Functionality of the nanoscopic crystalline Al/amorphous Al 50Ti50 surface embedded composite observed in the NaAlH4 + xTiCl3 system after milling

Article abstract of DOI:10.1016/j.jallcom.2011.11.045

The NaAlH₄ + xTiCl₃ (x < 0.1) system has been studied by a combination of X-ray synchrotron and neutron diffraction, and isotopic H₂/D₂ scrambling after the completion of the milling process, and the first thermal release of hydrogen (H). An in situ X-ray synchrotron diffraction study of the isochronal release of hydrogen from planetary milled (PM) NaAlH₄ + 0.1TiCl₃ shows that crystalline (c-) Al_1-xTi_x phases do not form until almost all H is released from the sample, demonstrating that the surface embedded nanoscopic crystalline Al/amorphous (a-) Al₅₀Ti₅₀ composite facilitates the release of H during the very first thermal desorption. Planetary milled (PM) NaAlH₄ + xTiCl₃ is observed to disproportionate at room temperature, with no NaAlH₄ remaining after ca. 200 days. A complete lack of ambient hydrogen release from PM NaAlH ₄ + 0.1Al (80 nm) measured over 200 days suggests that the nanoscopic a-Al₅₀Ti₅₀ phase is entirely responsible for the hydrogen release during thermal desorption of milled NaAlH₄ + xTiCl ₃. Isotopic H/D exchange has been observed by combined neutron and X-ray synchrotron diffraction on a PM NaAlD₄ + 0.04TiCl₃ sample, after exposing the milled sample to 20 bar H₂ at 50 °C for ca. 6 days. Under these pressure/temperature (P/T) conditions, disproportionation of NaAlD₄ is avoided, and ca. 32% of D atoms are exchanged with H atoms. Asymmetrically broadened reflections in the synchrotron data show peak splitting into two unit cell types, one expanded with H, the other remaining close to pure D based unit cell dimensions. The 2-phase model when fitted to the neutron data demonstrates that ca. 56% of D atoms in ca. 58% of all unit cells are exchanged with H, yielding a NaAl(H_0.56D _0.44)₄ composition for the expanded unit cells. HD scrambling (1 bar mixture of H₂ and D₂ at 23 °C) performed on desorbed H empty PM NaAlH₄ + 0.1TiCl₃ shows classic H₂ + D₂ ? 2HD equilibrium mixing, demonstrating that nanoscopic Ti containing Al_1-xTi_x surface embedded phases perform a H₂ dissociation/recombination function that unadulterated NaAlH₄ cannot.

Full text of DOI:10.1016/j.jallcom.2011.11.045

Journal of Alloys and Compounds 514 (2012) 163–169
Contents lists available at SciVerse ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Functionality of the nanoscopic crystalline Al/amorphous Al Ti surface
5
0
50
embedded composite observed in the NaAlH + xTiCl system after milling
4
3
a,c,∗
b
a
d
e
c
c
M.P. Pitt
, P.E. Vullum , M.H. Sørby , M.P. Sulic , H. Emerich , M. Paskevicius , C.E. Buckley ,
J.C. Walmsleyb,f, R. Holmestad , B.C. Hauback
Physics Department, Institute for Energy Technology, P.O. Box 40, Kjeller, NO-2027, Norway
Department of Physics, Norwegian University of Science and Technology, NO-7491, Trondheim, Norway
Department of Imaging and Applied Physics, Curtin University, GPO Box U1987, Perth 6845, Western Australia, Australia
Department of Chemistry, University of Hawaii, Honolulu, HI 96822, USA
b
a
a
b
c
d
e
f
Swiss-Norwegian Beam Line, European Synchrotron Radiation Facility, BP 220, Grenoble, Cedex, France
SINTEF Materials and Chemistry, NO-7465 Trondheim, Norway
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 5 August 2011
Received in revised form 7 November 2011
Accepted 9 November 2011
Available online 25 November 2011
The NaAlH + xTiCl (x < 0.1) system has been studied by a combination of X-ray synchrotron and neu-
4
3
tron diffraction, and isotopic H2/D2 scrambling after the completion of the milling process, and the first
thermal release of hydrogen (H). An in situ X-ray synchrotron diffraction study of the isochronal release
of hydrogen from planetary milled (PM) NaAlH4 + 0.1TiCl3 shows that crystalline (c-) Al1−xTix phases
do not form until almost all H is released from the sample, demonstrating that the surface embedded
nanoscopic crystalline Al/amorphous (a-) Al50Ti50 composite facilitates the release of H during the very
first thermal desorption. Planetary milled (PM) NaAlH4 + xTiCl3 is observed to disproportionate at room
temperature, with no NaAlH4 remaining after ca. 200 days. A complete lack of ambient hydrogen release
from PM NaAlH4 + 0.1Al (80 nm) measured over 200 days suggests that the nanoscopic a-Al50Ti50 phase
is entirely responsible for the hydrogen release during thermal desorption of milled NaAlH4 + xTiCl3. Iso-
topic H/D exchange has been observed by combined neutron and X-ray synchrotron diffraction on a PM
Keywords:
Metal hydrides
Amorphous materials
Transition metal alloys and compounds
Gas–solid reactions
Synchrotron radiation
Neutron diffraction
◦
NaAlD + 0.04TiCl sample, after exposing the milled sample to 20 bar H2 at 50 C for ca. 6 days. Under
4
3
these pressure/temperature (P/T) conditions, disproportionation of NaAlD4 is avoided, and ca. 32% of D
atoms are exchanged with H atoms. Asymmetrically broadened reflections in the synchrotron data show
peak splitting into two unit cell types, one expanded with H, the other remaining close to pure D based
unit cell dimensions. The 2-phase model when fitted to the neutron data demonstrates that ca. 56% of D
atoms in ca. 58% of all unit cells are exchanged with H, yielding a NaAl(H0.56D0.44)4 composition for the
◦
expanded unit cells. HD scrambling (1 bar mixture of H2 and D2 at 23 C) performed on desorbed H empty
PM NaAlH4 + 0.1TiCl3 shows classic H2 + D2 ↔ 2HD equilibrium mixing, demonstrating that nanoscopic
Ti containing Al1−xTix surface embedded phases perform a H2 dissociation/recombination function that
unadulterated NaAlH4 cannot.
©
2011 Elsevier B.V. All rights reserved.
1
. Introduction
location and functionality of the transition metal (TM)/rare earth
RE) containing phases is paramount to basic understanding and
(
The transition metal enhanced NaAlH4 system currently
potentially further engineering other classes of complex hydrides
such as the borohydride family [3,4]. Large quantities of hydrogen
remains as the prototypical example of obtaining hydrogen
reversibility and rapid absorption/desorption hydrogenation kinet-
ics in the complex hydride family. The early transition metals such
as Sc [1], Ti [2] and rare earths such as Ce [1,2] have proven to be the
most efficient additives for the NaAlH4 system. Understanding the
can be contained in phases such as LiBH and Mg(BH ) , containing
4
4 2
18.5 and 14.9 wt.% H, respectively, but these compounds are greatly
hindered as practical hydrogen storage materials by high ther-
mal and kinetic stabilities, and no suitable destabilizing catalysts
have been found to date. A common theme among the addition of
metal enhancing species to NaAlH4 is the formation of nanoscopic
Al1 TMx [5–7] or Al
surface of NaAlH4 powder grains [8].
In [8], the location of Ti was determined after the comple-
tion of the NaAlH4 + xTiCl3 milling process at short times of ca.
REx [2] phases that are embedded on the
−x
1−x
∗ Corresponding author at: Hydrogen Storage Research Group, Department of
Imaging and Applied Physics, Curtin University, Kent Street, Bentley, Perth, Western
Australia 6102, Australia. Tel.: +61 8 9266 3673; fax: +61 8 9266 2377.
E-mail address: mark.pitt@gmail.com (M.P. Pitt).
0
925-8388/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2011.11.045

1
64
M.P. Pitt et al. / Journal of Alloys and Compounds 514 (2012) 163–169
1
h. It was observed by combinatorial X-ray synchrotron diffrac-
tion and transmission electron microscopy (TEM) measurements
that Ti was located in a nanoscopic composite of Al nanocrystals
immersed in an amorphous Al50Ti50 matrix. This nano composite
is embedded on the surface of moderately defected (uncorrelated)
single crystals of NaAlH . From the study in [5], it is known that
4
nanoscopic crystalline (c-) Al1 Tix phases can be formed from the
−x
nano Al/a-Al50Ti50 composite by extended planetary milling (>10 h)
◦
producing c-Al80Ti20, isothermal vacuum annealing (at 150 C)
◦
producing c-Al88Ti12, isochronal annealing (2 C/min) producing
c-Al86Ti14, or from hydrogen (H) cycling (150 bar at 150 C) yield-
◦
ing c-Al85Ti15. This suggests that at some stage during the very
first thermal desorption of hydrogen from milled NaAlH + xTiCl ,
4
3
the amorphous Al50Ti50 in the nanoscopic Al/a-Al50Ti50 composite
must be depleted/transformed to a crystalline Al1 Tix structure.
−x
The matching of size ranges from dark field measurements for the
nanoscopic Al crystals (2–20 nm) in the Al/a-Al50Ti50 composite
in PM NaAlH + 0.1TiCl3 compared to the size range of c-Al85Ti15
4
nanocrystals (4–25 nm) in H cycled NaAlH + 0.1TiCl3 from the
4
study in [5] indicates that Ti atoms are sourced from the a-Al50Ti50
matrix, where they locally diffuse into the Al nanocrystals. This
process occurs within the nanoscopic Al/a-Al50Ti50 composite on
the NaAlH4 powder grain surface during the very first thermally
induced release of H.
It is the aim of this study to determine the temperature that Al
nanocrystals are converted to c-Al1 Tix structures, and in doing
−x
so demonstrate the temperature range over which the amorphous
Al50Ti50 may be considered functional in terms of H release. We
also utilise isotopic H /D2 scrambling in combination with neu-
2
tron diffraction to demonstrate that the surface embedded nano
Al/Al50Ti50 composite is not only capable of molecular H2 dissoci-
ation/recombination, but also promotes bulk diffusion of H within
the NaAlH4 structure.
2.
Experimental procedure
NaAlH4 was purchased from Albermarle Corporation (LOT NO.#: 22470404-01).
All halide precursors were purchased from Sigma–Aldrich Chemicals Inc. (>99.99%
purity). 80 nm Al (99.9% purity) was obtained from Nanostructured & Amorphous
Materials Inc. NaAlD4 was synthesised in accordance with previously reported wet
synthesis of NaAlH4 [9]. At all times, all powders have been handled under inert
Ar or N2 atmosphere in a dry glove box, with <1 ppm O2 and H2O. Milled NaAlH4
powders (pure and with TiCl3 and Al additive) were prepared in 1 g quantities in
a Fritsch P7 planetary mill, with ball to powder ratio (bpr) of 20:1, at 750 rpm for
a period of 1 h, under glove box atmosphere. Milled powders were taken directly
to the beamline after milling for diffraction measurements. HD scrambling studies
were performed with a dual focussing high resolution VG Model 70 SE Mass Spec-
trometer. The instrument was focussed on 4 amu (D2). Samples were introduced
Fig. 1. (a) Long term room temperature disproportionation of PM NaAlH4 + 0.02TiCl3
measured up to 196 days, showing an exponential decay in NaAlH4 phase propor-
tion. (b) Room temperature HD scrambling of PM only NaAlH4 + 0.1TiCl3 measured
up to 42 days against a starting pressure of 1 bar D2.
through a stainless steel inlet. Initially with the inlet at a working temperature of
◦
1
80 C, a mixture of H2 and D2 could be slowly and reproducibly hybridised. As such,
the inlet was cooled and head gas samples were prevented from remaining in the
inlet for too long a period. Samples were connected from glassware with a double
ended 20 gauge needle through a pair of septa. Accelerating voltage was 8000 V.
Resolution was set to 800, yielding excellent sensitivity at low AMU. Detection effi-
ciency for H2 was ca. 7% less than HD. Detection efficiency for HD was ca. 7% less
than for D2. HD measurements were performed typically up to 7 days at ambient
temperature. Powder X-ray diffraction data were recorded at the Swiss–Norwegian
Beamline (SNBL) at the European Synchrotron Radiation Facility (ESRF) in Grenoble,
France. Samples were contained in rotating 0.8 mm boron–silica glass capillaries.
diffraction analysis software RIETICA [11]. A mixed H/D model in space group I4 /a
was refined for the isotopically D/H exchanged NaAlD(H)4 structure type. A Voigt
1
profile function was used to separate size and strain contributions to the diffrac-
tion line shape. The strong background induced from the presence of H atoms in
the structure was modelled with a type I Chebyshev polynomial. X-ray synchrotron
diffraction patterns were analysed by the Rietveld method, using RIETICA. Diffrac-
tion lineshape profiles were fitted with a full Voigt function, with the instrumental
shape determined by a NIST LaB 660a lineshape standard, further annealed to
6
1800 ◦C.
−4
High resolution data (ꢀd/d ∼ 2 × 10 ) was typically collected at 295 K between
◦
◦
5
and 35 2ꢁ, in steps of 0.003–0.030 , depending on the sample broadening. A
wavelength of 0.4998 A˚ was obtained from a channel cut Si (1 1 1) monochroma-
3. Results and discussion
−3
tor. Medium resolution (ꢀd/d ∼ 3 × 10 ) in situ annealing data were collected on
◦
◦
a 2-D image plate (MAR345) over the 2ꢁ range 3–34 with step size 0.015 and
In typical 1 h PM NaAlH + xTiCl samples, the conditions on the
4
3
exposure time of 30 s. A wavelength of 0.7111 A˚ was used. Powder neutron diffrac-
tion data was collected using the PUS diffractometer at the JEEP II reactor at Kjeller,
Norway [10]. Monochromated neutrons with ꢂ = 1.5550 A˚ were obtained from a Ge
powder surface after milling are sufficient to allow room temper-
ature desorption of hydrogen from the sample, demonstrated in
Fig. 1(a). There is almost complete exponential decay of NaAlH4
into Na AlH and Al after 196 days of storage in an inert atmosphere
(
5 1 1) focussing monochromator. The detector unit consists of two banks of seven
3
◦
◦
position-sensitive He detectors, each covering 20 in 2ꢁ (binned in 0.05 steps).
Data was collected in the 2ꢁ range 10–130 . The as synthesised NaAlD4 and the
3
6
◦
glove box. It is clear that significant amounts of H are released from
the powder bulk where no Ti containing phases are present. This
indicates that under ambient conditions, the surface embedded Ti
isotopically D/H exchanged NaAl(D1−xHx)4 + 0.04TiCl3 powders were measured in a
sealed cyclindrical vanadium sample holder of 5 mm diameter, which was rotated
at room temperature. Neutron diffraction patterns were modelled with the Rietveld

M.P. Pitt et al. / Journal of Alloys and Compounds 514 (2012) 163–169
165
Fig. 2. (a) Neutron diffraction pattern of as synthesised pure NaAlD4. Data are represented by open circles, the calculated Rietveld fit by the solid line. The difference profile
between the data and the calculated pattern is below. Reflection markers for NaAlD4 are given by the vertical bars. (b) Neutron diffraction pattern of PM NaAlD4 + 0.04TiCl3
◦
after soaking under 20 bar H2 at 50 C for ca. 6 days. Reflection markers from top to bottom represent the H expanded phase NaAl(H0.56D0.44)4, NaCl, Al and the remaining
unexpanded NaAlD4 phase.
rich nano Al/a-Al50Ti50 composite performs a molecular H recom-
we assume also that at low temperature, the hydride/deuteride
plateau pressures are equivalent to each other. As such, we then
expect that disproportionation of NaAlH4 is avoided under 1 bar
D2 gas pressure. In Fig. 1(b), a clear exponential rise in HD frac-
tion is observed over the first 14 days, after which, the HD fraction
can be observed to artefactually decrease as the finite supply of
D2 is consumed. Even after nearly all D2 is consumed after 42
days, the H2 fraction does not appear to have plateaued, indicat-
2
bination function that unadulterated NaAlH cannot, and facilitates
4
the release of H bound in the NaAlH4 bulk. It is not yet clear if after
1
96 days, H will begin to be released from Na AlH .
3 6
We have utilised isotopic HD scrambling experiments to confirm
that the surface embedded Ti rich nano Al/a-Al50Ti50 composite
works in reverse, and also dissociates molecular H2 and allows
atomic H back into the NaAlH4 structure by allowing H trans-
port across the near surface NaAlH /a-Al50Ti50 interface. By near
ing that some NaAlH decomposition may have occurred to release
4
4
surface, we refer to the NaAlH /a-Al50Ti50 interface, which is on
H. As no isotherms exist for the PCT diagram at room temperature,
we must rely on the extrapolation of the van’t Hoff function and
the assumption of no isotope affect at low temperature. Clearly,
the starting pressure of 1 bar D2 is close to the predicted plateau
pressure for NaAlH4 at ca. 0.5 bar, indicating that HD scrambling
should be performed at higher pressure to observe H/D exchange
in NaAlH4.
4
average 10 nm below the outer powder grain surface [8]. It is at
this interfacial monolayer where Ti atoms may protrude into the
NaAlH4 unit cell, and very locally distort Al–H bonds immediately
in the vicinity of the interface. Fig. 1(b) shows a HD scrambling
experiment on PM NaAlH + 0.1TiCl , conducted at room tempera-
4
3
ture under a starting D2 pressure of 1 bar for 42 days. The starting
pressure of 1 bar is above the predicted plateau pressure of ca.
◦
Such an experiment has been conducted at 50 C and 20 bar
◦
0.5 bar at room temperature, based on an average extrapolation
of D2 on milled NaAlH + 0.04TiCl3 [16]. At 50 C, the predicted
4
of all known pressure–composition–temperature (PCT) data for
Ti enhanced NaAlH4 [12–15]. Based on equivalent plateau pres-
plateau pressure is ca. 1.7 bar, which is well below the 20 bar start-
ing pressure, again suggesting that NaAlH disproportionation will
4
◦
sures between TiCl enhanced NaAlH and NaAlD at 130 C in [12],
be avoided (and again assuming no significant isotope effect at low
3
4
4

1
66
M.P. Pitt et al. / Journal of Alloys and Compounds 514 (2012) 163–169
◦
Fig. 3. (a) High resolution X-ray synchrotron diffraction pattern of PM NaAlD4 + 0.04TiCl3 after soaking under 20 bar H2 at 50 C for ca. 6 days. A single phase calculated fit is
shown to demonstrate the artefactual positional misfit. (b) Single phase NaAl(H0.325D0.625)4 model showing artefactual positional misfit for the (004) reflection. (c) A 2-phase
model fit to the (0 0 4) doublet with misfit removed. The model contains 58% H expanded NaAl(H0.56D0.44)4 unit cells, and 42% pure NaAlD4 unit cells.
temperature). In [16], a steep decay in D /HD ratio can be observed
over an initial period of ca. 33 h, after which the ratio plateaus. In
phase than is observed experimentally for bulk micron-sized
powders [17]. Further, TDS analysis demonstrates an extremely
high hydrogen release temperature of >800 K for both a-Al50Ti50H36
and a-Al60Ti40H21 [17], indicating it is highly unlikely that H(D)
proceeds through the 5–10 nm thick surface embedded a-Al50Ti50
phase. Rather, atomic H is likely to diffuse along/around the surface
of the nanoscopic a-Al50Ti50 phase.
2
total, 1.3 of the 4 formula units of H of NaAlH are estimated to have
4
2
−
exchanged with D, on the basis of 2.13 × 10 mol of D atoms recov-
◦
ered at 180 C after annealing the pre-evacuated sample at the end
of the HD scrambling period. It is inferred that bulk exchange of H
with D in the NaAlH4 phase has occurred, without decomposition.
With knowledge of the surface structure from our previous
study [8], it is possible to further investigate the nature of this
apparently exchanged D. As Al50Ti50 is known to absorb hydrogen,
the H(D) absorption properties of the surface become relevant. A
comprehensive study of the H absorption properties of amorphous
In order to study the distribution of exchanged bulk H observed
in [16], we have emulated the experiment in reverse, by study-
ing a deuteride, PM NaAlD + 0.04TiCl , under hydrogen gas, H , at
4
3
2
◦
20 bar and 50 C. We have observed the pre and post exchanged
samples by neutron and X-ray synchrotron diffraction. As D and
H have opposite sign and differing magnitude neutron scattering
lengths, and as H also has a large incoherent scattering cross sec-
tion, the exchange of D with H in the deuterated bulk will show
strong reflection intensity reductions and a strong background
increase in a neutron diffraction pattern. Diffraction can also con-
firm that NaAlH4 decomposition has been avoided during the H/D
and crystalline Al1 Tix alloys demonstrates that a-Al50Ti50 and a-
−x
Al60Ti40 absorb significant quantities of H at room temperature,
up to Al60Ti40H21 and Al50Ti50H36, under 50 bar H2 pressure [17].
◦
As 96% of the exchanged D was recovered on annealing to 180 C
in [16], only a small fraction of D could possibly be absorbed by
the surface embedded a-Al50Ti50 phase. This is consistent with the
−
2
analysis that to store the recovered 2.13 × 10 mol of D atoms,
0 times as much H(D) would need to be stored in the a-Al50Ti50
exchange, or alternatively verify the presence of H in any Na AlD6
3
4
produced. Fig. 2(a and b) shows the dramatic changes in neutron

M.P. Pitt et al. / Journal of Alloys and Compounds 514 (2012) 163–169
167
◦
◦
Fig. 4. In situ isochronal (2 C/min) annealing data in the hydrogen release temperature range 51–228 C from PM NaAlH4 + 0.1TiCl3, across the most interesting range of
d-spacing from 2.28 to 3.10 A˚ .
diffraction patterns before and after soaking a NaAlD + 0.04TiCl
was fitted, with both the exchanged and un-exchanged unit cell
4
3
◦
sample under 20 bar H2 gas at 50 C for ca. 6 days. Maximum Bragg
reflected intensity of the NaAlD4 phase is reduced to ca. 13% of
the starting magnitude, and the incoherent background structure
is more than doubled. A minor instrumental reflection is observed
dimensions refined, yielding a = 5.0095 A˚ and c = 11.3106 A˚ for the
pure NaAlD4 cell, and a = 5.0163 A˚ and c = 11.3351 A˚ for the mixed
2
H/D unit cell, with final ꢃ = 2.059. This model yields an excellent fit
to the (2 0 0) doublet in the neutron data, with the final number of
affected unit cells in the sample standing at ca. 58%, yielding a final
mixed H/D unit cell of composition NaAl(H0.56D0.44)4. This final
2-phase model also demonstrates that the mixed H/D cell mechan-
ically dilates the un-exchanged NaAlD4 cells prismatically by ca.
1%. Fig. 3(b) shows the (0 0 4) reflection for the single phase model,
and Fig. 3(c) the final 2-phase model, with the artefactual posi-
tional misfit removed. Lineshape analysis indicates the mosaic of
the exchanged H/D unit cells is ca. 100 nm compared to ca. 250 nm
for the un-exchangedcells, demonstratingthat large regions of each
powder grain have experienced exchange and allowed long ranged
◦
at 31.18 in the pure NaAlD4 sample, which is excluded from the
refinement in Fig. 2(b). The reflection markers from top to bot-
tom in Fig. 2(b) represent the H expanded NaAlD4 phase, NaCl, Al
and the remaining unexpanded NaAlD phase. No Na Al{D
Hx}
4
3
1−x
6
or Na AlD phase is evident, demonstrating that H is exchanged
3
6
only with the NaAlD4 phase, and no disproportionation occurs,
which is consistent with the equilibrium plateau pressure for the
◦
NaAl{D Hx} phase being well below 20 bar at 50 C. Initially, a
1
−x
4
single phase mixed NaAl{D Hx} model was fitted to the neu-
1−x
4
tron data, with the best fit yielding ca. 32.5% occupancy of H, in
excellent agreement with [16], where 1.3 out of the 4 formula units
of H were exchanged with D. Although a statistically satisfactory
fit to the neutron data was obtained with the single phase model,
one experimental reflection remains almost completely underfit-
H/D diffusion to occur within the NaAlD structure on a depth scale
4
of 100 nm from the surface. Although the NaAl(H0.56D0.44)4 com-
position represents an average measurement, the mixed H:D ratio
suggests H and D can be mixed in individual AlD4 units. A signif-
icant fraction of the powder remains un-exchanged, with 42% of
◦
ted by the calculated model (2 0 0) (at 2ꢁ = 36.18 ). High resolution
X-ray synchrotron data of the D/H exchanged sample provided the
information necessary to produce calculated intensity in the (2 0 0)
reflection. Fig. 3(a) shows the X-ray synchrotron diffraction pattern
of the D/H exchanged sample with a single phase model fitted. Posi-
tional misfit is apparent everywhere in the difference profile, and on
close inspection, strong leading edge, high d-spacing asymmetry is
evident on every reflection, indicating the presence of two tetrahy-
dride phases. Rietveld fitting of the synchrotron data utilised the
following methodology: (i) a single phase model was fitted, with
unit cell dimensions intermediate between the pure NaAlH4 and
all unit cells remaining as pure NaAlD , which is likely because the
system has reached an equilibrium distribution.
Commensurate with the observation that milled pure NaAlH4
samples have produced spectral features in infra red and NMR data
[18,19] indicating perturbed Al–H bonds, we have also studied a
4
milled only pure NaAlH sample (no additive) with HD scrambling.
4
We observe a 10% rise in HD fraction over 7 days for NaAlH4 under
a 1 bar H /D2 mixture, considerably less HD than that observed
2
for TiCl3 enhanced NaAlH . We have not measured this sample
4
by diffraction, and cannot ascertain that bulk H/D exchange has
occurred, however, it is clear that the pure NaAlH4 surface after
milling is amenable to molecular dissociation/recombination of
pure NaAlD unit cells, with a = 5.0148 A˚ and c = 11.3247 A˚ and
4
2
ꢃ = 3.122. This model contributes zero intensity to the (2 0 0)
doublet in the neutron data; (ii) a 2-phase model was fitted,
H . The non-zero strain component in the pure milled NaAlH4
2
with the remaining un-exchanged NaAlD4 unit cell constrained
diffraction lineshape, combined with the single crystalline nature
of the powder grains strongly suggests that the strain compo-
nent is of a non-correlated nature, as no grain boundary network
exists. It is thus likely that the grain is moderately defected with
either free edge or screw dislocations. The termination of free
edge/screws on the powder grain surface may serve as dissocia-
at pure NaAlD4 dimensions, with a = 5.0084 A˚ and c = 11.3034 A˚ ,
and a mixed H/D cell with a = 5.0138 A˚ and c = 11.3257 A˚ , yield-
2
ing a ꢃ = 2.504. When this model is applied to the neutron data,
the (2 0 0) doublet intensity is ca. (3/4) fitted, and the number
of expanded unit cells determined by quantitative phase analysis
(
QPA) is slightly overestimated at ca. 73%; (iii) a 2-phase model
tion/recombination sites for molecular H , by virtue of the extra
2

1
68
M.P. Pitt et al. / Journal of Alloys and Compounds 514 (2012) 163–169
energy stored as line tension in the dislocation cores. After the first
thermal desorption and subsequent first absorption of hydrogen
◦
(
with typical H cycling conducted up to 150 C and 150 bar), the
NaAlH4 lineshape is considerably sharper, indicating that all the
free microstructure has been annealed out. As such, the uncorre-
lated microstructure may provide a minor role during the first low
temperature release of hydrogen before the NaAlH4 phase is com-
pletely decomposed, however, in subsequent H cycles it is of no real
significance. To eliminate the potential role of nanoscopic Al in dis-
sociation/recombination of molecular H /and or bulk release of H, a
2
NaAlH + 0.1Al (80 nm) sample was milled and stored in the glove
4
box at ambient temperature, and followed over 1.5 years, to test
if Al destabilized NaAlH . After 1.5 years, the diffraction patterns
4
are identical, with only NaAlH4 and Al present, with no indication
of Na AlH . While we might expect a reduced effect for the larger
3
6
8
0 nm Al compared to the 20 nm Al in the surface embedded nano
Al/a-Al50Ti50 composite, no bulk H is released at all, and it is a safe
assumption that the a-Al50Ti50 phase is directly responsible for the
ambient release of bulk H.
Considering the large amount of D/H exchange observed, and
the obvious H molecular dissociation/recombination properties
2
◦
at 23–50 C, knowledge of the crystallisation temperature of the
Fig. 5. HD scrambling of a starting 1 bar mixture of H2 and D2 on dehydrided
PM NaAlH4 + 0.1TiCl3. A 1.19:2.00:0.95 mixture of H2:HD:D2 is obtained after 7
days (normalised to HD), showing near classic equilibrium mixing according to
H2 + D2 ↔ 2HD.
surface embedded nanoscopic a-Al50Ti50 phase is desirable from
the point of view that H does not appear to be stored in it, nor
travel through the a-Al50Ti50 bulk. Fig. 4 shows isochronal anneal-
ing of a PM only NaAlH + 0.1TiCl sample, over the temperature
4
3
◦
◦
range 51–228 C. Most of the H from NaAlH is desorbed by 127 C,
4
◦
obtain the true exchange independent HD surface scrambling rate,
it is necessary to conduct HD scrambling at P/T conditions at which
the sample will not hydride (i.e., at pressures below the equilib-
rium plateau pressure), and after at least one thermal desorption
and most of the H from Na AlH6 by 203 C. Crystalline Al
Tix is
3
1−x
◦
observed to start forming at 177 C, with a strong asymmetry on Al
◦
(
1 1 1) evident by 203 C. As most of the H has been released from
◦
the sample by 203 C, it is clear that the a-Al50Ti50 matrix allows
has occurred to create crystalline Al1 Tix phase from the Al/a-
−x
molecular H recombination for almost the entire release of H from
2
Al50Ti50 composite, so that H absorption rate and HD scrambling
the sample, from both the NaAlH and Na AlH phases. The forma-
4
3
6
rate are referenced to the same c-Al1 Tix phase. As the surface
−x
tion of crystalline Al1 Tix phases from the a-Al50Ti50 matrix is not
−x
embedded c-Al1 Tix phase does not change composition after ca.
−x
a crystallisation of the amorphous matrix itself, which is expected
◦
2 H cycles [5], HD scrambling can be performed on twice hydro-
gen cycled H empty samples, from room temperature upwards
to crystallise in a single phase state at ca. 660 C for Al50Ti50 [20].
Rather, the formation of ca. 4–25 nm Al85Ti15 crystallites observed
at moderate H /D2 pressures to obtain the H/D exchange inde-
2
in H cycled NaAlH + xTiCl [5] is consistent with local diffusion of Ti
4
3
pendent temperature dependent surface only HD scrambling rate.
Fig. 5 shows HD scrambling conducted on an isothermally des-
atoms from the a-Al50Ti50 matrix into the 2–20 nm pure Al crystal-
lites. The growth of c-Al1 Tix appears to be limited, with very close
−x
orbed PM NaAlH + 0.1TiCl3 sample, for a 1 bar mixture of H /D2
4
2
matching of the size ranges, Al 2–20 nm: Al1 Tix 4–25 nm, and
−x
at room temperature, conducted over 7 days. Classic equilibrium
H + D ↔ 2HD mixing is observed, demonstrating that c-Al Tix
similar dislocation density in the 2–20 nm Al and 4–25 nm Al1 Tix
5].
The annealing data from Fig. 4 also highlight another experi-
−x
2
2
1−x
[
(
x < 0.25) phases formed during H cycling perform a classic H2 dis-
sociation/recombination function.
mental difficulty for conducting HD scrambling experiments on
the TiCl3 enhanced NaAlH4 system. It is stated in [16] that “H/D
exchange is much faster than rehydrogenation of decomposed
alanate”. This statement alludes to the origins of the kinetic rate
4. Concluding remarks
limiting feature for hydrogen absorption in the NaAlH + xTiCl3
The nanoscopic Al/a-Al50Ti50 composite embedded on the sur-
face of single crystalline NaAlH4 powder grains allows the release
of H from NaAlH4 and subsequent molecular H2 formation at
ambient temperature over ca. 200 days. Isotopic H/D exchange
observed over 6 days for PM NaAlD + 0.04TiCl under 20 bar H2
at 50 C indicates that the surface embedded nano Al/a-Al50Ti50
composite performs a molecular H2 dissociation/recombination
function, and facilitates bulk exchange of D with H. After the D/H
exchange period, 58% of all unit cells are expanded by H to form
the composition NaAl(H0.56D0.44)4. The mosaic of ca. 100 nm for
4
system, and implicitly assumes that molecular H2 dissocia-
tion/recombination is not rate limiting, which appears consistent
with the rate study in [12] indicating diffusion of AlHx entities are
the rate limiting step. There exist several problems with the HD
4
3
◦
◦
interpretation from [16]: (i) the D /HD decay at 50 C and 20 bar
2
observed in [16] occurs due to the presence of an a-Al50Ti50 phase
on the NaAlH4 surface. Comparing the absorption rate to milled
data is difficult as after the first thermal desorption to remove H,
c-Al88Ti12 ↔ c-Al86Ti14 forms [5] (as in Fig. 4) from the nanoscopic
Al/Al50Ti50 composite, and the rate of HD scrambling may not be the
same on these crystalline surfaces as the a-Al50Ti50 surface; (ii) the
rate of HD scrambling is highly likely pressure/temperature (P/T)
NaAl(H0.56D0.44) indicates that long ranged diffusion of H/D is pos-
4
◦
sible within the NaAlD4 structure at 50 C and 20 bar. Isochronal
annealing of PM only NaAlH + 0.1TiCl3 shows that almost all H
4
and mol% TiCl dependent, and (iii) conducting HD scrambling at
is released before crystalline Al86Ti14 is observed, demonstrat-
ing that the nano Al/a-Al50Ti50 composite performs the H release
function during the first thermal desorption. A lack of hydrogen
3
◦
5
0 C and 20 bar are sufficient P/T conditions to hydride a H empty
NaAlH + xTiCl sample, and as such two processes (H/D exchange
4
3
and direct un-exchanged surface HD scrambling) are contributing
to the final HD scrambling rate. Such features demonstrate that to
release from NaAlH + 0.1Al (80 nm) over 1.5 years strongly sug-
gests that a-Al50Ti50 is the phase that is functionally responsible
4

M.P. Pitt et al. / Journal of Alloys and Compounds 514 (2012) 163–169
169
for H release during the very first thermal desorption of milled
Swiss–Norwegian Beam Line for providing experimental assistance
and logistics during long-term attachments. The authors acknowl-
edge Prof. C.M. Jensen for discussions on the nature of HD isotope
scrambling in the TiCl3 enhanced NaAlH4 system, and facilitation
of mass spectroscopy measurements.
NaAlH + xTiCl . Ambient HD scrambling on post thermal desorbed
4
3
PM NaAlH + 0.1TiCl3 shows near classic 1:2:1 equilibrium mixing
4
according to H + D ↔ 2HD, indicating that both a-Al50Ti50 and c-
2
2
Al Ti perform a classic molecular H dissociation/recombination
86
14
2
function that unadulterated NaAlH4 cannot. The Al
Tix phases
1−x
also provide a thermally stable sink for Ti, which subsequently
provides a stable monolayer of Al1 Tix phase at the near surface
References
−x
[
[
1] B. Bogdanovic, M. Felderhoff, A. Pommerin, F. Schuth, N. Spielkamp, A. Stark, J.
Alloys Compd. 471 (2009) 383–386.
2] M.P. Pitt, M. Paskevicius, C.J. Webb, M.H. Sorby, S. Delleda, T.R. Jensen, B.C.
Hauback, C.E. Buckley, E.Mac A. Gray, Int. J. Hydrogen Energy 36 (2011)
NaAlH /Al Tix interface.
4
1−x
Our study serves as a model for TiCl3 addition to NaAlH4 from
points of view: (i) the addition of TiCl3 to NaAlH4 results in
2
8
403–8411.
Ti containing phases embedded on the NaAlH4 surface that are
nanoscopic in dimension, and that are efficient at molecular dis-
[
3] L.H. Rude, T.K. Nielsen, D.B. Ravnsbaek, U. Bosenberg, M.B. Ley, B. Richter, L.M.
Arnbjerg, M. Dornheim, Y. Filinchuk, F. Besenbacher, T.R. Jensen, Phys. Status
Solidi A 208 (2011) 1754–1773.
sociation/recombination of H , and (ii) the dispersion of these
2
nanoscopic Ti containing phases on the NaAlH surface implies that
[4] D.B. Ravnsbaek, Y. Filinchuk, R. Cerny, T.R. Jensen, Z. Kristallogr. 225 (2010)
57–569.
4
5
the interfacial area between Al1 Tix phases and NaAlH4 is only
−x
[
5] M.P. Pitt, P.E. Vullum, M.H. Sørby, M.P. Sulic, C.M. Jensen, J.C. Walmsley, R.
Holmestad, B.C. Hauback, Acta Mater. 56 (2008) 4691–4701.
small, but greatly efficient at releasing hydrogen from the NaAlH4
bulk, even at great distances from the surface where Ti is embedded.
The surface location of Ti also highlights the importance of reducing
the size of the NaAlH4 bulk to reduce bulk limiting AlH4 diffusion
[6] Y. Suttisawat, V. Jannatisin, P. Rangsunvigit, B. Kitiyanan, N. Muangsin, S. Kul-
prathipanja, J. Power Sources 163 (2007) 997–1002.
[
7] C. Weidenthaler, A. Pommerin, M. Felderhoff, B. Bogdanovic, F. Schuth, Phys.
Chem. Chem. Phys. 5 (2003) 5149–5153.
distances. Understanding of the near surface Al1 Tix/NaAlH inter-
face in promoting rate limited hydrogen release from the NaAlH4
bulk is of utmost importance. The surface embedding of well dis-
[8] M.P. Pitt, P.E. Vullum, M.H. Sørby, D. Blanchard, M.P. Sulic, H. Emerich, M. Paske-
vicius, C.E. Buckley, J. Walmsley, R. Holmestad, B.C. Hauback, J. Alloys Compd.
−x
4
(
2011), doi:10.1016/j.jallcom.2011.11.021.
[
9] J.P. Bastide, J.El. Hajri, P. Claudy, A.El. Hajbi, Synth. React. Inorg. Met. Org. Chem.
25 (1995) 1037–1047.
tributed and isolated nanoscopic Al1 Tix phases suggests that
−x
[
[
10] B.C. Hauback, H. Fjellvåg, O. Steinsvoll, K. Johansson, O.T. Buset, J. Jørgensen, J.
Neutron Res. 8 (2000) 215–232.
11] B.A. Hunter, Comm. Powder Diffr. Newslett. 20 (1998) 21.
[12] W. Lohstroh, M. Fichtner, Phys. Rev. B 75 (2007) 184106.
[13] G. Streukens, B. Bogdanovic, M. Felderhoff, F. Schuth, Phys. Chem. Chem. Phys.
AlH4 tetrahedra can only be perturbed in very limited numbers
locally at the near surface Al1 Tix/NaAlH4 interface, initiating a
−x
hydrogen release reaction which propagates into the NaAlH bulk.
4
−
Local charge imbalance of AlH4 units and Ti coordination in the
8
(2006) 2889–2892.
Al1 Tix/NaAlH interface is crucial to understanding how this reac-
−x
4
[
14] K. Gross, W. Luo, E. Majzoub, G. Roberts, S. Spangler, D. Dedrick D, T. Johnson,
J. Chan, DOE FY2003, Progress Report, 2003.
[15] B. Bogdonavic, R.A. Brand, A. Marjanovic, M. Schwickardi, J. Tolle, J. Alloys
Compd. 302 (2000) 36–58.
16] J.M. Bellosta von Colbe, W. Schmidt, M. Felderhoff, B. Bogdonavic, F. Schuth,
Angew. Chem. Int. Ed. 45 (2006) 3663–3665.
[17] K. Hashi, K. Ishikawa, K. Suzuki, K. Aoki, Mater. Trans. 43 (2002) 2734–2740.
tion propagates from the surface into the bulk. High resolution TEM
is the obvious tool to provide direct space images of this interface,
however, electron beam induced evaporation of Na and H which
destroys the interface precludes such an observation [21]. Density
functional theory models of Ti substitution in the NaAlH4 surface
[
[
[
[
[
[
18] S. Gomes, G. Renaudin, H. Hagemann, K. Yvon, M.P. Sulic, C.M. Jensen, J. Alloys
(
slab models) may be relevant in this regard [22], and provide at
Compd. 390 (2005) 305–313.
19] G. Majer, E. Stanik, L.E. Valiente Banuet, F. Grinberg, O. Kircher, M. Fichtner, J.
Alloys Compd. 404–406 (2005) 738–742.
20] Z. Jun-Hong, H. Bai-yun, H. Yue-hui, Z. Ke-chao, L. yong, Trans. Nonferrous Met.
Soc. China 12 (2002) 610–614.
21] P.E. Vullum, M.P. Pitt, J.C. Walmsley, B.C. Hauback, R. Homestad, J. Alloys Compd.
least a simplistic picture of how Ti atoms may be locally coordinated
by H in the Al1 Tix/NaAlH4 interface.
−x
Acknowledgements
509 (2011) 281–289.
22] O.M. Løvvik, S.M. Opalka, Phys. Rev. B 71 (2005) 054103.
This project has been funded by the Synchrotron Program of
the Research Council of Norway. MPP thanks the staff of the