Eur. Phys. J.

B (2013) 86: 350

DOI: 10.1140/epjb/e2013-40141-1 THE EUROPEAN
PHYSICAL JOURNAL B
Regular Article

Ab initio electronic band structure study of IIIVI layered

semiconductors
Daniel Olgun1,2, Alberto Rubio-Ponce3, and Andres Cantarero2,a
1
Departamento de Fsica, Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional,
07300 Mexico D.F., Mexico
2
Materials Science Institute, University of Valencia, P.O. Box 22085, 46071 Valencia, Spain
3
Departamento de Ciencias Basicas, Universidad Autonoma Metropolitana-Azcapotzalco, Av. San Pablo 180,
02200 Mexico D.F., Mexico

Received 20 February 2013 / Received in nal form 25 April 2013

Published online 7 August 2013 
c EDP Sciences, Societa Italiana di Fisica, Springer-Verlag 2013

Abstract. We present a total energy study of the electronic properties of the rhombohedral -InSe, hexag-
onal -GaSe, and monoclinic GaTe layered compounds. The calculations have been done using the full
potential linear augmented plane wave method, including spin-orbit interaction. The calculated valence
bands of the three compounds compare well with angle resolved photoemission measurements and a dis-
cussion of the small discrepancies found has been given. The present calculations are also compared with
recent and previous band structure calculations available in the literature for the three compounds. Fi-
nally, in order to improve the calculated band gap value we have used the recently proposed modied
Becke-Johnson correction for the exchange-correlation potential.

1 Introduction lattice showing four possible stacking arrangements of the

The family of IIIVI semiconductors, such as GaS, GaSe,
GaTe and InSe has been extensively studied for decades,
both from the basic aspects, as their photovoltaic and pho-
toelectronic properties [13], as from potential applica-
tions in nonlinear and optical bistable devices [4,5], in the
development of solar cells and solid-state batteries [69]
or in the fabrication of memory devices [10]. These com-
pounds crystallize in a layered structure, each layer being
composed of four monoatomic anion-cation-cation-anion
layers. The highly anisotropic behavior of their optical
and transport properties is a consequence of the layered
arrangement of these compounds.
The current renewed interest in the IIIVI compounds
comes from the successful attempts to grow these com-
pounds by molecular-beam epitaxy [11,12]. In particular,
the epitaxy of GaSe and InSe on silicon opens the use of
these materials as buer layers in electronic devices and
motivated recent and active research [1316].
In the layered structures, the cations are tetrahedrally
coordinated (three anions and one cation), while the an-
ions are bounded to three cations. Cation pairs in the
crystal saturate the bonding and are responsible for the
semiconducting behavior. The bonding between the layers
are due to van der Waals interaction, the distance between
the layers being called the van der Waals gap. Most of the
structures crystallize in the hexagonal or rhombohedral
a
e-mail: andres.cantarero@uv.es
complex layers, leading to four polytypes, named , , ,
and . All the polytypes are anisotropic given rise to a
dierence in the dielectric properties of the materials de-
pending on the direction (parallel or perpendicular to the
layers). Gallium telluride crystalizes in a lower symme-
try crystal system, the monoclinic, but it is still a layered
compound, as will be discussed later when comparing the
structures of the three compounds.
The electronic band structure for the IIIVI layered
compounds was calculated rstly in a two-dimensional
approximation by using an empirical tight-binding model
(ETB) [1720]. In the following years, the inter-layer inter-
action was introduced, either using ETB or pseudopoten-
tial methods, carried on mainly for the -polytype [2128].
The rst calculation of the electronic structure of InSe
was done by McCanny and Murray [20], using the ETB
method neglecting the interaction among the layers. The
calculation was improved by Bourdon et al. [29] by in-
cluding the interlayer interaction. Robertson [22] and Doni
et al. [24] calculated the electronic structure of GaSe, GaS
and InSe for the simplest -polytype. The electronic struc-
ture of the most common polytype -InSe had been pre-
viously calculated by means of an ETB approach, includ-
ing spin-orbit eects [25,26], and later on by Kunc and
Zeyher [30] and Gomes da Costa et al. [31] using ab ini-
tio pseudopotentials. More recently, Ulrich and collabora-
tors [32,33] calculated the band structure of InSe using a
full potential linear augmented plane wave method, while
Ferlat et al. [34] have done the calculation using the GW
Page 2 of 9
correction. On the other hand, there is also a recent the-
oretical work on -InSe [3538]. The most common poly-
type of InSe, -InSe, has a direct band gap (1.23 eV at
room temperature [32,33]) at the Z-point of the rhombo-
hedral Brillouin zone (BZ).
In the case of GaSe, the most common polytype is the
hexagonal -GaSe. It was rstly calculated in an ETB ap-
proach by Nagel et al. [21], and later with empirical pseu-
dopotentials by Depeursinge [25,26]. According to these
calculations and the comparison with experimental data
given by these authors, -GaSe is an indirect semiconduc-
tor with band gap of 2.10 eV at 77 K, and has a direct
band gap of 2.12 eV (the experimental optical measure-
ments are given in Tab. 5 of Ref. [21]). However, several re-
cent calculations have shown that -GaSe is a direct band
gap compound [3,39,40], although it becomes an indirect
band gap semiconductor under hydrostatic pressure [3].
The real crystal structure of GaTe is monoclinic. In
order to make the calculations more aordable an hexag-
onal lattice, more specically the -polytype [22,41], has
been used in the past. However, recently, rst principles
calculations for monoclinic GaTe have been performed,
showing that GaTe is a direct gap semiconductor [4246].
All these calculations neglect spin-orbit interaction. The
measured bandgap is 1.801 eV at low temperature [47]
and 1.796 eV at room temperature [43]. These two values
are too close (only 6 meV), since typical variation of the
gap due to electron-phonon interaction is of the order of a
few tenths of eV. It would be necessary to deeper analyze
the gap variation of GaTe with temperature to have more
condent data on this compound.
Since the development of the density functional the-
ory (DFT) [48], and the posterior formalism due to Kohn-
Sham (KS) [49], the improvement in the prediction of sev-
eral solid state properties have increased with time. It is
well-known, however [5053], that the common practice
of associate the calculated KS eigenvalues with the elec-
tronic band structure of the solid, in the case of semi-
conductors and insulators, leads to incorrect band gap
values [34,54,55]. It is found that the solution of the KS
equations, using either the local density approximation
(LDA) [48,56] or the generalized gradient approximation
(GGA) [57] to the exchange correlation potential and en-
ergy, underestimates the calculated band gap value, in av-
erage, by an amount of 50% (the so-called band gap
problem). It was suggested that, on the underestimation
on Eg , the error in the KS calculations comes from the
fact that we do not know exactly the exchange corre-
lation potential, Vxc , and great eort was done during
years to compute a more accurate Vxc [5052,57]. How-
LDA
ever, for Si, a comparison of the exact Vxc vs. Vxc
shows that these potentials are very similar, and it was
LDA
concluded that Vxc provided a good approximation to
the ground state charge density of the solid [52], thus the
exact exchange-correlation potential does not provide the
correct bandgap. Later on, it was suggested that the cor-
rection to Eg should be done using an energy dependent
self-energy, xc ([n] : , r), which lies beyond the Kohn-
Sham formalism [57]. Nevertheless, it should be stressed
Eur. Phys. J. B (2013) 86: 350
that neither complicated calculations as GW, nor the
new proposed hybrid functionals are able to completely
solve the band gap problem, although great improvements
have been found in several cases [34,55,5860]. These re-
sults correspond to the gap at 0 K. Additionally, the
gap variation with the temperature requires a more so-
phisticated treatment, since electron-phonon interaction
must be included [61], even at 0 K, due to quantum ef-
fects. An alternative improvement to the KS formalism
is the construction of simple but accurate energy poten-
tials [57]. One of these improvements is the called modied
Becke-Johnson (m-BJ) approach [62,63]. This approach
is concerned with the modication of the exchange eec-
tive potential, and has been shown to improve the calcu-
lated band gap value of several semiconductor compounds,
without complicated computations which demand also im-
portant computational resources.
In this work we have done total energy calcula-
tions for the rhombohedral -InSe, hexagonal -GaSe,
and monoclinic GaTe layered semiconductor compounds.
Our objective is to show a comparison of the calculated
band dispersion for the valence band region with avail-
able photoemission data for these compounds. In our
study we have used the relativistic full-potential linearized
augmented plane wave method [64], including spin-orbit
interaction. Then, as a by-product of our study we re-
port the calculated band gap of these compounds. In or-
der to improve the calculated band gap value, we have
used the modied Becke-Johnson correction implemented
in the Wien2k code. This approach modies the exchange
eective potential proposed by Becke and Johnson [62]
and allows us to obtain better band gap values in com-
parison to those obtained using the LSDA or GGA ap-
proaches [63]. For those authors who still doubt on the
accuracy or applicability of DFT calculations, it should be
stressed, however, that although this and other improve-
ments have been included in DFT based calculations, the
band gap problem is not yet totally closed, as can be found
in a recent work showing a test on the m-BJ for several
semiconductor compounds [65], and we should be warning
on the limitations of the theory.
2 Computational details
In our calculations, we have used the full potential lin-
earized augmented plane wave method (FP-LAPW) as
implemented in the Wien2k code [64]. In this method,
wave functions, charge density, and potential are expanded
in spherical harmonics within non-overlapping mun-tin
spheres and in plane waves in the remaining interstitial
region of the unit cell. In the code the core and va-
lence states are treated dierently. Core states are treated
within a multi-conguration relativistic Dirac-Fock ap-
proach, while valence states are treated in a scalar rel-
ativistic approach. The exchange-correlation energy was
calculated using the generalized gradient approximation
(GGA) correction of Perdew et al. [57]. The convergence
of the total energy in terms of the variational cuto-energy
parameter has been guaranteed by using an appropriate
Eur. Phys. J. B (2013) 86: 350 Page 3 of 9

b
a2 c
a1 a1 a2

Fig. 1. Unit cell of the three compounds studied in this work, including the atoms displaced one unit cell from the origin:
Hexagonal cell of rhombohedral -InSe (left), hexagonal cell of -GaSe (center), and monoclinic-GaTe (right). To show the layered
character of the studied compounds only the bondings between the interlayer atoms are shown (the layer-layer interaction is of
van der Waals type). The largest spheres (blue online) represent the anions (Se or Te) and the smaller ones (red online) the
cations (In or Ga). The direction of the main axis are indicated on the right part of the gures. Comparing the layers of InSe
(left gure) and GaSe (center gure) with that of GaTe (right gure), we can realize that the cation-cation bonds have two
dierent orientations in the last case.

set of k-points in all the calculations. In these calculations,
the In 4p-states and the Ga 3d-states were treated as va-
lence band states using the local orbital extension of the
LAPW method [64], and the spin-orbit corrections have
been taken into account.
The plane-wave cuto used was 9 Ry for InSe, 10 Ry
for GaSe, and 9 Ry for GaTe. A set of 120 k-points was
used to calculate the total energy for InSe, this k-points
set is equivalent to a 8 8 8 Monkhorst-Pack [66] grid of
the unit cell, 70 k-points were used for GaSe, equivalent to
a 13 13 2 Monkhorst-Pack [66] grid, and 132 k-points
were used for GaTe, equivalent to a 11411 Monkhorst-
Pack [66] grid, respectively.
In order to reach the optimal structure we have re-
laxed the internal atomic positions to obtain the mini-
mum of the total energy. After the relaxation process we
obtain that the nearest neighbor distances in InSe and
GaSe compare very well with experimental measurements,
as we have shown in references [32,33,67]. In the case of
monoclinic GaTe we have also found that our calculated
nearest neighbor distances and next nearest neighbor dis-
tances compare very well with experimental data [45,68],
as well as with recent ab initio calculations [46].
3 Results and discussion
3.1 InSe
The crystal structure of -InSe is rhombohedral (space
group R3m). The lattice parameters are a = 0.4002 and
c = 2.4946 nm [32,33]. The primitive cell contains three
layers and four atoms, two Se and two In, which can be
chosen on the z-axis (as it is schematically shown in Fig. 1
at the left). The cations are at InI : 23 ( 12 +z2 ), InII : 23 (1
z2 ), and the anions are at SeI : 23 z1 , SeII : 23 (1 z1 ) [21],
where the internal parameters, in units of c, are z1 =
0.09130 and z2 = 0.16645 [69].
Page 4 of 9
2 2
1 1
-InSe
0 0
-1 -1
-2 -2
Energy (eV)
-3 -3
-4 -4
-5 -5
-6 -6
-7 -7
-8 -8
K M Z split by around 0.1 eV.
Fig. 2. At the left, calculated electronic band structure of -
InSe. Black lines show the m-BJ calculated bands, while the
blue lines show the GGA calculations (see text for details).
The measured electronic bands are shown as symbols [70]. For
completeness, we show the BZ of the rhombohedral InSe on
the right part of the gure calculated with the real reciprocal
parameters. The notation is that given by Amokrane et al. [70]
for easier comparison.
The electronic band structure of -InSe (including
spin-orbit eects) is shown in Figure 2 (left) for the high
symmetry directions in the BZ. The fundamental gap is
direct and occurs at the Z-point of the BZ, as commented
previously. The calculated band gap using the GGA ap-
proach is 0.44 eV, more than one half of the experimental
value. However, using the m-BJ correction our calculated
value is 1.83 eV. As we have found, our GGA-calculations
are consistent with early ab initio pseudopotential calcu-
lations by Gomes da Costa et al. [31]. Comparison of the
calculated bands using the GGA approach (blue lines on
line) with the m-BJ calculations (black lines) is shown in
the gure. As it is expected, the m-BJ correction moves
up the minimum of the conduction band, in such a way
that the corrected value for the band gap is 1.83 eV. The
m-BJ correction improves the calculated band gap value,
although in our case the calculated value is about 49%
bigger than the experimental one. GW calculations on
Eur. Phys. J. B (2013) 86: 350
Table 1. Calculated band gap for the dierent III-VI semicon-
ductors studied in this work. We compare the known experi-
mental value at room temperature with our calculated value
using the GGA correction as well as the GGA + m-BJ correc-
tion. For completeness we show the GW values where they are
known, the gures in brackets show the discrepancy between
the experimental and calculated values.
System Egexp EgGGA EgGGA-mBJ EgGW
InSe 1.23 [32,33] 0.44 (64) 1.83 (49) 1.0 [34] (18)
GaSe 2.12 [21] 0.94 (45) 1.85 (15) 2.34 [39] (9)
GaTe 1.80 [47] 0.59 (33) 1.34 (34 )
the other hand reported a calculated band gap value of
1.0 eV, value that is 18% smaller than the experimental
one (see Tab. 1). At the same time, we have obtained that
at -point the m-BJ correction moves up the calculated
valence bands. The lower band shown in the gure, moves
up around 0.2 eV, while the band at the maximum of the
valence band is moved down around 0.15 eV. The remain-
ing bands, near the maximum of the valence band, are
The electronic band structure calculated in this work
for the valence band region for -InSe polytype is com-
pared in Figure 2 with the angle resolved photoemis-
sion measurements reported by Amokrane et al. [70]. The
comparison is done along the paths K- -M and -Z of
the rst BZ, which collects all the experimental measure-
ments. The -M direction corresponds to the x-axis, -K
to the y-axis, and -Z to the z-direction of the BZ. In their
work, Amokrane et al. compared their measured valence
band data values with that calculated by Gomes da Costa
et al. [31]. However, in the work of Gomes da Costa et al.
some of the calculated bands do not correspond to the
experimentally measured direction, thus Amokrane com-
pare the experiment with theoretical data calculated in
dierent directions. Instead of the layer plane (K- -M ),
Amokrane et al. compare the M - -M directions with the
calculations along A- -B from Gomes da Costa et al. [31],
i.e. instead of the directions in the plane, the directions
from the -point to the centers of the hexagonal (A-point)
and rectangular (B-point) faces of the BZ.
From Figure 2, the experimental data in the K-
are shifted with respect to the theoretical band struc-
ture (this direction is not compared in the work of
Amokrane et al. [70]). If we move up the experimental
points 0.50.8 eV, the t would be much better. Although
the uncertainty is not given in the work of Amokrane
et al. [70], we can deduce from the shift of the gap that
it would be of the order of 0.2 eV. Even this shift would
not be enough to explain the dierences. There are some
experimental points, however, which t better as soon as
we approach to the point. In the -M direction the
comparison with the experiment is good, in general, and
becomes better as soon as we approach to the M -point of
the BZ. It is interesting to notice that the dierence be-
tween the real measured direction -M does not deviate
appreciably from the A- or -B used for comparison in
the paper of Amokrane et al. [70]. The reason may be that
Eur. Phys. J. B (2013) 86: 350 Page 5 of 9

the BZ of -InSe is very at in the -Z direction, in other

2
words, the c parameter is very large in comparison to a. As -GaSe 2
commented previously, the BZ drawn in Figure 2 (right) -GaSe
was done using the real values of a and c. In the -Z 0 0
direction the agreement is again satisfactory, although it
would be better if we shift up in energy the experimental -2 -2
points. In any case, the experiment does not go to zero at
the Z-point. There are also experimental data which do -4 -4
not correspond to a real band, like that around 2.5 eV in

Energy (eV)
the work of Amokrane et al. [70], where the set of points -6 -6
B in their Figure 10 do not correspond to a real band. It is
worth to notice that our scale is amplied as compared -8 -8
to that given in the paper of Amokrane et al. [70], since our
vertical scale is very large (to distinguish well the energy -10 -10
dierences) and our horizontal scale is very small (propor-
Se (4s) Se (4s)
tional to the -Z direction of the BZ), while they amplify -12 -12
the horizontal scale. In our gure we see clearly that there
are two experimental bands around 6 to 7 eV, while -14 -14
in the graph shown in the paper of Amokrane et al. [70]
it seems that there is only one band as a consequence of
-16 Ga (3d) Ga (3d) -16
the scaling. In their gure there is only one band, while
K M M
the experiment shows two.
A
L H
3.2 GaSe
*

The crystal structure of -GaSe is hexagonal (space group M K

P 6m2), a = 0.3743 and c = 1.5919 nm being the lattice
parameters of the unit cell [71]. The primitive cell con-
tains three layers and twelve atoms, two Se and two Ga
per layer (see Fig. 1 for a schematic representation of the
hexagonal unit cell). The cations are at (a/3, 2a/3, rGac)
and (0, 0 sGa c), where rGa = 0.425 and sGa = 0.075 are
relative internal parameters (in units of c). The anions are
at (2a/3, a/3, rSec) and (0, 0 sSe c), where rSe = 0.15
and sSe = 0.35. Furthermore, the internal parameters
(r, s) fulll the relation r + s = 1/2 [71]. -GaSe, accord-
ing to early experimental data and empirical calculations
is an indirect band gap semiconductor [21,2528]. How-
ever, from our LAPW-calculations we nd that -GaSe is
a direct band gap semiconductor, as it is shown in Fig-
ure 3. The same results have been recently reported both
experimentally and theoretically [39,40]. Using GGA our
calculated band gap value at ambient pressure is 0.939 eV,
using the m-BJ correction our calculated band gap value
is 1.85 eV. While from GW calculations a band gap value
of 2.34 eV has been reported [39] (see Tab. 1). Further-
more we have found that -GaSe becomes an indirect band
gap semiconductor as a function of hydrostatic pressure,
at about 4 GPa [3]. The indirect gap is located at the
M -point in the conduction band, 0.329 eV higher than
the minimum of the conduction band at ambient condi-
tions, over the -point (see Fig. 3).
As in the case of -InSe, we have obtained that the
m-BJ correction moves up the calculated valence bands.
For GaSe we found that the m-BJ correction displaces up
the valence band at -point in average from 0.5 eV for
the bands located around 7.0 eV to 0.1 eV for the bands
located at the maximum of the valence band. Then, in
going from InSe GaSe GaTe the number of atoms in
Fig. 3. Calculated electronic valence band structure of GaSe
vs. (full lines) the measured electronic bands (empty cir-
cles) [74]. Left panel shows for comparison the calculated
bands along the -K-M and -M - direction as reported by
Plucinski et al. [74] in their Figures 13 and 14. At the right, we
show the hexagonal BZ calculated with the real parameters of
GaSe.
the unit cell increases signicantly, and from this fact also
the number of bands increases, in that case does not help
us to plot the GGA and m-BJ calculated bands to compare
with the experimental data, for this reason for GaSe and
GaTe we will comment only the m-BJ calculated bands.
Concerning with the orbital-atomic contribution to the
valence and conduction bands we found from our calcu-
lations that the cation (Ga) 3d states are about 15.5 eV
below the top of the valence band and show no dispersion
as it is expected, while the Se-4s are around 14.0 eV below
the maximum of the valence band and show a small dis-
persion. These states can be considered as core states. The
lowest band contributing to the bond is around 7.0 eV be-
low the maximum of the valence band. Actually, all states
shown in Figure 3 are 4s-4p cation states hybridized with
the 4p anion states. We have found also a small contri-
bution of the anion and cation d states. The lower band
shown in the gure (about 8.0 eV), comes mainly from
the cation s states. The lower conduction bands, on the
other hand, are mainly constituted of cation 4s-states hy-
bridized with cation and anion p-states, and a small con-
tribution of the anion d-state was also found. The large
Page 6 of 9
Table 2. Calculated internal parameters for monoclinic-GaTe
found in this work, after the minimization of the total energy.
x-param. y-param.
Ga1 (4i) 0.8625 0.0844
Ga2 (4i) 0.7648 0.1984
Ga3 (4i) 0.4342 0.4135
Te1 (4i) 0.8310 0.8316
Te2 (4i) 0.5446 0.8283
Te3 (4i) 0.8491 0.4670
number of bands is due to the fact that we have eight
atoms (72 electrons in the valence band) in the unit cell. In
general, our discussion on the symmetry of the lower con-
duction bands and upper valence bands agrees with that
of Kuroda and Nishina [72] and Choi et al. [73], who did
an ETB model. However, since we present here a full elec-
tron calculation then we are able to describe not only the
anion-cation s and p hybridization, but the anion-cation
d hybridization as well.
In Figure 3, we also compare the ARPES measure-
ments for GaSe(0001) direction reported in reference [74]
with our calculations. Figure 3 shows the -K-M and
-M - directions. In spite of the large number of bands
shown in the gure, we observe that our calculation re-
produces quite well the experimental data, the same in
number, as well as the overall dispersion of the bands in
the given directions. Finally, we found that our calculated
electronic valence bands, depicted in Figure 3, show a very
good agreement with recently reported ab initio pseudopo-
tentials calculations [39,40,74].
3.3 GaTe
The crystal structure of GaTe is monoclinic, space group
B2/m (see Fig. 1). The lattice parameters are a = 1.744,
b = 1.0456, c = 0.4077 nm, and the monoclinic angle
= 104.44 [75]. The initial internal parameters used in
our calculations are that given by Julien-Pouzol et al. [75].
In order to nd the stable crystal structure that minimizes
the total energy of monoclinic GaTe, we have relaxed the
internal parameters, while we have used the experimen-
tal lattice parameters. After the relaxation procedure we
have found the internal atomic values showed in Table 2,
these values are comparable with that listed by Julien-
Pouzol et al. [75] and the recently calculated values by
Yamamoto et al. [43]. Using our optimized internal atomic
values, we nd that the Ga-Te and Ga-Ga, nearest neigh-
bor distances, and Te-Te next nearest neighbor distances
compare very well with recent EXAFS data [68], and with
the data reported by Pal and Bose [45] (see Tab. 3). The
comparison with the measured values show that our cal-
culated nearest neighbor distances are less that 4% larger
than the experimental ones. The same is found in compar-
ison with recent LDA (GGA) ab initio calculations [46].
The crystallographic unit cell of monoclinic GaTe con-
tains six formula units (that is 108 valence electrons),
while only two formula units appear in the layer per
Eur. Phys. J. B (2013) 86: 350
Table 3. The Ga-Te and Ga-Ga nearest neighbor distances,
and Te-Te next nearest neighbor distances reported in the ex-
perimental works of Pellicer et al. [68], and Pal et al. [45] are
compared with the values calculated in this work. We also com-
pare our results with previous ab initio calculations performed
using LDA and GGA correction (this last in parenthesis) [46].
Experiment (A) Calculated (A)
Ga-Te 2.6372.686 [68] 2.4582.574
2.638 [45] 2.642.67 [46]
(2.692.72)[46]
Ga-Ga 2.43 [68] 2.4582.574
2.473 [45] 2.392.41[46]
(2.462.48)[46]
Te-Te 4.096[45] 4.267
unit cell in other IIIVI layer compounds crystallizing in
an hexagonal lattice. This explains why GaTe is one of
the less studied IIIVI layered semiconductor compounds.
The electronic band structure of GaTe has been calculated
in the past using an hexagonal model [22,25,26,41]. That
is, it has been studied in the -polytype and the exper-
imental results have been analyzed within this scheme.
However, due to the large number of valence electrons,
the calculated electronic band structure for monoclinic
GaTe is much more complex, as it is shown in Figure 4,
and it was recently reported in another ab initio calcula-
tions [4244,46], although neglecting relativistic eects.
Figure 4 shows our calculated electronic band struc-
ture for monoclinic GaTe, including spin-orbit eects.
From our calculations we found that monoclinic-GaTe is
a direct band gap semiconductor with the gap in the Z
point of the BZ, the direction perpendicular to the atomic
layers, in the same direction as -InSe. The calculated
GGA band gap value is 0.595 eV, while the m-BJ cor-
rection gives us the value 1.34 eV (the measured value at
10 K is 1.801 eV [47], and the room temperature value
is 1.796 eV [43]). These two values are too close (only
6 meV) to be realistic, since the typical variation of the
gap of a semiconductor due to electron-phonon interac-
tion is of the order of a few tenths of eV. The calculated
electronic valence band shows the same pattern as that of
pseudopotential calculations of Sanchez-Royo et al. [42],
the LMTO-calculations of Yamamoto et al. [43], as well
as with the bands obtained using the projector augmented
wave method of Rak et al. [44] and Rocha and Lordi [46].
Due to the large number of bands in this material, it is dif-
cult to compare them band to band to the photoemission
data. However, in general we observe the same pattern for
the calculated bands obtained in reference [43], as well
as the discussion of the calculated DOS given for these
authors.
As for the InSe and GaSe cases we found that using the
m-BJ correction moves up the calculated valence bands in
average from 0.5 eV for the lower bands, to 0.1 eV for the
upper bands. Although in the present case this displace-
ment will help us to reproduce properly the experimental
data, as will be discussed later.
Eur. Phys. J. B (2013) 86: 350
monoclinic GaTe
0
-2
-4
Energy (eV)
-6
-8
-10 Te (5s)
-12
-14 Ga (3d)
-16
H Z P Y'
Z We found that most of the experimental bands have a
P H
Y
*
Fig. 4. Calculated electronic band structure of GaTe. For a
direct comparison with the reported bands of Sanchez-Royo
et al. [42], the calculated bands are computed along the path
shown in the rst Brillouin zone to the right.
From an analysis of the density of states and electronic
bands, we have obtained that the Ga 3d electrons are lo-
cated at about 15.0 eV from the top of the valence band
(see Fig. 4). This value should be compared with the XPS
value of 18.2 eV of Antonangeli et al. [41]. The remaining
valence bands are constituted as follows: the lower states,
about 10.8 eV, are formed mainly from Te 5s states, as
we have three dierent Te atoms in the unit cell, there
are six of such states. The next set of six values, around
[7, 5] eV, come from Ga 4s states. The uppermost va-
lence band states are an hybridization of Ga 4p with Te
5p states, and a small contribution of Te 4d states and
Ga 3d states. The lowest conduction band has a strong
hybridization of Ga 4s and 4p states with Te 5p states,
with a few proportion of cation and anion d-states. In gen-
eral this discussion compares very well with that given by
Sanchez-Royo et al. [42], see also the Figure 2 and related
discussion given by Rak et al. [44].
Figure 5 shows the comparison of our calculated
electronic valence bands with ARPES data reported by
Sanchez-Royo et al. [42]. As we observe from the gure,
the general trend of the experimental data is reproduced in
our calculations. The left panel of Figure 5 shows that the
experimental bands are almost at in the -Z direction.
Page 7 of 9
Fig. 5. Calculated electronic valence band structure of GaTe
(full lines) vs. the measured electronic bands (empty circles).
The experimental data from ARPES reported by Sanchez-
Royo et al. [42].
theoretical counterpart in our calculations. Although for
the energy range from 3.0 to 4.0 eV small discrepancies
were found between the measured bands and our calcu-
lations. On the other hand, the authors of reference [42]
commented on the possibility that these at bands could
be related with surface or resonance states, which are bet-
ter known and studied in well-known IIVI [7678] and
IIIV [79] semiconductor surfaces. There are some works
on IIIVI semiconductor surfaces [13,8082], but they did
not discuss this particular point.
Figure 5, right panel, shows a comparison of our calcu-
lated electronic valence bands with that measured in the
H-Z direction. As in the -Z direction, there is a set of
measured bands, in the upper valence band region, that
have a counterpart with our calculated bands, where the
general trend of the measured dispersion is reproduced by
our calculations. Although we observe that for the middle
energies there are experimental bands showing discrepan-
cies with the calculated bands, mainly in the gap energy
interval from 3.0 to 4.0 eV. Then, we found that the bands
located at 5.4 and 6.4 eV are well reproduced in our calcu-
lations. It is in this sense that we think that the calculated
valence bands reproduce the correct dispersion shown for
the experimental data.
In our calculation we have included spin-orbit interac-
tion, as it is discussed in references [42,44], the splitting
of the bands is observed mainly for the maximum of the
valence bands in the -point (see Fig. 4 in Ref. [44]), in
the remaining bands spin-orbit eects are not responsible
for the dierences between the calculated and the mea-
sured bands, as already commented above for the -InSe
case. In this work, we report bulk band structure calcula-
tions, and we found that all the experimental bands could
be identied within the calculated bands. Thus, we do not
think that the measured bands located in the valence band
gaps should be related with surface or resonance states,
contrary to the comments of Sanchez-Royo et al. [42].
But, in order to clarify this point, more experimental and
Page 8 of 9
theoretical works analyzing the surface of these materials
are necessary.
4 Conclusions
We have performed a total energy calculation of the elec-
tronic band structure for the IIIVI layered compounds
-InSe, -GaSe, and monoclinic GaTe. In particular for
-GaSe, we found that our calculations reproduce prop-
erly experimental data, as well as recent theoretical calcu-
lations. Worthy to note is the calculated dispersion for the
valence band states, that compare very well with ARPES
data. We found that the m-BJ correction improves the
calculated dispersion of the valence bands, which compare
very well with experimental data. Although, in compar-
ison with GGA calculations, our m-BJ calculations im-
prove the calculated band gap values, a better improve-
ment should be done since the accuracy of the model it is
still decient.
The atomic orbital character of the calculated elec-
tronic bands, obtained from an analysis of the DOS for
-GaSe and monoclinic GaTe, show good agreement with
the previously published work.
For monoclinic-GaTe most of the measured bands
show a at dispersion. It is found that some of these bands
are located in the calculated valence band gaps of GaTe,
from these ndings there are some speculations that these
bands could be related with surface and resonance states,
although there are neither experimental nor theoretical re-
ports that conrm or discard these facts. Our calculations
support the fact that these bands are bulk states. How-
ever, and in order to clarify this point, more experimental
and theoretical work are necessary.
DO gratefully acknowledges nancial support from the
CINVESTAV-IPN and CONACYT-Mexico. We also acknowl-
edge the computer facilities of the University of Valencia, in
particular the use of the supercomputer TIRANT.
References
1. A. Segura, J.N. Besson, A. Chevy, M.S. Martin, Nuovo
Cimento Soc. Ital. Fis. B 38, 345 (1977)
2. C. Ulrich, M.A. Mroginski, A.R. Goni, A. Cantarero, U.
Schwarz, V. Munoz, K. Syassen, Phys. Status Solidi B 198,
121 (1996)
3. U. Schwarz, D. Olgun, A. Cantarero, M. Hanand, K.
Syassen, Phys. Status Solidi B 244, 244 (2007)
4. M.A. Hernandez, J.F. Sanchez, M.V. Andres, A. Segura,
V. Munoz, Opt. Pura Apl. 26, 152 (1993)
5. W.C. Eckho, R.S. Putnam, S. Wang, R.F. Curl, F.K.
Tittel, Appl. Phys. A 63, 437 (1996)
6. A. Segura, J.L. Valdes, F. Pomer, A. Cantarero, J.L.
Lloret, A. Chevy, in Proceedings of the Fifth E.C.
Photovoltaic Solar Energy Conference (D. Reidel Pub. Co.,
Dordrecht, 1984), pp. 927930
7. J.F. Sanchez-Royo, A. Segura, O. Lang, E. Schaar, C.
Pettenkofer, W. Jaegermann, L. Roa, A. Chevy, J. Appl.
Phys. 90, 2818 (2001)
Eur. Phys. J. B (2013) 86: 350
8. V.K. Lukyanyuk, M.V. Tivarnitskii, Z.D. Kovalyuk, Phys.
Status Solidi A 104, K41 (1987)
9. M. Balkanski, C. Julien, J.Y. Emery, J. Power Sources 26,
615 (1989)
10. N. Romeo, La Rivista del Nuovo Cimento 3, 103 (1073)
11. K.Y. Liu, K. Ueno, Y. Fujikawa, K. Saiki, A. Koma, Jpn
J. Appl. Phys. 32, L434 (1993)
12. V. Le Thanh, M. Eddrief, C. Sebenne, A. Sacuto, M.
Balkanski, J. Cryst. Growth 135, 1 (1994)
13. R. Rudolph, C. Pettenkofer, A. Klein, W. Jaegermann,
Appl. Surf. Sci. 167, 122 (2000)
14. S. El Monkad, M. Eddrief, J.P. Lacharme, K. Amimer,
C.A. Sebenne, Surf. Sci. 352354, 833 (1996)
15. T. Loher, K. Ueno, A. Koma, Appl. Surf. Sci. 130132,
334 (1998)
16. B. Abidri, M. Ghaour, A. Abdellaoui, M. Bouslama, S.
Hiadsi, Y. Monteil, Appl. Surf. Sci. 256, 3007 (2010)
17. G. Fisher, Helv. Phys. Acta 36, 317 (1963)
18. F. Bassani, G. Partori-Parravacini, Nuovo Cimento 50, 95
(1967)
19. H. Kamimura, K. Nakao, Phys. Rev. 22, 1313 (1969)
20. J.V. McCanny, R.B. Murray, J. Phys. C 10, 1211 (1977)
21. S. Nagel, A. Baldereschi, K. Maschke, J. Phys. C 12, 1625
(1979)
22. J. Robertson, J. Phys. C 12, 4777 (1979)
23. P. Gomes da Costa, M. Balkanski, R.F. Wallis, Phys. Rev.
B 43, 7066 (1991)
24. E. Doni, R. Girlanda, V. Grasso, A. Balzarotti, M.
Piacentini, Nuovo Cimento B 51, 154 (1979)
25. Y. Depeursinge, Nuovo Cimento B 64, 111 (1981)
26. Y. Depeursinge, Nuovo Cimento B 38, 153 (1977)
27. M. Schluter, Nuovo Cimento B 13, 313 (1973)
28. M. Schluter, J. Camassel, S. Kohn, J.P. Voitchovsky, Y.R.
Shen, M.L. Cohen, Phys. Rev. B 13, 3534 (1976)
29. A. Bourdon, A. Chevy, J.M. Besson, in Proceedings
of the 14th International Conference on the Physics of
Semiconductors, edited by B.L.H. Wilson, IOP Conf. Proc.
No. 43 (Institute of Physics and Physical Society, London,
1979), p. 1371
30. K. Kunc, R. Zeyher, Europhys. Lett. 7, 611 (1988)
31. P. Gomes da Costa, R.G. Dandrea, R.F. Wallis, M.
Balkanski, Phys. Rev. B 48, 14135 (1993)
32. C. Ulrich, D. Olgun, A. Cantarero, A.R. Goni, K. Syassen,
A. Chevy, Phys. Status Solidi B 221, 777 (2000)
33. D. Olgun, A. Cantarero, C. Ulrich, K. Syassen, Phys.
Status Solidi B 235, 456 (2003)
34. G. Ferlat, H. Xu, V. Timoshevskii, X. Blase, Phys. Rev. B
66, 085210 (2002)
35. K.E. Glukhov, N.K. Tovstyuk, Phys. Status Solidi B 247,
318 (2010)
36. M.O.D. Camara, A. Mauger, I. Devos, Phys. Rev. B 65,
125206 (2002)
37. M.O.D. Camara, A. Mauger, I. Devos, Phys. Rev. B 65,
205308 (2002)
38. M. Sznajder, K.Z. Rushchanskii, L.Yu. Kharkhalis, D.M.
Bercha, Phys. Status Solidi B 243, 592 (2006)
39. D.V. Rybkovskiy, N.R. Arutyunyan, A.S. Orekhov, I.A.
Gromchenko, I.V. Vorobiev, A.V. Osadchy, E.Yu. Salaev,
T.K. Baykara, K.R. Allakhverdiev, E.D. Obraztsova, Phys.
Rev. B 84, 085314 (2011)
40. V.N. Brudnyi, A.V. Kosobutsky, S.Yu. Sarkisov,
Semiconductors 44, 1158 (2010)
Eur. Phys. J. B (2013) 86: 350
41. F. Antonangeli, A. Balzarotti, E. Doni, R. Girlanda, V.
Grasso, M. Piacentini, Physica B 105, 59 (1981)
42. J.F. Sanchez-Royo, J. Pellicer-Porres, A. Segura, V.
Munoz-Sanjose, G. Tobas, P. Ordejon, E. Canadell, Y.
Huttel, Phys. Rev. B 65, 115201 (2002)
43. A. Yamamoto, A. Syouji, T. Goto, E. Kulatov, K. Ohno, Y.
Kawazoe, K. Uchida, N. Miura, Phys. Rev. B 64, 035210
(2001)
44. Zs. Rak, S.D. Mahanti, K.C. Mandal, N.C. Fernelius, J.
Phys: Condens. Matter 21, 015504 (2009)
45. S. Pal, D.N. Bose, Sol. State Commun. 97, 725 (1996)
46. C. Rocha Leao, V. Lordi, Phys. Rev. B 84, 165206 (2011)
47. H.S. Guder, B. Abay, H. Efeoglu, Y.K. Yogurtcu, J. Lumin.
93, 243 (2001)
48. P. Hohenberg, W. Kohn, Phys. Rev. 136, B864 (1964)
49. W. Kohn, L.J. Sham, Phys. Rev. 140, A1133 (1965)
50. L.J. Sham, M. Schluter, Phys. Rev. Lett. 51, 1888 (1983)
51. L.J. Sham, M. Schluter, Phys. Rev. B 32, 3883 (1985)
52. R.W. Godby, M. Schluter, L.J. Sham, Phys. Rev. Lett. 56,
2415 (1986)
53. J.P. Perdew, Int. J. Quantum Phys. S19, 497 (1986)
54. J.P. Perdew, M. Levy, Phys. Rev. Lett. 51, 1884 (1983)
55. M. Jain, J.R. Chelikowsky, S.G. Louie, Phys. Rev. Lett.
107, 216806 (2011)
56. J.P. Perdew, Y. Wang, Phys. Rev. B 45, 13244 (1992)
57. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77,
3865 (1996)
58. A. Fleszar, Phys. Rev. B 64, 245204 (2001)
59. A. Fleszar, M. Pottho, W. Hanke, Phys. Status Solidi C
4, 3270 (2007)
60. J. Heyd, J.E. Peralta, G.E. Scuseria, R.L. Martin, J. Chem.
Phys. 123, 174101 (2005)
61. D. Olgun, M. Cardona, A. Cantarero, Sol. State Commun.
122, 575 (2002)
62. A.D. Becke, E.R. Johnson, J. Chem. Phys. 124, 221101
(2006)
63. F. Tran, P. Blaha, Phys. Rev. Lett. 102, 226401 (2009)
64. P. Blaha, K. Schwarz, J. Luitz, WIEN97, Vienna
University of Technology 1997. [Improved and updated
Unix version of the original copyrighted WIEN code, which
was published by P. Blaha, K. Schwarz, P. Sorantin, S.B.
Trickey, Comput. Phys. Commun. 59, 339 (1990)]
65. J.A. Camargo-Martnez, R. Baquero, Phys. Rev. B 86,
195106 (2012)
Page 9 of 9
66. H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13, 5188 (1976)
67. U. Schwarz, A.R. Goni, K. Syassen, A. Cantarero, A.
Chevy, High Press. Res. 8, 396 (1991)
68. J. Pellicer-Porres, A. Segura, A. San Miguel, V. Munoz,
Phys. Status Solidi B 211, 389 (1999)
69. A. Likforman, D. Carre, J. Etienne, B. Bachet, Acta
Crystallogr. B 31, 1252 (1975)
70. A. Amokrane, F. Proix, S. El Monkad, A. Cricenti, C.
Barchesi, M. Eddrief, K. Amimer, C.A. Sebenne, J. Phys.:
Condens. Matter 11, 4303 (1999)
71. K. Schubert, E. Doerre, M. Kluge, Z. Metallkd. 46, 216
(1955)
72. N. Kuroda, Y. Nishina, Physica B 105, 30 (1981)
73. S.G. Choi, D.H. Levi, C. Martinez-Tomas, V. Munoz
Sanjose, J. Appl. Phys. 106, 053517 (2009)
74. L. Plucinski, R.L. Johnson, B.J. Kowalski, K. Kopalko,
B.A. Orlowski, Z.D. Kovalyuk, G.V. Lashkarev, Phys. Rev.
B 68, 125304 (2003)
75. M. Julien-Pouzol, S. Jaulmes, M. Guittard, F. Alapini,
Acta Crystallogr. B 35, 2848 (1979)
76. D. Olgun, R. Baquero, Phys. Rev. B 51 16891 (1995)
77. K.U. Gawlik, J. Brugmann, S. Harm, C. Janowitz, R.
Manzke, M. Skibowsky, C.H. Solterbeck, W. Schattke,
B.A. Orlowski, Acta Phys. Pol. A 82, 355 (1992)
78. D.W. Niles, H. Hochst, Phys. Rev. B 43, 1492 (1991)
79. J. Olde, G. Mante, H.-P. Barnscheidt, L. Kipp, J.-C. Kuhr,
R. Manzke, M. Skibowski, J. Henk, W. Schattke, Phys.
Rev. B 41, 9958 (1990)
80. J.F. Sanchez-Royo, A. Segura, J. Pellicer-Porres, A. Chevy,
Surf. Sci. 600, 3734 (2006)
81. J.F. Sanchez-Royo, J. Pellicer-Porres, A. Segura, S.J.
Gilliland, O. Safonova, A. Chevy, Surf. Sci. 601, 3778
(2007)
82. S.-W. Yu, T. Lischke, N. Muller, U. Heinzmann, C.
Pettenkofer, A. Klein, P. Blaha, J. Braun, J. Phys.:
Condens. Matter 11, 6715 (1999)
83. J. Pellicer-Porres, F.J. Manjon, A. Segura, V. Munoz, C.
Power, J. Gonzalez, Phys. Rev. B 60, 8871 (1999)
84. K.C. Mandal, R.M. Krishna, T.C. Hayes, P.G. Muzykov,
S. Das, T.S. Sudarshan, S.G. Ma, IEEE Trans. Nucl. Sci.
58, 1981 (2011)
85. R. Rudolph, C. Pettenkofer, A.A. Bostwick, J.A. Adams,
F. Ohuchi, M.A. Olmstead, B. Jaeckel, A. Klein, W.
Jaegermann, New J. Phys. 7, 108 (2005)