IOP PUBLISHING
NANOTECHNOLOGY
Nanotechnology 20 (2009) 275704 (5pp)
doi:10.1088/0957-4484/20/27/275704
Nanostructures of LiBH4: a
density-functional study
P Vajeeston1, P Ravindran and H Fjellvåg
Center for Materials Science and Nanotechnology, Department of Chemistry,
University of Oslo, Box 1033 Blindern, N-0315, Oslo, Norway
E-mail: ponniahv@kjemi.uio.no
Received 29 January 2009, in final form 15 May 2009
Published 17 June 2009
Online at stacks.iop.org/Nano/20/275704
Abstract
The phase stability and electronic structure of α -LiBH4 -derived nanostructures and possible low
energy surfaces of thin films have been investigated using the ab initio projected augmented
plane wave method. Structural optimizations based on total energy calculations predicted that,
for the α -LiBH4 phase, the (010) surface is the most stable of the possible low-energy surfaces.
The predicted critical sizes of the nano-cluster and nano-whisker for α -LiBH4 are 1.75 and
1.5 nm, respectively. Similarly, the bond distances in the surfaces of a nano-whisker are found
to be higher than that in the bulk material. The calculated hydrogen site energies suggest that it
is relatively easier to remove hydrogen from the surface of the clusters and nano-whiskers than
from bulk crystals.
(Some figures in this article are in colour only in the electronic version)
was lowered by about 30 ◦ C by partial cation substitution of
Li by Mg, which has a larger electronegativity. The hydrogen
evolution processes for most of the above-mentioned systems
utilize a similar general procedure, i.e. thermal decomposition
of LiBH4 destabilized by additives or their derivatives. Vajo
et al [7] revealed another approach to utilize the huge
amount of hydrogen in LiBH4 . They mechanically milled
LiBH4 + 1/2MgH2 together with a small amount of TiCl3
catalyst, thus producing a system that can reversibly store
up to 8.1 wt% hydrogen. Moreover, the adjustment of the
reaction pathway resulted in a decrease of 25 kJ mol−1 of
H2 in the hydrogenation/dehydrogenation enthalpy compared
with that of pure LiBH4 . Recent efforts have focused
on the reaction mechanism [8, 9], incorporating additives,
such as metals [5, 10], metal halides [11, 12] oxides [11],
sulfides [12], hydrides [7, 10], or, more recently, nanoporous
scaffolds [12, 13], to thermodynamically destabilize LiBH4
toward optimized (lowered) desorption temperatures. In order
to use LiBH4 as an energy carrier in mobile applications
one has to find the possible ways to decrease the hydrogen
desorption temperature further. In the present study we have
investigated the role of particle size and the nanophase effect
on stability, electronic structure, and chemical bonding in
LiBH4 .
1. Introduction
A safe, efficient, and affordable way to store hydrogen
still presents a major challenge to the attainability of a
viable hydrogen-based economy. Among various hydrogen
storage materials currently under study, complex hydrides
have attracted considerable interest since the discovery by
Bogdanovic and Schwickardi that a small amount of TiCl3
doped into NaAlH4 could facilitate accelerated and reversible
hydrogen release under moderate conditions [1, 2]. Because
of the large gravimetric capacity of hydrogen (18.3 wt%),
LiBH4 is regarded as one of the promising candidates for
safe and efficient hydrogen storage. However, it suffers from
unfavorable thermal stability and requires a temperature of
about 400 ◦ C to desorb hydrogen over 0.1 MPa [3].
Zuttel et al [4] claimed that the onset decomposition
temperature for LiBH4 can be reduced to around 200 ◦ C by
mixing LiBH4 with SiO2 . A recent investigation by Au et al [5]
showed that mechanical milling of LiBH4 with selected metal
oxides or metal chlorides not only produces a destabilized
hydride but also a reversible hydrogen storage system. With the
aid of an additive, the decomposed LiBH4 could be recharged
at 600 ◦ C under 7 MPa hydrogen. Another study by Orimo
et al [6] reported that the decomposition temperature for LiBH4
1 http://folk.uio.no/ponniahv.
0957-4484/09/275704+05$30.00
1
© 2009 IOP Publishing Ltd Printed in the UK
Nanotechnology 20 (2009) 275704
P Vajeeston et al
Table 1. Calculated surface energy (in J m−2 ) for LiBH4 in different
possible low-energy surfaces.
2. Methods
The quantum-mechanical calculations have been performed in
the frame of density-functional theory using the generalized
gradient approximation (GGA) [14] as implemented in the
VASP code [15]. The interaction between the ion and electron
is described by the projector augmented wave method [16].
For the present calculations we have used a plane wave cutoff
energy of 500 eV. The k-points were generated using the
Monkhorst–Pack method with a grid size of 8 × 8 × 12
and 8 × 8 × 1, for the bulk and surfaces, respectively.
Iterative relaxation of atomic positions was stopped when
the change in total energy between successive steps was less
than 1 meV/cell. With this criterion, the forces generally
−1
acting on the atoms were found to be less than 0.1 eV Å .
Nano-clusters and nano-whiskers of different sizes have been
constructed from an optimized bulk phase with respect to
different supercell sizes. The k-points were generated using
the Monkhorst–Pack method with a grid size of 1 × 1 × 1
and 2 × 2 × 1 for structural optimization of nano-clusters and
nano-whiskers, respectively. During the construction of the
nano-clusters/whiskers the LiBH4 stoichiometry was always
maintained. For nano-whisker construction the vacuum is
included only in the x and y directions. The vacuum thickness
considered was wide enough to prevent whisker-to-whisker or
cluster-to-cluster interactions and we found that a width of
12 Å was sufficient to ensure that the energy was converged
to less than 1 meV/atom.
Surface energy
Direction
Present
From [17]
(010)
(101)
(100)
(011)
(111)
(201)
(001)
(110)
0.110
0.112
0.115
0.212
0.231
0.303
0.336
1.210
0.119
0.125
0.116
0.347
them. The surface energy of a crystal can be calculated using
the following equation
E surf (n) =
E tot (n) − E bulk (n)
2A
(1)
where E tot and A are the total energy and total surface area,
respectively. E bulk refers to the energy of the bulk α -LiBH4
system containing the same number of molecular units in the
slab. Since the constructed supercell of the slab has two
surfaces, the energy difference is normalized by twice the area
of each surface in equation (1). From the calculated surface
energy as a function of layer thickness we have found that
in all the studied thin film geometries seven to nine layers
supercell (depending upon the surface) is sufficient to get the
well converged surface energy.
The calculated surface energies vary from 0.11 to
1.2 J m−2 (depending upon the surface; see table 1) and the
magnitudes are in the following sequence: (010) < (101) <
(100) < (011) < (111) < (201) < (001) < (110).
Further, the calculations show that the surface energy is almost
the same for the (010), (100), and (101) surfaces and the
variation with respect to (010) is only 0.005 and 0.002 J m−2
for (100) and (101) surfaces, respectively. This finding is
consistent with the other theoretical investigation [17]. In the
present study we have optimized the surfaces globally, while
in [17] only a few layers of surface atoms were allowed to
relax and the atoms in the center of the slab were fixed during
the structural optimization, which makes a difference in the
calculated surface energies. The present calculations suggest
that the 110 surface has a much higher surface energy than
the other surfaces. Hence, we believe that one can remove
the H from 110 surface relatively more easily than from other
surfaces. The creation of these surfaces is associated with
the breaking of H–Li bonds in bulk LiBH4 . The low surface
energy of these surfaces indicates that the energy cost to create
these surfaces is much lesser than that in α -MgH2 [18]. It
should be noted that the BFDH method did not list the lowenergy (010) surface, but listed only the other low-energy
surfaces. This is because the BFDH method only uses the
crystal lattice and symmetry to generate a list of possible
growth faces and does not take into account the energetics of
the system. The stronger the bonding effects in the crystal,
the less accurate the method becomes. In many cases one can
get good approximations, and the method is always useful for
identifying important faces in the growth process. However,
3. Results and discussion
At ambient conditions LiBH4 crystallizes with an orthorhombic structure (α -LiBH4 ) in which each [BH4 ]− anion is surrounded by four lithium Li+ cations and each Li+ by four
[BH4 ]− , both in tetrahedral configurations. In this study we
have concentrated on the ambient condition α -LiBH4 phase.
The possible low energy surfaces were identified with the help
of the Bravais–Friedel-Donnay–Harker (BFDH) method implemented in the MS Modeling package (version 4.2). The
main reason to use the BFDH method is to obtain a rough estimate of the faces that are likely to be important for the crystal
habit. This information has been used to pre-screen the face
list used as an input to more sophisticated VASP calculations.
According to the BFDH calculation (001), (101), (100), (201),
and (111) are possible low-energy surfaces. In order to validate the BFDH method we have also cleaved other possible
low-index (010), (011), and (110) surfaces. For the considered
surface models we have included an integer number of LiBH4
formula units and they are thus stoichiometric. We have also
avoided generating surface models that are significantly polar
and therefore artificially stable due to long-range electrostatic
forces.
For the surface calculations the unrelaxed slabs have been
cut from the optimized bulk crystal, where bulk structures have
been fully relaxed with respect to stress in the cell and forces
acting on each atom. All atoms in such created slabs have been
allowed to relax using the minimization of forces acting on
2
Nanotechnology 20 (2009) 275704
P Vajeeston et al
Li
B
H
(b)
(a)
Figure 1. Optimized stable (a) nano-whisker (size 3.2 nm) and (b) nano-cluster (size 2.57 nm) of LiBH4 derived from the α -LiBH4 structure.
to identify the correct low-energy surfaces one must perform
ab initio total energy calculations for all possible surfaces.
The optimized stable nano-whisker and nano-cluster are
displayed in figure 1. In order to identify the critical particle
size, we have calculated the total energy as a function of
the cluster size, as shown in figure 2. From figure 2 it
is evident that if the cluster size decreases the total energy
becomes more positive (i.e. the formation energy decreases
with decrease in the cluster size). In particular there is a
steep increase in the total energy when the size of the cluster
is below 1.75 nm. Similarly, the calculated total energy as a
function of nano-whisker diameter shows that when we reduce
the diameter below 1.5 nm the nano-whiskers become highly
unstable (not shown in figure). If one reduces the cluster
size and nano-whisker diameter, the formation energy of the
clusters/nano-whisker becomes less negative, indicating that
one can destabilize LiBH4 by preparing it in nano-phases.
This is a good indication for reducing the decomposition
temperature which is much needed to utilize complex hydrides
for energy storage applications. The surface-to-volume ratio
increases upon decreasing the cluster/nano-whisker size. Since
the surface atoms have a lower coordination, the average
number of bonds is lower for smaller clusters. For the MgH2
clusters a similar trend in desorption energy versus cluster
size was found with the DFT method [19]. Moreover, in
ultra small clusters and whiskers the hydrogen atoms are
generally found to occupy the less stable top and bridge sites at
the surfaces compared to the more stable three-dimensionally
coordinated sites commonly found in thicker clusters/whiskers
(diameter above 1.75/1.5 nm, respectively). The calculated
B–H distances versus number of bonds (the size used is
2.57 nm and 3.2 nm for clusters and whiskers, respectively;
see figure 3) for the relaxed biggest clusters/whiskers indicate
Figure 2. Calculated total energy (in eV/f.u.) as a function of LiBH4
cluster size (in nanometers).
that the values were very scattered compared to that in the
bulk phase. In particular several B–H bonds have longer bond
distances than in the bulk. This type of structural arrangement
is expected in nano- and amorphous phases with no threedimensional crystallinity. From figure 3 it is clear that most
of the B–H bonds have an interatomic distance of 1.23 Å,
corresponding to the B–H distance in the bulk LiBH4 .
It should be noted that when we increase the cluster/nanowhisker size above the critical size these nano-objects will
have core LiBH4 structural units which makes them quite
stable. Hence, one must reduce the particle size beyond the
critical size in order to easily remove the H from the LiBH4
3
Nanotechnology 20 (2009) 275704
P Vajeeston et al
Calculated H site energy (kJ/mol)
54
30.26
Bulk
Figure 3. Calculated interatomic distances between B and H in the
optimized LiBH4 nano-clusters (size 2.57 nm) and whisker (size
3.2 nm). The arrow mark indicates the theoretical B–H distance for
the bulk LiBH4 phase.
52.96
H4
H3
13.10
H2
8.39
H1
Cluster
53.12
29.85
15.19
8.95
Whisker
Figure 4. Schematic representation of the calculated hydrogen site
energies in the optimized LiBH4 nano-cluster (size 2.09 nm) and
whisker (size 2.25 nm) compared with that in bulk material. H1, H2,
H3, and H4 refer to H in between Li; H between Li and B; H
connected with B; and H in center of the cluster, respectively. All the
values are given in kJ mol−1 .
particles. To substantiate this observation we have calculated
the H site energy (HSE; E )in these nano-phases. In the
nano-clusters/whiskers H is situated in four different chemical
environments, namely, at the surface: H1 (in between Li);
H2 (between Li and B); H3 (H connected with B); and H4
(center of the cluster). The H site energy is calculated in the
following manner E = (E Hvac + 12 E Hmol ) − (E nano ), where
E Hvac and E nano refers to the energy of the nano-object with
and without a H vacancy, E Hmol is the total energy of a free H2
molecule calculated in a large box. For the HSE study we used
optimized sizes of 2.09 and 2.25 nm for clusters and whiskers,
respectively. The calculated HSE for clusters/whiskers are
scattered in wide energy range (see figure 4); which is highly
dependent upon the environment of the H sites. The calculated
H1, H2, H3, and H4 site energy values for the cluster are 8.93,
13.10, 30.26, and 52.96 kJ mol−1 , respectively. Similarly,
for whiskers, the calculated HSE values for H1, H2, H3,
and H4 sites are 8.95, 15.19, 29.85, and 53.12 kJ mol−1 ,
respectively. The corresponding HSE value in the bulk phase is
54 kJ mol−1 . This clearly indicates that the energy required to
remove H from the center of the nanophase (cluster/whisker)
is similar to that in the bulk material. Moreover, the small
values of hydrogen site energies in the surfaces of the nanophases compared with that in bulk material indicate that one
can remove hydrogen relatively easily from the nano-phases.
Similar to the bulk phase all the studied nano-phases and the
surfaces also have non-metallic character and the calculated
band gap ( E g ) value varies between 4.4 and 7.02 eV (7.07,
6.71, 5.6, and 4.6 eV for bulk, surface, whisker, and cluster,
respectively). In general, for small nanoparticles the band
seems markedly smaller than the bulk gap. Normally, very
small semiconductor particles show a higher E g value than
bulk band gaps due to quantum confinement. However, for
the nanowire we have obtained a lower E g value than the that
in bulk for the MgH2 phase [18]. Similarly, the nano-phases
based on LiBH4 also have a smaller E g value than the bulk
material. This might be expected because, for the carbon nanotubes, when we move from the ultra small wire to the bulk wire
the electronic structure change from metal → semiconductor
→ insulator. The present study suggests that due to weakening
of bonds in the outer surfaces of the nano-phases of LiBH4 the
calculated band gap values in the nano-phases are found to be
smaller than in the bulk materials.
4. Conclusions
In summary, a theoretical study of the possible low-energy
surfaces and stability of nano-clusters and nano-whiskers was
conducted using ab initio total energy calculations. The
calculated surface energy of the low-index surfaces shows that
the (010) surface is the most stable surface in LiBH4 . We
have predicted that the critical size of the nano-cluster and
nano-whisker of LiBH4 is 1.75 and 1.5 nm, respectively. If
one reduces the diameter below these critical sizes the stability
of the cluster/nano-whisker is drastically reduced. We have
identified that in such objects most of the atoms are exposed
to the surface. The bonding interactions in surface layers are
considerably weaker than at the center of the cluster/whisker.
As a result, one can expect that the removal of hydrogen from
the surface of the nano-phases is much easier than from the
bulk or from the inner part. In order to use LiBH4 as a hydrogen
4
Nanotechnology 20 (2009) 275704
P Vajeeston et al
storage materials one must reduce the particle size below the
critical size or try to find alternative ways to weaken the B–
H bond, for example by appropriate catalysts, mixed powders,
storing in carbon scaffolds etc.
[6] Orimo S, Nakamori Y, Kitahara G, Miwa K, Ohba N,
Towata S and Züttel A 2005 J. Alloys Compounds
404–406 427
[7] Vajo J J and Skeith S L 2005 J. Phys. Chem. B 109 3719
[8] Kang J K, Kim S Y, Han Y S, Muller R P and Goddard W A
2005 Appl. Phys. Lett. 87 111904
[9] Lodziana Z and Vegge T 2004 Phys. Rev. Lett. 93 145501
[10] Yang J, Sudik A and Wolverton C 2007 J. Phys. Chem. C
111 19134
[11] Au M and Jurgensen A 2006 J. Phys. Chem. B 110 7062
[12] Vajo J J, Salguero T T, Gross A F, Skeith S L and Olson G L
2007 J. Alloys Compounds 446/447 409
[13] Gross A F, Vajo J J, Atta S L V and Olson G L 2008 J. Phys.
Chem. C 112 5651
[14] Perdew J P 1991 Electronic Structure of Solids ed P Ziesche
and H Eschrig (Berlin: Akademie) p 11
Perdew J P, Burke K and Wang Y 1996 Phys. Rev. B 54 16533
Perdew J P, Burke S and Ernzerhof M 1996 Phys. Rev. Lett.
77 3865
[15] Kresse G and Hafner J 1993 Phys. Rev. B 47 558
Kresse G and Furthmuller J 1996 J. Comput. Mater. Sci. 6 15
[16] Blöchl P E 1994 Phys. Rev. B 50 17953
Kresse G and Joubert J 1999 Phys. Rev. B 59 1758
[17] Ge Q 2004 J. Phys. Chem. A 108 8682
[18] Vajeeston P, Ravindran P and Fjellvåg H 2008 Nanotechnology
19 275704
[19] Wagemans R W P, van Lenthe J H, de Jongh P E,
van Dillen A J and de Jong K P 2005 J. Am. Chem. Soc.
127 16675
Acknowledgments
This work was funded by European Union seventh framework
program under the ‘NanoHy’ (grant agreement no.: 210092)
project. PV gratefully acknowledge the Research Council
of Norway for providing the computer time at the
Norwegian supercomputer facilities and Karel Knı́žek for
useful communications.
References
[1] Bogdanovic B and Schwickardi M 1997 J. Alloys Compounds
253 1
[2] Bogdanovic B, Brand R A, Marjanovic A, Schwickardi M and
Tolle J 2000 J. Alloys Compounds 302 36
[3] Züttel A, Wenger P, Rentsch S, Sudan P, Mauron Ph and
Emmenegger Ch 2003 J. Power Sources 118 1
[4] Züttel A, Rentsch S, Fischer P, Wenger P, Sudan P,
Mauron Ph and Emmenegger Ch 2003 J. Alloys Compounds
356/357 515
[5] Au M, Jurgensen A and Zeigler K J 2006 Phys. Chem. B
110 26482
5