OPEN Evidence of the hydrogen release

SUBJECT AREAS:
mechanism in bulk MgH2
MATERIALS FOR ENERGY Kazuhiro Nogita1,2, Xuan Q. Tran1, Tomokazu Yamamoto2, Eishi Tanaka2, Stuart D. McDonald1,
AND CATALYSIS
Christopher M. Gourlay3, Kazuhiro Yasuda2 & Syo Matsumura2
CHEMICAL ENGINEERING

1
Nihon Superior Centre for the Manufacture of Electronic Materials, School of Mechanical and Mining Engineering, The University of
Received Queensland, Brisbane, QLD 4072, Australia, 2Department of Applied Quantum Physics and Nuclear Engineering and the
27 November 2014 Ultramicroscopy Research Center, Kyushu University, Fukuoka 819-0395, Japan, 3Department of Materials, Imperial College,
London. SW7 2AZ. UK.
Accepted
21 January 2015
Hydrogen has the potential to power much of the modern world with only water as a by-product, but storing
Published hydrogen safely and efficiently in solid form such as magnesium hydride remains a major obstacle. A
13 February 2015 significant challenge has been the difficulty of proving the hydriding/dehydriding mechanisms and,
therefore, the mechanisms have long been the subject of debate. Here we use in situ ultra-high voltage
transmission electron microscopy (TEM) to directly verify the mechanisms of the hydride decomposition of
bulk MgH2 in Mg-Ni alloys. We find that the hydrogen release mechanism from bulk (2 mm) MgH2 particles
Correspondence and is based on the growth of multiple pre-existing Mg crystallites within the MgH2 matrix, present due to the
requests for materials difficulty of fully transforming all Mg during a hydrogenation cycle whereas, in thin samples analogous to
should be addressed to nano-powders, dehydriding occurs by a ‘shrinking core’ mechanism.
K.N. (k.nogita@uq.
edu.au)

T
he storage of hydrogen in solid form as magnesium hydride is a promising technology that is being
developed for fuel cells for automotive and stationary applications and to enable safer hydrogen transporta-
tion1–3. Pure magnesium has a hydrogen carrying capacity of around 7.6 wt%, however, the resulting hydride
is relatively stable and the temperature must be increased to 300uC at 1 bar H2 for the hydrogen to be released with
acceptable release rate with realistic industry use2. Design goals for on-board storage (automotive) aim for a
temperature of hydrogen release of between 60 and 120uC2, however land-based consumer and industrial storage
systems do not have such stringent requirements4. Much research on Mg-based H-storage materials is based on
ball milled nano-scale powders due to their superior H-sorption properties over bulk powders2,5. Another
approach is to improve the H-sorption properties of bulk powders, for example, by treating liquid Mg-alloys
with alkali and alkali-earth metals prior to casting and powder production6. A significant challenge in research on
both nano-powders and bulk-powders has been the difficulty of proving the hydriding/dehydriding mechanisms
and, therefore, the mechanisms have long been the subject of debate1–3. Multiple models for the desorption of
MgH2 have been developed and these have been placed in several categories including ‘shrinking core’ models3,7–9,
‘nucleation and growth’ models3,7–9, ‘multiple step’ kinetic models3,7–9, ‘migration and coalescence’ (Greenwood
and Speight) model10, and an Ostwald ripening model, eg. Ref. 11. Of these models the first two contain largely
opposing ideas yet both approaches have been used with some success in the modeling of experimental results
from either pressure-composition-temperature (PCT) curves or thermal gravimetric analysis (TGA) and differ-
ential thermal analysis (DTA) data3. These techniques and others including in-situ heating X-ray diffraction
(XRD)2,12 and room temperature TEM13–16 do not give direct evidence of the operating mechanisms. Such
evidence requires 4-D data, ie., three dimensional observations over time. As such, in-situ temperature controlled
TEM observation is an attractive approach for determining the mechanism of hydrogen release. Three groups
have reported using in-situ TEM at 200 kV to observe hydrogen desorption15,17,18. However, conventional TEM
with an accelerating voltage of 200 kV has disadvantages including inelastic incident beam interactions with the
samples, and the sample dimensions (typically less than 100 nm in thickness) make surface effects more prom-
inent19. Recently Mooij and Dam20 demonstrated in-situ the two dimensional nucleation and growth of single
hydride domains of up to several millimeters in diameter by an optical transmission technique (hydrogenogra-
phy) for thin Mg film. They found that the nucleation and growth process affects the hysteresis between
absorption and desorption.
Here we show the hydrogen release behavior from MgH2 in real-time using in situ ultra-high voltage trans-
mission electron microscopy (TEM), and directly verify the mechanisms of the hydride decomposition of bulk

SCIENTIFIC REPORTS | 5 : 8450 | DOI: 10.1038/srep08450 1

 www.nature.com/scientificreports

Figure 1 | A TEM image and selected area electron diffraction patterns from Mg and MgH2 phases in a nominally fully hydrogenated sample.

MgH2 in Mg-Ni alloys. The ultra-high accelerating voltage energy retarding growth as MgH2-Mg interfaces grow into the last
(1,000 kV) enables bulk samples a few micrometers in thickness to small Mg islands surrounded by MgH2 (e.g. Figure 1). Therefore, a
be studied which minimizes surface effects19. The observations are key finding in this work is that some Mg phase is retained after a ‘full’
then compared with MgH2 decomposition in thin-samples (tens of hydrogenation cycle. Importantly, it is expected that some Mg will be
nm) with length scale analogous to nano-powders. retained also at the end of recharging of industrial H-storage systems
Figure 1 shows a TEM image and electron diffraction patterns, based on MgH2 bulk powder. This has a significant effect on dehy-
observed at an acceleration voltage of 300 kV, from a MgH2 particle driding mechanisms as shown in the next sections.
produced by hydriding a Mg-Ni-based H-storage alloy. They show In-situ TEM observations were performed on a ,2 mm particle of
the presence of MgH2 and also Mg phase as hexagonal shaped Mg bulk powder at an acceleration voltage of 1,000 kV with a heating
grains of around 30 to 60 nm size within the MgH2 matrix. From the holder and a high resolution video recorder. The temperature at the
low indexed electron diffraction patterns from MgH2 [101]* and Mg TEM sample grid over the time of observation is in the supporting
[001]* taken from the same selected area diffraction method, diffrac- data (Figure S1). The average heating rate is approximately 13uC/min
tion spots of (12-1) from MgH2 and (2210) from Mg overlap. That from 28 to 455uC. Concerns that the insulation effects between the
means the crystallographic orientation relationships (ORs) between TEM sample grid and the specimen may render the measured sample
MgH2 and Mg are MgH2 [101]*//Mg [001]* (beam direction) and temperature inaccurate can be alleviated by comparing the differ-
MgH2 (12-1)//Mg (2210) (plane) as well as MgH2 (101)//Mg (002) ential scanning calorimetry (DSC) and TEM data. A comparison of
since MgH2 [101]* and Mg [001]* are perpendicular to MgH2 (101) the in-situ hydrogen release observations in the Movie S1, shows the
and Mg (002). The differences of lattice spacings between MgH2 (12- MgH2 to Mg phase transformation is completed at around 420uC
1) and Mg(2210) as well as between MgH2 (101) and Mg (002) are and the DSC experiments in Figure S5 show that the hydrogen
both about 4%. This crystallographic orientation would represent a release peak temperature was 423uC at a similar heating rate of
low strain energy situation facilitating the hydriding phase trans- 15uC/min.
formation from Mg to MgH2. Also Figure 1 shows the hexagonal Figure 2(a)–(d) shows selected still frame TEM images taken at the
facets of the Mg grains have (100) and (1–10) habit planes within temperatures of (a) 300uC, (b) 420uC, (c) 430uC and (d) 455uC (the
MgH2. Schober et.al.21 reported the relationships as MgH2 [001]*// observation video is available in Movie S1). Several bright grains were
Mg [21–10] * and MgH2 (200)//Mg (002), which has a 15.5% lattice observed in the sample in the low temperature range up to 300uC,
mismatch. Paik et. al.18 measured MgH2 [001]*//Mg [2210]* and which correspond to Mg grains, similar to those shown in Figure 1.
MgH2 (2110)//Mg (002), which has a 18.4% mismatch. Those Mg grains subsequently grow and coalesce, with increasing
Note that the sample in Figure 1 was nominally fully-hydrided temperature (Figure 2). There are several large Mg grains at 455uC,
and, therefore, it is necessary to consider why a small volume fraction which are clearly shown in the low magnification image in
of nano-sized Mg grains remained in the MgH2 matrix. In the sup- Figure 2(e). These events in the bulk sample correspond well with
porting data, Figure S4 shows scanning electron microscopy (SEM) the DSC shown in the Figure S5 and corresponding Synchrotron
images prepared from samples quenched at selected times (e.g. 5, 8 XRD data under air and 0.1 MPa conditions in Figure S312. This
and 20 hours) during hydrogen absorption at 340uC and 1 MPa. A in-situ TEM result is direct evidence that dehydriding of ,2 mm
large number of small MgH2 nuclei formed around the Mg dendrite bulk MgH2 particles occurs by the growth of multiple pre-existing
in the early stages of hydrogenation and grew into the Mg phase with Mg grains within a MgH2 matrix.
time. The cracking of the growing MgH2 phase in Figure S4 (b) and A volume change (shrinkage) of around 30% occurs during hydro-
(c) is associated with the release of strain energy caused by the large gen release associated with the phase transformation from MgH2 to
volume change of transformation (30.4% at 340uC22). Even after a Mg (30.4% at 340uC reported by Ono et al22), and from Figure 2 and
prolonged period of hydrogenation (e.g. 20 hours in Figure S4 (c)), Movie S1, we can see some slight shrinkage of the bulk MgH2 particle
small islands of the Mg phase are still retained in the MgH2 matrix. (around 2 micrometers diameter) and Mg grains in the bulk MgH2
This is likely due to the impingement of growing MgH2 effectively particle with concave shape at 455uC where hydrogen release had
providing a barrier against further hydrogen diffusion because the stopped. However, there was an absence of any void formation
coefficient of hydrogen diffusion in MgH2 is at least three orders of within the bulk particle. It is likely the strain associated with volu-
magnitude less than that in Mg23, and may also be due to strain metric contraction is accommodated by deformation (including con-

SCIENTIFIC REPORTS | 5 : 8450 | DOI: 10.1038/srep08450 2

 www.nature.com/scientificreports

Figure 2 | Selected still frame TEM images from in-situ video of high voltage (1,000 kV) TEM of a ,2 mm bulk MgH2 particle taken at (a) 3006C, (b)
4206C, (c) 4306C and (d) 4556C, and (e) a low magnification bright field image of the sample (a single bulk powder particle) at 4556C.

traction in the transverse direction). At the temperatures involved From the direct in-situ observations of hydrogen release behavior
this deformation appears to be more favourable than the nucleation from the bulk (thick) sample presented in this work, the observations
and growth of porosity, which requires substantial energy. are schematically shown in Figure 4(a), which is closest to the mul-
For comparison, in-situ TEM was performed at 200 kV TEM tiple ‘nucleation and growth’ model3, with the important difference
(Figure 3 and Movie S2). In this case, the thickness in the observation that small (tens of nm) Mg grains were pre-existing in the sample
area is a few tens of nanometers as required for imaging at 200 kV (Figure 1) and, therefore, nucleation was not a pre-requisite for the
and is at the edge of the sample. Figure 3 shows selected still frame transformation. In this case, the transformation is growth controlled,
TEM images from the in-situ video in Movie S2 taken at the tem- where hydrogen atoms diffuse from MgH2 with the driving force for
peratures of (a) 50uC, (b) 150uC, (c) 250uC, and (d) 380uC. Mg grains grain boundary movement being derived from the free energy dif-
form at the thinnest edge of the sample, and progress toward the ference between atoms in adjacent grains. Also from Figure 2(e),
inside of the MgH2 sample at ,150uC. This result indicates the several areas near the grain surface have remained as MgH2 and
mechanism of hydrogen release in this thin section of the sample is oxide, indicating that even at 455uC, some volume still contains
quite different from the bulk sample results obtained with high- hydrogen as the MgH2 phase. The existence of both Mg and MgH2
voltage TEM in Figure 2 and Movie S1. Also, the hydrogen release phases after both charging and discharging is likely to play an
temperature is much lower than that obtained by high voltage TEM important role in hydriding/dehydriding kinetics of bulk powder.
observations (Figure 2 and Movie S1). The influence of conventional The effect of pre-existing Mg grains on the dehydrogenation of
200 kV TEM on dehydriding mechanisms includes, (1) the thin (less MgH2 was reported by Tanniru et. al.24, using scanning electron
than 100 nm) sample thickness allowing surface effects to have a microscopy. They concluded that the nucleation barrier for hexa-
disproportionate effect, and (2) the inelastic incident beam inter- gonal close-packed Mg plays an important role in establishing the
action with the sample atoms being stronger than at high voltage hydrogen release temperature. When the magnesium powders are
and, as a result, more pronounced ‘electronic excitation’ occurs. hydrogenated such that the surface is completely covered by the

SCIENTIFIC REPORTS | 5 : 8450 | DOI: 10.1038/srep08450 3

 www.nature.com/scientificreports

Figure 3 | Selected still frame TEM images from in-situ video of conventional (200 kV) TEM through a thinned region (a few tens of nm) of a MgH2
particle taken at (a) 506C, (b) 1506C, (c) 2506C, and (d) 3806C.

hydride phase, the desorption temperature is found to be high, owing nano-powders as well as election beam heating effect due to the
to the energy required for the nucleation of the Mg on the surface. inelastic incident beam interactions with the samples19. In this case,
Antisari et. al.25 have confirmed the difficulty of Mg nucleation in the formation of Mg at the sample edge followed by growth into the
MgH2, and the presence of some retained Mg phase is likely to MgH2 is consistent with ‘shrinking core’ models, where nucleation
accelerate dehydriding kinetics. If there were no pre-existing Mg and growth of Mg phase happen from the edge/surface of the sample.
grains, surface nucleation would be easier than nucleation within Based on ‘shrinking core’ models, Ouyang et. al.26 found the hydrid-
the volume due to the volume change. However, in the presence of ing/dehydriding process in nano-grained (around 30 nm in dia-
pre-existing Mg grains within the volume, growth from within is meter) of Mg is catalyzed by the combination of in situ formed
more favourable than nucleation and growth from the surface. extremely fine CeH2/CeH2.73 and Ni to Mg/MgH2. The experiments
The thin sample (tens of nm thick) studied by conventional TEM performed were not isothermal and did not enable a comparison of
at 200 kV in Figure 3 and Figure 4(b) may reflect the behavior of fine the kinetics of hydrogen transportation through the surrounding

Figure 4 | Schematic hydrogen release mechanisms from a MgH2 grain: (a) multiple ‘nucleation and growth’ model for bulk MgH2 grains and
(b) ‘shrinking core’ model for thin MgH2 TEM samples.

SCIENTIFIC REPORTS | 5 : 8450 | DOI: 10.1038/srep08450 4

 www.nature.com/scientificreports

MgH2 phase between the thin and bulk samples. While the overall 13. Danaie, M. & Mitlin, D. TEM analysis and sorption properties of high-energy
milled MgH2 powders. J. Alloys Compd. 476, 590–598 (2009).
thickness of the thin samples will be the dominant geometric factor 14. Tanaka, K., Miwa, T., Sasaki, K. & Kuroda, K. TEM studies of nanostructure in
influencing the desorption rate the spatial distribution of any pre- melt-spun Mg–Ni–La alloy manifesting enhanced hydrogen desorbing kinetics.
existing Mg nuclei and in particular the distance of these nuclei from J. Alloys Compd. 478, 308–316 (2009).
the free surface will play an important role in determining the 15. Isobe, S. et al. Study on reaction mechanism of dehydrogenation of magnesium
desorption rate of the bulk samples. hydride by in situ transmission electron microscopy. Appl. Phys. Lett. 96, 223109-
223101-223103 (2010).
From in-situ TEM imaging, we conclude that the hydrogen release 16. Fujii, H. et al. Remarkable hydrogen storage, structural and optical properties in
mechanism from bulk (2 mm) MgH2 particles is based on the growth multi-layered Pd/Mg thin films. Materials Transactions 43, 2721–2727 (2002).
of multiple pre-existing Mg grains (crystallites within the MgH2 17. Danaie, M., Tao, S. X., Kalisvaart, P. & Mitlin, D. Analysis of deformation twins
matrix) present due to the difficulty of fully transforming all Mg and the partially dehydrogenated microstructure in nanocrystalline magnesium
hydride (MgH2) powder. Acta Mater. 58, 3162–3172 (2010).
during a hydrogenation cycle whereas, in thin samples analogous 18. Paik, B. et al. MgH2 - Mg phase transformation driven by a high-energy electron
to nano-powders, dehydriding occurs by a ‘shrinking core’ beam: An in situ transmission electron microscopy study. Philos. Mag. Lett. 90,
mechanism. 1–7 (2010).
19. Tanaka, M. et al. Sequential multiplication of dislocation sources along a crack
front revealed by high-voltage electron microscopy and tomography. J. Mater.
Methods Res. 26, 508–513 (2011).
MgH2 sample preparation. To prepare samples for hydrogen absorption they were 20. Mooij, L. & Dam, B. Hysteresis and the role of nucleation and growth in the
machined to fine chips in air using a drill press. The unreacted chips were of the order hydrogenation of Mg nanolayers. PCCP 15, 2782–2792 (2013).
of 0.1 mm thick and a few mm in length. Hydrogen sorption was conducted using an 21. Schober, T. The magnesium-hydrogen system: Transmission electron
automated gravimetric testing apparatus PCTM-5000A (Technosystem Ltd., Japan)27 microscopy. Metall. Trans. A 12, 951–957 (1981).
using laboratory grade high purity H2 gas (99.98% purity). The apparatus levitates 22. Ono, S. et al. Phase transformation and thermal expansion of Mg-Ni alloys in a
approximately one gram of the sample material and directly records the weight hydrogen atmosphere. Journal of the Less Common Metals 88, 57–61 (1982).
change using a balance. The details of this machine are described elsewhere6,27. The 23. Čermák, J. & Král, L. Hydrogen diffusion in Mg–H and Mg–Ni–H alloys. Acta
samples were ‘hydrided’, under conditions of 2 MPa, 350uC for 20 hours. During Mater. 56, 2677–2686 (2008).
hydriding, MgH2 and Mg2NiH4 formed and significant cracking occurred, reducing 24. Tanniru, M., Tien, H.-Y. & Ebrahimi, F. Study of the dehydrogenation behavior of
the particle size to a few mm. All TEM results presented in the main paper are from magnesium hydride. Scripta Mater. 63, 58–60 (2010).
powder particles containing MgH2 and no discernable Mg2NiH4. 25. Antisari, M. V. et al. Scanning electron microscopy of partially de-hydrogenated
Sample preparation for high voltage (1,000 kV) TEM involved selecting single MgH2 powders. Intermetallics 17, 596–602 (2009).
particles of MgH2 from the reaction product and loading them on a temperature 26. Ouyang, L. Z. et al. Enhanced hydrogen storage kinetics and stability by synergistic
controlled TEM sample holder with a 3 mm diameter grid for high temperature use. effects of in situ formed CeH2.73 and Ni in CeH2.73-MgH2-Ni nanocomposites.
Note that particles were examined without any further mechanical/chemical pre- The Journal of Physical Chemistry C 118, 7808–7820 (2014).
parations which would be difficult using conventional TEM with 200 kV. 27. Yamada, A. & Funatsu, T. Measuring system for absorbed hydrogen using
Conventional 200 kV TEM was also performed for comparison with the 1,000 kV magnetic levitation system. Fuel Cell (Japan) 4, 25–29 (2004).
experiments and to explore the influence of sample thickness and beam heating on
dehydriding mechanisms. To produce thin samples for conventional TEM, MgH2
particles were selected from the hydride powder and fractured in an agate mortar
producing fragments with edges a few tens of nanometers in thickness. Acknowledgments
This research was supported by a JSPS Fellowship for Research in Japan (S11730), a Kyushu
TEM observations. For the detailed crystallography of Mg and MgH2 and their University Short-term Young Scholar Exchange Program, a Smart Futures Fellowship from
relative orientation relationships, a transmission electron microscopy (TEM), JEM- the Queensland Government, Australia and by Hydrexia Pty. Ltd. We thank Mr. Takahiro
3200FSK (JEOL, Japan) at acceleration voltage of 300 kV with an Omega filter was Nogita at Film School, Griffith University for video editing, Dr. Qinfen Gu at the Australian
used. In-situ TEM observations were performed using a JEM-1000 (JEOL, Japan) at Synchrotron for assistance with XRD experiments (AS093/PD1726, AS102/PD/QLD/
an acceleration voltage of 1,000 kV with an EM-HSTH (JEOL, Japan) heating holder FI2488 and AS131/PD5782), and the ANFF-Q and CMM at the University of Queensland
and high resolution video recorder. for use of the DSC and SEM facilities.
For comparison with conventional in-situ TEM observations, we used a JEM-
2100HCLM (JEOL, Japan) at an acceleration voltage of 200 kV with a Model 652
(GATAN, U.S.A.) double tilt heating holder. Author contributions
K.N. contributed to the planning of all experiments, participating TEM and Synchrotron
XRD experiments and writing the article. X.Q.T. contributed to the discussion and
1. Reilly, J. J. & R, H. Wiswall, J. The Reaction of hydrogen with alloys of magnesium conducted DSC and SEM experiments and Synchrotron XRD data analysis. T.Y.
and nickel and the formation of Mg2NiH4. Inorg. Chem. 7, 2254–2256 (1968). contributed to the discussion and TEM experiments and analysis. E.T. contributed to the
2. Sakintuna, B., Lamari-Darkrim, F. & Hirscher, M. Metal hydride materials for discussion and TEM experiments and analysis. S.D.M. contributed to the planning of all
solid hydrogen storage: A review. Int. J. Hydrogen Energy 32, 1121–1140 (2007). experiments, discussion, and writing the article. C.M.G. contributed to the interpretation,
3. Stepura, G., Rosenband, V. & Gany, A. A model for the decomposition of titanium discussions, and writing the article. K.Y. contributed to the planning of all experiments,
hydride and magnesium hydride. J. Alloys Compd. 513, 159–164 (2012). TEM data analysis and discussion. S.M. contributed to the planning of all experiments,
4. Moroz, S. et al. Systems based on hypo-eutectic Mg–Mg2Ni alloys for medium to TEM data analysis, discussion, and writing the article.
large scale hydrogen storage and delivery. J. Alloys Compd. 580, S329–S332
(2013). Additional information
5. Dornheim, M., Eigen, N., Barkhordarian, G., Klassen, T. & Bormann, R. Tailoring Supplementary information accompanies this paper at http://www.nature.com/
hydrogen storage materials towards application. Adv. Eng. Mater. 8, 377–385 scientificreports
(2006).
6. Nogita, K. et al. Engineering the Mg-Mg2Ni eutectic transformation to produce Competing financial interests: K.N. holds a patent on modification of hypoeutectic Mg-Ni
improved hydrogen storage alloys. Int. J. Hydrogen Energy 34, 7686–7691 (2009). alloys with alkali and alkali-earth metals (K. Nogita and A. K. Dahle, ‘‘Magnesium Alloys for
7. Wang, P., Wang, A. M., Wang, Y. L., Zhang, H. F. & Hu, Z. Q. Decomposition Hydrogen Storage’’, International Patent Application No: WO 2006/060851, 2/12/2005)
behavior of MgH2 prepared by reaction ball-milling. Scripta Mater. 43, 83–87 which resulted in a spin-off company Hydrexia Pty. Ltd. which is working to build a
(2000). commercial H-storage system from this technology. This letter is about the fundamental
8. Castro, F. J. & Meyer, G. Thermal desorption spectroscopy (TDS) method for mechanisms of dehydriding of MgH2 and is not related to the claims of the patent and has
hydrogen desorption characterization (I): theoretical aspects. J. Alloys Compd. no direct financial implications on the work of Hydrexia Pty. Ltd. K.N. also supervises an
330–332, 59–63 (2002). ongoing research project at the University of Queensland funded by Hydrexia Pty. Ltd.
9. Gabis, I., Evard, E., Voyt, A., Chernov, I. & Zaika, Y. Kinetics of decomposition of How to cite this article: Nogita, K. et al. Evidence of the hydrogen release mechanism in
erbium hydride. J. Alloys Compd. 356–357, 353–357 (2003). bulk MgH2. Sci. Rep. 5, 8450; DOI:10.1038/srep08450 (2015).
10. Greenwood, G. W. & Speight, M. V. Grain boundary mobility and its effects in
materials containing inert gases. Philosophical Magazine 9, 683–689 (1964). This work is licensed under a Creative Commons Attribution 4.0 International
11. Goodhew, P. J. Cavity growth mechanism maps. Scripta Mater. 18, 1069–1073 License. The images or other third party material in this article are included in the
(1984). article’s Creative Commons license, unless indicated otherwise in the credit line; if
12. Nogita, K., McDonald, S. D., Duguid, A., Tsubota, M. & Gu, Q. F. Hydrogen the material is not included under the Creative Commons license, users will need
desorption of Mg-Mg2Ni hypo-eutectic alloys in Air, Ar, CO2, N2 and H2. J. Alloys to obtain permission from the license holder in order to reproduce the material. To
Compd. 580, S140–S143 (2013). view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS | 5 : 8450 | DOI: 10.1038/srep08450 5