i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
journal hom epa ge: www.elsev ier.com/locate/he
a
b
b , d b
a , b , * a , b
c
c
b
a
Dept. of Mechanical Engineering, Texas A&M University, College Station, TX 77843, USA b Dept. of Materials Science and Engineering, Texas A&M University, College Station, TX 77843, USA c Dept. of Chemistry, Texas A&M University, College Station, TX 77843-3255, USA d Dept. of Electrical Engineering, Texas A&M University, College Station, TX 77843-3128, USA a r t i c l e i n f o
Article history:
Received 13 June 2013
Received in revised form
18 October 2013
Accepted 2 December 2013
Available online 7 January 2014
Keywords:
Orthorhombic Mg hydride
Destabilization
Size effect
Stress effect
Porous Mg films
Mg pillars a b s t r a c t
Mg is a promising light-weight material that has superior hydrogen storage capacity.
However H
2 storage in Mg typically requires high temperature, w 500 e
600 K. Furthermore it has been shown that there is a peculiar film thickness effect on H
2 sorption in Mg films, that is thinner Mg films desorb H
2 at higher temperature
[1]
. In this study we show that the morphology of DC magnetron sputtered Mg thin films on rigid SiO
2 substrate varied from a continuous dense morphology to porous columnar structure when they grew thicker.
Sputtered Mg films absorbed H
2 at 373 K and evolved into a metastable orthorhombic Mg hydride phase. Thermal desorption spectroscopy studies show that thinner dense MgH
2 films desorb H
2 at lower temperature than thicker porous MgH
2 films. Meanwhile MgH
2 pillars with greater porosity have degraded hydrogen sorption performance contradictory to general wisdom. The influences of stress on formation of metastable MgH
2 phase and consequent reduction of H
2 sorption temperature are discussed.
Copyright ª 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Hydrogen is a promising alternative fuel for renewable energy applications due to its abundance, high efficiency and zero carbon emission when used in proton exchange membrane
(PEM) fuel cells
[2,3] . In order to economically apply hydrogen
fuels for automobile applications, light-weight hydrogen storage materials with high uptake capacity and low desorption temperature are necessary. In addition, hydrogen recycling should be performed reversibly under PEM fuel cell operation temperatures. Compared to the 700 bar gas cylinders currently used by most hydrogen vehicle manufacturers, hydrogen stored in metal hydrides offers the advantage of a much higher volumetric density
[4,5]
. Among numerous hydrogen storage candidates, magnesium hydride is an attractive material because of its high H
2 storage capacity
(7.6 wt.%) and economical availability as the third most abundant element on earth.
Bulk Mg hydride, however, has tetragonal phase (referred to as T-MgH
2 hereafter) that is thought to be unsuitable for automotive application because of its high formation
* Corresponding author . Dept. of Materials Science and Engineering, Texas A&M University, College Station, TX 77843, USA. Tel.: þ 1 979
845 2143.
E-mail address: zhangx@tamu.edu
(X. Zhang).
0360-3199/$ e see front matter Copyright ª 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhydene.2013.12.017

2598 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7 enthalpy ( D H ¼ 75 kJ/mol H
2
[6]
). Due to the high thermodynamic stability of Mg hydride, hydrogen absorption/ desorption typically occurs at w 673 K in bulk Mg, much higher than practical applications (333 e
353 K)
[2,7,8] . In order to
release hydrogen at moderate temperature for PEM fuel cell applications, Mg hydride must be destabilized. Additionally hydrogen sorption kinetics must be improved to achieve practical refueling time for vehicle applications. Nanostructured materials have a large surface to volume ratio, provide more nucleation sites for hydrogen sorption, promote rapid hydrogen diffusion, and hence may lower the kinetic barriers for hydrogenation
[9 e
11] . Recent reports showed that
ball-milled nanocrystalline Mg powders can expedite hydrogen desorption process
[12
e
14] . However, nanograins in
these materials coarsen over succeeding thermal cycles and are highly susceptible to oxidation. Numerous attempts, via ball milling method, have been made to improve the kinetics of hydrogen sorption in pure Mg and Mg mixed with transition metals, with limited reduction on hydrogen sorption temperatures
[10,15,16] . Vajo et al. reported thermodynamic
destabilization in more complex hydrides of Mg alloys
[17 e
20]
.
Besides bulk nanostructured Mg and its alloys, Mg based thin films are also actively studied as model systems to understand hydrogen sorption kinetics
[1,21 e
24]
. Among these studies, Mg e
Pd is a notable thin film system with peculiar H
2 sorption properties. Pd is a noble metallic catalyst that dissociates H
2 molecules to H atoms at room temperature. It has been shown that sputtered Mg thin film with a Pd cap layer can absorb and desorb under conditions suitable for PEM fuel cells (373 K, 0.1 MPa)
[1]
. The authors used a “cooperative phenomenon” to explain low temperature H
2 sorption in Mg/
Pd multilayer films. Basically the absorption of H
2 in Pd
(catalyst) leads to tensile stress in Mg and thus facilitates the hydrogenation of Mg film. During desorption, the opposite process occurs, that is desorption of H
2 from Pd leads to compressive stress that enables rapid removal of H
2 from Mg hydride
[1] . If interfacial stress indeed plays a critical role on H
sorption as stated in the “cooperative phenomenon”, then one would anticipate that thinner films desorb H
2 at a lower temperature as such a mechanism implies lower H
2 sorption temperature in thinner films. However it was reported that thicker Mg films desorbed H
2 at lower temperature. On the other hand, studies on Pd/Mg/Pd trilayer films
[25]
revealed insignificant residual stress evolution before and after hydrogen loading.
Meanwhile interfacial energy may contribute to destabilization of Mg hydride as proposed by
Mooij et al.
[26] . Pd-decorated Mg blades prepared by oblique
angle thermal deposition also have shown a low H desorption temperature, w 373 K
[27] .
Other than noble metal catalyst (such as Pd), transition metal catalysts were added to Mg films to expedite H
2 sorption kinetics in T-MgH
2
[28 e
34] . Furthermore certain transition
metal catalysts incorporated by co-sputtering with Mg have led to reduction of the hydrogen formation energy
[35 e
38] . Co-
sputtered Mg e
Ti thin film showed cyclic capability, but hydrogen formation energy is unknown
[37 e
39]
. Mg e
Ni e
Ti ternary thin film exhibited hydride formation energy of
40 kJ/mol H
2
[40,41]
. Co-sputtered Mg-25 at.% Nb thin film showed the formation of a metastable bcc alloy with the hydride formation enthalpy of 52 kJ/mol H
2
[42]
. However, the hydrided Mg e
Nb film desorbed hydrogen only at 448 K and deterioration of the Mg and Nb (a gradual segregation) was observed after extended cycling studies. We have recently reported that H
2 sorption in 1.6
m m Mg film grown on rigid SiO
2 substrate and free standing Mg/Nb multilayers led to the formation of a metastable orthorhombic MgH
2
(O-MgH
2
) phase, which has an ultra-low hydride formation energy, 37 kJ/mol
H
2
[43] , compared to
75 kJ/mol H
2
MgH
2 desorbs H
2 of stable T-MgH
2
. Thus Oat w 373 K or lower. These experimental results compared well with those obtained by first principle calculations
[43] . In parallel TEM studies have shown that bcc
metastable Mg formed on top of Nb
[44]
and DFT calculations were performed to gain insight on the stabilization of such a metastable bcc Mg phase
[45] . The potential tie between bcc
Mg and the formation of O-MgH
2 investigations.
requires further
In this paper we show that thinner Mg films indeed desorb
H
2 at lower temperature than thicker films. Film thickness has a direct influence on porosity of Mg films, namely thinner film
( w 200 nm) formed a continuous dense layer, beyond which it evolved into a porous columnar structure. Mg pillars with different porosity also have drastically different H
2 sorption performance. Contradictory to our intuitive anticipation, greater porosity degrade H
2 sorption performance of Mg pillars. The evolution of stress during H
2 induced formation of O-MgH
2 sorption and stressplay major roles on hydrogen sorption properties of Mg films. This study provides an important step forward towards in-depth understanding of destabilization of Mg hydride via stress-driven formation of metastable phases.
2.
Experimental
Mg films with layer thickness t varying from 200 to 1600 nm were deposited by DC magnetron sputtering on oxidized Si
(100) substrates with 1 m m thermal oxide layer at room temperature. All Mg films were protected by 25 nm Pd cap layer, which catalyzes the dissociation of hydrogen molecules and protect Mg films from oxidation. Mg and Pd targets with
99.99% purity were used for magnetron sputtering. The base pressure of the chamber was better than 6.6
10 6 Pa prior to deposition. The films were grown under w 0.8 mTorr Argon pressure with 99.999% purity. The deposition rates for Mg and
Pd were w
2 and 0.3 nm/s, respectively. To fabricate Mg nanopillars, Si wafers were mounted on a custom-designed substrate holder at an angle of 5 relative to the direction of the sputtering source (i.e. using a configuration of glancing angle deposition). During glancing angle deposition, the substrate was not rotated. In parallel, inclined deposition at 45 was also performed without substrate rotation. After deposition, samples were transported directly into an attached high vacuum chamber for hydrogen absorption and desorption studies. The base pressure of hydrogen chamber was w
6.66
10
7
Torr before hydrogen loading. All Mg films were hydrided at a pressure of 0.25 MPa with ultra high purity gas mixture (96% Ar þ 4% H
2
) at 373 K for 24 h. Hydrogen desorption studies were performed by thermal desorption spectroscopy (TDS) technique enabled by a quadruple mass spectrometer in a residual gas analyzer (RGA). During

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
2599 desorption studies, Mg specimens were heated from room temperature up to a maximum of 623 K at a constant heating rate of 4 K/min. X-ray diffraction (XRD) experiment was performed on a Bruker-AXS D8 Bragg e
Brentano X-ray diffractometer. Morphology and structure of as-deposited and hydrided films were examined by scanning electron microscopy (SEM) using a JEOL JSM-7500 field emission scanning electron microscope operated at 5 kV. Mg films with 400 nm in thickness and 25 nm Pd cap layers were used to monitor stress evolution during hydrogen loading experiments (373 K up to
24 h). Stress was obtained by measuring the curvature of substrates by using a Dektak 150 surface profilometer. Multiple stress bars (a minimum of 3) were used to obtain an average stress value during hydrogenation. The stress evolution in films due to hydrogen loading was investigated up to
24 h, by which the film was fully hydrided. The porosity of Mg films before and after hydrogen loading was measured by using ASAP 2020 volumetric physisorption analyzer. The stresses in films before and after hydrogen loading were measured by examining the curvature of substrate using a
Dektak 150 Stylus profilometer with a Z height resolution of w 0.5 nm. The stress can then be obtained from the Stoney’s equation
[46] :
s ¼
Mt
6 h
2 1
R
1
1
R
0
;
(1) where s is the residual stress, M is the biaxial modulus of substrate, t and h are the substrate and film thickness, respectively.
R
1 and R
0 are the respective radii of substrate after and before depositions.
8.0x10
4
6.0x10
4
4.0x10
4
2.0x10
4
0.0
25
Bulk
Mg (0002) a
Mg 200nm d = 2.58
30 35
2 Theta (degree)
40 45
4.0x10
4
O-MgH
2
(110)
Mg (0002) b
2.0x10
4
0.0
25
d = 3.16
30 35 40
2 Theta (degree)
200nm
45
4.0x10
4 c
T-MgH
2
(110)
O-MgH
2
(110)
3.
Results
The XRD plots of Mg films in
Fig. 1
a show that all Mg films exhibit (0002) texture and the peak intensity increases proportionally as the layer thickness increases.
Fig. 1 b shows XRD
profiles of Mg films on rigid Si substrate after hydrogen loading (373 K/24 h). Most of the as-deposited Mg films transformed to orthorhombic Mg hydride (referred to as the
“O-MgH
2
” thereafter) similar to CaCl
2 with a (110) texture.
Detailed crystal structure analysis for O-MgH
2 can be found elsewhere
[43] .
Fig. 1
c shows that Mg 1600 nm has distorted O-
MgH
2 structure, different from that of the stable tetragonal
MgH
2
(T-MgH
2
) position. Thinner films have (110) interplanar spacing (again as O-MgH
2
) with much more deviation from the
T-MgH
2
.
Fig. 2
a shows XRD patterns of 5 m m tall Mg pillars
(morphology will be shown later) deposited at different angles
(5 , 45 ) before and after hydrogen loading. Both as-deposited
Mg pillars showed highly textured (0002) orientation, while Mg pillars with 5 deposition angle (referred to as “5 pillar film” hereafter) also had a (101 1) secondary peak. After hydrogen loading (373 K/24 h), the 45 pillar film was fully hydrided and transformed to O-MgH
2
, while the 5 pillar film had both Mg hydride peaks and hcp Mg (0002) peak. At higher magnification (
Fig. 2
b), a distortion from stable T-MgH
2
(011) manifested by a peak shift towards O-MgH
2 was observed in both pillar films, instead of forming two drastically different O-MgH
2 and
T-MgH
2 peaks.
2.0x10
4
0.0
2 5
Mg 200nm d = 3.16
2 Theta (degree)
3 0
Fig. 1 e
X-ray diffraction (XRD) patterns of Mg films of different thickness ( t [ 200 e
1600 nm with a 25 nm thick
Pd cap layer) before and after hydrogen loading at 373 K for
24 h. (a) All as-deposited Mg films had (0002) texture and its peak intensity increased with increasing t . A slightly reduced d -spacing of Mg (0002) was observed in thinner Mg film ( t [ 200 nm) compared to that of bulk Mg. (b) After hydrogen loading the intensity of Mg hcp (0002) peaks diminished significantly. Meanwhile orthorhombic MgH
2
(O-MgH
2
) (110) peak emerged with stronger texture in thicker MgH
2 films. (c) Comparison of XRD profiles for hydrogen-loaded Mg films shows that all hydride films have O-MgH
2 phase, different from traditional T-MgH
2 phase (typically observed in bulk Mg hydride). The (110) interplanar spacing of thinner O-MgH
2 deviation from (110) T-MgH
2
.
film has greater

2600 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
25
O-MgH
2
(110)
O-MgH
2
(011) a
2
2
Mg (0002)
Mg (1011)
30 35
45° pillars
40
2Theta (degree)
45
T-MgH
2
(011)
2
2 b
O-MgH
2
(110)
26 28
2Theta (degree)
30
Fig. 2 e
(a) XRD pattern of 5 m m thick Mg pillars obtained with different inclined deposition angle (5 , 45 ) before and after hydrogen loading. Both as-deposited Mg pillars showed highly (0002) textured Mg peak and Mg pillars deposited with 5 glancing angle (referred to as Mg 5
L pillars) showed a Mg (101 1) peak. After hydrogen loading at
373 K/24 h, Mg 45 pillars were fully hydrided by forming
O-MgH
2 structure. Mg 5 pillars showed both Mg hydride peaks and hcp-Mg peak, implying incomplete hydrogen loading in the pillars. (b) Magnified XRD profiles of hydrided Mg pillars deposited at an incidence angle of 5 and 45 . Both hydrided pillar films have (110) O-MgH
2 peaks that are different from the stable T-MgH
2
.
XSEM micrographs of all corresponding hydrogen-loaded
Mg samples are shown in
Fig. 3
a
0 e e
0
. Mg hydride films were substantially thicker than as-deposited parent films. For instance the hydride film thickness increased to 250 and
450 nm (
Fig. 3 a
0 and b
0
) compared to 200 and 400 nm thick
parent films ( Fig. 3
a and b). The porous 800 nm thick Mg films
became nearly fully dense ( Fig. 3
c
0
), but the thickness increased moderately to merely 850 nm. Compared to the asdeposited 45 pillar film, the hydrided pillar film in
Fig. 3 d
0 appeared much denser. On the other hand, the morphology of hydrided 5 pillars film (
Fig. 3 e
0
) had little variation in porosity compared to the as-deposited films.
Fig. 4 a compares the TDS spectra of hydrogen for hydro-
genated Mg films with various film thicknesses. In the 800 and
1600 nm Mg hydride films, the TDS spectra appear to be consisted of two peaks. The lower temperature peak shows comparable intensity to the high temperature peak in the TDS spectrum of hydrided Mg 800 nm film. In the 1600 nm Mg film, however, the high temperature peak, w 410 K, has much greater peak intensity than the low temperature peak, w 395 K.
At smaller layer thickness, t ¼ 200 and 400 nm, only one H
2 desorption peak was observed. The hydrogen desorption peak temperature gradually shifted to lower temperatures as t decreased. For instance, the hydrided Mg 200 nm film reached a peak temperature at w 370 K. Both onset and peak desorption temperatures of all Mg films, summarized in
Fig. 4 b,
decreased gradually with decreasing Mg film thickness. We notice that both porous films (800 and 1600 nm thick Mg) had two desorption peaks, whereas thinner films typically had only one desorption peak at lower temperatures.
Fig. 5
compares TDS spectra for H
2 loaded 45 and 5 pillar films. Both specimens have two groups of desorption peaks.
The low temperature group in both specimens started hydrogen desorption at w
363 K and exhibited peak desorption at w 410 K. In the 45 pillar films, the intensity of the first (low temperature) group of peaks is much greater than that of the second (higher temperature) group. Whereas in the hydrided
5 pillar film, the intensity of the second group of peaks
(consisting of multiple peaks) is at least comparable to that of the first group of peaks.
Cross-section scanning electron microscopy (XSEM) micrographs in
Fig. 3
a e e show the evolution of film morphology of various Mg films. All films were sputtered on Si substrates with thermally grown Si oxide layers. 200 nm thick Mg film
(
Fig. 3 a) had fully dense, continuous structure and the plan-
view SEM micrograph of the film (not shown here) shows a featureless, shiny surface. However, when film thickness ( t ) is greater than 200 nm, surface roughness and porosity appeared (
Fig. 3 b).
Fig. 3 c reveals an initial
w
200 nm dense layer on substrate followed by porous structure thereafter in thick Mg film ( t ¼ 800 nm).
Fig. 3
d e e shows that the 45 and 5
Mg pillar films were both porous and the 5 pillar film had extremely high porosity and dispersed columnar structures.
The columnar diameters are similar in both pillar films.
4.
Discussion
2
We recently reported the formation of orthorhombic Mg hydride (O-MgH
2
) in hydrided Mg films on rigid substrate and in
Mg/Nb multilayers
[43] . This is surprising as bulk O-MgH
2 has been observed only under high pressure, several GPa
[47,48] .
The formation of O-MgH
2 in single layer Mg deposited on rigid substrate is a consequence of stress-induced by substrate constraint. During hydrogen loading process, the formation of
MgH
2 leads to substantial volume expansion in Mg films.
However the rigid Si substrate has little volume expansion as it barely absorbs H
2
. Thus the large stress arising from volume expansion incompatibility (constrained by rigid substrate) leads to the formation of O-MgH
2
[43] .
In
Fig. 6
, the structure transformations from hexagonal close packed (hcp) Mg to Mg hydrides are compared from

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
2601 three- and two-dimensional view. Mg with hcp structure is shown in
Fig. 6
a.
Fig. 6 b presents the conventional unit cell of
stable T-MgH
2 with a ¼ b ¼ 4.475
A and c ¼ 2.998
A.
Fig. 6
c is the unit cell of O-MgH
2 observed in this study with a ¼ 3.851
A, b ¼ 5.464
A and c ¼ 3.144
A. Details on measurement of lattice parameters via in situ X-ray diffraction technique at advanced photon source e
Argonne National Laboratory can be found elsewhere
[43]
. Comparison of Mg and its hydrides has been illustrated in two-dimensional projected view in
Fig. 6
d e f. The blue plane in
Fig. 6
d corresponds to a {2110} plane shown in
Fig. 6 a. Lattice distortion and atomic shuffling occurs when
hcp Mg absorbs hydrogen to form Mg hydrides. During the formation of T-MgH
2
( Fig. 6 e), the
< 0002 > and < 0110 > directions of hcp Mg expand and transform to < 110 > and < 110 > of T-MgH
2 respectively. While contraction occurs in < 2110 > direction in Mg, and this direction transforms to < 001 > direction of T-MgH
2 after hydrogen loading. In contrast during the formation of O-MgH
2 in this study, the
<
0002
>
,
<
0110
>
, and < 2110 > directions of hcp Mg evolve to < 210 > , < 210 > and
< 001 > of O-MgH
2
, respectively as shown in
Fig. 6
f. The difference in magnitude of distortion along certain characteristic orientations during the formation of both types of hydrides has been labeled in
Fig. 6
e and f.
In Mg/Nb multilayers, a different type of O-MgH
2
( a ¼ 4.903
A, b ¼ 3.789
A and c ¼ 3.215
A) was identified
[43]
. In absence of rigid substrate (free standing multilayers), a large stress arises in multilayers due to volume expansion incompatibility between Mg hydride and Nb hydride. During the formation of Mg hydride there is a volume expansion of 32.8%.
In contrast only 13.8% of volume expansion occurs during the formation of NbH. Since Mg, Nb and their hydrides are rigidly bonded through layer interfaces (as reflected from reversible microstructure change and retention of layer interfaces during hydrogenation
[43]
), such a large volume expansion disparity during hydrogenation of Mg and Nb in multilayers results in large stress, and consequently the formation of O-
MgH
2
. Synchrotron XRD experiments confirmed the formation of this metastable phase. It shall be noted that stress arising from phase transformation is different from residual growth stress.
The evolution of residual stress in hydrided Mg 400 nm films was monitored as a function of hydrogen loading time
(at 373 K) in Fig. S1 As-deposited Mg 400 nm films had a negligible tensile residual growth stress, w 10 MPa. During the initial loading stage, compressive stress quickly developed and reached a maximum of w 200 MPa within the first 1 h of hydrogen loading. The stress in hydrided films then decayed quickly to w 30 MPa by 3 h of hydrogen loading. Further increase of hydrogen loading time led to slow but progressive reduction of compressive stress. By 24 h the hydrided films were nearly stress free, with merely a slight residual tensile stress.
At the beginning of hydrogen loading process ( < 1 h), a large amount of hydrogen atoms are introduced into Mg film.
Consequently the film shall expand both laterally and vertically. However as Si substrate does not absorb hydrogen, the volume expansion of Mg films will lead to the development of a large compressive stress. To avoid delamination from substrate (delamination was not observed in the hydrided Mg films on Si substrate), further hydrogenation led to more volume expansion along the vertical (film normal) direction rather than lateral expansion and consequently relieved compressive stress in the hydrided films. Thickness variation by w 25% or less has been observed in the hydrided Mg films.
Upon complete hydrogenation, an ‘equilibrium’ stress state is achieved, where a slight tensile residual stress is observed in
MgH
2 films. Although there is still a need for in-depth analysis of these results, such preliminary study does indicate that a significant compressive stress develops quickly during the initial hydrogen loading process due to volume expansion incompatibility between Mg films and Si substrates.
Chung et al. used both in situ stress measurement and ex situ XRD to compare stress evolution before and after hydrogen loading
[25] . Their in situ stress measurement also
shows rapid development of compressive stress in the early hydrogen loading stage, but within a much shorter transient period. Ex situ XRD studies show little stress variation in Mg film. It should be noted that their multilayer stacking has a trilayer configuration (Pd/Mg/Pd), different from what we used in this study (Mg with a very thin Pd cap layer).
Fig. 7
shows the evolution of percentage of thickness increase
D t / t (after hydrogenation) with the initial thickness ( t o
) of Mg films. The horizontal dash line indicates the expected thickness increase after hydrogen loading, w
21% due to variation of lattice parameters (shown in
Fig. 6 f) from Mg to O-MgH
2
.
After hydrogen loading, Mg 200 nm expanded by 25% along the plane normal direction, slightly exceeding the predicted value. However, the value of D t / t o decreased rapidly, with increasing Mg film thickness, down to w
5% in Mg 1600 nm films. This somewhat surprising result (large deviation from expected values) is reconciled when examining the evolution of microstructure during hydrogenation.
XSEM studies show that the morphology of Mg films evolves with their thickness, namely thinner films (when t 200 nm) are fully dense, whereas porosity increases significantly in thicker films. The porosity of Mg films was measured before and after hydrogen loading. As shown in
Fig. S2 , the surface area (proportional to porosity) in asdeposited films increased sharply in as-deposited thicker Mg films. After hydrogen loading, the surface area of Mg 200 nm barely changed since the as-deposited Mg 200 nm films already had nearly full density. However, after hydrogenation, the porosity in porous films (Mg 800 and 1600 nm films) decreased drastically due to substantial volume expansion.
This result is in good agreement with SEM micrographs that illustrate morphological evolution in Mg films after hydrogenation in
Fig. 3 .
As shown schematically in
Fig. 8 , hydrogenation of fully
dense films ( t ¼ 200 nm) leads to significant increase of film thickness to accommodate the volume expansion during formation of Mg hydrides. However, during hydriding of thicker porous Mg films, there is sufficient free space to accommodate volume expansion by filling up porous films.
Hence the evolution of film thickness in porous Mg is much less dramatic than that in dense thin Mg films.

2602 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
2603
1.0x10
-5
8.0x10
-6 a
200nm
1st peak
6.0x10
-6
400nm
4.0x10
-6
800nm
2.0x10
-6
0.0
300
1600nm
350 400
Temperature (K)
2nd peak
450
450
425
400
1st peak
375
350
325
0
2nd peak onset b
400 800 1200 1600
Mg film thickness (nm)
Fig. 4 e
(a) TDS spectra of Mg hydride films show the evolution of hydrogen pressure during desorption of hydride films with different film thickness. A thinner
(200 nm for Mg, and 250 nm after hydrogen loading) Mg hydride film desorbed hydrogen at 373 K and showed only one prominent desorption peak, while thicker specimen desorbed hydrogen at higher temperature with two detectable peaks. (b) The H
2 desorption temperature for the onset, first and second peaks decreased with decreasing
Mg film thickness.
Fig. 5 e
TDS spectra of hydrided (373 K/24 h) Mg 5 and 45 pillar films. In Mg 45 pillar films, most hydrogen was desorbed during the first intense peak. Mg 5 pillar film shows several additional desorption peaks at higher temperatures.
to the substrate, O-MgH
2 in close proximity to substrate is likely to be more unstable comparing to that forms away from substrate/film interface, and consequently two hydrogen desorption peaks were observed. The thermodynamic stability is better characterized by the onset temperature during H
2 desorption.
Fig. 4 b shows the thinner films also have a clear size
dependent reduction of onset temperature for H desorption.
The magnitude of lattice distortion of O-MgH
2 is less significant in thicker Mg films as porosity provides fertile open volume to accommodate hydrogenation induced lattice expansion. Correspondingly the magnitude of stress during hydrogen loading is much lower, and consequently the thermal stability of distorted Mg hydride improves. The clear reduction of D t / t o with Mg film thickness in
Fig. 7
suggests the same conclusion: that is thicker porous films desorb hydrogen at higher temperature due to the lack of lattice constraint. Our previous pressure-composition isothermal (PCI) measurement shows that the formation enthalpy of O-MgH
2 w 37 kJ/mol. H
2
[43]
, much less than w 75 kJ/mol. H
2 is for T-
MgH
2
, and hence hydrogen desorption in O-MgH
2 occurs at a much lower temperature, w 400 K or less, than that in stable T-
MgH
2
, w
673 K. The drastic destabilization of Mg hydride due to stress-induced O-MgH
2 is thus an appealing approach to achieve low operation temperature in this light-weight metal hydride for vehicle applications.
The TDS spectra revealed size (film thickness) dependence on the hydrogen desorption temperature, that is, thinner Mg films desorb hydrogen at lower temperature than thicker films. In addition, double desorption peaks were observed in thicker
(porous) Mg films ( t > 400 nm). In Mg 800 nm films, the first peak at w
386 K may arise from the O-MgH
2 transformed from the underlying dense Mg layer. The other desorption peak at slightly higher temperature, w 400 K, could be due to the formation of O-MgH
2 from slightly porous Mg upper layer. As the volume expansion incompatibility (between hydrided films and unaffected rigid substrate) is more prominent in Mg films close
Our results are notably different from the work reported by
Fujii et al. who performed study on a similar Mg/Pd system
[1] .
Fig. 3 e
Cross-sectional SEM micrographs of Mg films and pillars before (a e e) and after hydrogen loading (a
0 e e
0
). (a) Mg
200 nm film had dense structure with continuous and smooth surface. (b) Mg 400 nm film started to show a porous structure as films grew thicker. (c) Mg 800 nm showed porous columnar structure with a dense 200 nm thick layer on the substrate.
(d e e) show Mg pillars produced by glancing angle deposition. 45 pillar film had columns grown toward the sputter plasma.
5 pillar film had extremely porous structure. After hydrogen loading, all MgH
2 increase and pores were filled up due to volume expansion. (d
0 films (a
0 e c
0
) showed substantial thickness
) After hydrogen loading Mg 45 pillar film had dense structure. (e
0
) Mg 5 pillar film remained pillar morphology after H
2 loading.

2604 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
Fig. 6 e
Three-dimensional structures of (a) hcp Mg, (b) T-MgH
2 and (c) O-MgH
2
. Comparison of two-dimensional views projected on (d) a (2110) plane of hcp Mg (e) a (001) plane of T-MgH
2
, and (f) a (001) plane of O-MgH
2
. The transparent blue planes represent the corresponding planes projected on a (2110) plane of hcp Mg. The transparent red planes represent the projected (001) plane in the conventional unit cells of T-MgH
2 and O-MgH
2
. The angle q changes from 70.2
to 90 when hcp
Mg transforms to Mg hydrides. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
In the previous study, Fujii et al. showed that for a Pd50 nm/
Mg/Pd50 nm trilayer system, the hydrogen desorption peak temperature decreases with increasing Mg film thickness from 385 K ( t
Mg
¼ 200 nm) to 350 K ( t
Mg
¼ 800 nm). Although cooperative phenomenon has been proposed to explain remarkable hydrogen sorption performance in Mg/Pd system, this model does not explain their reverse size dependent variation of desorption temperature. The differences between these two studies could be understood from several perspectives. First, we notice that their Mg films were sandwiched between two Pd layers, a cap and a buffer layer right on top of substrate. While having Pd on both ends might accelerate H
2 desorption, the usage of Pd as a buffer layer could also bring some challenge. The poor adhesion between Pd and glass substrate could lead to delamination of hydrided trilayer films
(or a stress-free Mg film)
[1] , and thus lose the advantage of
constraint (stress) from rigid substrate. Second, Pd has lower elastic modulus, w 117 GPa
[49] , compared to a biaxial elastic
modulus of 180 GPa
[49]
for Si (100) substrate. Hence there is consistently greater influence (constraint) arising from Si substrate in our study. Third, there is difference in morphology of as-deposited Mg films. Fujii et al. obtained dense and continuous films prepared by RF assisted deposition, while we generated a porous structure when t > 200 nm.
Fig. 7 e
The thickness increase of Mg films after hydrogen loading. Mg 200 nm film shows 25% expansion in out-ofplane direction. As film thickness increases, MgH
2 film thickness expansion decreases. The estimated values
(shown by dash line) were calculated assuming fully dense
Mg films.
The general perception is that Mg films with larger surface areas should have better H
2 sorption behavior. Hence porous
Mg pillars may have improved performance over continuous
Mg films. However our studies revealed the opposite phenomenon. Hydrogen sorption performance of the Mg pillars can also be explained by porosity and stress effect. The XRD studies in
Fig. 2
show that the 45 pillar film was fully hydrided to O-MgH
2 phase while the 5 pillar film was not, as indicated by the residual hcp-Mg peak. The highly porous and dispersed columnar structure in the 5 pillar film prevents the introduction of large stress from rigid substrate. The TDS spectrum in
Fig. 5
illustrates how porosity in pillars may impact hydrogen desorption kinetics. The 45 pillars film has a prominent low temperature desorption peak at w 408 K, and a weak, broad high temperature peak, spreading from 474 to 573 K. The low temperature desorption peak arises from dense Mg hydride layer adjacent to substrate, whereas the high temperature peak arises from H

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
2605
Fig. 8 e
Schematics illustrating the impact of stress and porosity on destabilization of Mg hydride. (a) Mg 200 nm film exhibits dense structure in both as-deposited and hydrogen-loaded states and film expanded 25% along the out-of-plane direction. (b) Porous thick Mg film (800 nm) had columnar structures with a thin initial dense layer. After hydrogen loading, pores in the film were filled up and the film showed only 6 e
7% thickness increase.
desorption from less dense Mg hydride (subjected to less stress effect). As shown in
Fig. 1 c, the formation of O-MgH
2 leads to a
(110) diffraction peak that has increasing difference (distortion) from (110) T-MgH
2 at smaller layer thickness. Such distortion is manifested by a gradual peak shift. XRD profiles in
Fig. 2
b shows this gradual peak shift in both pillar films, rather than two drastically different O-MgH
2 and T-MgH
2 peaks. Most hydrogen was however desorbed by w 425 K in the 45 pillar films. In contrast, in the 5 pillar film, the high temperature H
2 sorption peak ranging from 490 to 540 K dominated the desorption process, as the impact of stress is significantly reduced due to a much greater porosity in these highly porously pillar films.
Thus the pillar studies confirm the idea that constraint (stress) from rigid substrate is critical to achieve O-MgH
2 with superior
H
2 sorption performance in Mg.
Acknowledgments
We acknowledge financial support by NSF e
CBET, Energy for
Sustainability Program, under grant no. 0932249 and partial support by NSF-CMMI 1129065. Access to the microscopes at the Materials Characterization Facility at Texas A&M University is also acknowledged.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.ijhydene.2013.12.017
.
5.
Conclusions
We explored the hydrogen storage performance of porous and dense Mg films prepared by magnetron sputtering under various conditions. Thinner dense Mg films ( t ¼ 200 nm) have lower hydrogen desorption temperature than thicker porous films. Mg 45 pillar films also shows improved hydrogen sorption performance than the more porous 5 pillar films. Stress arising from rigid substrate during hydrogenation of Mg films plays a critical role to induce metastable orthorhombic Mg hydride that can easily desorb H
2 at low temperatures. The stress concept validated via numerous experiments could have general implications on destabilization of other types of metal hydrides for H
2 storage application.
r e f e r e n c e s
[1] Fujii H, Higuchi K, Yamamoto K, Kajioka H, Orimo S,
Toiyama K. Remarkable hydrogen storage, structural and optical properties in multi-layered Pd/Mg thin films. Mater
Trans 2002;43:2721 e
7 .
[2] Eberle U, Felderhoff M, Schu¨th F. Chemical and physical solutions for hydrogen storage. Angew Chem Int Edit
2009;48:6608 e
30 .
[3] Schlapbach L, Zuttel A. Hydrogen-storage materials for mobile applications. Nature 2001;414:353 e
8 .
[4] Jain IP, Vijay YK, Malhotra LK, Uppadhyay KS. Hydrogen storage in thin film metal hydride d a review. Int J Hydrogen
Energy 1988;13:15 e
23 .

2606 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
[5] Barkhordarian G, Klassen T, Bormann R. Catalytic mechanism of transition-metal compounds on Mg hydrogen sorption reaction. J Phys Chem B 2006;110:11020 e
4 .
[6] Zaluska A, Zaluski L, Stro¨m e
Olsen JO. Nanocrystalline magnesium for hydrogen storage. J Alloy Compd
1999;288:217 e
25 .
[7] Bogdanovi c B, Ritter A, Spliethoff B. Active MgH
2 e
Mg systems for reversible chemical energy storage. Angew
Chem Int Edit 1990;29:223 e
34 .
[8] Shao H, Xin G, Zheng J, Li X, Akiba E. Nanotechnology in Mgbased materials for hydrogen storage. Nano Energy
2012;1:590 e
601 .
[9] Aguey-Zinsou K-F, Ares-Fernandez J-R. Hydrogen in magnesium: new perspectives toward functional stores.
Energ Environ Sci 2010;3:526 e
43 .
[10] Fichtner M. Nanotechnological aspects in materials for hydrogen storage. Adv Eng Mater 2005;7:443 e
55 .
[11] Pundt A. Hydrogen in nano-sized metals. Adv Eng Mater
2004;6:11 e
21 .
[12] Huot J, Liang G, Schulz R. Mechanically alloyed metal hydride systems. Appl Phys A 2001;72:187 e
95 .
[13] Schwarz R. Hydrogen storage in magnesium-based alloys.
MRS Bull 1999;24:40 e
4 .
[14] Aguey-Zinsou KF, Ares Fernandez JR, Klassen T, Bormann R.
Effect of Nb
2
O
5 on MgH
2 properties during mechanical milling. Int J Hydrogen Energy 2007;32:2400 e
7 .
[15] Alapati SV, Johnson JK, Sholl DS. Identification of destabilized metal hydrides for hydrogen storage using first principles calculations. J Phys Chem B 2006;110:8769 e
76 .
[16] Hanada N, Ichikawa T, Fujii H. Catalytic effect of Ni nanoparticle and Nb oxide on H-desorption properties in MgH
2 prepared by ball milling. J Alloy Compd
2005;404 e
406:716 e
9 .
[17] Li W, Vajo JJ, Cumberland RW, Liu P, Hwang S-J, Kim C, et al.
Hydrogenation of magnesium nickel boride for reversible hydrogen storage. J Phys Chem Lett 2009;1:69 e
72 .
[18] Vajo JJ, Li W, Liu P. ChemInform abstract: thermodynamic and kinetic destabilization in LiBH
4
/Mg
2
NiH
4
: promise for borohydride-based hydrogen storage. Chem Commun
2010;46:6687 e
9 .
[19] Vajo JJ, Olson GL. Hydrogen storage in destabilized chemical systems. Scripta Mater 2007;56:829 e
34 .
[20] Vajo JJ, Salguero TT, Gross AF, Skeith SL, Olson GL.
Thermodynamic destabilization and reaction kinetics in light metal hydride systems. J Alloy Compd 2007;446:409 e
14 .
[21] Qu J, Wang Y, Xie L, Zheng J, Liu Y, Li X. Hydrogen absorption e desorption, optical transmission properties and annealing effect of Mg thin films prepared by magnetron sputtering. Int J Hydrogen Energy 2009;34:1910 e
5 .
[22] Bouhtiyya S, Roue´ L. Pd/Mg/Pd thin films prepared by pulsed laser deposition under different helium pressures: structure and electrochemical hydriding properties. Int J Hydrogen
Energy 2009;34:5778 e
84 .
[23] Qu J, Wang Y, Xie L, Zheng J, Liu Y, Li X. Superior hydrogen absorption and desorption behavior of Mg thin films. J Power
Sources 2009;186:515 e
20 .
[24] Remhof A, Borgschulte A. Thin-film metal hydrides. Chem
Phys Chem 2008;9:2440 e
55 .
[25] Chung CJ, Lee S-C, Groves JR, Brower EN, Sinclair R,
Clemens BM. Interfacial alloy hydride destabilization in Mg/
Pd thin films. Phys Rev Lett 2012;108:106102 .
[26] Mooij LPA, Baldi A, Boelsma C, Shen K, Wagemaker M,
Pivak Y, et al. Interface energy controlled thermodynamics of nanoscale metal hydrides. Adv Energy Mater 2011;1:754 e
8 .
[27] Yu L, Gwo-Ching W. Air stability of low-temperature dehydrogenation of Pd-decorated Mg blades.
Nanotechnology 2012;23:025401 .
[28] Hanada N, Ichikawa T, Fujii H. Catalytic effect of nanoparticle 3d-Transition metals on hydrogen storage properties in magnesium hydride MgH
2 prepared by mechanical milling. J Phys Chem B 2005;109:7188 e
94 .
[29] Liang G, Huot J, Boily S, Van Neste A, Schulz R. Catalytic effect of transition metals on hydrogen sorption in nanocrystalline ball milled MgH
2 e
Tm (Tm ¼ Ti, V, Mn, Fe and Ni) systems. J Alloy Compd 1999;292:247 e
52 .
[30] Pozzo M, Alfe` D. Hydrogen dissociation and diffusion on transition metal ( ¼ Ti, Zr, V, Fe, Ru, Co, Rh, Ni, Pd, Cu, Ag)-doped
Mg(0001) surfaces. Int J Hydrogen Energy 2009;34:1922 e
30 .
[31] Pasturel M, Slaman M, Schreuders H, Rector JH, Borsa DM,
Dam B, et al. Hydrogen absorption kinetics and optical properties of Pd-doped Mg thin films. J Appl Phys
2006;100:023515 e
8 .
[32] Gremaud R, Borgschulte A, Chacon C, van Mechelen JLM,
Schreuders H, Zu¨ttel A, et al. Structural and optical properties of Mg x
Al
1 x
H y gradient thin films: a combinatorial approach. Appl Phys A 2006;84:77 e
85 .
[33] Richardson TJ, Slack JL, Armitage RD, Kostecki R, Farangis B,
Rubin MD. Switchable mirrors based on nickel e magnesium films. Appl Phys Lett 2001;78:3047 e
9 .
[34] Niessen RAH, Notten PHL. Electrochemical hydrogen storage characteristics of thin film MgX (X ¼ Sc, Ti, V, Cr) compounds. Electrochem Solid St 2005;8:A534 e
A538 .
[35] Ingason A´S, Eriksson AK, O
Mg:C thin films. J Alloy Compd 2007;446 e
447:530 e
3 .
[36] Zahiri B, Amirkhiz BS, Mitlin D. Hydrogen storage cycling of
MgH
2 thin film nanocomposites catalyzed by bimetallic Cr Ti.
Appl Phys Lett 2010;97:083106-3 .
[37] Baldi A, Gremaud R, Borsa DM, Balde´ CP, van der Eerden AMJ,
Kruijtzer GL, et al. Nanoscale composition modulations in
MgyTi
1 y
H x thin film alloys for hydrogen storage. Int J
Hydrogen Energy 2009;34:1450 e
7 .
[38] Borsa DM, Gremaud R, Baldi A, Schreuders H, Rector JH,
Kooi B, et al. Structural, optical, and electrical properties of
Mg y
Ti
1 y
H x thin films. Phys Rev B 2007;75:205408 .
[39] Baldi A, Gonzalez-Silveira M, Palmisano V, Dam B,
Griessen R. Destabilization of the Mg e
H system through elastic constraints. Phys Rev Lett 2009;102:226102 .
[40] Gremaud R, Broedersz CP, Borsa DM, Borgschulte A,
Mauron P, Schreuders H, et al. Hydrogenography: an optical combinatorial method to find new light-weight hydrogenstorage materials. Adv Mater 2007;19:2813 e
7 .
[41] Zheng Q, Pivak Y, Mooij LPA, van der Eerden AMJ,
Schreuders H, de Jongh PE, et al. EXAFS investigation of the destabilization of the Mg e
Ni e
Ti (H) system. Int J Hydrogen
Energy 2012;37:4161 e
9 .
[42] Tan X, Wang L, Holt CMB, Zahiri B, Eikerling MH, Mitlin D.
Body centered cubic magnesium niobium hydride with facile room temperature absorption and four weight percent reversible capacity. Phys Chem Chem Phys 2012;14:10904 e
9 .
[43] Ham B, Junkaew A, Arroyave R, Chen J, Wang H, Wang P, et al. Hydrogen sorption in orthorhombic Mg hydride at ultra-low temperature. Int J Hydrogen Energy
2013;38:8328 e
41 .
[44] Ham B, Zhang X. High strength Mg/Nb nanolayer composites. Mater Sci Eng A 2011;528:2028 e
33 .
[45] Junkaew A, Ham B, Zhang X, Talapatra A, Arro´yave R.
REPORT: stabilization of bcc Mg in thin films at ambient pressure: experimental evidence and ab initio calculations.
Mater Res Lett 2013:1 e
7 .
[46] Stoney GG. The tension of metallic films deposited by electrolysis. Proc R Soc Lond A 1909;82:172 e
5 .
[47] Moriwaki T, Akahama Y, Kawamura H, Nakano S,
Takemura K. Structural phase transition of rutile-type MgH
2 at high pressures. J Phys Soc Jpn 2006;75:074603 .

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 9 ( 2 0 1 4 ) 2 5 9 7 e
2 6 0 7
2607
[48] Vajeeston P, Ravindran P, Kjekshus A, Fjellva˚g H. Pressureinduced structural transitions in MgH
2
. Phys Rev Lett
2002;89:175506 .
[49] Simmons G, Wang H. Single crystal elastic constants and calculated aggregate properties: a handbook. Cambridge,
Mass: MIT Press; 1971 .