Probing the structure, stability and hydrogen storage properties of calcium dodecahydro-closo-dodecaborate

Article abstract of DOI:10.1016/j.jssc.2010.03.026

Calcium borohydride can reversibly store up to 9.6 wt% hydrogen; however, the material displays poor cyclability, generally associated with the formation of stable intermediate species. In an effort to understand the role of such intermediates on the hydrogen storage properties of Ca(BH₄)₂, calcium dodecahydro-closo-dodecaborate was isolated and characterized by diffraction and spectroscopic techniques. The crystal structure of CaB₁₂H₁₂ was determined from powder XRD data and confirmed by DFT and neutron vibrational spectroscopy studies. Attempts to dehydrogenate/hydrogenate mixtures of CaB₁₂H₁₂ and CaH₂ were made under conditions known to favor partial reversibility in calcium borohydride. However, up to 670 K no notable formation of Ca(BH₄)₂ (during hydrogenation) or CaB₆ (during dehydrogenation) occurred. It was demonstrated that the stability of CaB₁₂H₁₂ can be significantly altered using CaH₂ as a destabilizing agent to favor the hydrogen release.

Full text of DOI:10.1016/j.jssc.2010.03.026

ARTICLE IN PRESS
Journal of Solid State Chemistry 183 (2010) 1133–1140
Contents lists available at ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Probing the structure, stability and hydrogen storage properties of calcium
dodecahydro-closo-dodecaborate
Ã
Vitalie Stavila a, , Jae-Hyuk Her c,d, Wei Zhou c,d, Son-Jong Hwang e, Chul Kim e,1
,
Leigh Anna M. Ottley b, Terrence J. Udovic c
a Sandia National Laboratories, Livermore, CA 94551, United States
b Sandia National Laboratories, Albuquerque, NM 87106, United States
c NIST Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, MD 20899, United States
d Department of Materials Science and Engineering, University of Maryland, College Park, MD 20742, United States
e Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Calcium borohydride can reversibly store up to 9.6 wt% hydrogen; however, the material displays poor
cyclability, generally associated with the formation of stable intermediate species. In an effort to
understand the role of such intermediates on the hydrogen storage properties of Ca(BH4)2, calcium
dodecahydro-closo-dodecaborate was isolated and characterized by diffraction and spectroscopic
techniques. The crystal structure of CaB12H12 was determined from powder XRD data and confirmed by
DFT and neutron vibrational spectroscopy studies. Attempts to dehydrogenate/hydrogenate mixtures of
CaB12H12 and CaH2 were made under conditions known to favor partial reversibility in calcium
borohydride. However, up to 670 K no notable formation of Ca(BH4)2 (during hydrogenation) or CaB6
(during dehydrogenation) occurred. It was demonstrated that the stability of CaB12H12 can be
significantly altered using CaH2 as a destabilizing agent to favor the hydrogen release.
Published by Elsevier Inc.
Received 27 October 2009
Received in revised form
16 March 2010
Accepted 16 March 2010
Available online 20 March 2010
Keywords:
Calcium dodedecahydro-closo-
dodecaborate
Hydrogen storage
Metal borohydrides
X-ray and neutron techniques
Density functional theory
1. Introduction
Recently Ozolins et al. [12] used theoretical first-principle
methods to predict the enthalpies for the following two
decomposition pathways for Ca(BH4)2:
Metal borohydrides have attracted increasing interest as
promising materials with significant potential for hydrogen
storage applications [1–4]. However, there are a number of
factors that limit their utility, including high dehydrogenation
temperatures and slow kinetics. Vajo et al. [5] proposed a
destabilization approach to modify the thermodynamic properties
of LiBH4 using MgH2. A number of other additives including CaH2,
ScH2, TiH2, and various metals were found to display similar
destabilizing effects [6,7]. While there has been some progress in
decreasing the dehydrogenation temperatures of borohydrides by
destabilization, the problem of poor cycling characteristics
Ca(BH4)2-(2/3)CaH2+(1/3)CaB6+(10/3)H2
Ca(BH4)2-(5/6)CaH2+(1/6)CaB12H12+(13/6)H2
(I)
(II)
The two reactions turn out to have almost identical enthalpies,
E40 kJ/mol H2. A slightly lower value of 32 kJ/mol H2 for the
enthalpy of reaction (I) was reported from calculations by Miwa
et al. [13]. While the thermodynamic characteristics of the two
reactions seem to be similar, the amounts of hydrogen released
differ significantly in (I) and (II) and correspond to 9.6% and 6.3%
mass fraction of hydrogen. Using the prototype electrostatic
ground state (PEGS) method, the authors predicted the crystal
structures of CaB12H12, MgB12H12, and Li2B12H12 [12]. The crystal
structures are critical for the evaluation of the thermodynamic
characteristics of the reactions involving these species [12,14].
Unfortunately, no experimental data were available in the
literature concerning the Ca and Mg salts to confirm the predicted
structures.
remains largely unsolved. There have been reports indicating
2ꢀ
that [B12H12
]
species are formed as intermediates during the
thermal decomposition of metal borohydrides [8–11].
Ã
Corresponding author at: Sandia National Laboratories, Livermore, CA 94551,
United States. Fax: +925 294 3231.
E-mail address: vnstavi@sandia.gov (V. Stavila).
1
The reverse reaction of (I) has been shown to occur in
moderate yield at 713 K and 70 MPa when CaH2 and CaB6 reacted
Current address: Department of Chemistry, Hannam University, Daejeon,
South Korea.
0022-4596/$ - see front matter Published by Elsevier Inc.
doi:10.1016/j.jssc.2010.03.026

ARTICLE IN PRESS
1134
V. Stavila et al. / Journal of Solid State Chemistry 183 (2010) 1133–1140
in the presence of TiCl3/Pd [15]. Aoki et al. [16] reported that the
plateau pressure for Ca(BH4)2 at 593 K is around 0.6 MPa and
the amount of H2 desorbed is 5.9 wt%. It has to be noted that the
amount of H2 released in reaction (II) can be achieved at E653 K
in pure Ca(BH4)2 [17] and at slightly lower temperatures when
various additives are used [18]. At higher temperatures, Kim et al.
[19] found that about 8.3% mass fraction of H2 is released in the
presence of NbF5 and suggested that CaB6 and NbB2 might be
among the decomposition products. Based on in situ XRD analysis,
the hydrogen release reaction was described as follows [17]:
published procedures [22,23,25]. The reported infrared data were
recorded on a Varian 800 FT-IR spectrometer using attenuated
total reflectance. Elemental analyses were done at Galbraith
Laboratories, Inc. TGA and DSC measurements were performed on
Mettler Toledo instruments. SEM and energy dispersive X-ray
spectroscopy (EDX) experiments were performed on
a FEI
scanning electron microscope. Solid-state MAS NMR spectra were
measured using a Bruker Avance 500 MHz spectrometer with an
11.7 T magnet and employing a boron-free Bruker 4 mm CPMAS
probe. Neutron vibrational spectra were measured using the BT-4
filter analyzer neutron spectrometer [26] at the NIST Center for
Neutron Research. First-principles calculations were performed
within the plane-wave implementation of the generalized
gradient approximation to DFT using a Vanderbilt-type ultrasoft
potential with Perdew–Burke–Ernzerhof exchange correlation
(Quantum-ESPRESSO package [27]).
Ca(BH4)2-CaH2+2B+3H2
(III)
Recently, Riktor et al. [20] reported that upon heating Ca(BH4)2
to 643–673 K under dynamic vacuum an intermediate phase
CaB2Hx (x most probably equals 2) can be isolated. The decom-
position process can be represented in this case by a hypothetical
reaction as follows:
2.2. Synthesis of [Ca(H2O)5(MeCN)2][B12H12], 3
Ca(BH4)2-Ca(B2H2)+3H2
(IV)
Cesium dodecahydro-closo-dodecaborate, Cs2B12H12 (932 mg,
2.0 mmol) was dissolved upon heating in 60 mL H2O and the
warm solution was passed through an Amberlite IR-120 ion-
exchange column in H+ form. The acidic fraction represents an
aqueous solution of (H3O)2[B12H12]. The solution was reacted
with aqueous Ca(OH)2 at room temperature until pH¼7 was
reached. Most of the solvent was removed on a rotary evaporator,
and then 20 mL acetonitrile was added. In 2 weeks, crystalline
compound 3 was formed in ꢂ30% yield. Anal., Found % (Calcd. %)
for C4H28B12CaN2O5: C, 13.25 (13.57); H, 8.20 (7.97); N, 7.64
(7.91). ATR (cmꢀ1): 3590, 3535, 3508, 3355, 3224, 2478, 2302,
2269, 1615, 1407, 1369, 1204, 1069, 928, 713, 669.
The authors suggested that the CaB2Hx intermediate phase can
serve as a precursor for CaB12H12 [20]. Since the [B12H12
2ꢀ
]
salts
are known to display remarkable thermal and chemical stability
[21], it remains unclear whether CaH2 and CaB12H12 can be
converted into Ca(BH4)2 (reverse reaction of (II)). It also remains
to be determined whether reverse reactions (III) and/or (IV) are
feasible. The formation of a single solid product (reaction (IV))
may be advantageous in order to avoid phase separation between
Ca and B species (reactions (I)–(III)). Despite the appearance of
several recent reports on the hydrogen storage properties of
Ca(BH4)2, the mechanism by which H2 is released is still
uncertain, and some controversy surrounds the actual products
formed during the thermal decomposition, which depends on the
temperature regime and additives used.
2.3. Synthesis of CaB12H12, 1
Possible occurrence of polyborane clusters among the decom-
position products, as well as the uncertainty about their
rehydriding behavior, has garnered great interest in the scientific
community. For that reason we have chosen to explore the
synthesis of CaB12H12 (1) and study its hydrogen storage proper-
ties. Two precursor complexes, [Ca(H2O)7][B12H12] ꢁ H2O [22,23]
(2) and [Ca(H2O)5(MeCN)2][B12H12] (3), have been employed to
produce crystalline 1 upon dehydration/desolvation. A Rietveld
refinement showed that the experimental structure of 1 is largely
in accord with the PEGS predicted low-energy monoclinic C2/c
model [12] with similar unit cell parameters, but a different
unique b-axis. Our density functional theory (DFT) calculations
confirm the experimentally refined C2/c structure and correlate
well with the phonon density of states (DOS) measured via
neutron vibrational spectroscopy (NVS). Hydrogen sorption and
desorption studies on mixtures of CaB12H12 and CaH2 were
performed to understand the different rationales for the dehy-
drogenation/hydrogenation reactions and the role of the inter-
mediates on the cycling characteristics of Ca(BH4)2.
The anhydrous CaB12H12 salt was obtained by removal of
solvent molecules from
2 or 3 upon heating. In a typical
experiment, 750 mg of 3 were loaded into a Schlenk tube and
heated in vacuum as follows: (i) at 398 K for 1 h, then (ii) at 498 K
for 2 h, and finally (iii) at 573 K for 4 h. The isolated yield was
ꢂ85% due to some loss of the fine powder upon heating in
vacuum. Anal., Found % (Calcd. %) for CaB12H12: Ca, 21.1 (22.03);
B, 70.2 (71.32); H, 6.79 (6.65). ATR (cmꢀ1): 3683, 3522, 2304,
2473, 1607, 1112, 1081, 753, 730.
2.4. Hydrogenation and dehydrogenation experiments
The hydrogenation attempts were made at 670 K and 100 MPa
H2 pressure. CaB12H12 and CaH2 were ground together in a
mortar, then placed in a 25-mL hardened steel grinding vial
containing two tungsten-carbide balls and milled on a SPEX 8000
high-energy mill for 1 h. The powders were pressed into thin
pellets and subjected to 100 MPa pressure H2 at 670 K for 72 h.
The temperature programmed desorption (TPD) experiments
were performed on a milled 1:1 CaB12H12:CaH2 sample. After
mechanical milling, the samples were transferred into a stainless
steel sample holder, which was subsequently attached to a
Sieverts apparatus. Pressure changes were monitored with
calibrated pressure transducers and recorded by a LabView-based
software program.
2. Experimental
2.1. Materials, equipment, and characterization details
Preparations and handling of air-sensitive materials were
carried out under a dry Ar atmosphere using standard Schlenk
and glovebox techniques. All of the chemicals were of reagent
grade and were obtained from Sigma-Aldrich or Strem [24].
Compound 2 and anhydrous MB12H12 (M¼Mg, Ca, and Ba) and
M2B12H12 (M¼Li, Na, K, and Cs) salts were prepared by previously
2.5. Powder X-ray diffraction
X-ray diffraction (XRD) data were obtained at room tempera-
ture with
a Rigaku Ultima III sealed-tube (1.6 kW, CuKa)
diffractometer. The TOPAS-Academic program was used to index

ARTICLE IN PRESS
V. Stavila et al. / Journal of Solid State Chemistry 183 (2010) 1133–1140
1135
11 B MAS NMR
the pattern and to solve the structure. Yet the first attempt failed,
ω r =13-15 kHz
probably due to the non-optimal quality powder XRD data. The
successful initial structure solution was inspired by the theore-
tically predicted structure model of Ozolins et al. [12]. Comparing
unit-cell parameters, we noticed that the unique axis of the
M(B12H12 n
)
monoclinic lattice was incorrectly assigned, since the
very close to 901. After changing the assumed unique axis, a new
search was performed with molecular (rigid-body) [B12H12
b angle was
Ba
Ca
Mg
K
2ꢀ
]
and Ca2+ in direct space. The simulated annealing methods
successfully converged and subsequent structure refinement was
done with the Rietveld method. Due to the intrinsic deficiency of
information in the powder data and the insensitivity of X-rays for
H atoms, the rigid body constraints were kept in the final
refinement step, but B–B and B–H distances were allowed to vary
while the icosahedral symmetry was preserved. All displacement
parameters were treated isotropically, and all B and H atoms were
constrained to have the same values.
Na
Li
40
30
20
10
0
-10
-20
-30
-40
-50
-60
ppm
2.6. Single-crystal X-ray structure information
Fig. 1. Center band of 11B MAS NMR spectra of Mn[B12H12] (M¼Li, Na, K, Mg, Ca,
Ba; n¼1, 2) salts. The samples were dehydrated by heating under vacuum except
the Mg-compound.
Single-crystals of 3 suitable for X-ray crystallography were
separated as small needles from concentrated aqueous solutions
of 1 in the presence of MeCN. Crystals were mounted onto a glass
fiber from a pool of FluorolubeTM and immediately placed in a
cold N2 vapor stream at 173(2) K, on a Bruker AXS diffractometer
equipped with a SMART 1000 CCD detector using graphite
corresponds to water protons (Fig. S3). A comparison of the 11B
MAS NMR powder spectra for various MnB12H12 compounds
(M¼Li, K, Na, Mg, Ca, Ba; n¼1, 2) is shown in Fig. 1. While the
center peak position at ꢂ15.5 ppm is similar for all the alkali
metals, slight down field shift to ꢂ14.5 ppm is seen for the series
of alkaline-earth metals in the current study. The quadrupole
coupling of the 11B nucleus also turned out to vary slightly. The
fitting of MAS NMR spectra yields quadrupole coupling constants
monochromatized MoK
a radiation. Lattice determination and
data collection were carried out using SMART Version 5.054
software. Data reduction was performed using SAINTPLUS Version
6.01 software and corrected for absorption using the SADABS
program within the SAINT software package. The structure was
solved by direct methods that yielded the heavy atoms, along
with a number of the lighter atoms. The hydrogen atoms were
fixed in positions of ideal geometry and refined using SHELXS
software. The final refinement of each compound included
anisotropic thermal parameters for all non-hydrogen atoms.
CQ and ZQ and shift anisotropy parameters (CSA), which reflect the
electronic environment around the boron nuclei. The resulting
parameters are compiled in Table 1. The calcium salt revealed a
smaller quadrupole coupling constant for 11B compared to other
cations (ꢂ700 kHz) [29], indicating that the icosahedra
experience less of field gradient with Ca2+ than with alkali
metals (see list in Table 1 of Ref. [29]).
3. Results and discussion
Tiritiris and Schleid [22] reported
a thermogravimetric
analysis of 2, which showed several dehydration steps with the
formation of anhydrous 1 at 497 K. The TGA/DSC analysis of 3
indicates loss of both water and acetonitrile molecules to generate
anhydrous 1 at 563 K. The remaining material retains a constant
weight between 563 and 693 K. Thereafter, it gradually loses
weight, which is attributed to the decomposition of the
anhydrous product. The obtained residue at 870 K shows no
diffraction in the XRD pattern. The exact nature of the product is
not clear, but it shows some residual B–H stretches in the IR
spectrum. Solid state NMR characterization of anhydrous 1 after
heat treatments at 673 and 723 K under vacuum for 4 h revealed
significant changes in 1H MAS NMR signal intensity (100%-93%
at 673 K and 100%-75% at 723 K), which may suggest the loss of
one or several hydrogen atoms per icosahedral cage. 11B MAS
(Fig. 2a) and 1H NMR (Fig. 2b) spectra show no significant change
in the center band of anhydrous 1 heated at 673 K. Noticeable
broadening of the center band peak was observed when the
temperature was raised to 723 K. The overall quadrupole powder
pattern of heated samples showed a gradual change from the
unheated anhydrous 1 to lose the characteristic line shapes of
11B nuclei in the icosahedron, an indication of formation of a
disordered state as confirmed by the amorphous nature of the
product. While additional investigation is needed to elucidate
the structural changes upon heating, we believe that the NMR
results may reflect the loss of hydrogen upon heating over 673 K
to yield a disordered state as a result of dimer or oligomer
formation.
Aqueous solutions of CaB12H12 can be prepared easily by
reaction of (H3O)2[B12H12] with CaCO3 [22] or Ca(OH)2 (this
work). The corresponding octahydrate 2 previously reported in
the literature was isolated from concentrated aqueous solutions
[22,23], while in the presence of acetonitrile, compound 3 was
isolated. Completely anhydrous 1 was obtained by heating the
solvated compounds 2 or 3 in vacuum up to 573 K. The crystal-
linity of the anhydrous product was found to depend on the
heating rate; slower heating generally resulted in a more crystal-
line material. The most prominent features of the FTIR vibrational
2ꢀ
spectra of the [B12H12
]
groups in 1–3 are the n(B–H) peaks in
the region 2500–2470 cmꢀ1 (singlets for 2 and 3 and doublets
for 1), (B–B) near
(B–H) peaks around 1080 cmꢀ1 and
715–730 cmꢀ1 (Fig. S1, Supplementary material). Complex 3
displays two additional weak bands at 2269 and 2302 cmꢀ1
d
n
,
which can be assigned, respectively, to the stretching CRN
vibration and a combination mode of CH3 deformation and C–C
stretching vibrations in the acetonitrile molecule. These bands are
shifted to higher values by 15 and 9 cmꢀ1 from the corresponding
peaks at 2254 and 2293 cmꢀ1 in pure MeCN.
1H and 11B MAS NMR of 1 as well as 11B solution NMR of the
aqueous solutions prepared from 1 contain characteristic features
2ꢀ
for [B12H12
]
anions (Fig. S2). The solution NMR of 1 confirms
2ꢀ
the presence of the [B12H12
]
units with JBꢀH¼125 Hz [28]. The
hydration is shown to narrow both 1H and 11B NMR signals
dramatically; the 4.5 ppm peak in the 1H NMR spectrum

ARTICLE IN PRESS
1136
V. Stavila et al. / Journal of Solid State Chemistry 183 (2010) 1133–1140
Table 1
NMR parameters of 11B in the CaB12H12 sample compared to literature data [8,29].
Sample
diso (ppm)
dcsa (ppm)
gcsa
CQ (kHz)
gQ
Lorentzian
Gaussian
broadening (Hz)
broadening (Hz)
CaB12H12 (present work)
K2B12H12 [8]
ꢀ14.4970.03
ꢀ15.0
ꢀ133711
0.270.1
0.15
–
600
713
686
0.2570.11
7317238
22307346
92
–
0.2
–
1200
–
1204
–
K2B12H12 [29]
ꢀ15.5
Table 2
11 B MAS NMR
Fractional coordinates of
relaxation results.
1 (monoclinic, C2/c) from powder XRD and DFT
Atoms
x (XRD)
y (XRD)
0.832(1)
z (XRD)
x (DFT)
y (DFT)
z (DFT)
ω r = 15 kHz
Ca
0
0.25
0
0.8297 0.25
B1
B2
B3
B4
B5
B6
H1
H2
H3
H4
H5
H6
ꢀ0.1721
ꢀ0.1721
0.0677
0.0677
0.0032
0.2159
ꢀ0.2966
ꢀ0.2966
0.1167
0.1167
0.0055
0.3721
0.2518
0.0992
0.2989
0.0520
0.1755
0.1755
0.3070
0.0440
0.3882
ꢀ0.0372
0.1755
0.1755
0.3242
0.3242
0.3216
0.3216
0.4116
0.3199
0.3779
0.3779
0.3733
0.3733
0.5286
0.3705
ꢀ0.156
ꢀ0.1586
0.083
0.234
0.3336
0.3336
0.0924
0.2785 0.3107
0.0494 0.3118
0.0821
0.0356
0.2293
ꢀ0.2663
ꢀ0.271
0.1441
0.1464
0.0645
0.3933
0.1635 0.4022
0.1635 0.2978
0.2848 0.3961
0.0429
0.3602
0.3937
0.3537
0.3589
ꢀ0.0305
0.1634 0.511
0.1626 0.3344
The lattice parameters were optimized during the DFT relaxation. The B and H
positions were not freely refined. B–B and B–H distances were refined with
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70
ppm
˚
constrains to 1.826(8) and 1.26(4) A, respectively.
1H MAS NMR
ωr =15 kHz
20
15
10
5
0
-5
-10
ppm
Fig. 2. (a) Center band of 11B MAS NMR spectra for anhydrous 1 (black), heated at
400 1C (blue), and heated at 450 1C (green). (b) Center band of 1H MAS NMR spectra
for the same samples as in (a). (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
Fig. 3. XRD pattern of anhydrous, polycrystalline CaB12H12 (1) and the Rietveld
refinement result. Black: raw data points. Red: calculated profile. Blue: difference
curve (observedꢀcalculated) on the same scale. Vertical tick marks: allowed Bragg
reflection positions. After 2hE201, the additional pattern and difference curve are
both presented with 3 ꢃ magnification. The arrow at ꢂ151 denotes the position of
an additional Lorentzian impurity line attributed to the strongest reflection of the
.
trihydrated phase, CaB12H12 3H2O, which was included in the refinement process.
Anhydrous 1 is soluble in water and can be recrystallized as an
octahydrate from aqueous solutions. The products obtained via the
dehydration/desolvation of 2 and 3 display identical IR spectra.
Powder XRD studies demonstrated that the two products are
virtually identical and represent anhydrous 1. Compound 1 was
(For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)
equatorial plane centered on each Ca2+ cation, while the other two
2ꢀ
anions are located in axial positions. Each Ca2+ cation
˚
found to crystallize in a monoclinic unit cell (C2/c, a¼7.242(1) A,
[B12H12
]
3
2ꢀ
˚
˚
˚
b¼11.971(3) A, c¼10.744(2) A,
b
¼89.82(3)1, V¼931.5(3) A , Z¼4;
interacts with eight neighboring H atoms from the five [B12H12]
(see Table 2 and Table S1 in Supplementary material). The structure
is similar to the theoretical model predicted by Ozolins et al. [12];
however, the unique b-axis turned out to be different. The crystal
structure displays Ca2+ cations surrounded by five icosahedral
anions. Two H atoms are from each of the three equatorial
2ꢀ
[B12H12
[B12H12
]
anions and one H atom is from each of the two axial
anions. The experimental, fitted, and difference profiles
2ꢀ
]
of the XRD patterns are shown in Fig. 3. The structure of 1 and its
packing scheme are presented in Figs. 4 and 5. Using DFT, structural
2ꢀ
2ꢀ
[B12H12
]
anions. Three [B12H12
]
anions form the triangular

ARTICLE IN PRESS
V. Stavila et al. / Journal of Solid State Chemistry 183 (2010) 1133–1140
1137
optimizations were performed with respect to atomic positions and
lattice parameters. The room-temperature structure from XRD data
was numerically relaxed at 0 K. The optimized structure was close
to that experimentally observed (see Table 2), suggesting that
there is no structural phase transition between 0 K and room
temperature.
that for B or Ca, spectral features are dominated by the various
internal vibrational modes of the dodecahydro-closo-dodecabo-
rate anions involving hydrogen displacements. For comparison,
the phonon density of states (DOS) was calculated from the DFT-
optimized structure using the supercell method (2 ꢃ 2 ꢃ 1 cell
size) with finite displacements [30,31] and was appropriately
weighted to take into account the H, B, and Ca total neutron
scattering cross sections. The resulting weighted DOS (broadened
The neutron vibrational spectrum for 1 was measured at 4 K.
Since the total scattering cross section for H is much larger than
Fig. 6. NV spectrum (black, vertically offset for clarity) of 1 at 4 K compared with
the calculated (weighted) phonon density of states (red) of the C2/c structure.
Horizontal bars represent the full-width-at-half-maximum instrumental resolu-
tion and vertical bars show71r
(N.B.: 1 meVE8.066 cmꢀ1). (For interpretation of
Fig. 4. Crystal structure of 1 in the monoclinic (C2/c) unit cell viewed along the
c-axis (perspective).
the references to color in this figure legend, the reader is referred to the web
version of this article.)
Fig. 5. The packing scheme in 1. Yellow: Ca, green: B, gray: H, partially transparent red triangle: equatorial plane consisting of three counter anions. Blue dashed lines:
˚
˚
˚
˚
Ca?H distances of 2.04(3) A, 2.22(2) A, 2.38(4) A, and 2.46(2) A. (For interpretation of the references to color in this figure legend, the reader is referred to the web version
of this article.)

ARTICLE IN PRESS
1138
V. Stavila et al. / Journal of Solid State Chemistry 183 (2010) 1133–1140
by the known instrumental resolution) is in good agreement with
the experimental NVS results (see Fig. 6), providing additional
corroboration for the refined CaB12H12 structure. Previous reports
CaB12H12 form, namely trihydrated CaB12H12 ꢁ 3H2O. The structure
in Fig. 7 displays the Ca2+ cation coordinated by three H2O
2ꢀ
molecules equatorially and two tridentate [B12H12
]
anions
suggest that NVS measurements and phonon DOS calculations
occupying the axial positions in the coordination sphere. The
2ꢀ
2ꢀ
are indeed sensitive to the structural details of the [B12H12
]
cationic Ca(H2O)3 units and anionic [B12H12
]
groups form a
lattice arrangements, as exemplified for Cs2B12H12 [32] and
Li2B12H12 [33].
one-dimensional chain structure along the c-axis of the unit cell.
Interestingly enough, in 1, each Ca+2 cation displays bi-
tetrahedral surroundings that are reminiscent of the
coordination scheme for trihydrated CaB12H12 ꢁ 3H2O, in which
The crystal structures of several anhydrous metal salts of
2ꢀ
[B12H12
]
have been reported to date. M2B12H12 salts (M¼K, Rb,
Cs and Tl) crystallize in a cubic lattice with M+ cations in the
tetrahedral interstices of the ccp-arranged icosahedral [B12H12
the three water molecules are replaced by three bi-dentate
2ꢀ
2ꢀ
]
[B12H12
]
anions.
anions. The primitive cubic crystal structure of Li2B12H12 contains
Compound 3 was found to crystallize in the monoclinic space
group P21/c (see additional structural data in Tables S1, S3, and
S4; Supplementary material). The structural parameters of 2 and 3
are quite similar except that in 3, two water molecules are
Li+ cations in a strongly distorted octahedral environment formed
2ꢀ
by six hydrogen atoms of three [B12H12
]
groups [33]. The
coordination environment of the Na+ cations in the recently
reported monoclinic P21/n structure of Na2B12H12 can be viewed
as either an octahedron or a cube [34]. Relatively short alkali
metal–hydrogen distances are found; for example, Li?H dis-
˚
tances in Li2B12H12 measure 2.08 and 2.22 A [33], while the
corresponding M?H distances in Na2B12H12 [34] and K2B12H12
˚
[25] are 2.22–2.86 and 2.94 A, respectively. The Ca?H distances
˚
in 1 are 2.04(3), 2.22(2) and 2.46(2) A for equatorial interactions
˚
and 2.38(4) A for axial interactions. Such distances suggest
significant alkali-metal?H interactions, which seem to be a
common feature for anhydrous dodecahydro-closo-dodecaborate
2ꢀ
salts. [B12H12
]
is one of the least coordinating anions and in
most of the compounds is not part of the first coordination sphere
of the metal cation [35,36]. There are very few exceptions, usually
involving transition metals in low oxidation state, as in [(PhMe2-
P)3RuB12H12] [37] or (Bu4N)2{[1,2-C6F4Hg)3]2[B12H12]} [38].
Compound 1 can be stored in a dry atmosphere; in the
presence of moisture it tends to absorb water. Thus, about a
month later after the initial powder XRD data collection, we
revisited the vacuum-grease-sealed capillary specimen of 1 and
found that the XRD pattern was completely changed. We
attributed this to an imperfect seal allowing a slow uptake of
water into the hygroscopic dodecahydro-closo-dodecaborate salt.
The new powder pattern (Fig. S4) was auto-indexed with a
˚
˚
rhombohedral unit cell (R3c, a¼11.231(3) A, c¼16.590(4) A in
3
˚
hexagonal setting, V¼1812(1) A , Z¼6; see Tables S1 and S2,
Fig. 8. A view of the crystal structure of 3 along the b-axis.
Supplementary material) and revealed as partially hydrated
Fig. 7. (Left) CaB12H12 ꢁ 3H2O structure in the hexagonal (R3c) unit cell viewed along the c-axis (perspective). (Right) The 1D packing scheme of CaB12H12 ꢁ 3H2O. Yellow:
Ca2+, green: B, gray: H, red: O2ꢀ, blue dashed lines: Ca–H bond distances of 2.59(1) A. The Ca cation is surrounded by three equatorial H2O molecules (Ca–O: 2.333(7) A)
2+
˚
˚
2ꢀ
and two axial tridentate [B12H12
]
anions. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

ARTICLE IN PRESS
V. Stavila et al. / Journal of Solid State Chemistry 183 (2010) 1133–1140
1139
substituted with two acetonitrile molecules and the complex has
no crystallization water molecules. Fig. 8 shows the unit cell
structure of 3. The compound contains two Ca2+ and two
2ꢀ
[B12H12
]
ions in the asymmetric unit. The B–B bond distances
and angles in 3 are consistent with the corresponding values
reported for the icosahedral clusters in the octahydrate [22]. The
7-fold coordination of the Ca atom (Fig. S5, Supplementary
Material) in
3 is defined by five Ca–O bonds with bond
˚
distances between 2.339(1) and 2.465(1) A and two Ca–N bonds,
˚
2.518(1) and 2.550(1) A. The mean Ca–O distance, 2.399(1) is only
slightly shorter than the sum of ionic radii for Ca2+ and O2ꢀ
,
˚
2.44 A [39].
We are currently investigating the hydrogen absorption
properties of CaB12H12 ball-milled with various metal hydrides.
Ball-milling is known to facilitate both hydrogen release and
absorption in metal hydrides [40–43]. Our rehydrogenation
attempts indicate that the CaB12H12 sample ball-milled with 5
equiv. of CaH2 does not form any noticeable amounts of calcium
borohydride at 670 K and 100 MPa hydrogen pressure. Interesting
enough, when the CaB12H12:5CaH2 mixture is heated in vacuum
at 833 K prior to hydrogenation at 100 MPa H2 pressure and 670 K,
small amounts of crystalline Ca(BH4)2 are produced. Although the
mechanism by which Ca(BH4)2 is generated requires additional
studies, it is possible that CaB6 is generated in situ, and then reacts
with CaH2 to form the borohydride. Based on first-principles
calculations, Ozolins et al. [12] predicted that CaB12H12 can react
with 1 equiv. of CaH2 to generate H2 (6.3% mass fraction capacity)
and CaB6:
Fig. 10. XRD powder profile of the product generated from ball-milled CaB12H12
:
CaH2 (1:1) upon heating to 870 K and its comparison to the CaB6 phase, JCPDS No.
00-031-0254. The inset shows the SEM image of the as-obtained powder.
evidence of CaB6 being formed. The product is hygroscopic as
evidenced by the mass increase of the material exposed to moist air.
EDX spectroscopy of the sample revealed that in addition to the
main Ca and B peaks, the spectrum displays a characteristic oxygen
peak. No such peak was detected when CaB12H12 was decomposed
in the presence of CaH2 to form the air-stable CaB6 (Fig. S7).
We believe that high kinetic barriers may be responsible for
poor hydrogen release and absorption in reactions involving
CaB12H12 up to 670 K. This result is consistent with experimental
observation of decreasing hydrogen storage capacity of M[BH4]m
CaB12H12+CaH2-2CaB6+7H2
(V)
systems under similar conditions as desorption/absorption cycles
2ꢀ
Our TPD studies on a 1:1 mixture of CaB12H12 and CaH2 revealed
that at 670 K only 0.8% mass fraction H2 is released (Fig. S6), in
contrast with Ca(BH4)2, which, under these conditions, releases 8.1%
mass fraction of hydrogen. An ex situ XRD analysis of the product
showed no CaB6 formation up to 670 K. We applied TGA to study
the thermal behavior of the materials up to 870 K. Heating the ball-
milled CaB12H12:CaH2 sample up to 870 K resulted in ꢂ6% mass
loss. The weight loss is in good agreement with the calculated
amount of H2 released in reaction (V). Under the same conditions,
the mass loss from pure CaB12H12 is less than 1.5% (Fig. 9). The XRD
of the sample resulting from dehydriding the 1:1 CaB12H12:CaH2
composite confirmed that CaB6 was formed as a crystalline product
of the reaction (Fig. 10). SEM revealed that the resulting material
consists of submicron particles interconnected into larger
agglomerates (inset in Fig. 10). In contrast, CaB12H12 thermally
treated at 870 K displays an amorphous XRD pattern with no
are repeated. Accumulation of [B12H12
]
intermediates leads to
significant function loss, presumably due to the high stability of
the dodecahydro-closo-dodecaborate clusters.
4. Conclusions
In the present paper, we demonstrate the synthesis of
CaB12H12 from [Ca(H2O)7][B12H12] ꢁ H2O and [Ca(H2O)5(-
MeCN)2][B12H12] and investigate its structure and reactivity. The
crystal structure of CaB12H12 was refined from powder XRD data.
NVS data are in agreement with the DFT calculated phonon
density of states (DOS) and support the choice of the monoclinic
C2/c structure. The rehydrogenation tests on the ball-milled
mixtures of CaB12H12 and CaH2 did not show any borohydride
formation at 100 MPa H2 pressure and 670 K. Similar to alkali
metal dodecahydro-closo-dodecaborates, CaB12H12 displays
a
remarkable thermal stability. We show that the addition of
CaH2 has a destabilizing effect on CaB12H12 and favors hydrogen
release. The formation of dodecahydro-closo-dodecaborate spe-
cies during the decomposition of Ca(BH4)2 can be disadvanta-
geous with respect to reversible hydrogen storage applications.
The accumulation of such stable intermediate species in the
reaction mixtures not only decreases the capacity, but also the
cyclability of the material. This study highlights a possible
mitigation of this problem by using destabilization approaches
which modify the reaction pathways and can favor the hydrogen
release.
Acknowledgments
The authors would like to acknowledge the many
valuable suggestions made by Drs. Ewa Ro¨nnebro, Eric Majzoub,
Timothy Boyle and Leonard Klebanoff. We are grateful to
Fig. 9. TGA data for CaB12H12 and ball-milled CaB12H12:CaH2 (1:1).

ARTICLE IN PRESS
1140
V. Stavila et al. / Journal of Solid State Chemistry 183 (2010) 1133–1140
Dr. Satish S. Jalisatgi for generously providing the Li2B12H12 and
K2B12H12 samples and to Jeff Chames for help with the SEM/EDX
experiments. We gratefully acknowledge financial support from
the US Department of Energy, Office of Energy Efficiency and
Renewable Energy, in the Hydrogen, Fuel Cells & Infrastructure
Technologies Program under Contract nos. DE-AC04-94AL85000,
DE-AI-01-05EE11104, and DE-AI-01-05EE11105. The NMR facility
at Caltech was supported by the National Science Foundation
(NSF) under Grant no. 9724240 and partially supported by
the MRSEC Program of the NSF under Award Number
DMR-0520565.
[13] K. Miwa, M. Aoki, T. Noritake, N. Ohba, Y. Nakamori, S. Towata, A. Zu¨ttel,
S. Orimo, Phys. Rev. B 74 (2006) 155122.
[14] N. Ohba, K. Miwa, M. Aoki, T. Noritake, S. Towata, Y. Nakamori, S. Orimo,
A. Zu¨ttel, Phys. Rev. B 74 (2006) 075110.
[15] E. Ro¨nnebro, E.H. Majzoub, J. Phys. Chem. B 111 (2007) 12045–12047.
[16] M. Aoki, K. Miwa, T. Noritake, N. Ohba, M. Matsumoto, H.W. Li, Y. Nakamori,
S. Towata, S. Orimo, Appl. Phys. A 92 (2008) 601–605.
[17] J.H. Kim, S.A. Jin, J.H. Shim, Y.W. Cho, J. Alloys Compd. 461 (2008) L20–L22.
[18] J.H. Kim, S.A. Jin, J.H. Shim, Y.W. Cho, Scr. Mater. 58 (2008) 481–483.
[19] J.H. Kim, J.H. Shim, Y.W. Cho, J. Power Sources 181 (2008) 140–143.
[20] M.D. Riktor, M.H. Sorby, K. Chlopek, M. Fichtner, B.C. Hauback, J. Mater. Chem.
19 (2009) 2754–2759.
[21] I.B. Sivaev, V.I. Bregadze, S. Sjo¨berg, Collect. Czech. Chem. Commun. 67 (2002)
679–727.
[22] I. Tiritiris, T. Schleid, Z. Anorg. Allg. Chem. 627 (2001) 1836–1845.
[23] K.A. Solntsev, N.T. Kuznetsov, V.I. Ponomarev, Inorg. Mater. 12 (1976)
874–878.
[24] Certain commercial suppliers are identified in this paper to foster under-
standing. Such identification does not imply our recommendation or
endorsement nor does it imply that the materials or equipment identified
are necessarily the best available for the purpose.
[25] I. Tiritiris, T. Schleid, Z. Anorg. Allg. Chem. 629 (2003) 1390–1402.
[26] T.J. Udovic, D.A. Neumann, J. Leao, C.M. Brown, Nucl. Instrum. Methods Phys.
Res. A 517 (2004) 189–201.
Appendix A. Supplementary material
Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.jssc.2010.03.026.
[27] /http://www.pwscf.org/S.
[28] S. Hermanek, Chem. Rev. 92 (1992) 325–362.
References
[29] I. Tiritiris, T. Schleid, K. Muller, Appl. Mag. Res. 32 (2007) 459–481.
[30] G. Kresse, J. Furthmuller, J. Hafner, Europhys. Lett. 32 (1995) 729–734.
[31] T. Yildirim, Chem. Phys. 261 (2000) 205–216.
[32] D.G. Allis, B.S. Hudson, J. Phys. Chem. A 110 (2006) 3744–3749.
[33] J.-H. Her, M. Yousufuddin, W. Zhou, S.S. Jalisatgi, J.G. Kulleck, J.A. Zan,
S.-J. Hwang, R.C. Bowman, T.J. Udovic, Inorg. Chem. 47 (2008) 9757–9759.
[34] J.-H. Her, W. Zhou, V. Stavila, C.M. Brown, T.J. Udovic, J. Phys. Chem. C 113
(2009) 11187.
[1] A. Zu¨ttel, A. Borgschulte, S.I. Orimo, Scr. Mater. 56 (2007) 823–828.
[2] B. Sakintuna, F. Lamari-Darkrim, M. Hirscher, Int. J. Hydrogen Energy 32
(2007) 1121–1140.
[3] Y. Nakamori, H.W. Li, K. Kikuchi, M. Aoki, K. Miwa, S. Towata, S. Orimo,
J. Alloys Compd. 446 (2007) 296–300.
[4] S.I. Orimo, Y. Nakamori, J.R. Eliseo, A. Zu¨ttel, C.M. Jensen, Chem. Rev. 107
(2007) 4111–4132.
[35] S.H. Strauss, Chem. Rev. 93 (1993) 927–942.
[5] J.J. Vajo, S.L. Skeith, F. Mertens, J. Phys. Chem. B 109 (2005) 3719–3722.
[6] J. Yang, A. Sudik, C. Wolverton, J. Phys. Chem. C 111 (2007) 19134–19140.
[7] J. Purewal, S.J. Hwang, R.C. Bowman, E. Ro¨nnebro, B. Fultz, C. Ahn, J. Phys.
Chem. C 112 (2008) 8481–8485.
[8] S.J. Hwang, R.C. Bowman, J.W. Reiter, J. Rijssenbeek, G.L. Soloveichik, J.C. Zhao,
H. Kabbour, C.C. Ahn, J. Phys. Chem. C 112 (2008) 3164–3169.
[9] S.I. Orimo, Y. Nakamori, N. Ohba, K. Miwa, M. Aoki, S. Towata, A. Zu¨ttel, Appl.
Phys. Lett. 89 (2006) 021920.
[10] G.L. Soloveichik, Y. Gao, J. Rijssenbeek, M. Andrus, S. Kniajanski, R.C. Bowman
Jr., S.-J. Hwang, J.-C. Zhao, Int. J. Hydrogen Energy 34 (2009) 916–928.
[11] H.W. Li, K. Kikuchi, Y. Nakamori, N. Ohba, K. Miwa, S. Towata, S. Orimo, Acta
Mater. 56 (2008) 1342–1347.
[36] I.A. Zakharova, Coord. Chem. Rev. 43 (1982) 313–324.
[37] M. Elrington, N.N. Greenwood, J.D. Kennedy, M. Thorntonpett, J. Chem. Soc.
Dalton Trans. (1987) 451–456.
[38] E.S. Shubina, I.A. Tikhonova, E.V. Bakhmutova, F.M. Dolgushin, M.Yu. Antipin,
V.I. Bakhmutov, I.B. Sivaev, L.N. Teplitskaya, I.T. Chizhevsky, I.V. Pisareva, V.I.
Bregadze, L.M. Epstein, V.B. Shur, Chem. Eur. J. 7 (2001) 3783–3790.
[39] R.D. Shannon, Acta Crystallogr. A 32 (1976) 751–767.
[40] A. Andreasen, T. Vegge, A.S. Pedersen, J. Solid State Chem. 178 (2005)
3672–3678.
[41] H. Shao, H. Xu, Y. Wang, X. Li, J. Solid State Chem. 177 (2004) 3626–3632.
[42] R.J. Newhouse, V. Stavila, S.-J. Hwang, L.E. Klebanoff, J.Z. Zhang, J. Phys. Chem.
C 114 (2010) 5224–5232, doi:10.1021/jp9116744.
[12] V. Ozolins, E.H. Majzoub, C. Wolverton, J. Am. Chem. Soc. 131 (2009)
230–237.
[43] H. Li, K. Kikuchi, Y. Nakamori, K. Miwa, S. Towata, S. Orimo, Scr. Mater. 57
(2007) 679–682.