Chloride substitution in sodium borohydride

Article abstract of DOI:10.1016/j.jssc.2011.05.030

The dissolution of sodium chloride and sodium borohydride into each other resulting in formation of solid solutions of composition Na(BH₄) _1-xCl_x is studied. The dissolution reaction is facilitated by two methods: ball milling or combination of ball milling and annealing at 300 °C for three days of NaBH₄-NaCl samples in molar ratios of 0.5:0.5 and 0.75:0.25. The degree of dissolution is studied by Rietveld refinement of synchrotron radiation powder X-ray diffraction (SR-PXD) data. The results show that dissolution of 10 mol% NaCl into NaBH₄, forming Na(BH₄)_0.9Cl_0.1, takes place during ball milling. A higher degree of dissolution of NaCl in NaBH₄ is obtained by annealing resulting in solid solutions containing up to 57 mol% NaCl, i.e. Na(BH₄)_0.43Cl_0.57. In addition, annealing results in dissolution of 1020 mol% NaBH₄ into NaCl. The mechanism of the dissolution during annealing and the decomposition pathway of the solid solutions are studied by in situ SR-PXD. Furthermore, the stability upon hydrogen release and uptake were studied by Sieverts measurements.

Full text of DOI:10.1016/j.jssc.2011.05.030

Journal of Solid State Chemistry 184 (2011) 1858–1866
Contents lists available at ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Chloride substitution in sodium borohydride
Dorthe B. Ravnsbæk, Line H. Rude, Torben R. Jensen n
Center for Materials Crystallography (CMC), Interdisciplinary Nanoscience Center (iNANO), Department of Chemistry, Aarhus University,
˚
Langelandsgade 140, DK-8000 Arhus C, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history:
The dissolution of sodium chloride and sodium borohydride into each other resulting in formation of
solid solutions of composition Na(BH4)1ꢀxClx is studied. The dissolution reaction is facilitated by
two methods: ball milling or combination of ball milling and annealing at 300 1C for three days of
NaBH4–NaCl samples in molar ratios of 0.5:0.5 and 0.75:0.25. The degree of dissolution is studied by
Rietveld refinement of synchrotron radiation powder X-ray diffraction (SR-PXD) data. The results show
that dissolution of 10 mol% NaCl into NaBH4, forming Na(BH4)0.9Cl0.1, takes place during ball milling.
A higher degree of dissolution of NaCl in NaBH4 is obtained by annealing resulting in solid solutions
containing up to 57 mol% NaCl, i.e. Na(BH4)0.43Cl0.57. In addition, annealing results in dissolution of
10–20 mol% NaBH4 into NaCl. The mechanism of the dissolution during annealing and the decomposi-
tion pathway of the solid solutions are studied by in situ SR-PXD. Furthermore, the stability upon
hydrogen release and uptake were studied by Sieverts measurements.
Received 3 March 2011
Received in revised form
16 May 2011
Accepted 23 May 2011
Available online 30 May 2011
Keywords:
Hydrogen storage
Solid solutions
Anion substitution
X-ray diffraction
Solid state diffusion
& 2011 Elsevier Inc. All rights reserved.
1. Introduction
for a range of new bimetal borohydrides, e.g. MSc(BH4)4 (M¼Li, Na
and K) [13–16], NaZn(BH4)3, MZn2(BH4)5 (M¼Li or Na) [17] as well
Developments of new environmentally friendly energy concepts
are essential for the future of the modern industrialized society and
hydrogen has been suggested as a future carrier of renewable
energy. Unfortunately, storage of hydrogen remains an unsolved
and extremely challenging task for a future hydrogen economy,
especially for utilization of hydrogen in mobile applications. Solid
state hydrogen storage as light element hydrides is a potential
solution, which may provide sufficiently high energy density and
efficiency [1,2]. Within this class of materials, the metal borohydrides
have received increasing interest in the past decade due to their
high gravimetric and volumetric hydrogen densities. Unfortunately,
many of these materials are too stable, i.e. decompose at too high
temperatures for practical applications [3–5]. Therefore, much
research has focused on synthesis of novel materials and on altering
the properties of known materials [6].
A wide range of novel metal borohydrides have been synthesized
and characterized structurally, physically and chemically. An empiri-
cal correlation between the decomposition temperature or the
enthalpy of formation and the Pauling electronegativity of the cation
has been found [7]. The studies of the novel metal borohydrides
suggest that the introduction of some degree of covalent character
in the M–BH4 interaction facilitates release of hydrogen at lower
temperatures and also the formation of more open structures [5],
e.g. Mg(BH4)2 [8], Mn(BH4)2 [9] and Y(BH4)3 [10–12]. This also holds
as the first structurally ordered heteroleptic borohydride, KZn
(BH4)Cl2 [18].
One approach to improve the properties of known borohydrides
is by chemical reactions as realized for reactive hydride composites,
for which the thermodynamic properties may be significantly
improved as observed for the 2LiBH4–MgH2 system [19–24].
Furthermore, the kinetic and possibly also thermodynamic proper-
ties of reactive hydride composites can be improved by nanocon-
finement or using catalytically active additives [25–27].
A recently developed method suggested for altering the thermo-
dynamic properties of borohydride materials is anion substitution,
where a halide anion i.e. Fꢀ, Clꢀ, Brꢀ or Iꢀ is substituted for either
Hꢀ or BH4ꢀ in the metal borohydride structure. Fast dissolution of
lithium chloride in the hexagonal structure of lithium borohydride,
h-LiBH4 followed by slow segregation of lithium chloride from the
solid solution Li(BH4)1ꢀxClx was recently observed [28,29]. The
heavier halides appear to stabilize lithium borohydride, however,
theoretical studies suggest that the more electronegative fluoride ion
may introduce destabilization [30]. Furthermore, three new solid
solutions have been found in the Ca(BH4)2–CaI2 system [31].
Dissolution of MCl into MBH4, M¼Li, Na or K, have also been
observed in samples containing novel metal borohydrides synthe-
sized by ball milling, where MCl or other chloride containing salts are
formed as a byproduct, e.g. MCl is slowly dissolved in MBH4 during
the thermal decomposition of MSc(BH4)4, M¼Na or K [15,16].
Therefore, anion substitution might also play an important role in
the understanding of the thermal decomposition of the novel metal
borohydrides in such samples.
n Corresponding author. Fax: þ45 8619 6199.
E-mail address: trj@chem.au.dk (T.R. Jensen).
0022-4596/$ - see front matter & 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2011.05.030

D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
1859
In this study we investigate the dissolution of NaCl in NaBH4.
We utilize two different dissolution methods: ball milling and
annealing. The mechanism for the dissolution and the degree of
anion substitution obtained by the different methods is studied
by Rietveld refinement of synchrotron radiation powder X-ray
diffraction (SR-PXD) data. Furthermore, the effect of the anion
substitution on the thermal stability is studied by Sieverts
measurements.
sealed inside a glovebox. The temperature was controlled with a
thermocouple placed in the sapphire tube 1 mm from the sample
[33]. All obtained raw images were transformed to 2D-powder
patterns using the FIT2D program [34], which was also used for
calibration using measurements of the standard NIST LaB6 sample
and masking diffraction spots from the single-crystal sapphire
sample holder. Uncertainties of the integrated intensities were
calculated at each 2y-point by applying Poisson statistics to the
intensity data, considering the geometry of the detector [35].
For the study of the anion substitution mechanism, sample S1
was heated from RT to 300 1C, kept at this temperature for 30 min
and cooled to 30 1C. Both heating and cooling were conducted with
heating and cooling rates of 10 1C/min. This procedure was repeated
three times in total in order to investigate the stability of the formed
solid solutions. To study the effect of the anion substitution on the
decomposition pathways samples S1A and S2A was heated from RT
to 600 1C with a heating rate of 10 1C/min.
2. Experimental
2.1. Sample preparation
Samples of NaBH4–NaCl in molar ratios of 0.5:0.5 and
0.75:0.25, denoted sample S1 and S2, respectively, and a sample
containing only NaBH4, denoted S3 were ball milled (BM) using
the same conditions comprised of 60 times 2 min of milling each
intervened by 2 min breaks to avoid heating of the samples. Ball
milling was conducted under inert conditions (argon atmosphere)
using a Fritsch Pulverisette P4 planetary mill using 80 mL tung-
sten carbide (WC) containers and a sample to balls mass ratio of
approximately 1:35 (balls WC, o.d. 10 mm). After ball milling,
approximately 0.5 g of each sample was transferred to a corun-
dum (Al2O3) crucible, placed inside a sealed argon filled quartz
capillary, and annealed in a furnace kept at a fixed temperature of
300 1C for 3 days. This is referred to as annealing (A). After
annealing the samples are denoted S1A, S2A and S3A, respectively.
Information about the sample composition and treatment is given
in Table 1. The chemicals used were: NaBH4 (Z98.5%, Aldrich)
and NaCl (499.5%, Ferak). The preparation and manipulation of
all samples were performed in an argon-filled glovebox with a
circulation purifier.
2.3. Laboratory powder X-ray diffraction (PXD)
In house PXD was measured for sample S1A 12 months after
annealing. The sample had been stored in the glove box at room
temperature (RT) during these 12 months. The PXD measurement
was performed in Debye–Scherrer transmission geometry using a
Stoe diffractometer equipped with a curved Ge(1 1 1) monochro-
mator (Cu Ka1 radiation) and curved position sensitive detector.
Data were collected at RT between 41 and 1271 2y with a counting
time of ꢁ960 s per step. The sample was mounted in a glovebox
in a 0.5 mm glass capillaries sealed with glue.
2.4. Rietveld refinements
Rietveld refinements were performed using the Fullprof pro-
gram [36]. The backgrounds were described by linear interpola-
tion between selected points, while pseudo-Voigt profile
functions were used to fit the diffraction peaks. In the refinements
of single SR-PXD datasets at selected temperatures, unit cell
parameters, zero shift, profile parameters and the overall tem-
perature factors, Bov were refined. The chloride substitution in
sodium borohydride was modeled by introducing Cl on the
position of B in NaBH4 and the occupancies of Cl, B and H were
refined. The overall occupancy of Cl and BH4 was constrained to
one. For the samples S1A and S2A the BH4 substitution in NaCl
was modeled in a similar way, introducing B on the position of Cl.
2.2. In situ synchrotron radiation powder X-ray diffraction
In situ SR-PXD was collected at beamline I711 of the synchro-
tron MAX II, Lund, Sweden at the research laboratory MAX-Lab
with a MAR165 CCD detector system and selected wavelengths of
˚
l
¼0.945(2) or 1.072(2) A [32]. The X-ray exposure time was 30 s.
The sample cell was specially developed for studies of gas/solid
reactions and allows elevated pressure and temperature to be
applied. The powdered sample was mounted in a sapphire (Al2O3)
single-crystal tube (o.d. 1.09 mm, i.d. 0.79 mm) in an argon-filled
glovebox with a circulation purifier. The sample holder was
Table 1
Sample composition prior to treatment, utilized treatment to obtain anion substitution (BM: ball milling for 120 min, A: Annealed at 300 1C for 3 days), sample
composition after treatment and unit cell parameter, a for the solid solutions, NaBH4 and NaCl. Substitution degrees, mol%, wt% and unit cell parameter after treatments
are determined by Rietveld refinement of SR-PXD data.
Sample
Before treatment
Composition
After treatment
Treatment
Composition
˚
a (A)
S1
NaBH4
NaCl
50 mol%
50 mol%
40 wt%
60 wt%
BM
Na(BH4)0.90Cl0.10
NaCl
50.3(5) mol%
49.7(4) mol%
40.9(5) wt%
59.1(5) wt%
6.1208(5)
5.6421(3)
S1A
NaBH4
NaCl
50 mol%
50 mol%
40 wt%
60 wt%
BMþA
Na(BH4)0.43Cl0.57
Na(BH4)0.10Cl0.90
NaCl
63.9(18) mol%
14.5(8) mol%
9.3(4) mol%
59.4(17) wt%
15.3(8) wt%
10.2(4) wt%
15.1(4) wt%
5.9579(2)
5.7800(11)
5.6876(7)
–
NaBO2
12.2(3) mol%,
S2A
NaBH4
NaCl
75 mol%
25 mol%
66 wt%
34 wt%
BMþA
Na(BH4)0.68Cl0.32
Na(BH4)0.20Cl0.80
NaBO2
80.0(12) mol%
8.9(4) mol%
74.6(11) wt%
10.1(5) wt%
15.3(6) wt%
6.03416(19)
5.6506(8)
–
11.1(4) mol%
S3
NaBH4
NaBH4
100 mol%
100 mol%
100 wt%
100 wt%
BM
NaBH4
NaBH4
100.0 mol%
100.0 mol%
100.0 wt%
100.0 wt%
6.15714(9)
S3A
BMþA
6.15928(11)

1860
D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
The hydrogen atoms were not included in the latter model, since
they are very weak X-rays scatters and the orientation of BH4
units i.e. the positions of the hydrogen might not be ordered.
For the in situ SR-PXD data measured for sample S1 a sequential
refinement was carried out by splitting the data into two series of
PXD no. 1 to 200 containing NaCl and Na(BH4)1ꢀxClx and PXD no.
200 to 422 containing NaCl, Na(BH4)1ꢀxClx and NaBO2. Small
amounts of NaBO2 is observed before scan 200, however, the
amounts are too small for reliable fits. The first pattern of each
series was refined as described above, however, with all linearly
interpolated background points varied and the preliminarily deter-
mined zero shifts fixed. For the sequential refinement all unit cell
parameters, profile parameters, overall temperature factors, Bov and
occupancies of Cl, B and H in Na(BH4)1ꢀxClx were refined.
The standard uncertainties of the occupancies given by the
Rietveld refinements are unrealistically low, i.e. below 1% of the
refined values. These are likely to be higher mainly due to the
correlation between the occupancies and the atomic temperature
factors not being taken into account. Therefore the standard
uncertainties of the substitution degrees are not given. Further-
more, the sample compositions extracted by Rietveld refinement
are only approximate, since these might be affected by presence
of amorphous phases and/or micro-adsorption.
of the solid solution formed by ball milling is Na(BH4)0.90Cl0.10.
This reflects that a reaction according to reaction scheme (1) takes
place during ball milling.
ð1ꢀxÞNaBH4 þxNaCl-NaðBH4Þ1ꢀxClx
ð1Þ
Reflections from NaCl exhibit similar 2
y
position and intensity
as compared to the sample prior to ball milling, which suggests
that no or very limited amounts of NaBH4 is dissolved in NaCl
during ball milling.
Rietveld refinement of SR-PXD data measured for sample S1
using a model without chloride substitution was also conducted
but resulted in a poorer fit giving agreement factors of Rp¼8.43
and Rwp¼10.3 as compared to those for the model implementing
chloride substitution of Rp¼4.55 and Rwp¼5.73 (corrected for
background). Furthermore, a Rietveld model implementing BH4
substitution in NaCl was also tried for sample S1, however, the
occupancy of Cl refined to close to zero, i.e. no measurable
substitution of NaBH4 in NaCl was observed due to ball milling.
The SR-PXD data and Rietveld refinement of the annealed
NaBH4–NaCl (0.5:0.5, S1A) sample (S1A) is shown in Fig. 1b. Visual
inspection of the SR-PXD data, Fig. 1a and b, suggests significant
differences between the two samples S1 and S1A, prepared by ball
milling and annealing, respectively. The chloride substituted sodium
˚
borohydride is found to have a unit cell parameter of a¼5.9579(2) A
and the composition Na(BH4)0.43Cl0.57, i.e. more than half the anions
in this solid solution are Clꢀ. This suggests that the formation of the
solid solution by reaction scheme (1) is more efficiently induced by
heating (annealing) than by pressure (ball milling).
2.5. Sieverts measurements
Sieverts measurements (temperature programmed desorption,
TPD) were recorded for samples S1A, S3 and sample S3A using a
PCTpro 2000 instrument from Hy-Energy [37]. The samples were
loaded in an autoclave and sealed in argon atmosphere. For the
first TPD experiment for each sample, a multistep temperature
program was used i.e. the sample was heated from RT to 450 1C
with a heating rate of 0.5 1C/min followed by a dwell time of 10 h
at 450 1C. Subsequently, the samples were heated to 500 1C with a
heating rate of 200 1C/min followed by a dwell time of 25 h at
500 1C. At last, the samples were heated to 550 1C with a heating
ramp of 200 1C/min and the measurement was continued at
550 1C until desorption was completed. The second and third
TPD experiments were performed with a continuous temperature
program from RT to 550 1C using a heating rate of 0.5 1C/min. For
all desorption experiments a hydrogen back pressure of 1 bar was
used to suppress the formation of diborane. Hydrogen absorp-
tions were conducted at a fixed temperature of 450 1C for
40–94 h. A hydrogen pressure of ca. 90 bar was initially applied.
In addition, a substitution of BH4ꢀ into NaCl is also found to
have occurred in sample S1A according to reaction scheme (2).
yNaClþð1ꢀyÞNaBH4-NaðBH4Þ1ꢀyCly
ð2Þ
This results in a solid solution of composition Na(BH4)0.10Cl0.90
˚
.
The unit cell parameter of a¼5.886(3) A is significantly larger
˚
than that found for the unsubstituted NaCl of a¼5.6421(3) A
(refined value using SR-PXD data for sample S1 at RT). The Bragg
peaks originating from Na(BH4)0.10Cl0.90 are relatively broad,
indicating that the degree of substitution is distributed with an
average composition y¼0.90.
The fact that BH4ꢀ substitution in NaCl is not observed in the
ball milled samples prior to annealing suggests that this process is
less favorable compared to Clꢀ substitution into NaBH4, which is
more favorable and hence occurs during ball milling. This might
be due to the larger radii of BH4ꢀ compared to Clꢀ i.e. the
compound with the smaller anion, NaCl is more easily dissolved
in the compound with the larger anion, NaBH4. This causes
contraction of the unit cell and increased lattice energy.
3. Results and discussion
Only 9.3 mol% unsubstituted NaCl is present in sample S1A, i.e.
the annealing causes the substitution reactions (reaction schemes
(1) and (2)) to take place to a high extent. Furthermore, sodium
borate, NaBO2 is observed to form (12.1 mol%) during annealing.
This may be due to water and/or oxygen contamination of the
sample. This may possibly be avoided by thoroughly drying the
starting materials prior to sample preparation to avoid presence
of hydrous compounds such as NaBH4 ꢂ 2H2O [39].
3.1. Substitution methods
Anion substitution is obtained using two different methods:
ball milling (sample S1) or ball milling and annealing, i.e. heating
the samples at a fixed temperature of 300 1C for 3 days (samples
S1A and S2A). The samples were examined by Rietveld refinement
of SR-PXD data (Table 1).
The ball milled sample NaBH4–NaCl (0.5:0.5, S1) reveals
substitution of chloride into the structure of sodium borohydride,
i.e. formation of a solid solution Na(BH4)1ꢀxClx (Fig. 1a). The anion
substitution causes a decrease in the unit cell parameter, due tꢀo
In order to investigate possible full substitution (100%), i.e.
formation of a single homogeneous solid solution, a sample
containing 25 mol% NaCl (see Table 1) was ball milled and
annealed (sample S2A) in a similar fashion as sample S1A. The
Rietveld refinement of the SR-PXD data shown in Fig. 1c reveal
formation of two solid solutions of composition Na(BH4)0.68Cl0.32
(80 mol%) and Na(BH4)0.20Cl0.80 (8 mol%) with unit cell para-
the smaller ionic radii of Clꢀ (1.81 A) as compared to that of BH4
˚
˚
(2.05(19) A) [38], i.e. the cell parameter for the Na(BH4)1ꢀxClx is
˚
found to be a¼6.1208(5) A. In comparison the unit cell parameter
˚
˚
found in a sample of ball milled NaBH4 (S3) is a¼6.15714(9) A
meters of a¼6.03416(19) and 5.6505(5) A, respectively. Again,
(Fig. 1d). Furthermore, significant changes in the relative peak
intensities are observed in the PXD data and dissolution of
10 mol% NaCl in NaBH4 is clearly observed, i.e. the composition
NaBO2 have formed (12 mol%). It is interesting that BH4ꢀ
substitution into NaCl is observed even though only 32 mol%
Clꢀ is substituted into NaBH4, which is far less than the 57 mol%

D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
1861
˚
Fig. 1. SR-PXD data measured at RT and Rietveld refinement profiles plotted as a function of 1/d for (a) NaBH4–NaCl (0.5:0.5, S1,
l¼1.072(2) A), (b) NaBH4–NaCl (0.5:0.5,
˚
˚
˚
S1A,
l¼0.945(2) A), (c) NaBH4–NaCl (0.75:0.25, S2A, l¼0.945(2) A) and (d) NaBH4 (S3, l¼0.945(2) A). Circles: SR-PXD data (Yobs); Upper curves: Rietveld refinement
profiles (Ycalc); Lower curves: Residual (Yobs–Ycalc); Marks: (from top) in (a): NaCl and Na(BH4)0.90Cl0.10, (b): NaCl, Na(BH4)0.10Cl0.90, Na(BH4)0.43Cl0.57 and NaBO2, (c):
Na(BH4)0.20Cl0.80, Na(BH4)0.68Cl0.32 and NaBO2 and (d): NaBH4.
observed for sample S1A. This suggests, that substitution of Clꢀ
for BH4ꢀ in NaBH4 is a fast process as compared to BH4ꢀ substitu-
tion in NaCl and that both reactions are diffusion limited. This
likely means, that the smaller amount of NaCl in the sample, the
smaller amount of Clꢀ substitution into NaBH4, before BH4ꢀ
substitution into NaCl is initiated. This suggests that prolonged
annealing may produce a single homogeneous solid solution
Na(BH4)1ꢀxClx. However, the fact that this is not obtained by
annealing at 300 1C for 3 days suggests that the diffusion of the
anions is relatively slow.
Fig. 2 shows a plot of the observed unit cell parameters of
Na(BH4)1ꢀxClx solid solutions at RT as a function of substitution
degree, x. The plot shows an approximate linear relationship
according to Vegard’s law. It should be noted that this relation-
ship is only approximate due to the difficulty of determining the
exact composition of the solid solutions due to correlation
between the occupancies and the atomic temperature factors. In
contrast, the unit cell parameters are very reliable experimental
observations. However, unit cell parameters of the solid solutions
formed from dissolution of NaBH4 into NaCl is not included, since
Fig. 2. Observed unit cell parameters of Na(BH4)1ꢀxClx solid solutions at RT as a
function of substitution degree, x.

1862
D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
3
˚
they exhibit broad Bragg reflections lying close to those of NaCl,
i.e. the cell parameters are likely to be less accurate.
identical (179.606(4) and 179.599(13) A , respectively). This sug-
gests that no dissolution of NaBH4 into NaCl takes place during the
short time of the experiment (only 90 min at 300 1C).
For the annealed sample NaBH4–NaCl (0.5:0.5, S1A) in house
PXD data was measured 12 month after annealing in order to
investigate possible solid state phase segregation. The sample still
contains NaCl and two solid solutions: Na(BH4)1ꢀxClx and
Na(BH4)1ꢀyCly with y4x. The relative small amount of NaBO2
observed in the SR-PXD data is not observed in the in house PXD
data. The quality of the data is not sufficient to determine the
degree of substitutions, x and y however, the cell parameters,
determined by Rietveld refinement, are found to be similar to
those observed directly after annealing. This suggests that no
significant phase segregation is taking place, which is in contrast
to the significant phase segregation observed for solid solutions
Li(BH4)1ꢀxClx, for which x decreases from 0.42 to 0.09 during 13
month at RT [29].
An initial expansion of the unit cell volume is also observed for
Na(BH4)1ꢀxClx during the first heating ramp from 32 to ꢁ240 1C. As
the temperature reaches 240 1C during the first heating ramp, the
unit cell volume of Na(BH4)1ꢀxClx decreases rapidly and continues
to do so for approximately 20 min. In this time range the unit cell
˚
parameter is found to decrease from 6.2099(3) to 6.1295(7) A, while
the degree of substitution, x changes from approximately 0.15 to
0.50. However, it should be noted that the substitution degrees at
elevated temperatures obtained from Rietveld refinement are only
approximate due to the correlation between the occupancies and
the atomic temperature factors.
During the last 10 min of the first annealing period at 300 1C
the substitution rate decreases significantly and reaches a plateau
value, i.e. the decrease in unit cell volume of Na(BH4)1ꢀxClx
becomes linear. During the second and third annealing periods
at 300 1C the unit cell volume of Na(BH4)1ꢀxClx decreases linearly
with similar rates. By fitting linear equations to the change in unit
cell parameter during these annealing periods (PXD no. 58–85,
165–208 and 289–332 corresponding to t¼46–57, 112–142 and
197–226 min, respectively), the unit cell parameter change is
3.2. Mechanism for anion substitution
In order to investigate the mechanism of the substitution process
during annealing, the ball milled sample NaBH4–NaCl (0.5:0.5, S1)
was investigated by in situ SR-PXD during three consecutive cycles
of heating to 300 1C, annealing for 30 min and cooling to RT. The
obtained data are shown in Fig. 3. The unit cell volumes of NaCl and
the solid solution Na(BH4)1ꢀxClx, obtained by Rietveld refinement of
422 PXDs, are shown as a function of time together with the
temperature profile for the experiment in Fig. 4.
found to be ꢀ5.53(8) ꢃ 10ꢀ4 A/min. By relating this to the unit
˚
cell volume change as a function of substitution degree (from
Fig. 2) the rate of NaCl substitution into NaBH4 at 300 1C is found
to be 0.105(3) mol% Cl/min. By similar calculations for the fast
substitution during the first 20 min of the first annealing period
(PXD no. 36–47 corresponding to t¼24–32 min), the initial
substitution rate is found to be 1.48(8) mol% Cl/min.
As the temperature of the sample is changed, the unit cell
volume of NaCl increases and decreases linearly due to thermal
expansion and contraction, respectively. The linear thermal expan-
During the temperature ramps expansion and contraction of the
solid solution Na(BH4)1ꢀxClx is observed. The linear thermal expan-
sion coefficients,
a for NaCl during the first heating ramp is
2.562(6) ꢃ 10ꢀ4 Kꢀ1 (T¼32–300 1C). The reported
a
for the same
sion coefficient,
a (T¼32–200 1C) is found to decrease with the
temperature range is 2.515(4) ꢃ 10ꢀ4 Kꢀ1 (calculated from the
equation for cell parameter, a for NaCl as a function of T reported
by Pathak and Vasavada [40]). No changes are observed in the unit
cell volume of NaCl during the three annealing, i.e. the unit cell
volumes at T¼32 1C at the start and end of the experiment are
increasing degree of substitution, x, i.e.
a
¼5.07(2) ꢃ 10ꢀ4
,
4.828(18) ꢃ 10ꢀ4, 4.70(3)ꢃ 10ꢀ4 and 4.55 ꢃ 10ꢀ4 Kꢀ1 for the four
ramping periods (t¼0–17, 68–102, 152–186 and 236–253 min,
respectively). For comparison, the linear thermal expansion coeffi-
cient for NaBH4 (extracted by Rietveld refinement of the in situ
Fig. 3. In situ SR-PXD data for BM NaBH4–NaCl (S1, 0.5:0.5) heated from RT to 300 1C, kept at 300 1C for 30 min, cooled from 300 to 30 1C (
D
T/D
t¼10 1C/min). This is
˚
J
repeated by three cycles. Symbols:
NaCl, K NaBH4 and & NaBO2
(
l
¼1.072(2) A).

D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
1863
Fig. 4. Unit cell volumes for NaCl (lower curve) and the solid solution Na(BH4)1ꢀxClx (upper curve) shown as function of time. The data are extracted from the in situ
SR-PXD data shown in Fig. 3 by Rietveld refinement. The temperature profile for the experiment is shown as the thin curve.
SR-PXD measured on sample S3, see Fig. 5b) is found to be
5.36(6) ꢃ 10ꢀ4 Kꢀ1 (T¼46–200 1C). Furthermore, the fact that the
decrease in unit cell volume of Na(BH4)1ꢀxClx is linear upon cooling
suggests that NaCl do not segregate from NaBH4 on the time scale of
the experiment.
At RT after the third annealing period the unit cell parameter
˚
of the solid solution has decreased to 5.9368(8) A and the final
composition is found to be Na(BH4)0.43Cl0.57. Interestingly, the
final composition of the solid solution, Na(BH4)0.43Cl0.57 is the
same as in the annealed sample, S1A. However, in the latter
sample substitution of BH4 into NaCl was also observed. This
suggests that substitution of BH4ꢀ into NaCl is slower than
substitution of Clꢀ into NaBH4. This is likely due to limited solid
state diffusion of the bigger anion, BH4ꢀ compared to that of Clꢀ
.
However, different nucleation and growth rates of the two solid
solutions might also hamper the substitution of BH4ꢀ into NaCl.
3.3. Effect of substitution on thermal decomposition
The decomposition pathway for the solid solution Na(BH4)1ꢀx
Clx was investigated using in situ SR-PXD measurements con-
ducted on the annealed NaBH4–NaCl samples (S1A and S2A) and
on the ball milled NaBH4 (S3) for comparison. Fig. 5a and b shows
the in situ SR-PXD for sample S1A and S3, respectively. The in situ
SR-PXD data for the thermal decomposition of sample S2A is
similar to that of sample S1A and is not included.
In situ SR-PXD data (Fig. 5b) for the ball milled NaBH4 (S3)
reveal formation of NaBO2 at ꢁ300 1C. At ꢁ495 1C the diffraction
from NaBH4 decreases rapidly and vanishes due to melting and
thermal decomposition. At this temperature the diffraction from
NaBO2 also vanishes, possibly due to dissolution in the NaBH4
melt. No diffraction is observed from decompositions products.
However, metallic Na might have formed as a dehydrogenation
product from NaBH4 due to the lack of a hydrogen back-pressure
during the dehydrogenation (NaH is reported to decompose at
ꢁ425 1C [3]). Sodium (Tm¼98 1C, Tb¼883 1C) might evaporate at
the relative high temperatures. The decomposition facilitates a
significant decrease in the background intensity, which indicates
that less material is exposed to the X-ray beam.
The sample of NaBH4–NaCl (0.5:0.5, S1A) after ball milling
and annealing contains as mentioned two solid solutions:
Na(BH4)0.43Cl0.57 and Na(BH4)0.1Cl0.9, besides NaCl and NaBO2. The
Fig. 5. In situ SR-PXD data for (a) NaBH4-NaCl (0.5:0.5, S1A) and (b) NaBH4 (S3)
measured from RT to 600 1C (
DT/
D
t¼10 1C/min). Symbols:
Na(BH4)1ꢀxClx, ’
˚
J
Na(BH4)1ꢀyCly,
NaCl, K NaBH4 and & NaBO2 (l
¼0.945(2) A).

1864
D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
diffracted intensity from NaCl shows a slow increase starting at
and for comparison; ball milled NaBH4 (S3) and annealed NaBH4
(S3A). The results are summarized in Table 2.
The first desorption measurement (TPD) is shown in Fig. 6.
For the annealed NaBH4 (S3A) a total hydrogen release of
2.6 wt% is observed. This corresponds to 32% of the calculated
ꢁ380 1C, simultaneously with broadening and small changes in the
2y
positions of the Bragg peaks from the two solid solutions. In fact,
the Bragg peaks from the two solid solutions move towards each
other and tend to merge at temperatures T4350 1C. This indicates
that a relatively fast solid state diffusion takes place and that
prolonged annealing at this temperature might result in formation
of a single homogeneous solid solution. As the temperature reaches
ꢁ380 1C the diffracted intensity from NaCl decrease, i.e. is consumed
by dissolution into the solid solutions and the diffracted intensity
from both Na(BH4)1ꢀxClx and Na(BH4)1ꢀyCly increase rapidly. Upon
further heating (T4415 1C) the compositions of the two cubic solid
solutions changes fast, i.e. the unit cell parameter, a for Na(BH4)1ꢀxClx
decreases due to increasing Clꢀ content, while a for Na(BH4)1ꢀyCly
increases due to increasing BH4ꢀ content. At ꢁ500 1C only one
Na(BH4)1ꢀxClx solid solution is present with relatively narrow Bragg
peaks, indicating that the solid solution is relatively homogeneous. At
ꢁ530 1C Na(BH4)1ꢀxClx begins to decompose simultaneously with
disappearing of the NaBO2 peaks. The decomposition results in
capacity,
rm(H2)¼7.992 wt%. The majority of the gas-release is
observed in the temperature range from 500 to 550 1C, however, a
small amount of gas is released in the temperature range
from RT to 300 1C. A sample of ball milled NaBH4 (S3) was
measured for comparison. A release of 3.6 wt% (45% of the
calculated capacity) is observed for the same period of time
(60 h). After 108 h a total of 6.6 wt% is released from the ball
milled NaBH4 (S3).
The sample of annealed NaBH4–NaCl (0.5:0.5, S1A) shows a
first desorption of 1.92 wt% hydrogen release after 60 h
of measurement (Tr550 1C). This corresponds to 61% of the
calculated capacity,
rm(H2)¼3.14 wt%. It should be noted that
the SR-PXD measurements as mentioned show that a small
amount of NaBO2 forms during annealing. Hence release of the
full calculated hydrogen capacity for this sample cannot be
expected.
changes of the 2y peak positions towards those for NaCl, which is
due to extraction of NaBH4 from the solid solution. Sodium borohy-
dride likely melts, dissolve NaBO2 and decompose to amorphous
boron and molten sodium, which may partly evaporate. After the
temperature reached 600 1C, the sample was cooled to RT and the
remaining solid solution was found to consists of 6.7 mol% BH4ꢀ, i.e.
The properties for the cycling of hydrogen release and uptake
were investigated by rehydrogenation of the annealed NaBH4
(S3A) and the sample of annealed NaBH4–NaCl (0.5:0.5, S1A) and
subsequently measuring the temperature desorption profiles.
Two such cycles were performed for each sample, i.e. the second
and third desorptions were measured (see Fig. 7). The annealed
NaBH4 (S3A) shows gas release of 1.2 and 1.0 wt% (15% and 13% of
the calculated capacity) for the second and third desorptions,
respectively. The second hydrogen desorption measurement
performed for the annealed NaBH4–NaCl (0.5:0.5, S1A) shows a
the composition is Na(BH4)6.7Cl93.4
.
3.4. Hydrogen release and uptake in the solid solution
Sieverts measurements were used to investigate the hydrogen
release and uptake for the annealed NaBH4–NaCl (0.5:0.5, S1A)
Table 2
Calculated hydrogen content, rm(H2) and observed hydrogen release measured by the Sieverts method for annealed NaBH4 (S3A), ball milled NaBH4 (S3) and annealed
NaBH4-NaCl (0.5:0.5, S1A). The observed hydrogen release is shown for three consecutive TPD measurements. The percentage of the full capacity and the absorption time
is written in the brackets. Experiments with a complete desorption is denoted END.
Sample
q
m(H2) (wt%)a
Cycle 1 obs. mass loss (wt%)
Cycle 2 obs. mass loss (wt%)
Cycle 3 obs. mass loss (wt%)
60 h
END
END
END
NaBH4 (S3A)
7.99
7.99
3.14
2.6 (32%)
3.6 (45%)
1.9 (61%)
–
1.2 (15%, Abs 47 h)
1.2 (15%, Abs 60 h)
2.8 (90%, Abs 94 h)
1.0 (13%, Abs 74 h)
1.4 (18%, Abs 56 h)
2.3 (74%, Abs 45 h)
NaBH4 (S3)
6.6 (82%)
1.9 (61%)
NaBH4–NaCl (S1A)
a
Assuming the decomposition reaction: NaBH4-NaHþBþ3/2H2.
Fig. 6. Temperature pressure desorption measurement of the 1. desorption cycle, conducted from RT to 450 1C (DT/Dt¼0.5 1C/min) for annealed NaBH4 (S3A, solid line),
ball milled NaBH4 (S3, dot-dash) and NaBH4-NaCl (0.5:0.5) (S1A, dashed line). The temperature profile is shown as dots. The hydrogen release (wt%) is shown on the left
axis and the temperature scale on the right.

D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
1865
Fig. 7. Temperature pressure desorption measurement of the 2 and 3 desorption cycles, conducted from RT to 450 1C (
NaBH4–NaCl (0.5:0.5) (S1A, dashed line). The temperature profile is shown as dots. The hydrogen release (wt%) is shown on the left axis and the temperature scale on the right.
DT/Dt¼0.5 1C/min) for annealed NaBH4 (S3A, solid line) and
total hydrogen release of 2.8 wt% (90% of the calculated capacity).
Thus the hydrogen release is greater in the second cycle com-
pared to the first, indicating that some hydrogen was released
during the annealing at 300 1C performed prior to the Sieverts
measurements. The third desorption shows a release of 2.3 wt%
H2 (74% of the calculated capacity), which is a decrease compared
to the second cycle. This indicates that either the sample was not
fully loaded with hydrogen during the 45 h of absorption or that
the reversibility is reduced with the number of hydrogen release/
uptake cycles performed.
References
[1] L. Schlapbach, Nature 460 (2009) 809.
[2] S.-I. Orimo, Y. Nakamori, J.R. Eliseo, A. Zu¨ttel, C.M. Jensen, Chem. Rev. 107
(2007) 4111.
[3] W. Grochala, P.P. Edwards, Chem. Rev. 104 (2004) 1283.
[4] A.
Z
u¨ttel, S. Rentsch, P. Fischer, P. Wenger, P. Sudan, Ph. Mauron,
Ch. Emmenegger, J. Alloys Compd. 356–357 (2003) 515.
[5] D.B. Ravnsbæk, Y. Filinchuk, R. Cerny, T.R. Jensen, Z. Kristallogr. 225 (2010) 557.
^
´
[6] L.H. Rude, T.K. Nielsen, D.B. Ravnsbæk, U. Bo¨senberg, M.B. Ley, B. Richter, L.M.
Arnbjerg, M. Dornheim, Y. Filinchuk, F. Besenbacher and T.R. Jensen, Phys.
Status Solidi, in press, doi:pssc.201001214.R1.
[7] Y. Nakamori, K. Miwa, A. Ninomiya, H.-W. Li, N. Ohba, S. Towata, A. Zu¨ttel,
S. Orimo, Phys. Rev. B 74 (2006) 045126.
ˇ
[8] Y. Filinchuk, R. Cerny´ , H. Hagemann, Chem. Mater. 21 (2009) 925.
ˇ
4. Conclusion
[9] R. Cerny´, N. Penin, H. Hagemann, Y. Filinchuk, J. Phys. Chem. C 113 (2009) 9003.
ˇ
[10] D.B. Ravnsbæk, Y. Filinchuk, R. Cerny´, M.B. Ley, D. Haase, H.J. Jakobsen,
J. Skibsted, T.R. Jensen, Inorg. Chem. 49 (2010) 3801.
[11] C. Frommen, N. Aliouane, S. Deledda, J.E. Fonneløp, H. Grove, K. Lieutenant,
I. Llamas-Jansa, S. Sartori, M.H. Sørby, B.C. Hauback, J. Alloys Compd. 496
(2010) 710.
Chloride substitution in sodium borohydride is investigated in
samples of NaBH4–NaCl. The substitution has been obtained by
mechanical (ball milling) and thermal (annealing) treatment and
the resulting solid solutions, Na(BH4)1ꢀxClx have been studied by
SR-PXD. Ball milling causes formation of Na(BH4)0.90Cl0.10, while
prolonged annealing results in solid solutions of higher Clꢀ
contents, e.g. Na(BH4)0.43Cl0.57. This suggests that the formation
of the solid solution is more efficiently induced by heating
(annealing) than by mechanical treatment (pressure). Further-
more, prolonged annealing results in dissolution of NaBH4 into
NaCl. The in situ SR-PXD study of the substitution mechanism
shows that fast dissolution of NaCl into NaBH4 is initiated at
ꢁ240 1C and that the substitution rate decreases after 20 min of
annealing at 300 1C, after which substitution proceeds at a
constant rate. These studies also indicate that the substitution
of BH4ꢀ into NaCl is taking place at a much lower rate. Investiga-
tion of the decomposition pathway of the Na(BH4)1ꢀxClx solid
solutions shows that at higher temperature, i.e. ꢁ415 1C the rate
of solid state diffusion is significantly increased such that a single
and homogeneous solid solution is formed before slow decom-
position of the NaBH4–part of the solid solution at ꢁ530 1C.
[12] T. Jaron, W. Grochala, Dalton Trans. 39 (2010) 160.
[13] H. Hagemann, M. Longhini, J.W. Kaminski, T.A. Wesolowski, R. Cerny´,
ˇ
N. Penin, M.H. Sørby, B.C. Hauback, G. Severa, C.M. Jensen, J. Phys. Chem. A
112 (2008) 7551.
[14] 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. A 112 (2008) 3164.
ˇ
´
[15] R. Cerny, G. Severa, D.B. Ravnsbæk, Y. Filinchuk, V. d’Anna, H. Hagemann,
D. Haase, C.M. Jensen, T.R. Jensen, J. Phys. Chem. C 114 (2010) 1357.
ˇ
´
[16] R. Cerny, D.B. Ravnsbæk, G. Severa, Y. Filinchuk, V. d’Anna, H. Hagemann,
D. Haase, J. Skibsted, C.M. Jensen, T.R. Jensen, J. Phys. Chem. C 114 (2010) 19540.
[17] D. Ravnsbæk, Y. Filinchuk, Y. Cerenius, H.J. Jakobsen, F. Besenbacher,
J. Skibsted, T.R. Jensen, Angew. Chem. Int. Ed. 48 (2009) 6659.
[18] D.B. Ravnsbæk, L.H. Sørensen, Y. Filinchuk, D. Reed, D. Book, H.J. Jakobsen,
F. Besenbacher, J. Skibsted, T.R. Jensen, J. Eur. Inorg. Chem. 2010 (1608).
[19] J.J. Vajo, S.L. Skeith, F. Mertens, J. Phys. Chem. B 109 (2005) 3719.
[20] U. Bo¨senberg, S. Doppiu, L. Mosegaard, G. Barkhordarian, N. Eigen, A. Borgschulte,
T.R. Jensen, Y. Cerenius, O. Gutfleisch, T. Klassen, M. Dornheim, R. Bormann, Acta
Mater. 55 (2007) 3951.
[21] Y.W. Cho, J. Shim, B. Lee, CALPHAD 30 (2006) 65.
[22] U. Bo¨senberg, D.B. Ravnsbæk, H. Hagemann, V. D’Anna, C.B. Minella,
C. Pistidda, W. van Beek, T.R. Jensen, R. Bormann, M. Dornheim, J. Phys.
Chem. C 114 (2010) 15212.
[23] T.E.C. Price, D.M. Grant, V. Legrand, G.S. Walker, J. Int., Hydrogen Energy 35
(2010) 4154.
[24] T.E.C. Pricea, D.M. Granta, I. Telepenia, X.B. Yub, G.S. Walker, J. Alloys Compd.
472 (2009) 559.
[25] T.K. Nielsen, U. Bo¨senberg, R. Gosalawit, M. Dornheim, Y. Cerenius, F.
Besenbacher, T.R. Jensen, ACS Nano, in press, doi:10.1021/nn1006946.
[26] U. Bo¨senberg, J.W. Kim, D. Gosslar, N. Eigen, T.R. Jensen, J.M. von Colbe,
Y. Zhou, M. Dahms, D.H. Kim, R. Gu¨nther, Y.W. Cho, K.H. Oh, T. Klassen,
R. Bormann, M. Dornheim, Acta Mater. 58 (2010) 3381.
Acknowledgment
The authors are grateful to the Danish National Research Founda-
tion (Centre for Materials Crystallography), the Danish Strategic
Research Council (Centre for Energy Materials), The European
Commission (contract NMP-2008-261/FLYHY), the Danish Research
Council for Nature and Universe (Danscatt) and the Carlsberg
Foundation for funding. We thank the staff of beamline I711, MAX-
Lab for experimental support and for providing the beam time.
[27] T.K. Nielsen, F. Besenbacher, and T.R. Jensen, Nanoscale (2011) 10.1039/
C0NR00725K.
[28] L. Mosegaard, B. Møller, J.E. Jørgensen, Y. Filinchuk, Y. Cerenius, J.C. Hanson,
E. Dimasi, F. Besenbacher, T.R. Jensen, J. Phys. Chem. C 112 (2008) 1299.
[29] L.M. Arnbjerg, D.B. Ravnsbæk, Y. Filinchuk, R.T. Vang, Y. Cerenius, F. Besenbacher,
J.E. Jørgensen, H.J. Jakobsen, T.R. Jensen, Chem. Mater. 21 (2009) 5772.
[30] L. Yin, P. Wang, Z. Fang, H.M. Cheng, Chem. Phys. Lett. 450 (2008) 318.

1866
D.B. Ravnsbæk et al. / Journal of Solid State Chemistry 184 (2011) 1858–1866
[31] L.H. Rude, Y. Filinchuk, M.H. Sørby, B.C. Hauback, F. Besenbacher, T.R. Jensen,
J. Phys. Chem. C (2011) 10.1021/jp111473d.
[35] S. Vogel, L. Ehm, K. Knorr, G. Braun, Adv. X-ray Anal. 45 (2002) 31.
[36] J. Rodriguez-Carvajal, FULLPROF SUITE, (2003) LLB Sacley & LCSIM Rennes, France.
[37] PCTPro-2000—Calorimetry and thermal analysis, /http://www.setaram.
com/PCTPro-200.htmS.
˚
[32] Y. Cerenius, K. Sta˚ hl, L.A. Svensson, T. Ursby, A. Oskarsson, J. Albertsson,
A. Liljas, J. Synch. Rad. 7 (2000) 203.
[33] T.R. Jensen, T.K. Nielsen, Y. Filinchuk, J.-E. Jørgensen, Y. Cerenius, E. Mac,
A. Gray, C.J. Webb, J. Appl. Cryst. (2010) 43.
[38] D.R. Lide, CRC Handbook of Chemistry and Physics, 88th ed., CRC, Boca Raton,
2007, pp. 12–27.
[34] A.P. Hammersley, S.O. Svensson, M. Hanfland, A.N. Fitch, D. Ha¨usermann,
High Pressure Res. 14 (1996) 235.
[39] Y. Filinchuk, H. Hagemann, Eur. J. Inorg. Chem. (2008) 3127.
[40] P.D. Pathak, N.G. Vasavada, Acta Cryst. A26 (1970) 655.