO was excluded
from this analysis since the change in phase fraction is
small and it do not quantify the desorption of H2 from
MgH2 . The phase fractions are calculated by
(t) = [Ii (t) − Imin ]/[Imax − Imin ].
I (t) is the integrated intensity for reflection i at time
t, and Imin and Imax are the minimum and maximum
observed intensity, respectively. The background inten-
sity is estimated by, Imin . Phase fractions typically in
the range 0.2–0.8 are used for the kinetic investiga-
tion using an empirical rate law for isothermal phase
formation
time, to , can be observed prior to the reaction, which
can be related to a nucleation time given as
(t) = exp(−(k(t − to ))n ). (4)
Fitting Eqs. (3) and (4) to the experimental data provide
the rate constant, the Avrami exponents and the apparent
nucleation time. An apparent activation energy, EA , and
a pre-factor, A, for the overall reaction can be calculated
(2)
using Arrhenius law
k = A exp(−EA /RT ), (5)
where R is the gas constant and T the absolute temper-
ature. The delay, to , can be interpreted as a nucleation
time and an Arrhenius plot (− ln(to ) vs. 1/T ) gives an
apparent activation energy for the nucleation process,
EA0 [38].

(t) = exp(−(kt)n ). (3)

The parameters describing growth rates and nucle-
ation are contained in a rate constant, k, and the
Avrami exponent, n [37]. Eq. (3) is often denoted the
Johnson–Mehl–Avrami (JMA) model. To eliminate the
temperature variations of the Avrami exponent a fixed
value can be chosen, e.g. n = 4. In some cases a delay
3. Results
The decomposition of Ni-doped MgH2 , Mg(Ni)H2 ,
is illustrated by 40 consecutive powder X-ray diffrac-
tion patterns in Fig. 2 showing the dehydrogenation of
MgH2 by the disappearance of the (1 1 0) and (1 0 1) re-
flections and the corresponding appearance of Mg(1 0 0)
2056 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062

Fig. 2. Time-resolved in situ powder X-ray diffraction of dehydrogenation of Mg(Ni)H2 . A stack of 40 consecutive diffraction patterns
measured at T = 390 ◦ C and with acquisition time t = 182 s.

Fig. 4. Phase fractions determined by Rietveld refinement of MgH2 ,

Fig. 3. Experimental and calculated powder X-ray diffraction patterns Mg and MgO as a function of time for Mg(Ni)H2 dehydrogenated
for a partly dehydrogenated sample of Mg(Ni)H2 (T = 370 ◦ C). at 370 ◦ C.
The difference between observed and calculated intensities is also
included. The X, # and O markers correspond to MgH2 , Mg and
MgO, respectively (Rp = 6.37%, Rwp = 9.41%).

Phase fractions derived by Rietveld refinement,

and Mg(0 0 2) reflections. Rietveld refinement of a se-
lected pattern of Mg(Ni) reveal the existence of three
phases during dehydrogenation of the sample as shown
in the Fig. 3.
Fig. 4, for Mg and MgH2 vs. time have a sigmoidal
shape, whereas the fraction of MgO is only slightly in-
creasing. Dehydrogenation curves, i.e. phase fractions
of MgH2 vs. time have been constructed for all exper-
iments as shown in Fig. 5 for the Ni-doped sample.
 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062 2057

similar formation curves and there were no signs of pre-

Fig. 5. Dehydrogenation curves for pure Mg(Ni)H2 extracted from
the in situ PXD data (points) measured at fixed temperatures of 350,
370, 380 and 390 ◦ C. The theoretical curves are a fit using Eq. (3)
and with the results given in Table 3.
ferred crystallite orientation in the samples. An empiri-
cal rate law (JMA type), Eqs. (3) and (4), can describe
the data as shown in Fig. 5 (curves). Avrami exponents,
n, and rate constants, k, are presented in Table 3 using
Eq. (3) and in Table 4 rate constants, k  , are given for a
fixed value of, n=4. Rate constants, k  , and delay time,
to , extracted with Eq. (4) and a fixed exponent n = 2
are also given in Table 4. An Arrhenius plot to provide
activation energies for PXD data is shown in Fig. 6.
In Fig. 7 the DSC data is depicted, which shows
increasing peak temperature, Tp , with increasing heating
rate, . From observations of the heating rate and the
corresponding peak temperature the apparent activation
energy can be extracted using the Kissinger equation
[32], see Fig. 8.

4. Discussion

The phase fractions determined by Rietveld refinement Faster dehydrogenation kinetics with higher temper-
and those by integration of Bragg reflection intensities atures are clearly visible along with the sigmoidal shape
were identical within the experimental accuracy. Inte- of all curves for each sample, as illustrated in Fig. 5,
gration of reflections with different Miller indices gave in accordance with previous investigations [42,43].

Table 3
Kinetic parameters for the dehydrogenation of both pure and nickel-doped MgH2 obtained using Eq. (3)

No. MgH2 No. Mg(Ni)H2

T /◦ C k/s −1 n T /◦ C k/s −1 n

1 360 3.93 × 10−5 2.26 5 350 2.72 × 10−5 4.20

2 380 1.53 × 10−4 6.99 6 370 1.00 × 10−4 6.29
3 390 4.53 × 10−4 3.29 7 380 2.00 × 10−4 6.47
4 400 1.13 × 10−3 4.41 8 390 5.30 × 10−4 14.9

EA /kJ/mol 299 249

Table 4
Kinetic parameters for the dehydrogenation of pure and nickel-doped MgH2 obtained using Eq. (3), with a fixed value of the exponent, n = 4
(the rate constant, k  ), and Eq. (4), with n = 2 (the rate constant, k  and apparent nucleation time, to )

No. MgH2 No. Mg(Ni)H2

T /◦ C k  /s−1 k  /s−1 to /s T /◦ C k  /s−1 k  /s−1 to /s

1 360 4.12 × 10−5 4.73 × 10−5 3.99 × 103 5 350 2.71 × 10−5 3.88 × 10−5 1.13 × 104
2 380 1.50 × 10−4 2.20 × 10−4 2.22 × 103 6 370 9.99 × 10−5 1.43 × 10−4 3.27 × 103
3 390 4.57 × 10−4 6.19 × 10−4 577 7 380 1.92 × 10−4 2.75 × 10−4 1.68 × 103
4 400 1.13 × 10−3 1.58 × 10−3 258 8 390 4.99 × 10−4 7.35 × 10−4 66

EA /kJ/mol 291 308 241 249

EAO /kJ/mol 241 216
2058 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062

Fig. 7. Differential scanning calorimetry for the dehydrogenation

Fig. 6. Arrhenius plot of the kinetic data given in Table 3 for the
dehydrogenation of both pure MgH2 and Mg(Ni)H2 .
of Mg(Ni)H2 showing the heat flux as a function of temperature
measured from RT to 490 ◦ C at heating rates, , of 2, 3.5, 5
and 10 ◦ C/ min in a flow of He. Each curve has been scaled for
illustrative reasons.

Dehydrogenation of the Ni-doped sample is faster at

each temperature, see Fig. 6. The observed data seem to
indicate a nucleation time, to , prior to the onset of dehy-
drogenation, e.g. by visual inspection of Fig. 2. A fixed
Avrami exponent, n = 2, was chosen in order to extract
to . The data was fitted both with free and fixed Avrami
exponent, n, as seen in Tables 3 and 4. The fluctuations
observed for free n can be due to deviation from pure
‘nucleation and growth’ kinetics that is described by the
model. The Avrami exponent was observed to increase
from n = 2 to 4 for H2 desorption from MgH2 due a
temperature increase in accordance with previous work
[44,45]. However, it is generally agreed that the temper- Fig. 8. The DSC data (Fig. 7) displayed as a Kissinger plot providing
an apparent activation energy for the dehydrogenation of Mg(Ni)H2 .
ature dependency of the rate constant is not sensitive to
the choice of function describing the time dependence
[46,47]. The X-ray data was also analysed using a fixed 4.1. Comparison with related work
value of n = 4. The apparent activation energy of both
pure and Ni-doped samples are listed in Tables 3 and 4 In the present work, the focus is on the apparent acti-
showing values of ca. 300 and 250 kJ/mol, respectively, vation energy for the overall dehydrogenation process.
for the dehydrogenation. The estimated apparent acti- This reaction, metals with H2 , consists of several steps:
vation energy for the nucleation process, EAO , is less (i) nucleation and growth of Mg, (ii) diffusion of H
than EA for the overall reaction in contrast to previ- atoms in bulk Mg/MgH2 , (iii) surface penetration, i.e.
ous work [38]. From a Kissinger plot of the DSC data through MgO for oxidised samples, (iv) 2D surface dif-
(Fig. 8) the apparent activation energy for the dehy- fusion of H, (v) recombination of 2H to H2 a catalysed
drogenation of Mg(Ni)H2 is found to be 270 kJ/mol in process. Furthermore, (vi) for alloys cation migration
agreement with the activation energy of, ca. 250 kJ/mol can also be of importance during phase formation/phase
found from the in situ PXD experiments. Note, that the separation induced by hydrogenation or dehydrogena-
two experimental techniques are different in nature: the tion. The rate-determining step (the slowest) has the
former operated with continuous heating and the latter largest contribution to the apparent activation energy
at isothermal conditions. when EA 
H .
 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062 2059

Table 5
Apparent activation energy for hydrogenation and dehydrogenation of magnesium and some selected alloys measured by different techniques

Sample Technique EA /(kJ/mol) Ref.

Hydrogenation–magnesium
Mg (powder) 1 100 A [44]
Mg (vapour deposit) 3 138 [48]
Mg (foil) 2 270–314 D [42]

Dehydrogenation
MgH2 (powder) 1 120 A [49]
MgH2 (powder) 2 120–126 A [47]
MgH2 (vapour deposit) 3 132 A [48]
MgH2 (powder) 4 142 A [43]
MgH2 (powder) 1 156 A [49]
Mg (powder) 1 160 A [44]
MgH2 (powder) 2 233a D [31]
MgH2 (powder) 5 299 D This work

Hydrogenation—alloys of magnesium
Mg0.98 Ni0.02 (powder) 1 66 A [50]
Mg/Mg2 Cu (powder) 3 100 [48]
Mgx Niy Hz (powder)b 1 120 A [51]

Dehydrogenation
Mg0.95 V0.05 H2 1 52–99 A [52]
Mg0.95 V0.05 H2 1 62 A [53]
0.99MgH2 · 0.01Nb2 O3 1 62 A [54]
Mg0.95 Nb0.05 H2 1 62–77 A [25]
Mg/Mg2 CuHz (powder) 3 106 [48]
Mg0.87 Al0.07 Ni0.03 Mn0.03 H2 6 100–115 A [55]
MgH2 /Mg2 Cu 5 108 D [56]
MgH2 /MgCu2 5 160 D [56]
MgAlHx 5 160 D [57]
Mg(Ni)H2 (powder) 5 249 D This work
Mg(Ni)H2 (powder) 6 270 D This work

Activated samples, A, and deactivated, D. Technique: (1) Sievert-type apparatus, (2) high-pressure micro-balance, (3) pressure sweep method,
(4) thermal desorption spectroscopy, (5) time-resolved in situ X-ray diffraction and (6) TG/DSC.
a E was extracted by reanalysing the data in Ref. [31].
A
b The exact composition is not given in Ref. [51].

An overview of experimentally determined apparent
activation energies for formation and decomposition of
MgH2 and selected hydrides of Mg-based alloys is pre-
sented in Table 5. This table covers a range of differ-
ent materials and experimental techniques and reveals
a significant variation between the apparent activation
energies measured. There is a clear difference in EA for
activated samples measured after several cycles of hy-
drogenation/dehydrogenation and deactivated samples
exposed to air and measured after the first cycle.
4.2. The influence of surface oxidation
The diffracted intensity indicates an increasing
amount of crystalline MgO (see Fig. 4) possibly
due to crystallisation of amorphous MgO, reaction
of Mg with the quartz capillary (SiO2 ) or with gas
impurities. Mg powders with different particle size
were previously characterised by both neutron activa-
tion and wet-chemical analysis in order to measure the
amount of MgO [58]. These results suggest that ca.
4 w/w% oxygen might be expected for the particle size
used in this investigation assuming all oxygen being
present in a surface oxide layer.
MgO is expected to act as a barrier between bulk
MgH2 , gas phase and Ni, due to its dense rock-salt
structure limiting the diffusion of H2 atoms [59,60].
For both samples investigated diffusion of atomic H2
through Mg/MgH2 and MgO is contributing to the over-
all rate, whereas two-dimensional surface diffusion of
H on MgO is assumed to have negligible activation en-
ergy and thus a high rate. The apparent activation ener-
gies reported in this work are somewhat higher; possibly
due to exposure of the samples to the atmosphere prior
2060 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062
to PXD measurements that has created an oxidised sur-
face layer of MgO on the particles of MgH2 . This work
serves as an estimate of the maximal apparent activation
energy and the influence of MgO surface layers.
A freshly hydrogenated material can be considered
as activated with defects in the surface oxide layer. Dif-
fusion of H2 atoms to the gas/solid interface is thereby
facilitated. When the activated sample is exposed to air
it is deactivated by oxidation of Mg forming a more
perfect surface oxide layer. The results from the mate-
rials characterisation by EDX indicate that additional
contamination by oxygen was introduced after hydro-
genation.
On the other hand, Mg is known to have a high vapour
pressure that is expected to lead to considerable sinter-
ing of Mg particles. Mg and H2 was found to evaporate
simultaneous from thin films of MgH2 . The apparent
desorption energy of Mg from thin films of Mg and Mg
covered by MgO was found to be 146 and 314 kJ/mol,
respectively. Therefore, surface layers of MgO can be
considered as retarding the release and uptake of H2 but
also preventing Mg sublimation and thereby increase
the cyclic stability of, e.g. Mg nanoparticles. Thereby,
MgO has a stabilising effect [61,62].
4.3. The catalytic effect of nickel
It is generally agreed that Mg does not promote dis-
sociation of H2 molecules, H2 , to atoms on the metal
surface [63,64]. NZrskov et al. calculated the potential
energy surface (PES) of H2 adsorption and dissociation
on the Mg(0 0 0 1) close-packed surface, which is the
most thermodynamically stable [65]. These calculations
showed an activation barrier for H2 dissociation of ca.
0.5 eV (∼ 50 kJ/mol) in qualitative agreement with ex-
periments [66]. The bridge site (i.e. position between
two identical atoms) has the lowest dissociation bar-
rier, 0.4 eV determined by LDA-calculations [67]. Re-
cent theoretical work determine the activation energy for
H2 dissociation on a Mg(0 0 0 1) surface in the bridge
site to be EA ≈ 100 kJ/mol, without the presence of
a molecular chemisorbed state acting as a precursor to
dissociation [68]. In contrast, calculations and experi-
mental work show that the dissociation of H2 on Ni has
a negligible activation barrier around 0–10 kJ/mol, i.e.
the addition of Ni is expected to lower the dissociation
barrier significantly [69–71].
Furthermore, assuming microscopic reversibility the
same effect would apply for recombination of atomic H
into H2 molecules.
The experimentally observed difference in EA for
the dehydrogenation of MgH2 and Mg(Ni)H2 (300
and 250 kJ/mol, respectively) is indeed 50 kJ/mol. Fur-
thermore, the two samples of MgH2 and Mg(Ni)H2
have similar particle sizes and oxygen content sug-
gesting that the overall bulk diffusion of atomic H
through Mg/MgH2 /MgO are comparable for the
samples.
This might indicate that the dissociation of H2 and
recombination of H to H2 mainly occur on Ni due to its
lower barrier compared to pure Mg even with as little
as 2 w/w% Ni present. This may suggest that Ni mainly
catalyses the dissociation/recombination reaction bar-
rier for H2 and has minor influence on other processes
as mentioned earlier. Ni, like Mg, is oxidised in air to
the oxide, NiO. In contrast to MgO, NiO is readily re-
duced by H2 gas [72].
4.4. Comparison with magnesium alloys
Table 5 reveals a large span in EA values for Mg and
its alloys, supporting the hypothesis that the overall re-
action consists of several individual steps [73]. Alloying
Mg or including additives can lead to more complex re-
action schemes for the metal–H2 interaction. Extremely
fast kinetics was observed for the additives Nb2 O5 and
V where up to 6 w/w% H2 was absorbed or desorbed
within a few minutes [11,53,54]. The very fast kinetics is
due to the catalytic effect of the additive as well as ball-
milling of the samples prior to the measurements. In this
work, the catalytic effect of Ni was clearly detectable by
in situ PXD. In some cases alloying Mg can influence
thermodynamics, e.g. for Mg0.87 Al0.07 Ni0.03 Mn0.03 H2 ,
and the enthalpy for H2 desorption was
H = 55 kJ/
mol [55].
In a parallel study we investigated the alloy
Mg0.85 Cu0.15 . Hydrogenation at low- and high-H2 pres-
sure provided samples of (A) MgH2 + Mg2 Cu and (B)
MgH2 + MgCu2 , respectively, and in situ PXD reveal
EA for the dehydrogenation of deactivated samples to
be 108 and 160 kJ/mol, respectively [56]. The differ-
ence is possibly due to the phase formation reaction
during dehydrogenation, 3Mg + MgCu2 → 2Mg2 Cu
for (B). The same experimental set-up was utilised for
investigation of a Mg–Al alloy, Mg17 Al12 , which phase
separated to MgH2 +Al during hydrogenation [57]. De-
hydrogenation of this air-exposed (deactivated) sample
was followed by phase formation by cation migration
to form an MgAl-alloy, with EA = 160 kJ/mol.
Despite three decades of intensive research the com-
plexity of Mg and its alloys reaction with H2 is still
not fully understood. Deeper knowledge of all the in-
dividual steps in the overall reaction might reveal new
ideas for possible applications. Mg is still considered
 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062
as having potential applications as H2 storage material
in future.
5. Conclusion
The present results demonstrates that in situ powder
X-ray diffraction is a useful technique for kinetic inves-
tigation of dehydrogenation of Mg and Mg-doped al-
loys, in this case with only 2 w/w% Ni, Mg(Ni)H2 . The
method gives information on the phase composition of
the sample as a function of time and temperature. Fur-
thermore, phase fractions can be extracted as a func-
tion of time as demonstrated in this work and utilised
for kinetic analysis providing apparent activation ener-
gies. The method is sensitive and can clearly probe the
difference in dehydrogenation kinetics of MgH2 and
Mg(Ni)H2 , i.e. EA ≈ 300 and 250 kJ/mol, respectively.
The values are relatively high compared to previous
work possibly because the samples were exposed to air
prior to the investigations. The lower activation energy
for the Ni-doped sample suggests that Ni catalyses the
recombination of H to H2 .
Acknowledgements
The Danish Technical Research Council is acknowl-
edged for funding the center of excellence Towards the
hydrogen-based society. TRJ is grateful to the Danish
Natural Research Council for at Steno stipend and to
Carlsberg Fondet, and T. Vegge to the Danish Research
Agency and Danfoss Corporate Ventures A/S (Grant #
2013-01-0043).
References
[1] Schlapbach L, Züttel A. Nature 2001;414:353.
[2] Ritter JA, Ebner AD, Wang J, Zidan R. Mater Today 2003;
(Sept):18.
[3] http://www.ieahia.org/tasks/task17.html .
[4] US Department of Energy. http://www.eere.energy.gov/
hydrogenandfuelcells/hydrogen/-storage.html .
[5] Zaluska A, Zaluski L, Ström-Olsen. J Appl Phys A 2001;72:157.
[6] Gross KJ, Spatz P, Züttel A, Schlapbach L. J Alloys Compd
1997;261:276.
[7] von Zeppelin F, Reule H, Hirscher M. J Alloys Compd
2002;330–332:723.
[8] Tran NE, Imam MA, Feng CR. J Alloys Compd 2003;259:225.
[9] Rivoirard S, de Rango P, Fruchart D, Charbonnier J, Vempaire
D. J Alloys Compd 2003;356–357:622.
[10] Wang J, Wu F, Shan ZhQ. J Alloys Compd 2003;359:315.
[11] Liang G, Huot J, Boily S, van Neste A, Schulz R. J Alloys
Compd 1999;292:247.
2061
[12] Reiser A, Bogdanović B, Schlichte K. Int J Hydrogen Energy
2000;25:425.
[13] Schlapbach L, editor. Hydrogen in intermetallic compounds, II:
surface and dynamic properties, applications. Topics in applied
physics, vol. 67. Berlin: Springer; 1992.
[14] Léon A, Knystautas EJ, Huot J, Schulz R. J Alloys Compd
2002;345:158.
[15] Schimmel HG, Johnson MR, Kearley GJ, Ramirez-Cuesta AJ,
Huot J, Mulder FM. J Alloys Compd 2005;393(1–2):1.
[16] Tessier JP, Palau P, Huot J, Schulz R, Guay D. J Alloys Compd
2004;376(1–2):180.
[17] Yavari AR, LeMoulec A, de Castro FR, Deledda S, Frederichs
O, Botta WJ, et al. Scripta Mater 2005;52(8):719.
[18] Dehouche Z, Klassen T, Oelerich W, Goyette J, Bose TK,
Schulz R. J Alloys Compd 2002;347:319.
[19] Oelerich W, Klassen T, Bormann R. J Alloys Compd
2001;315:237.
[20] Hirata T, Matsumoto T, Amano M, Sasaki Y. J Phys F—Met
Phys 1981;11:521.
[21] Notten PHL, Daams JLC, De Veirman AEM, Staals AA. J
Alloys Compd 1994;209:85.
[22] Sun D, Gingle F, Nakamura Y, Enoki H, Bououdina M, Akiba
E. J Alloys Compd 2002;333:103.
[23] Pelletier J, Huot J, Sutton M, Schulz R, Sandy A, Lurio L,
‘ et al. Phys Rev B 2001;63:052103.
[24] Huot J, Pelletier JF, Liang G, Sutton M, Schulz R. J Alloys
Compd 2002;330–332:727.
[25] Huot J, Pelletier JF, Lurio LB, Sutton M, Schulz R. J Alloys
Compd 2003;348:319.
[26] Schimmel HG, Huot J, Chapon LC, Tichelaar FD, Mulder FM.
J Am Chem Soc 2005;127:14348.
[27] Charbonnier J, de Rango P, Fruchart D, Miraglia S, Skryabina
N, Huot J, et al. J Alloys Compd 2005;404–406:541.
[28] Brinks HW, Hauback BC, Norby P, Fjellvåg H. J Alloys Compd
2003;351:222.
[29] Gross KJ, Sandrock G, Thomas GJ. J Alloys Compd
2002;330–332:691.
[30] Balogh MP, Tibbetts GG, Pinkerton FE, Meisner GP, Olk CH.
J Alloys Compd 2003;350:136.
[31] Pedersen A, KjZller J, Larsen B, Vigeholm B. Int J Hydrogen
Energy 1983;8(3):205.
[32] Kissinger HE. Anal Chem 1957;29:1702.
[33] Clausen BS, Steffensen G, Fabius B, Villadsen J, Feidenhans’l
R, TopsZe H. J Catal 1991;132:524.
[34] Young RA, editor. The Rietveld method. Oxford: Oxford
University Press; 1993.
[35] McCusker L, Von Dreele R, Cox D, Louër D, Scardi P. J Appl
Crystallogr 1999;32:36.
[36] WINPOW, a Rietveld refinement program written by Kenny
Ståhl, Department of Chemistry, Technical University of
Denmark, can be downloaded from http://struktur.kemi.dtu.dk/
kenny/powder_program.html .
[37] Kelton K. Mater Sci Eng A 1997;226–228:142.
[38] Christensen AN, Jensen TR, Norby P, Hanson JC. Chem Mater
1998;10:1688.
[39] Straumanis M. J Appl Phys 1949;20:726.
[40] Ellinger F, Holley Jr. C, McInteer B, Pavone D, Potter R,
Staritzky E, et al. J Am Chem Soc 1955;77:2647.
[41] Karen P, Kjekshus A, Huang Q, Karen V. J Alloys Compd
1999;282:72.
[42] Pedersen A, Jensen K, Larsen B, Vigeholm B. J Less-Common
Met 1987;131:31.
2062 T.R. Jensen et al. / International Journal of Hydrogen Energy 31 (2006) 2052 – 2062
[43] Han J, Pezat M, Lee J-Y. J Less-Common Met 1987;
130:395.
[44] Fernandez JF, Sanchez CR. J Alloys Compd 2002;340:189.
[45] Fernandez JF, Sanchez CR. J Alloys Compd 2003;356–357:348.
[46] Vigeholm B, KjZller J, Larsen B, Pedersen A. J Less-Common
Met 1983;89:135.
[47] Stander C. J Inorg Nucl Chem 1977;39:221.
[48] Karty A, Grunzweig-Genossar J, Rudman P. J Appl Phys
1979;50(11):7200.
[49] Huot J, Liang G, Boily S, Van Neste A, Schulz R. J Alloys
Compd 1999;293–295:495.
[50] Friedlmeier G, Groll M. J Alloys Compd 1997;253–254:550.
[51] Martin M, Gommel G, Borkhardt C, Fromm E. J Alloys Compd
1996;238:193.
[52] Liang G, Huot J, Boily S, Schulz R. J Alloys Compd
2000;305:495.
[53] Liang G, Huot J, Boily S, Van Neste A, Schulz R. J Alloys
Compd 1999;291:295.
[54] Barkhordarian G, Klassen T, Bormann R. J Alloys Compd
2004;364:242.
[55] Delchev P, Solsona P, Drenchev B, Drenchev N, Spassov T,
Baro MD. J Alloys Compd 2005;388:98.
[56] Andreasen A, SZrensen MB, Burkarl R, MZller B, Molenbroek
AM, Pedersen AS, et al. Appl Phys A 2006;82:515.
[57] Andreasen A, SZrensen MB, Burkarl R, MZller B, Molenbroek
AM, Pedersen AS, et al. J Alloys Compd 2005;404–406:323.
[58] Vigeholm B. Chemical energy storage based on metal hydrides.
Technical Report RisZ-M-2608, Risoe National Laboratory,
DK-4000 Roskilde, Denmark, 1989 [in Danish].
[59] Hjort P, Krozer A, Kasemo B. J Alloys Compd 1996;237:74.
[60] Gonzalez R, Chen Y, Tsang KL. Phys Rev B 1982;26:4637.
[61] Ostenfeld CW, Chorkendorff I. Surf Sci 2006, submitted for
publication.
[62] Ostenfeld CW, Davis JC, Vegge T, Chorkendorff I. Surf Sci
2005;584(1):17.
[63] Sprunger PT, Plummer EW. Surf Sci 1994;307–309:118.
[64] NZrskov JK, Besenbacher F. J Less-Common Met
1987;130:475.
[65] NZrskov JK, HoumZller A, Johansson PK, Lundquist BI. Phys
Rev Lett 1981;46(4):257.
[66] Sprunger PT, Plummer EW. Chem Phys Lett 1991;187:559.
[67] Bird DM, Clarke LJ, Payne MC, Stich I. Chem Phys Lett
1993;212:518.
[68] Vegge T. Phys Rev B 2004;70:035412.
[69] NZrskov JK, Hammer B. Nature 1995;376:238.
[70] Harris J, Anderson S. Phys Rev Lett 1985;55:1583.
[71] Robota HJ, Vielhaber W, Lin MC, Segner J, Ertl G. Surf Sci
1985;155:101.
[72] Rodriguez JA, Hanson JC, Frenkel AI, Kim JY, Pérez M. J Am
Chem Soc 2002;124(2):346.
[73] Andreasen A, Vegge T, Pedersen AS. J Phys Chem B
2005;109:3340.