PHYSICAL REVIEW B 73, 094122 共2006兲
Structural phase stability and bonding behavior of BAlH5 „B = Mg, Ba…
from first-principles calculations
A. Klaveness,* P. Vajeeston, P. Ravindran, H. Fjellvåg, and A. Kjekshus
Department of Chemistry, University of Oslo, Box 1033 Blindern, N-0315 Oslo, Norway
共Received 24 October 2005; revised manuscript received 9 January 2006; published 28 March 2006兲
The ground-state structures of MgAlH5 and BaAlH5 have been subjected to full structural optimization
considering 50 different potential atomic arrangements as inputs for accurate density-functional total-energy
calculations. The experimentally known crystal structure and structural parameters for BaAlH5 are reproduced,
and the crystal structure of MgAlH5 is predicted. At 0 K and ambient pressures MgAlH5 and BaAlH5 crystallize in monoclinic 共CaFeF5 type, P21 / c兲 and orthorhombic 共prototype, Pna21兲 structures, respectively. In
addition to the ground-state MgAlH5 phase 共here designated ␣-MgAlH5兲, it is also predicted a metastable
modification 共termed ␤-MgAlH5, CaCrF5 type, Cc兲. The structures comprise isolated, highly distorted AlH6
octahedra, which form one-dimensional chains along the 关001兴 direction. In ␣- and ␤-MgAlH5 these chains are
fairly linear, while BaAlH5 exhibits distinct zigzag chains. ␣-MgAlH5 and BaAlH5 are nonmetallic phases with
estimated band gaps of 2.48 and 2.73 eV, respectively. Analyses of the density of states, charge density,
Mulliken population, and Born effective charge indicate that the interaction between Al and H is polar covalent
blended with an ionic woof, while Ba and Mg can be considered as virtually divalent ions.
DOI: 10.1103/PhysRevB.73.094122
PACS number共s兲: 81.05.Je, 71.15.Nc, 71.20.⫺b
I. INTRODUCTION
The future for energy storage appears to follow a hydrogen path. The lack of safe, cheap, and lightweight storage
materials which can be used to construct standardized storage units is one of the great obstacles which has to be overcome. Currently, pressurized or liquidized hydrogen are the
chosen means of hydrogen storage in prototype cars. Solidstate storage has been heavily researched, but such materials
are currently characterized as impractical, from constructiontechnical, financial, and energy-conserving points of view.
This has led to a search for new unexplored classes of hydrogen storage materials, such as carbon nanotubes, carbon
nanofibers, microporous substances, and complex hydrides.
Complex hydrides have higher hydrogen-storage capacity at
moderate temperatures and pressures as well as lower cost
than conventional hydride systems based on intermetallic
phases. However, a serious problem with these materials is
poor kinetics and lacking reversibility with respect to hydrogen absorption and desorption. Improved understanding of
the processes which occur in these materials during uptake
and release of hydrogen is of considerable interest. Recent
experimental evidence shows that mechanochemical processing under ambient conditions in the presence of certain
transition-metal catalysts1–4 is able to increase the absorption
and desorption properties for certain materials. On this background, alkali- and alkaline-earth-metal aluminum hydrides
present themselves as possible on-board hydrogen-storage
materials for the future. However, various basic processing
issues need to be addressed before these hydrides and derivatives thereof can find practical use as reliable solid-state
hydrogen-storage media. An improved understanding of the
hydrogen absorption and desorption of such phases requires
reliable and detailed crystal-structure data.
The crystal structures of binary hydrides have been frequently studied and are by and large well characterized. On
1098-0121/2006/73共9兲/094122共7兲/$23.00
turning to ternary hydrides, however, the amount of knowledge is considerably less extensive, and for quaternary and
multicomponent phases the structural knowledge is poor.
Owing to the often complex structural arrangements and difficulties involved in establishing hydrogen positions in intermetallic matrices 共by x-ray diffraction methods兲, structural
information on hydrides is very limited. With constituents
from groups I to III of the periodic table, one can formulate
several series of hydrides which deserve further attention. In
this and following contributions, focus will be on the BCH5
共B = group-II element, C = group-III element兲 series, which
represent a continuation of previous studies5–12 of the structural stability of the BH2 and ACH4 共A = group-I element兲
series, where a number of phases were identified. There are
few experimental facts available on BCH5 phases. In this
article we report on the crystal structures and bonding properties of MgAlH5 and BaAlH5, of which only the BaAlH5
structure is properly established by experiment 共determined
for BaAlD5 by powder neutron diffraction13兲. BaAlH5 is reported to form as an impure phase on hydrogenation of
Ba7Al13,13 whereas MgAlH5 is reported to form from
Mg2共AlH4兲2 upon heating.14 The decomposition of
Mg共AlH4兲2 is said to occur in three stages according to the
reaction scheme
120–155 °C
2Mg共AlH4兲2
→
2MgAlH5 + 2Al + 3H2 ,
共1兲
2MgH2 + 2Al + 3H2 ,
共2兲
210–255 °C
2MgAlH5
→
380–420 °C
MgH2 + 2Al
→
MgAl2 + H2 .
共3兲
However, repeated15 experiments were unable to reproduce
these findings. One of the specific motivations for this study
094122-1
©2006 The American Physical Society
PHYSICAL REVIEW B 73, 094122 共2006兲
KLAVENESS et al.
was to identify the ground-state structure for MgAlH5, which
enters as a key piece in this reaction sequence.
II. COMPUTATIONAL DETAILS
Total energies and density of states 共DOS兲 have been calculated by the projector-augmented-wave 共PAW兲 method16
as implemented by the Vienna ab initio simulation package
共VASP兲.17 The generalized gradient approximation 共GGA兲 of
Perdew, Burke, and Ernzerhof18,19 is used to obtain accurate
exchange and correlation energies for a particular atomic
configuration. The atoms are relaxed toward equilibrium until the Hellmann-Feynman forces are less than 10−3 eV Å−1.
Brillouin-zone integration is performed with a Gaussian
broadening of 0.1 eV during all relaxations. For the experimentally reported13 BaAlH5 structure we have used 150 k
points in the whole Brillouin zone. A similar density of k
points was used for the other structures considered. All calculations are performed with a 500-eV plane-wave cutoff. In
order to avoid ambiguities regarding the free-energy results
we have always used the same energy cutoff and a similar
k-grid density for convergence for all the considered structure variants. The present type of theoretical approach has
been successfully applied for a large variety of hydrides5,6,8
to reproduce ambient- and high-pressure phases. Values for
the bulk modulus have been obtained using a universal equation of state20 to fit the total energy versus volume relationship. The Born-effective-charge tensors were calculated with
VASP utilizing Berry-phase calculations with the help of locally developed codes for pre- and post-processing.
Charge densities and Mulliken populations have been calculated by the Hartree-Foch self-consistent-field 共SCF兲
method as implemented by the CRYSTAL code.21 Basis sets
with 8-61G,22 3-1-1G,23 85-11G,24 and 5-11G25 have been
used for Mg, Ba, Al, and H, respectively.
Fifty composition-related potential structure types were
considered for the theoretical simulation 共space-group notations in parentheses兲: AgBH5 共P4 / n兲, AuSCl5 共P21 / n兲,
BaAlH5 共Pna21兲, BaAlF5 共P21 / n兲, BaAlF5 共P21兲, BaAlF5
共P212121兲, BaFeF5 共P21兲, BaGaF5 共P212121兲, BaGdCl5
共C2 / c兲, CaFeF5 共P21 / c兲, CaAlF5 共C2 / c兲, CaCrF5 共Cc兲,
CaTiO5 共I2 / c兲, CdMnF5 共P21 / n兲, Cr2F5 共C2 / c兲, CsTbF5
共Cmca兲, CuAuO5 共P1̄兲, CUO5 共Pmmn兲, FeAlF5 共Immm兲,
KAuS5 共Ibam兲, La2Br5 共P21 / m兲, MnCrF5 共C2 / c兲, MnAlF5
共Cmcm兲, MnAlH5 共Ama2兲, NbPO5 共P21 / c兲, NbPO5
共Pnma兲, NbPO5 共P4 / nmm兲, NiTaTe5 共Cmcm兲, NpIO5
共Pna21兲, PAsO5 共P212121兲, PWO5 共Pna21兲, RbHfF5
共P21 / c兲,
RbSF5 共Pbnm兲,
RbTeF5 共Pnma兲,
SeUO5
共P21 / m兲, SrAlF5 共P21 / n兲, SrAlF5 共I4兲, SrFeF5 共P21 / c兲,
SrSbF5 共Pbcm兲, SrVF5 共P21 / c兲, TaPO5 共P21 / c兲, Te2O5
共P21兲, TeUO5 共Pca21兲, TiSO5 共C2 / c兲, TlTeF5 共Pnma兲,
TlZrF5 共P21 / c兲, UMoO5 共Pcca兲, VPO5 共Pnma兲, VPO5
共P4 / nb兲, and VSO5 共Pnma兲. From the chosen structural
starting points, full geometry optimization has been carried
out with only symmetry constrains on atomic coordinates
and unit-cell parameters. As structural relaxation may change
a structure drastictly, the final structure may not resemble the
starting structure. The structure types quoted are therefore
merely used as a reference to symmetry and starting structure. The structures utilized for this purpose were taken from
the inorganic crystal structure database26 unless otherwise
stated in the text.
Estimation of heats of formation were obtained with
equivalent parameters to those referred to above. Effects of
zero-point motion were not considered, since estimation of
these requires access to phonon modes which are not addressed within the scope of the present study. However, our
experience suggests that the neglect of zero-point-motion effects does not invalidate the present conclusions on phase
stability.8,11
The Born-effective-charge concept has been used to analyze the bonding character in solids. In a purely ionic compound, off-diagonal components of this tensor will be small
and the diagonal components will carry information about
how much charge is transferred from one site to another. We
used the King-Smith–Vanderbilt27 method to calculate the
polarization in perturbed cells, and from this information the
Born-effective-charge-tensor elements for the constituents
were derived.
III. RESULTS AND DISCUSSION
A. Crystal structure of BaAlH5
It is convenient to start the presentation with the findings
for the BCH5 series with BaAlH5, since the crystal structure
of this member is experimentally well established. For this
compound the outcome of the calculations shows that the
reported structure 关Table I, Fig. 1共c兲兴 has the lowest total
energy 关Fig. 2共b兲兴. In order to have a good general view for
the topmost competitors for the ground state, we have displayed total energy versus volume data for only the four
configurations with the lowest energy. The calculated unitcell dimensions and positional parameters at 0 K and ambient pressures are in excellent agreement with the roomtemperature experimental findings13 for BaAlD5. The
deviations in the unit-cell dimensions are almost zero 共less
than ±0.4%兲 which are remarkably good for densityfunctional calculations.
The crystal structure of BaAlH5 关Fig. 1共c兲兴 contains
corner-sharing AlH6 octahedra, which form one-dimensional
zigzag chains along the crystallographic c axis. These chains
are surrounded by Ba atoms which form a distorted hexagonal network, and such networks are found 共more or less pronounced兲 in all the energetically closely related structure arrangements. This suggests that this feature is an important
structural element for the stability of BaAlH5. The calculated
positional parameters for BaAlH5 make the AlH6 octahedra
highly distorted. The H-Al-H bond angles range between
81.72°and 97.96°and the Al-H bond lengths between 1.70
and 1.82 Å 共1.69 and 1.85 Å according to the experimental
data13兲. The average Al-H bond length 共1.77 Å兲 is close to
that in Li3AlH6 关1.73 Å 共Ref. 28兲 1.75 Å 共Ref. 29兲兴 and
Na3AlH6 关1.76 Å 共Ref. 30兲兴. The closest shell of H atoms
around Ba resides at distances ranging from 2.65 to 3.07 Å
and consists of 14 H atoms. The shortest H-H separation in
BaAlH5 is 2.25 Å and complies accordingly with the 2 Å
rule.31
094122-2
PHYSICAL REVIEW B 73, 094122 共2006兲
STRUCTURAL PHASE STABILITY AND BONDING¼
TABLE I. Optimized equilibrium structural parameters, bulk modulus 共B0兲, and the derivative of bulk modulus 共B⬘0兲 for ␣-MgAlH5,
␤-MgAlH5 共metastable兲, and BaAlH5. Structure type refers to the input for the structural relaxation computation.
Compound
共structure type, space group兲
Unit cell
共Å or deg兲
␣-MgAlH5
共CaFeF5, P21 / c兲
a = 4.7499
b = 8.8127
c = 6.6281
␤ = 109.75
␤-MgAlH5
共CaCrF5, Cc兲
a = 7.8033
b = 5.7251
c = 6.7393
␤⫽115.39
BaAlH5
共prototype, Pna21兲
a = 9.1568 关9.194共1兲兴a
b = 7.0718 关7.0403共9兲兴a
c = 5.1039 关5.1061共6兲兴a
aExperimental
x
Mg共4e兲:
Al共4e兲:
H1共4e兲:
H2共4e兲:
H3共4e兲:
H4共4e兲:
H5共4e兲:
Mg共4a兲:
Al共4a兲:
H1共4a兲:
H2共4a兲:
H3共4a兲:
H4共4a兲:
H5共4a兲:
Ba共4a兲:
Al共4a兲:
H1共4a兲:
H2共4a兲:
H3共4a兲:
H4共4a兲:
H5共4a兲:
0.527
0.092
0.400
0.349
0.121
0.197
0.130
0.542
0.000
0.008
0.201
0.771
0.027
0.246
0.686
0.041
0.008
0.584
0.578
0.357
0.708
Positional parameters
y
z
x
0.985
0.245
0.121
0.390
0.592
0.862
0.305
0.025
0.000
0.924
0.289
0.969
0.299
0.031
0.156
0.846
0.946
0.844
0.786
0.695
0.545
0.253
0.395
0.444
0.495
0.201
0.142
0.156
0.257
0.000
0.256
0.034
0.882
0.979
0.130
0.256
0.229
0.905
0.025
0.504
0.233
0.214
共0.687
共0.049
共0.006
共0.576
共0.572
共0.353
共0.711
y
0.156
0.847
0.939
0.846
0.805
0.697
0.541
z
B0 共GPa兲
B⬘0
35.2
2.7
47.1
4.9
34.3
4.4
0.250兲a
0.233兲a
0.919兲a
0.019兲a
0.497兲a
0.240兲a
0.209兲a
value from Ref. 13.
B. Crystal structure of MgAlH5
Among the tested structural variants for MgAlH5, the
monoclinic CaFeF5-type arrangement 关Fig. 1共a兲兴 共designated
␣-MgAlH5兲 is found to have the lowest total energy 关Fig.
2共a兲兴, and the unit-cell dimensions and positional parameters
at 0 K and ambient pressures for this phase are given in
Table I. The ␣-MgAlH5 structure has some similarities with
BaAlH5, comprising AlH6 octahedra and capped MgH7 octahedra. The AlH6 octahedra share corners and edges with
capped MgH7 octahedra.
The Al-H and Mg-H bond distances in ␣-MgAlH5 fall in
the ranges 1.68–1.78 and 1.86– 2.31 Å, respectively. The
H-Al-H bond angles 共81.72°–97.96°兲 demonstrate that also
the AlH6 octahedra of ␣-MgAlH5 are highly distorted. The
H-Mg-H angles in the MgH7 polyhedra take values between
62.02°and 99.56°. One interesting structural feature is the
MgH7 configuration which distinguishes ␣-MgAlH5 from the
AAlH4 series. In the latter series, A cannot be ascribed meaningful coordinations. This suggests that ␣-MgAlH5 displays
a somewhat different bonding situation for the hydrogen atoms than in the AAlH4 series.
In addition to the ground-state ␣-MgAlH5 phase 共V
= 66.14 Å3 f.u.−1兲, Fig. 2共a兲 shows that there occurs a metastable CaCrF5-type modification very close in energy and
somewhat expanded volume 共V = 67.57 Å3 f.u.−1兲. These
findings suggest that one can expect that the latter phase
共designated ␤-MgAlH5兲 will form under appropriate condi-
tions. In fact, the energy difference between the ␣ and ␤
forms of MgAlH5 is so small 共only 3.7 meV f.u.−1兲 that the
competition between the two types of structure arrangements
will be very sensitive to synthesis conditions and examination temperature. Unit-cell dimensions and positional parameters for ␤-MgAlH5 are included in Table I.
In the A3AlH6 共A = Li, Na, K兲 series which also comprises
octahedral AlH6 units,28,29,32 the AlH6 octahedra are well
separated, whereas they share corners and form chains in
␣-MgAlH5 and BaAlH5. In ␣-MgAlH5 these chains are
fairly linear, while BaAlH5 exhibits zigzag chains. Already
from these observations one may suspect that the two series
will exhibit different bonding characteristics.
A comparison of the bulk moduli for the ACH4 series with
␣- and ␤-MgAlH5 and BaAlH5 共Table I兲 shows that the latter
phases exhibit higher B0 and are hence relatively harder materials than the ACH4 phases. However, the magnitude of B0
classifies ␣- and ␤-MgAlH5 and BaAlH5 as easily compressible materials. The soft character of these materials arises
from the ionic bonding between Mg/Ba and the AlH6 chains.
Although these materials are soft, we expect that appreciable
energy will be required to strip the H atoms from BAlH5
phases.
C. Stability of MgAlH5 and BaAlH5
The enthalpies of formation of MgAlH5 and BaAlH5 are
calculated from the reaction BH2 + AlH3 → BAlH5. Similarily
094122-3
PHYSICAL REVIEW B 73, 094122 共2006兲
KLAVENESS et al.
FIG. 1. 共Color online兲 Crystal structure of 共a兲 ␣-MgAlH5 and
共b兲 ␤-MgAlH5. AlH6 octahedra are shown in gray tone, and capped
MgH7 octahedra are shown in blue color. 共c兲 Crystal structure of
BaAlH5. The corner-sharing one-dimensional chains of AlH6 octahedra along 关001兴 are emphasized.
the enthalpy of formation for Mg共AlH4兲2 is estimated from
the reaction 2AlH3 + MgH2 → Mg共AlH4兲2 by utilizing the
structure predicted by Løvvik and Molin33 for Mg共AlH4兲2.
The enthalpies of formation are attained by adding the calculated reaction energy differences 关12, −80, and 29
kJ mol−1 for the formation reactions of MgAlH5, BaAlH5,
and Mg共AlH4兲2, respectively兴 to the respective experimental
enthalpies of formation of MgH2 共−76.15 kJ mol−1兲,34 BaH2
共−177 kJ mol−1兲,35 and AlH3 共−11.7 kJ mol−1兲.36 This results
in enthalpies of formation of ⬃−76, −224, and −70
kJ mol−1 for ␣-MgAlH5, BaAlH5, and Mg共AlH4兲2, respectively.
The selection of the above reactions is made to obtain the
most realistic estimate for the standard enthalpies of formation at ambient conditions, without performing phonon calculations. Recalling that the constituents on both sides of the
reaction equation are insulators and postulating that the reactants together carry similar bonding characteristics to the
products, essential temperature effects are likely to cancel.
Consequently, these estimates are primarily based on the assumptions that the experimental enthalpies of formation are
accurate and that the phonon spectra for the constituents of
the right and left sides of the reaction equation are similar. If
one considers reactions which include metals and/or gases,
this procedure may not be appropriate since the reactants
and/or products now may involve imbalanced electronic heat
capacity terms and/or severely different phonon bands and/or
vibration modes.
According to a proposed guideline37 for stable hydrides at
ambient conditions, dehydrogenation reactions should be
spontaneous for enthalpy changes smaller than 39 kJ per mol
H2. The reaction enthalpy of the decomposition reaction
FIG. 2. Calculated cell volume versus total energy curves for 共a兲
MgAlH5 and 共b兲 BaAlH5 in different possible structure arrangements. Only the four lowest-energy alternatives are displayed.
Structure types are specified in the illustrations.
MgAlH5 → MgH2 + Al+ 3 / 2H2 is 0 kJ per mol H2, and this
phase should consequently be metastable. Similarly, the estimated reaction enthalpy for the assumed reaction
Mg共AlH4兲2 → MgH2 + 2Al+ 3H2 is −2 kJ per mol H2, and
also this phase should be metastable. Hence, low temperatures and very high pressures are likely to be needed to stabilize MgAlH5 and Mg共AlH4兲2. BaAlH5, however, should be
stable with respect to dehydrogenation to BaH2 共with an estimated reaction enthalpy of 61 kJ per mol H2兲.
D. Chemical bonding
The calculated total and site-projected DOS for ␣- and
␤-MgAlH5 and BaAlH5 are displayed in Fig. 3. The total
DOS have distinct energy gaps between the valence band
共VB兲 and conduction band 共CB兲. These hydrides are accordingly proper insulators at 0 K with estimated band gaps of
2.48, 2.55, and 2.73 eV, respectively. The GGA-computed
band gaps of the A3AlH6 series 共which have octahedral AlH6
building blocks兲 are slightly larger 共2.9– 3.5 eV兲,29,32
whereas the ACH4 series 共which have CH4 tetrahedral building blocks兲 exhibit much larger band gaps 共ca. 5 eV兲.8,11 This
094122-4
PHYSICAL REVIEW B 73, 094122 共2006兲
STRUCTURAL PHASE STABILITY AND BONDING¼
FIG. 3. Calculated total and partial DOS for 共a兲 ␣-MgAlH5, 共b兲 ␤-MgAlH5, and 共c兲 ␣-BaAlH5. The shading marks s states, dashed line
marks p states, and solid line in an unshaded field marks d states. The Fermi level is indicated by a vertical dashed line.
distinction is partially due to the difference in the coordination of Al 共6 versus 4兲 in the latter series. The change in
coordination alters the Al-H distance and consequently shifts
the location of the VB and CB through variations in the
orbital mixing.
The site-projected DOS 共PDOS兲 for sites H1–H4 of ␣MgAlH5 are almost identical, and hence only the data for H1
are depicted as representative in Fig. 3共a兲. The PDOS for the
H5 site is, on the other hand, considerably different from the
former set. Similarly for ␤-MgAlH5 关Fig. 3共b兲兴, H1 has a
similar PDOS to H2 and H4, while PDOS for H3 is similar
to H5. In BaAlH5 关Fig. 3共c兲兴, H2 and H4 exhibit virtually
identical PDOS to H3 and H5, respectively.
The PDOS for the Al site in ␣-MgAlH5 reveals signs of
hybridization. Further, the charge-density plot in Fig. 4共a兲
shows a distinct degree of covalent bonding between Al and
H, while nearly complete ionic behavior is established at the
Mg site. The findings indicate a relatively much stronger
covalent bonding between Al and H than between Mg and H.
The Born-effective-charge tensors in Table II display average
diagonal elements of 2.14, 2.54, and −0.93 for Mg, Al, and
H, respectively, while the corresponding Mulliken-population values are 2.0, 2.2, and −0.8. From this information we
conclude that the AlH6 units exhibit 共in elementary chemistry language兲 polar covalent d2sp3-hybridized bonding with
an ionic woof, while the Mg atoms can be considered as
fairly idealized divalent ions. Comparison of the PDOS for
H1 and H5 reveals a stronger bonding for H5, which is to be
expected, since it is the connecting atom in the AlH6 chains
in ␣-MgAlH6 structure.
094122-5
PHYSICAL REVIEW B 73, 094122 共2006兲
KLAVENESS et al.
FIG. 4. 共Color online兲 Calculated charge density for 共a兲 ␣MgAlH5 and 共b兲 BaAlH5. The minimum and maximum values
within the maps are 0 and 1 / 2 and the separation between two
successive contours is 0.065 e Å−3. White is set at zero in the color
representation. Stars indicate three atoms whose positions define the
plane.
The PDOS for BaAlH5 in Fig. 3共c兲 show that the subbands are separated, making the simple d2sp3-hybridized picture for AlH6 octahedra invalid for this compound. On the
basis of the PDOS for the Al site, we interpret them as bands
originating from s / p / d orbitals, to be correlated with the
energetically lowest, middle, and highest bands in the VB
region of Fig. 3共c兲. The charge-density plot in Fig. 4共b兲
shows a similar charge distribution to ␣-MgAlH5. The Borneffective-charge tensors show average diagonal elements of
2.67, 2.23, and −0.98 for Ba, Al, and H, respectively, and
corresponding Mulliken-population values 2.1, 2.1, and −0.8.
The Born effective charge fails in this case even to provide
an approximate estimate of the valence of Ba, indicating that
the movement of Ba induces considerable charge displacements in the rest of the structural framework. All in all, the
bonding within the AlH6 units is also in this case of the polar
covalent type with an ionic woof, while Ba is considered as
a virtually ideal divalent ion. The s character is most advanced at the H1 sites, which are the corner-sharing atom in
the zigzag chains, indicating a stronger Al-H bond to that
atom.
As the size of Ba requires a higher coordination than that
of Mg, the AlH6 arrangement is forced into zigzag chains in
BaAlH5. The consequently different environment of the hydrogen atoms in the AlH6 unit destroys the simple d2sp3
hybridization picture. The BAlH5 group of compounds is
therefore predicted to comprise two types, one found in the
AlH6 zigzag chains of BaAlH5 and the other in the fairly
linear AlH6 chains of ␣-MgAlH5. The determining factor is
the size of the alkaline-earth ion.
IV. CONCLUSION
Structural phase stability, electronic structure, and bonding characteristics for BAlH5 共B = Mg, Ba兲 have been systematically studied. The crystal structure of BaAlH5 has been
successfully reproduced by use of the total energy minimization technique on 50 different possible closely related test
TABLE II. Calculated Born-effective-charge-tensor elements 共Z*兲 and Mulliken population 共MP兲 for the constituents of ␣-MgAlH5 and
BaAlH5. Data for NaH as reference material are included for convenience.
␣-MgAlH5
BaAlH5
NaH
Mg
Al
H1
H2
H3
H4
H5
Ba
Al
H1
H2
H3
H4
H5
Na
H
xx
yy
zz
xy
yz
zx
xz
zy
yx
MP
2.096
2.433
−0.936
−0.873
−1.015
−1.103
−0.593
2.770
1.998
−0.811
−0.940
−0.929
−1.043
−1.017
0.955
−0.955
2.144
2.496
−1.000
−1.057
−0.977
−0.898
−0.695
2.602
2.178
−1.086
−1.131
−0.874
−0.765
−0.929
0.955
−0.955
2.174
2.679
−0.967
−0.991
−0.666
−0.661
−1.571
2.629
2.501
−1.255
−0.898
−1.031
−0.997
−0.991
0.955
−0.955
−0.079
0.030
0.344
0.357
−0.276
−0.329
−0.110
0.365
−0.146
0.108
0.084
0.014
−0.145
−0.021
0.000
0.000
0.001
0.286
−0.056
0.070
−0.031
0.015
0.041
−0.098
0.202
0.392
0.000
−0.128
−0.085
−0.057
0.000
0.000
−0.007
0.009
0.027
−0.052
0.013
−0.005
−0.002
0.047
0.070
−0.159
0.003
0.075
−0.001
0.069
0.000
0.000
0.006
0.000
0.080
−0.089
−0.046
0.051
−0.006
0.020
−0.033
−0.171
0.006
0.045
−0.026
0.120
0.000
0.000
0.000
0.020
−0.040
0.076
0.000
−0.007
−0.006
−0.097
−0.029
0.393
−0.078
−0.122
−0.038
−0.045
0.000
0.000
0.001
0.286
−0.056
0.070
−0.031
0.015
0.041
−0.098
0.202
0.392
0.000
−0.128
−0.085
−0.057
0.000
0.000
1.985
2.220
−0.876
−0.868
−0.827
−0.826
−0.809
2.068
2.076
−0.835
−0.839
−0.831
−0.820
−0.818
094122-6
PHYSICAL REVIEW B 73, 094122 共2006兲
STRUCTURAL PHASE STABILITY AND BONDING¼
structures. The crystal structure of MgAlH5 has been predicted by the same approach. At ambient conditions MgAlH5
and BaAlH5 crystallize in monoclinic 共CaFeF5 type, P21 / c兲
and orthorhombic 共prototype, Pna21兲 structures, respectively. The theoretically established equilibrium structural
parameters for BaAlH5 agree well with experimental findings. A common feature of the two structures is highly distorted, linear, or zigzag chains of AlH6 octahedra, which are
corner shared with neighboring AlH6 subunits to form onedimensional chains along the c direction. ␣-MgAlH5 and
BaAlH5 exhibit finite band gaps between the valence and
conduction bands, and these compounds are accordingly to
ACKNOWLEDGMENT
The authors gratefully acknowledge the Research Council
of Norway for financial support and for computer time at the
Norwegian supercomputer facilities.
19
*Electronic address: arnekla@kjemi.uio.no
1 B.
be classified as insulators at 0 K. From partial-density-ofstates, charge-density, Born-effective-charge, and Mullikenpopulation analyses it is concluded that these compounds
have polar covalent bonding blended with ionic woof between Al and H, while the alkaline-earth atoms are in virtually ideal ionic states.
Bogdanovic and M. Schwickardi, J. Alloys Compd. 253/254, 1
共1997兲.
2 A. Zaluska, L. Zaluski, and J. O. Ström-Olsen, J. Alloys Compd.
298, 125 共2000兲.
3 V. P. Balema, J. W. Wiench, K. W. Dennis, M. Pruski, and V. K.
Pecharsky, J. Alloys Compd. 329, 108 共2001兲.
4 J. Chen, N. Kuriyama, Q. Xu, H. T. Takeshita, and T. Sakai, J.
Phys. Chem. B 105, 11214 共2001兲.
5 P. Vajeeston, P. Ravindran, A. Kjekshus, and H. Fjellvåg, Phys.
Rev. Lett. 89, 175506 共2002兲.
6 P. Vajeeston, P. Ravindran, R. Vidya, A. Kjekshus, and H.
Fjellvåg, Appl. Phys. Lett. 82, 2257 共2003兲.
7
P. Vajeeston, P. Ravindran, R. Vidya, H. Fjellvåg, and A. Kjekshus, Phys. Rev. B 68, 212101 共2003兲.
8
P. Vajeeston, P. Ravindran, R. Vidya, H. Fjellvåg, and A. Kjekshus, Cryst. Growth Des. 4, 471 共2004兲.
9
P. Vajeeston, P. Ravindran, A. Kjekshus, and H. Fjellvåg, Appl.
Phys. Lett. 84, 34 共2004兲.
10
P. Vajeeston, P. Ravindran, A. Kjekshus, and H. Fjellvåg, J. Alloys Compd. 363, L7 共2004兲.
11 P. Vajeeston, P. Ravindran, A. Kjekshus, and H. Fjellvåg, J. Alloys Compd. 387, 97 共2005兲.
12 P. Vajeeston, P. Ravindran, A. Kjekshus, and H. Fjellvåg, Proceedings of the World Hydrogen Energy Conference, Yokohama,
Japan A1386, 2004.
13
Q.-A. Zhang, Y. Nakamura, K. Oikawa, T. Kamiyama, and E.
Akiba, Inorg. Chem. 41, 6941 共2002兲.
14
T. N. Dymova, N. N. Mal’tseva, V. N. Konoplev, A. I. Golovanova, D. P. Aleksandrov, and A. S. Sizareva, Russ. J. Coord.
Chem. 29, 385 共2003兲.
15 M. Fichtner, O. Fuhr, and O. Kircher, J. Alloys Compd. 356/357,
418 共2003兲.
16
P. E. Blöchl, Phys. Rev. B 50, 17953 共1994兲; G. Kresse and D.
Joubert, ibid. 59, 1758 共1999兲.
17
G. Kresse and J. Hafner, Phys. Rev. B 47, R558 共1993兲; G.
Kresse and J. Furthmuller, Comput. Mater. Sci. 6, 15 共1996兲.
18
J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77,
3865 共1996兲.
J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 78,
1396 共1997兲.
20
P. Vinet, J. Ferrante, J. R. Smith, and J. H. Rose, J. Phys. C 19,
L467 共1986兲; P. Vinet, J. H. Rose, J. Ferrante, and J. R. Smith, J.
Phys.: Condens. Matter 1, 1941 共1989兲.
21 V. R. Saunders, R. Dovesi, C. Roetti, R. Orlando, C. M. ZicovichWilson, N. M. Harrison, K. Doll, B. Civalleri, I. Bush, Ph.
D’Arco, and M. Llunell, CRYSTAL2003 Users Manual 共University of Torino, Torino, 2003兲.
22 M. Causa, R. Dovesi, C. Pisani, and C. Roetti, Phys. Rev. B 33,
1308 共1986兲.
23 M. P. Habas, R. Dovesi, and A. Lichanot, J. Phys.: Condens.
Matter 10, 6897 共1998兲.
24
M. Catti, G. Valerio, R. Dovesi, and M. Causa, Phys. Rev. B 49,
14179 共1994兲.
25
R. Dovesi, C. Ermondi, E. Ferrero, C. Pisani, and C. Roetti, Phys.
Rev. B 29, 3591 共1984兲.
26
Inorganic Crystal Structure Database, Gmelin Institut, Germany,
2001.
27
R. D. King-Smith and D. Vanderbilt, Phys. Rev. B 47, 1651
共1992兲.
28 H. W. Brinks and B. C. Hauback, J. Alloys Compd. 354, 143
共2003兲.
29 P. Vajeeston, P. Ravindran, A. Kjekshus, and H. Fjellvåg, Phys.
Rev. B 69, 020104共R兲 共2004兲.
30 E. Ronnebro, D. Noreus, K. Kadir, A. Reiser, and B. J.
Bogdanovic, J. Alloys Compd. 299, 101 共2000兲.
31 W. Zhong, R. D. King-Smith, and D. Vanderbilt, Phys. Rev. Lett.
72, 3618 共1994兲.
32 P. Vajeeston, P. Ravindran, A. Kjekshus, and H. Fjellvåg, Phys.
Rev. B 71, 092103 共2005兲.
33 O. M. Løvvik and P. N. Molin, Phys. Rev. B 72, 073201 共2005兲.
34 M. W. Chase, J. Phys. Chem. Ref. Data Monogr. 9, 1 共1998兲.
35
CRC Handbook of Chemistry and Physics, 83rd ed., edited by D.
R. Lide 共CRC Press, New York, 2002兲.
36
G. C. Sinke, L. C. Walker, F. L. Oetting, and D. R. Stull, J. Chem.
Phys. 47, 2759 共1967兲.
37
K. H. J. Buschow, P. C. P. Bouten, and A. R. Miedema, Rep.
Prog. Phys. 45, 937 共1982兲.
094122-7