Journal of Alloys and Compounds 493 (2010) 281–287
Contents lists available at ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Effect of additives on the synthesis and reversibility of Ca(BH4 )2
C. Rongeat a,b,∗ , V. D’Anna c , H. Hagemann c , A. Borgschulte b , A. Züttel b , L. Schultz a , O. Gutfleisch a
a
IFW Dresden, Institute for Metallic Materials, Helmholtzstrasse 20, D-01069 Dresden, Germany
Empa, Swiss Federal Laboratories for Materials Testing and Research, Laboratory 138, Hydrogen and Energy, Überlandstrasse 129, CH-8600 Dübendorf, Switzerland
c
Université de Genève, Département de Chimie Physique, 30 Quai Ernest-Ansermet, CH-1211 Genève 4, Switzerland
b
a r t i c l e
i n f o
Article history:
Received 28 October 2009
Received in revised form
11 December 2009
Accepted 15 December 2009
Available online 23 December 2009
Keywords:
Hydrogen storage
Calcium borohydride
Reactive ball milling
Raman and infrared spectroscopy
a b s t r a c t
Metal borohydrides are potential materials for solid-state hydrogen due to their high gravimetric and
volumetric hydrogen densities. Among them, Ca(BH4 )2 is particularly interesting because of the predicted suitable thermodynamic properties. In this work, we investigate a new synthesis route using
high-pressure reactive ball milling. Starting from CaH2 and CaB6 with a TiCl3 or TiF3 as additive, a reaction yield of 19% is obtained after 24 h milling at room temperature and 140 bar H2 . The presence of
Ca(BH4 )2 is confirmed by the presence of the stretching mode of the [BH4 ]− group in the infrared spectra of the as-milled samples. Using in situ XRD, we observe the recrystallisation of a poorly crystallised
Ca(BH4 )2 phase present after milling. The reversible decomposition/formation of Ca(BH4 )2 is obtained
with higher yield (57%) using higher temperature and TiF3 as additive but not with TiCl3 despite its similar electronic structure. The differences observed using different additives and the influence of the anion
are discussed.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
Hydrogen as energy carrier, combined with fuel cells or for direct
combustion, is one solution for the future energy challenge of our
society [1]. In order to build a “hydrogen economy”, safe and reliable storage is needed. Solid-state storage in hydrides appears to
be a valuable alternative, in particular for mobile applications [1].
High gravimetric and volumetric storage densities can be achieved
using complex hydrides. Metal borohydrides M(BH4 )x , with M a
light cation, are of particular interest. For example, LiBH4 contains 18.4 wt% of hydrogen but its thermodynamic properties do
not enable the release of hydrogen at temperatures required for
application (100–150 ◦ C) [2]. More suitable performances can be
achieved with Ca(BH4 )2 , hydrogen content 11.6 wt%, because of significantly lower predicted enthalpy of reaction. Miwa et al. [3] have
proposed Eq. (1) as the decomposition reaction and calculated an
enthalpy of reaction of 32 kJ mol−1 H2 .
3Ca(BH4 )2 ↔ 2CaH2 + CaB6 + 10H2
(1)
The total hydrogen release for this reaction is theoretically
9.6 wt%. Later, Siegel et al. [4] have reported a value of 41.4 kJ mol−1
H2 for the enthalpy of reaction (1) based on first principle calculations. The values reported for the enthalpy of reaction are similar
and lie in the range of interest for hydrogen storage applications.
∗ Corresponding author at: IFW Dresden, Institute for Metallic Materials,
Helmholtzstrasse 20, D-01069 Dresden, Germany. Tel.: +49 351 4659 669.
E-mail address: [email protected] (C. Rongeat).
0925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2009.12.080
The lower decomposition temperature of Ca(BH4 )2 compared to
LiBH4 has also been predicted by Nakamori et al. [5], as shown
in their plot of the decomposition temperature vs. the electronegativity of the different cations.
Different polymorphs of Ca(BH4 )2 are observed depending
on the synthesis method or heat treatment. The first calculated
structure of Ca(BH4 )2 , which is observed experimentally at room
temperature, is an orthorhombic Fddd structure [3,6,7] and is called
␣ phase. Another space group F2dd has been proposed for the
␣ phase more recently [8,9] based on synchrotron analysis. This
structure fits better the overall diffraction pattern of the pure ␣
phase but the main diffraction peaks are the same as previously
reported. Another polymorph named ␤-Ca(BH4 )2 is also observed
with a tetragonal P42 /m structure [7,10,11]. This phase is obtained
at room temperature [12] but is also formed by transformation of
the ␣ phase upon heating [13]. For this ␤ phase also, a different
space group P-4 has been reported [8,14]. Again, the main diffraction peaks are similar to the previous structure and only minor
diffraction peaks differ. Additional polymorphs ␣ , ␥ and ␦ (hightemperature phase) have been proposed/observed in the last years
by different groups [8,11–14] and can be obtained depending on
the synthesis method.
The overall decomposition reaction described in Eq. (1) has been
confirmed by recent works [15,16]. The reaction is more complex
than described by Eq. (1) and proceeds in two steps as observed
for other borohydrides. It has been reported that the first step of
decomposition leads to the formation of CaH2 and an intermediate phase and the second to the decomposition of the intermediate
compound into CaH2 and CaB6 (or “amorphous” B) [15,17,18]. The
282
C. Rongeat et al. / Journal of Alloys and Compounds 493 (2010) 281–287
exact nature of this intermediate phase is not known yet. Riktor et
al. [19] have proposed a CaB2 Hx compound as intermediate, however, the appearance of this phase has been found to depend on
the heating treatment and analysis techniques. The use of TiF3 ,
TiCl3 , NbF5 or NbCl5 leads to a decrease of the decomposition temperature [20,21], however, the same decrease of temperature is
obtained by ball milling Ca(BH4 )2 only. Nevertheless, the use of
appropriate additives is necessary to facilitate the re-absorption
reaction. Best results have been obtained using NbF5 : 5 wt% H2
(50–60% reaction yield) is absorbed with the highest rate under
90 bar H2 and 350 ◦ C for 24 h [21]. The better kinetics obtained
may be related to the low melting point of NbF5 (77 ◦ C), implying its melting during milling, that allows a better dispersion in
Ca(BH4 )2 during milling because of the liquid state. The formation
of CaH2−x Fx phases during decomposition may also play a role in
the enhancement of the sorption properties.
One of the first methods used is a metathesis reaction [22] from
LiBH4 (or NaBH4 ) and CaCl2 . Nevertheless, LiCl is formed as a byproduct and should be separated from the borohydride in a second
step. Another method is by wet chemical synthesis but this necessitates the use of large quantity of solvent, which has to be removed in
additional purification steps [23]. Rönnebro and Majzoub [16] have
demonstrated the possibility to synthesise Ca(BH4 )2 by hydrogenation of CaH2 and CaB6 . They have used conditions at 700 bar H2 and
440 ◦ C for 48 h. The reaction proceeds only when adding a mixture
of TiCl3 and Pd; with 8 wt% of these additives, the reaction yields
60% of Ca(BH4 )2 .
In this study, we show that the synthesis of Ca(BH4 )2 is possible
by reactive ball milling of CaH2 and CaB6 under mild conditions.
The mechano-chemical synthesis method has already been successfully used to prepare novel rare earth intermetallics [24] and
various complex hydrides [25–27]. Different characterisation techniques are combined to obtain a clear picture about the reaction
products. In particular, Raman and infrared (IR) spectroscopy give
complementary information about the composition of the samples
analysed. The importance of the additive for hydrogen absorption
will also be discussed. Up to now, the effect of the additive has
only been considered for the decomposition reaction, for the reabsorption reaction, it is only stated that the use of additive is
necessary [21]. In this paper, more information about Ca(BH4 )2
and corresponding reaction paths is obtained by investigating thoroughly the synthesis reaction and the effect of different additives
in detail.
2. Experimental
2.1. Sample preparation
Reactive ball milling is used for the synthesis of Ca(BH4 )2 through reaction (1)
from CaH2 (99.99%, Sigma–Aldrich) and CaB6 (99.5%, Alfa Aesar) with 2:1 mol ratio
under hydrogen pressure. Starting from the decomposition products has the advantage to avoid the formation of by-products. For mechano-chemical synthesis, we
used a stainless steel vial designed for milling under high-pressure (max 150 bar).
In addition, temperature and pressure sensors are installed in the lid of the vial
and can be used to monitor the reaction in situ during milling (evico-magnetics).
The addition of 8 mol% of additive to the starting mixture was done to promote the
reaction kinetics. Two compositions were used with two different additives. The
choice of TiCl3 (99.999%, Sigma–Aldrich) was guided by the very good efficiency
observed for sodium alanate [25]. It has already been used for Ca(BH4 )2 [16,20].
TiF3 (Sigma–Aldrich) has been proposed more recently [21] and has a positive effect
during decomposition of Ca(BH4 )2 . It is expected that this effect can also help the
reverse reaction. A limited contamination by iron was detected in the samples originating from the vial walls and balls. However, the presence of iron is likely to be
identical in the samples prepared with TiCl3 or TiF3 and should not hamper the comparison between the effect of both additives. The vial with 2 g of starting mixture
and 37 stainless steel balls (10 mm, ball-to-powder ratio 75:1) was pressurised with
140 bar H2 and then milled for 24 h without interruption in a Fritsch Pulverisette 6
planetary mill.
Cycling was done using a Sieverts’ type apparatus (Hy-Energy PCTPro 2000).
Around 200 mg of loose powder was loaded in a stainless steel cylinder with glass
wool on top to avoid powder escape. This cylinder was then installed in a stainless
steel sample holder with packing material to reduce the free volume as much as
possible. The sample holder is heated through an external heating jacket. Temperature of the sample is recorded by a thermocouple placed close to the bottom of the
cylinder. For each samples using TiCl3 or TiF3 as additive, one cycle was performed
at 350 ◦ C in static vacuum for desorption, and then in 90 bar H2 for absorption. The
sample temperature was elevated to 350 ◦ C during the first desorption step and then
kept constant for 10 h during desorption and for the all absorption step following.
Powder handling is done under Ar atmosphere in a glove box with H2 O and O2
contents less than 1 ppm.
2.2. Characterisation
Room temperature X-ray diffraction (XRD) analysis was performed using a STOE
STADI MP diffractometer (Mo K␣ radiation). The measurements were done in transmission using a Debye–Scherrer geometry and a Position Sensitive Detector (PSD).
In situ XRD measurements during heating were performed on a Brucker D8 Advance
apparatus with Cu K␣ radiation. For both type of measurement, the powder was filled
in a glass capillary (0.7 mm external diameter) prepared in the glove box and protected by paraffin. The capillary is then sealed by melting the end of the capillary. For
in situ XRD, the capillary is installed inside a MRI (Materials Research Instruments)
high-temperature capillary furnace.
To allow a direct comparison of the XRD patterns measured at different wavelengths, the diffracted intensities are plotted as a function of the wave vector Q
rather than the angle 2. The wave vector is defined by:
Q =
2
d
(2)
where d is the interplanar distance obtained from the Bragg’s law.
Raman spectroscopy was performed at room temperature using a Renishaw
Ramascope 2000 (Renishaw plc, Gloucestershire, U.K.) with a spectral resolution of
1 cm−1 . The 633 nm line of the HeNe laser was focused on the sample through the
magnifying objective (50×) of the microscope. Around 10 mg of sample were placed
in a Linkam THMS600 cell (Linkam Scientific Instruments, U.K.) directly adapted to
the microscope of the instrument. The cell allows the use of controlled atmosphere
and temperature. The glass window of the cell is 0.1 mm thick. The samples were
placed under a small Ar flow to avoid contamination during measurement.
For infrared (IR) measurements, all samples were handled in a nitrogen filled
glove box. The samples were introduced in a Specac Golden Gate ATR cell. The IR
spectra were recorded at room temperature using BIO-RAD EXCALIBUR spectrometer with 1 cm−1 of resolution.
Pure CaH2 , CaB6 and Ca(BH4 )2 were also analysed in the same conditions for
comparison. For Raman measurement, the Ca(BH4 )2 reference compound (Aldrich,
dry powder) is composed mainly of the ␤ phase with presence of small quantity of
the ␣ phase. For IR measurement, the Ca(BH4 )2 is composed mainly of the ␣ phase
(from drying Ca(BH4 )2 ·2THF, Aldrich).
3. Results and discussion
3.1. Synthesis by high-pressure reactive milling
The evolution of the temperature and the pressure recorded during the milling of the mixture 2 CaH2 + CaB6 + 0.08 TiCl3 are given
in Fig. 1a and from these values the hydrogen content in the vial is
calculated using the ideal gas law and shown in Fig. 1b. Note that
thorough external controls were performed during the complete
process to ensure that no hydrogen leak occurred. The variation of
hydrogen pressure measured during milling is only related to the
reaction taking place inside the vial. It has already been shown for
NaAlH4 that the variation of H2 pressure inside the vial gives reliable information on the reaction completion [28,29]. High-pressure
of 140 bar is used for milling to obtain the highest driving force
possible for the reaction [28,30]. A high number of balls and high
rotational speed are also used to increase the energy of the milling
[28,31] and improve the effectiveness of the process. At the beginning of the process, we observe an increase of the temperature due
to the friction between the balls and to the vial walls that leads to
an increase of the H2 pressure. A maximum temperature of 55 ◦ C
is measured in the vial lid and is reached after 5 h milling. Note
that the real temperature upon powder particle collision and the
temperature of the surrounding gas are higher than the recorded
temperature in the lid. This could lead to an overestimation of the
decrease of H2 determined from the pressure drop in the vial, especially at the beginning of the process when the temperature and
pressure change because of the mechanical activation and of the
C. Rongeat et al. / Journal of Alloys and Compounds 493 (2010) 281–287
283
Fig. 2. XRD patterns of as-milled samples 2 CaH2 + CaB6 + 0.08 additive (after 24 h
milling): (a) TiCl3 and (b) TiF3 .
Fig. 1. Evolution of (a) the pressure (solid line) and temperature (squares) recorded
in the vial and (b) the quantity of hydrogen in the vial (solid line) and of the temperature (squares) during milling of 2 CaH2 + CaB6 + 0.08 TiCl3 .
exothermic absorption of hydrogen by the mixture. The variation of
the quantity of H2 should be then seen as a qualitative view of the H2
absorption during the milling process. After 25 h, the rate of hydrogen absorption determined from the pressure decrease is very low
and the milling is stopped to analyse the compound(s) formed.
Considering the initial and final pressure in the vial (at room temperature), we obtain a consumption of 0.0186 mol H2 . The total H2
amount consumed for reaction (1) would be 0.0992 mol, considering the starting quantity of reactants, indicating that a reaction
yield of 19% can be obtained during milling. Although small, this
yield is quite encouraging considering that the reaction takes place
at low temperature (60% was previously obtained at 700 bar and
440 ◦ C, after 48 h [16]). Similar behaviour is obtained for the sample prepared with TiF3 (not shown). From this first experiment, the
additives seem to have little influence on the synthesis reaction. The
reaction yield is calculated assuming that the hydrogen absorbed
has reacted to form Ca(BH4 )2 and not any intermediate phase. This
assumption needs to be confirmed. The samples are therefore analysed by XRD to identify the hydride phase formed during milling
and/or to evidence the presence of new compounds.
The XRD patterns of the milled powders are very similar whatever the additive used and show mainly peaks related to CaH2 and
CaB6 (Fig. 2). For the sample milled with TiCl3 (Fig. 2a), the peaks
related to CaH2 are very weak and the low intensities observed can
also be related to a large broadening due to crystallite size reduction during milling. No other hydride phase formed during milling
can be detected, it is likely very disordered or poorly crystallised
and cannot be seen by XRD. For the sake of clarity in the following discussion, we call it amorphous hydride phase. In addition, no
trace of the additive itself is found in the patterns despite they are
present in relatively large amount in the starting mixture. There is
also no compound found that could originate from the decomposition of the additive TiCl3 . On the other hand, a small amount of CaF2
can be detected from the XRD pattern of the as-milled sample with
TiF3 . This phase may be formed by the decomposition of TiF3 by
reaction with CaH2 . Other CaF2−x Hx compounds with similar structure as CaF2 have been reported by Kim et al. [21], however, from
the XRD pattern, we do not detect the additional diffraction peaks
(compared to CaF2 ) of such phases (Fig. 2b); therefore, we only
consider the formation of CaF2 for the following discussion. In general, the additive itself or any compound formed from it should be
finely distributed in the powder (nanocomposite) or with a highly
disordered structure, so hardly detectable by XRD.
The borohydrides compounds are often difficult to analyse
by XRD because of their complex structures and sometime disordered/amorphous state, especially after milling. For example,
Agresti and Khandelwal [32] have evidenced by XRD the formation
of LiBH4 from milling of LiH and B only after removing un-reacted
compounds using a solvent and heating the products. Raman spectroscopy is often used as an additional characterisation technique
for these types of compounds [33–35], especially to detect the presence of borohydride phases while they cannot be identified in XRD
patterns [5,16]. Interestingly, it is possible to distinguish the different polymorphs of Ca(BH4 )2 by Raman spectroscopy [10]. Raman
spectroscopy is sensitive to the vibration of the molecules. Infrared
(IR) spectroscopy is also sensitive to the vibrations of molecules
and shows comparable spectra. However, the principles and selection rules are different for both techniques meaning that they give
complementary information on a sample [36]. The Raman effect is
based on a change of the polarisability of the molecule whereas the
IR spectroscopy is sensitive to the change of the dipole moment of
the molecule. These different criterions imply that different compounds and bonding are detected by each technique.
The Raman spectrum of pure Ca(BH4 )2 (Fig. 3a) is composed
mainly of vibration lines arising around 2300 cm−1 that corresponds [10] to the stretching modes of the B–H bonding in [BH4 ]− .
The corresponding bending modes appear in the 1000–1400 cm−1
range. For CaH2 , different lines arise at low wave numbers [37]:
130, 180, 206 cm−1 and we observe additional lines at 727 and
1000 cm−1 in Fig. 3b. For pure CaB6 , the spectrum [16,38] is composed of three lines at 755, 1117 and 1259 cm−1 (Fig. 3c). The Raman
spectra obtained for the samples prepared with TiCl3 or TiF3 (Fig. 3d
and e) show mainly the presence of CaB6 with the three vibration
lines as observed for the pure compound. No vibration line corresponding to CaH2 is observed for the sample containing TiCl3 , while
there might be a small contribution from CaH2 in the spectrum of
the sample milled with TiF3 (see ca. 150 cm−1 ). There is also no trace
of another vibration mode. It is not possible to identify the hydride
phase formed during milling. Infrared spectroscopy is therefore
used because of its higher sensitivity (higher cross section) that
can be beneficial in the case of a small amount of compound as it is
the case. The results of IR measurements on the as-milled samples
284
C. Rongeat et al. / Journal of Alloys and Compounds 493 (2010) 281–287
Fig. 5. In situ XRD measurements upon heating from samples milled with (a) TiCl3
and (b) TiF3 . The hump centred on 1.5 Å−1 is related to the diffraction of the glass
capillary. The phases present are: CaH2 (), CaB6 (), CaF2 (×) and ␤-Ca(BH4 )2 (␤).
Fig. 3. Raman spectra for reference compounds Ca(BH4 )2 (a), CaH2 (b), CaB6 (c) and
for the as-milled samples with TiCl3 (d) or TiF3 (e) as additive.
are given in Fig. 4.
The IR spectrum obtained for pure Ca(BH4 )2 (Fig. 4a) can be
compared with the Raman spectrum in Fig. 3a. There are two main
regions of vibrations, around 2300 cm−1 for the stretching modes
and for 1000–1300 cm−1 for the bending modes. The peak observed
at ca. 1600 cm−1 is related to a contamination by water. On the
other hand, the spectrum obtained for CaH2 (Fig. 4b) is different
Fig. 4. IR spectra for reference compounds Ca(BH4 )2 (a), CaH2 (b), CaB6 (c) and for
the as-milled samples with TiCl3 (d) or TiF3 (e) as additive.
with only a large feature around 1000 cm−1 . The spectrum of pure
CaB6 (Fig. 4c) does not show any vibration line in contrast with
Raman spectroscopy. The bands observed around 2330–2370 cm−1
are related to gaseous CO2 contamination. Similar IR spectra are
obtained for the both as-milled samples (Fig. 4d and e). We only
observe two broad components around 1000 and 2200 cm−1 . Comparing with the spectrum obtained for pure CaH2 , it is clear that
the component observed around 1000 cm−1 is related to this compound. The broad peak around 2200 cm−1 may correspond to the
vibration in the stretching mode of the B–H bonding in [BH4 ]− ,
which is observed in the spectrum of pure Ca(BH4 )2 . Note that the
wavelengths of the vibrations for the stretching modes are independent of the Ca(BH4 )2 polymorphs. It is therefore not possible to
determine which structure has been formed during milling. In the
1000–1500 cm−1 range, it is also possible to observe smaller features which may be related to the bending mode of B–H (besides
CaH2 contribution). Considering the little amount of borohydride
expected from the quantity of hydrogen absorbed during milling
(19%), it is not surprising that the signal detected is very low. The
very broad peaks obtained can reasonably be ascribed to amorphous/disordered Ca(BH4 )2 . The formation of another borohydride
is unlikely since Ca based compounds are the main starting products. The formation of Ti(BH4 )3 by reaction with the additive [39]
cannot be excluded (presence of CaF2 ). However, this compound
is very unstable and would have decomposed immediately during milling. In addition, the formation of Ti(BH4 )3 is reported only
by reaction between a borohydride and a Ti-based compound, not
from the “elements”. By combining different characterisation techniques, we can have a good indication that part of the milled
samples is Ca(BH4 )2 .
In situ XRD measurements were performed on both samples
prepared with Ti-based additives to analyse the behaviour of the
phases formed during heating to 300 ◦ C (before decomposition).
The patterns obtained at different temperatures are given in Fig. 5.
At low temperatures, the patterns are similar to the one obtained
after milling with the presence of CaH2 and CaB6 mainly (Fig. 2).
For the sample milled with TiCl3 (Fig. 5a), it is possible to see small
C. Rongeat et al. / Journal of Alloys and Compounds 493 (2010) 281–287
285
Fig. 7. XRD patterns of the samples milled with (a) TiCl3 and (b) TiF3 as additive
after the absorption step at 350 ◦ C and 100 bar H2 for 24 and 40 h, respectively. The
phases present are ␣-Ca(BH4 )2 (␣), ␤-Ca(BH4 )2 (␤), CaH2 (), CaB6 (), CaF2 (×),
CaCl2 (). Peaks marked with a question mark (?) could not be indexed.
Fig. 6. (a) Desorption step during heating to 350 ◦ C in static vacuum and (b) absorption step at 350 ◦ C and 100 bar H2 for samples milled with TiCl3 (open square) or TiF3
(open circle) as additive measured with a Sieverts’ type apparatus. The evolution of
the temperature during desorption is given by the solid line in (a).
peaks arising which correspond to ␤-Ca(BH4 )2 (high-temperature
phase) at 240 ◦ C. For the sample milled with TiF3 , these peaks arise
at 210 ◦ C (Fig. 5b). These peaks are growing when the temperature
increases. The appearance of the ␤ phase is in agreement with previous reports on phase transformations in Ca(BH4 )2 which show
that the room temperature polymorphs ␣ or ␥ transform into the
␤ phase above 200 and 270 ◦ C, respectively [13]. The growth of the
␤-Ca(BH4 )2 phase in our sample above 200 ◦ C is therefore not surprising and confirms that a Ca(BH4 )2 phase is already present in
the milled sample. The presence of an intermediate phase in the
milled sample, which transforms to Ca(BH4 )2 , is unlikely since the
samples were heated under Ar in the capillary. To form Ca(BH4 )2
from any intermediate phase, the presence of H2 is required (and
probably higher temperature). This observation together with the
results of IR spectroscopy are a reasonable proof of the formation
of Ca(BH4 )2 during milling.
To summarise this part, reactive ball milling under high hydrogen pressures leads to the formation of Ca(BH4 )2 in small quantity.
The combination of different characterisation techniques allows its
unambiguous identification, hardly possible by XRD only since it is
poorly crystallised. This result indicates that the quantity of hydrogen absorbed during milling is used for the formation of Ca(BH4 )2
with a reaction yield of 19% and no difference is observed between
the additives that show similar influence on the milling process.
3.2. Cycling capability
As a next step, we have performed one cycle desorption/absorption in a Sieverts’ type apparatus in order to study the
reversibility of the reaction. In addition, the formation of an ordered
Ca(BH4 )2 phase should be more favourable at high temperature
(350 ◦ C) as seen from in situ XRD measurements. The first desorption steps leads to similar results for both samples (Fig. 6a)
with a release of ca. 1.6 wt% of hydrogen in agreement with the
amount of gas absorbed during milling. The decomposition started
at 270–300 ◦ C for both samples during the heating ramp to 350 ◦ C
(hydrogen backpressure of about 0.2 bar). The release of H2 is
slightly slower and proceeds in two steps for the sample milled
with TiF3 . The results of the following absorption step are shown in
Fig. 6b. For both samples, we observe a first part (t < 2 h) with fast
hydrogen absorption and then a continuous and slow increase of
the hydrogen uptake for the remaining time. For the sample with
TiF3 , a capacity of 4.8 wt% H2 is achieved after 24 h whereas only
2.8 wt% H2 is reached in the sample containing TiCl3 . The capacity
obtained using TiCl3 remains far from the theoretical value of 9 wt%
H2 taking into account the weight of the additive (8.8 wt% if we
consider that TiCl3 has entirely reacted with CaH2 to form CaCl2 ).
For longer absorption time (not shown), the maximum capacity of
the sample containing TiF3 obtained is 5.3 wt% H2 after 40 h, corresponding to a reversibility yield of 57% considering the weight
of the additive (60% considering the maximum possible weight of
by-product CaF2 from the reaction between CaH2 and TiF3 ). This
value is in agreement with the reversible absorption of hydrogen
obtained by Kim et al. [21] in similar conditions but directly from
the decomposition products of pure Ca(BH4 )2 . To understand the
difference observed between the two additives, structural characterisation is performed by XRD and IR spectroscopy.
Significant different phases are observed in the XRD patterns
depending on the additive used (Fig. 7). For both samples, we
observe un-reacted CaH2 and CaB6 in agreement with the noncompletion of the reaction observed during kinetic measurements
(Fig. 6b). In addition, it is possible to observe the formation of CaCl2
or CaF2 from the reaction of the additive likely with CaH2 . The CaCl2
phase fraction is difficult to evaluate because the peaks indicated
by in Fig. 7a, at position expected for CaCl2 , have very high relative intensities; it is therefore very likely that there is overlapping
with peaks of other compounds. The main difference between the
samples prepared with TiCl3 or TiF3 is the clear appearance of the
␣ and ␤ phases of Ca(BH4 )2 in the sample milled with TiF3 whereas
in the sample with TiCl3 , we obtain only a weak peak related to
␣-Ca(BH4 )2 and other more important peaks (indicated with question marks in Fig. 7a) that could not be indexed. These new peaks
do not correspond to any intermediate phases [17,19] reported for
the decomposition of Ca(BH4 )2 or any polymorphs reported for
Ca(BH4 )2 itself. The formation of CaHCl was found in additional
samples prepared by millings with TiCl3 and performed in different conditions (not shown). Therefore, the un-indexed XRD peaks
may be related to a similar Ca-H-Cl(-B?) phase, as suggested by
286
C. Rongeat et al. / Journal of Alloys and Compounds 493 (2010) 281–287
Fig. 8. IR spectra for reference compounds Ca(BH4 )2 (a), CaH2 (b), CaB6 (c) and for
the cycled samples with TiCl3 (d) or TiF3 (e) as additive.
Kim et al. [20] after ball milling Ca(BH4 )2 , that is consistent with
the weak peaks of CaH2 and the overlap with the CaCl2 peaks. The
formation of this unknown phase limits the formation of Ca(BH4 )2
during absorption. Note also that no intermediate phase for the formation of Ca(BH4 )2 in sample with TiF3 is detected even though the
reaction is not complete. We also cannot detect any phase with Ti
from the decomposed additives.
IR spectroscopy is performed to check for phases not detected
by XRD (Fig. 8). The IR spectrum of the sample with TiCl3 shows
weak features related to CaH2 (Fig. 8d), while Ca(BH4 )2 vibration
lines appear clearly for the sample with TiF3 (Fig. 8e). This is a confirmation that after cycling, no or little Ca(BH4 )2 is formed in the
sample with TiCl3 . The phase formed in this case may be a Ca-HCl(-B?) phase as already supposed from the XRD measurement:
the small peaks around 1200 cm−1 observed in the IR spectrum
(Fig. 8d) might be its IR sign. The formation of this phase likely
prevents the formation of Ca(BH4 )2 and may be favoured by high
temperature since we do not see it after milling. In contrast, for
the sample with TiF3 , it is possible to form Ca(BH4 )2 during hydrogenation and no side products hinders reaction (1). The formation
of the unknown phase may be explained by the reactivity of TiCl3
towards borohydride compounds, for example, Fang et al. [39] have
ascribed the higher reactivity of TiCl3 vs. TiF3 towards lithium borohydride to the higher driving force of the reaction between LiBH4
and TiCl3 (than TiF3 ) to form LiCl, B and TiH2 . Similar effect might
be expected for the reaction of TiCl3 with Ca(BH4 )2 formed during
milling. The absence of a by-product is a clear advantage of TiF3
but this compound can have also another role in the absorption
reaction.
Little information is available on the sorption reactions and doping in Ca(BH4 )2 . Different possible mechanisms can be proposed
depending on the influence of Ti species or fluoride compounds
formed by (partial) decomposition of TiF3 observed here and/or the
effect of the F anion. Considering the thermodynamic point of view,
it may be that F is substituted in Ca(BH4 )2 and changes the enthalpy
of the absorption reaction. It has been demonstrated for sodium
alanate [40,41], by calculation verified with experiments, that the
fluorine anion F can substitute to H in Na3 AlH6 . This exchange
reduces the capacity of the compound but can change the thermodynamic properties to more suitable ones, e.g. by decreasing the
enthalpy of reaction. It has been predicted by density functional
theory that the same concept can be applied to LiBH4 [42]. From the
characterisation of the milled and cycled samples, it is not possible
to know whether some H atoms are substituted by F in Ca(BH4 )2 but
we assume that the same concept can be applied to this compound.
Considering the improvement of the kinetic properties, it is also
possible that the catalytic activity of Ti species, already demonstrated for sodium alanate [43], can also promote the H2 absorption
in the CaH2 and CaB6 mixture. The Ti species should be formed
by the reaction between the Ti-based additive and another phase
present. It can be assumed that TiF3 is more easily decomposed
because of the higher reactivity of F towards CaH2 to form CaF2
(observed after milling) due to the “easy” substitution of H atoms
by F atoms [44], while TiCl3 may have higher affinity to the borohydride phase. A higher quantity of Ti-based catalytic compounds
may then be formed in this sample explaining the better efficiency
of the hydrogen absorption. By analogy with the formation of ZrB2
reported after milling/cycling LiBH4 + MgH2 composite [45], we
can propose that here TiB2 is formed in the milled and/or cycled
samples. It has been demonstrated that TiB2 can decrease the
decomposition temperature of MgH2 and has a positive effect on
the hydrogen desorption at the near surface of MgH2 [46]. Nevertheless, this effect may not be directly applicable to complex
hydrides but we can suppose that the reaction between Ti and
B (from Ca(BH4 )2 or CaB6 ) weakens the B–H or B–B bonds and
favour the reaction rates. In addition, we can also consider the
role of CaH2−x Fx phase(s) (CaF2 here), also reported by Kim et al.
[21], formed during milling and cycling and their possible effect
on H2 absorption. Nevertheless, no mechanism has been proposed
up to now involving these phases and new concepts need to be
developed. The role of this fluoride phase(s) and the possible formation of Ti-based compounds like TiB2 is a key point to understand
the improvement of the sorption properties and further investigations are needed using different analytical methods to establish the
catalytic and reaction mechanism for Ca(BH4 )2 .
4. Conclusion
Ca(BH4 )2 can be synthesised by reactive ball milling of its
decomposition products CaH2 and CaB6 under 140 bar H2 near
room temperature. The reaction yield is 19% after milling which
is promising since the process is done at low temperature and 60%
yield can be achieved during re-hydrogenation. The compound is
in a poorly crystallised state. The combination of Raman and IR
spectroscopy with in situ XRD is therefore necessary for a complete
analysis of the samples. The comparison of the cycling behaviour
of TiCl3 - and TiF3 -doped Ca(BH4 )2 reveals that TiF3 is a better additive. The use of TiCl3 does not allow the reversible formation of
Ca(BH4 )2 , while by cycling the sample milled with TiF3 , Ca(BH4 )2 is
obtained already at mild hydrogenation conditions. The enhanced
reversibility can originate from a purely catalytic effect from Tibased compounds or CaF2−x Hx (CaF2 ) phases and/or be related to
the partial substitution of H by F in Ca(BH4 )2 changing the stability
of the hydride phase. Further investigations are needed to clarify
the chemical state of the additive and the location of F atoms that
should help to understand the H2 sorption and doping mechanisms
in Ca(BH4 )2 .
Acknowledgements
This work is financially supported by the Marie-Curie European Research Training Network (EU-RTN) COSY, by the FuncHy
(HGF) project and partly by the Swiss National Science Foundation.
C. Rongeat et al. / Journal of Alloys and Compounds 493 (2010) 281–287
Thanks are due to M. Herrich, B. Gebel, A. Remhof, I. Lindemann and
C. Geipel for experimental assistance.
References
[1] L. Schlapbach, A. Züttel, Nature 414 (2001) 353–358.
[2] A. Züttel, P. Wenger, P. Sudan, Ph. Mauron, Ch. Emmenegger, J. Power Sources
118 (2003) 1–7.
[3] K. Miwa, M. Aoki, T. Noritake, N. Ohba, Y. Nakamori, S. Towata, A. Züttel, S.
Orimo, Phys. Rev. B 74 (2006) 155122.
[4] D.J. Siegel, C. Wolverton, V. Ozolins, Phys. Rev. B 76 (2007) 134102.
[5] Y. Nakamori, H.W. Li, K. Kikuchi, M. Aoki, K. Miwa, S. Towata, S. Orimo, J. Alloys
Compd. 446–447 (2007) 296–300.
[6] P. Vajeeston, P. Ravindran, H. Fjellvag, J. Alloys Compd. 446–447 (2007)
44–47.
[7] Y.S. Lee, Y. Kim, Y.W. Cho, D. Shapiro, C. Wolverton, V. Ozolins, Phys. Rev. B 79
(2009) 104107.
[8] Y. Filinchuk, E. Rönnebro, D. Chandra, Acta Mater. 57 (2009) 732–738.
[9] L. George, V. Drozd, S.K. Saxena, E. Gil Bardaji, M. Fichtner, J. Phys. Chem. C 113
(2009) 15087–15090.
[10] M. Fichtner, K. Chlopek, M. Longhini, H. Hagemann, J. Phys. Chem. C 112 (2008)
11575–11579.
[11] F. Buchter, Z. Lodziana, A. Remhof, O. Friedrichs, A. Borgschulte, Ph. Mauron, A.
Züttel, D. Sheptyakov, G. Barkhordarian, R. Bormann, K. Chlopek, M. Fichtner,
M.H. Sorby, M.D. Riktor, B.C. Hauback, S. Orimo, J. Phys. Chem. B 112 (2008)
8042–8048.
[12] M.D. Riktor, M.H. Sorby, K. Chlopek, M. Fichtner, F. Buchter, A. Züttel, B.C.
Hauback, J. Mater. Chem. 17 (2007) 4939–4942.
[13] F. Buchter, Z. Lodziana, A. Remhof, O. Friedrichs, A. Borgschulte, Ph. Mauron, A. Züttel, D. Sheptyakov, L. Palatinus, K. Chlopek, M. Fichtner, G.
Barkhordarian, R. Bormann, B.C. Hauback, J. Phys. Chem. C 113 (2009)
17223–17230.
[14] E. Majzoub, E. Rönnebro, J. Phys. Chem. C 113 (2009) 3352–3358.
[15] J.H. Kim, S.A. Jin, J.H. Shim, Y.W. Cho, J. Alloys Compd. 461 (2008) L20–L22.
[16] E. Rönnebro, E. Majzoub, J. Phys. Chem. B: Lett. 111 (2007) 12045–12047.
[17] 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.
[18] Y. Kim, D. Reed, Y.S. Lee, J.Y. Lee, J.H. Shim, D. Book, Y.W. Cho, J. Phys. Chem. C
113 (2009) 5865–5871.
[19] M.D. Riktor, M.H. Sorby, K. Chlopek, M. Fichtner, B.C. Hauback, J. Mater. Chem.
19 (2009) 2754–2759.
[20] J.H. Kim, S.A. Jin, J.H. Shim, Y.W. Cho, Scripta Mater. 58 (2008) 481–483.
[21] J.H. Kim, J.H. Shim, Y.W. Cho, J. Power Sources 181 (2008) 140–143.
287
[22] Y. Nakamori, H.W. Li, M. Matsuo, K. Miwa, S. Towata, S. Orimo, J. Phys. Chem.
Solids 69 (2008) 2292–2296.
[23] R. Köster, Angew. Chem. 3 (1957) 94.
[24] O. Gutfleisch, M. Kubis, A. Handstein, K.H. Müller, L. Schultz, Appl. Phys. Lett.
73 (1998) 3001–3003.
[25] C. Rongeat, C. Geipel, I. Llamas-Jansa, L. Schultz, O. Gutfleisch, J. Phys.: Conf. Ser.
144 (2009) 012022.
[26] J.M. Bellosta von Colbe, M. Felderhoff, B. Bogdanovic, F. Schüth, C. Weidenthaler,
Chem. Commun. 37 (2005) 4732–4734.
[27] S. Doppiu, L. Schultz, O. Gutfleisch, J. Alloys Compd. 427 (2007) 204–208.
[28] C. Rongeat, I. Llamas-Jansa, S. Oswald, L. Schultz, O. Gutfleisch, Acta Mater. 57
(2009) 5563–5570.
[29] C. Rongeat, I. Llamas-Jansa, S. Doppiu, S. Deledda, A. Borgschulte, L. Schultz, O.
Gutfleisch, J. Phys. Chem. B 111 (2007) 13301–13306.
[30] D. Sun, S.S. Srinivasan, G. Chen, C.M. Jensen, J. Alloys Compd. 373 (2004)
265–269.
[31] M. Magini, C. Colella, A. Iasonna, F. Padella, Acta Mater. 46 (1998) 2841–2850.
[32] F. Agresti, A. Khandelwal, Scripta Mater. 60 (2009) 753–755.
[33] G. Renaudin, S. Gomes, H. Hagemann, L. Keller, K. Yvon, J. Alloys Compd. 375
(2004) 98–106.
[34] A. Borgschulte, A. Züttel, P. Hug, A.M. Racu, J. Schoenes, J. Phys. Chem. A 112
(2008) 4749–4753.
[35] A.R. Emery, R.C. Taylor, J. Chem. Phys. 28 (1958) 1029–1032.
[36] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds, 4th edition, Wiley, New York, 1986.
[37] J.S. Tse, D.D. Klug, S. Desgreniers, J.S. Smith, R. Flacau, Z. Liu, J. Hu, N. Chen, D.T.
Jiang, Phys. Rev. B 75 (2007) 134108.
[38] T.T. Xu, J.G. Zhen, A.W. Nicholls, S. Stankovich, R.D. Piner, R.S. Ruoff, Nano Lett.
4 (2004) 2051–2055.
[39] Z.Z. Fang, L.P. Ma, X.D. Kang, P.J. Wang, P. Wang, H.M. Cheng, Appl. Phys. Lett.
94 (2009) 044104.
[40] H.W. Brinks, A. Fossdal, B.C. Hauback, J. Phys. Chem. C 112 (2008) 5658–5661.
[41] L.C. Yin, P. Wang, X.D. Kang, C.H. Sun, H.M. Cheng, Phys. Chem. Chem. Phys. 9
(2007) 1499–1502.
[42] L.C. Yin, P. Wang, Z.Z. Fang, H.M. Cheng, Chem. Phys. Lett. 450 (2008) 318–321.
[43] J.M. Bellosta von Colbe, W. Schmidt, M. Felderhoff, B. Bogdanovic, F. Schüth,
Angew. Chem. Int. Ed. 45 (2006) 3663–3665.
[44] G.B. Wilson-Short, A. Janotti, A. Peles, C.G. Van de Walle, J. Alloys Compd. 484
(2009) 347–351.
[45] U. Boesenberg, U. Vainio, P.K. Pranzas, J.M. Bellosta von Colbe, G. Goerigk, E.
Welter, M. Dornheim, A. Schreyer, R. Bormann, Nanotechnology 20 (2009)
204003.
[46] V.D. Dobrovolsky, O.G. Ershova, Y.M. Solonin, O.Y. Khyzhun, V. Paul-Boncour, J.
Alloys Compd. 465 (2008) 177–182.