Vous êtes sur la page 1sur 8

Elucidating the High-k Insulator -Al2 O3 Direct/Indirect Energy Band Gap Type

Through Density Functional Theory Computations


R. C. R. Santos,1 E. Longhinotti,1 V. N. Freire2 , R. B. Reimberg3 , E. W. S. Caetano3
1

Departamento de Qumica Analtica e Fsico-Qumica,


Universidade Federal do Cear
a, 60455-900 Fortaleza, CE, Brazil
2
Departamento de Fsica, Universidade Federal do Cear
a, 60455-760 Fortaleza, CE, Brazil and
3
Instituto Federal de Educaca
o, Ciencia e Tecnologia do Cear
a, 60040-531 Fortaleza, CE, Brazil
(Dated: June 10, 2015)
The measured energy band gap of the high-k insulator -Al2 O3 is direct ( ) and equal to
8.80 eV, while DFT-calculated estimates are not conclusive regarding its direct/indirect character,
with values ranging from 6.29 to 9.2 eV. In view of this, we have performed state of the art density functional theory (DFT) computations of the structural and electronic properties of -Al2 O3
employing several exchange-correlation functionals, namely the local density approximation (LDA),
the generalized gradient approximation (GGA: PBE, RPBE, PBESOL); hybrid functionals (PBE0,
HSE06, B3LYP); and the LDA screened exchange approach (sX-LDA). In order to evaluate the
band gaps from the pure DFT functionals (LDA, GGA), the -sol scheme was also adopted as it
allows to improve band gap estimates with low computational cost. For the functionals tested in this
work, we have obtained -sol gaps ranging from 7.703 eV (direct, RPBE) to 8.530 eV (direct, LDA),
while for the hybrid functionals the best figure in comparison with experiment was obtained for the
B3LYP gap (8.674 eV, direct). The sX-LDA calculation, in contrast, predicts that -Al2 O3 is an
indirect gap material with band gap of 8.826 eV, but with a very close direct energy of 8.835
eV (a difference of only 9 meV), in excellent agreement with experimental data. The dependence of
the band gap type of -Al2 O3 as a function of the lattice parameters was investigated as well, and
our results suggest a subtle direct to indirect gap transition in -Al2 O3 for temperatures above 400
K due to thermal expansion.

The aggressive scaling of gate length and equivalent


gate oxide thickness demanded by the International Technology Roadmap for Semiconductors [1] have turned aluminum oxide -Al2 O3 (sapphire or corundum) into an interesting material to replace standard SiO2 technologies.
It has an experimental wide direct band gap of 8.8 [2
6] and a 2.8 eV conduction band offset (similar to SiO2 )
[7, 8]. Efforts to increase its modest dielectric constant
( 9.0) by appropriate doping are being undertaken [9].
Among its applications, -Al2 O3 is used as a substrate
for the growth of silicon and gallium nitride, and as an
active medium in lasers when doped with chromium or
titanium. A detailed description of the -Al2 O3 crystal structure was given by Batra [10]: its primitive unit
cell is rhombohedral (space group R-3C), containing two
Al2 O3 units, but it can also be described using a hexagonal unit cell [11, 12] with lattice parameters a = 4.756
A
and c = 12.982
A [13] at 4.5 K and containing 12 sixcoordinated Al atoms and 18 four-coordinated O atoms.
Barrier height (associated to the gap energy and band
offset) and charge carrier effective masses are key parameters to model the tunneling mechanism which describes
charge injection through a perfect dielectric layer, being estimated through electronic band structure calculations. In the case of -Al2 O3 , the pioneer calculations
for the crystal were non-self-consistent or semiempirical
in nature, but some were developed for small clusters,
as reviewed briefly by Xu and Chin [14]. The first Al2 O3 self-consistent band structure was obtained in a
preliminary way by French [16], soon followed by the
landmark work of Xu and Chin [14] (see also [15]) in
which the first-principles orthogonalized linear combina-

tion of atomic orbitals method (OLCAO) in the local


density approximation (LDA) within the density functional formalism (DFT) was used. They have obtained
an indirect band gap of 6.29 eV involving a transition
from the top of the valence band at a point at 20% of
the X line and the bottom of the conduction band at
[14]; effective masses obtained for the electrons were
0.16 along the X direction, 0.45 along Z, 0.40
along A and 0.38 along D (values given in units
of the free electron mass, mo ), while the hole effective
masses were not evaluated due to the flatness of the valence band [14]. Theoretical DFT works on the -Al2 O3
electronic structure published afterwards have obtained
similar estimates for the gap energy (all of them, however, with direct main gaps) and carrier effective masses
[5, 6, 1721]. Samantaray et al. have found a direct
band gap of 6.1 eV for -Al2 O3 at the point [4] using
the LDA exchange-correlation potential, while Mousavi
et al. [22], using the Full Potential Linearized Augmented
Plane Wave (FP-LAPW) approach employing the EngleVosko approximation (EVA) have found a direct band
gap of 7.2 eV. The electronic structure calculated for a
thin film of -Al2 O3 , by the way, has revealed a band
gap of only 2.6 eV [23], much smaller than the bulk.
The electronic structure of amorphous alumina was investigated using classical molecular dynamics and X
exchange potentials [24], being shown a local oxygen deficiency around Al atoms. On the other hand, GW calculations performed by Marinopoulos and Gr
uning predict
a direct band gap of 9.36 eV for -alumina [25], while
another study by Choi et al. [26], using the hybrid HSE
exchange-correlation functional, predicted a band gap of

2
9.2 eV. A calculated B3LYP band gap of 8.5 eV for Al2 O3 , in contrast, was obtained by Muscat et al. [27],
while a mBJ band gap of approximately 9.0 eV was found
by Lima et al. [9].
In this work, we have performed a series of simulations within the density functional theory formalism to
improve the understanding of the nature of the -Al2 O3
main energy band gap. At room temperature, all experimental data available indicate that this material has a
wide direct band gap, and all theoretical studies agree
that the top of the valence band of -Al2 O3 is very flat.
Such valence band flatness seems to imply that the exact
nature of the band gap of alumina is strongly sensitive to
variations in its lattice parameters. As a matter of fact,
some theoretical reports have found an indirect band gap
[1416] for -Al2 O3 .
The computations we carried out used as initial structure the unit cell lattice parameters at 4.5 K for alumina
found in reference [13], which employed the Mossbauer
wavelength standard to achieve very precise measurements. Geometry optimizations were performed taking into account four parametrizations for the exchangecorrelation functional: the local density approximation
(LDA) model of Ceperley and Alder [28], and the generalized gradient approximation (GGA) models of PerdewBurke-Ernzerhof [29] (PBE), the revised PBE model
(RPBE [30]), and the PBESOL model [31] (optimization for the solid state). As it is known, the Kohn-Sham
band gaps obtained from pure DFT calculations are usually well below experimental values [32]. In order to
address this issue, single point energy calculations were
worked out for the optimized structures using the hybrid functionals PBE0 [33] (PBE optimized geometry),
HSE06 [34](PBESOL optimized unit cell), and B3LYP
[35] (PBESOL optimized unit cell) in order to improve
the band gap estimate (hybrid functional simulations,
while computationally much more expensive than pure
DFT implementations, predict much better band gaps,
comparable to those obtained using more sophisticated
methods such as the QMC and GW approaches [27]). We
have also employed the screened exchange LDA scheme
(sX-LDA [32]) to estimate more accurately the band gap
of the LDA-optimized geometry. In the sX-LDA method,
the exchange contribution to the total energy is splitted
into two parts: a screened, nonlocal and a local density
component. Lastly, we have also applied the -sol technique [36] to obtain gap estimates directly from the pure
DFT exchange-correlation functionals. Overall, we have
found a total of 12 band gap estimates to be discussed
and compared with the experimental value.
In order to perform all the calculations, the plane-wave
basis set CASTEP code [37] was used together with on
the fly generated ultrasoft pseudopotentials (applied for
the geometry optimizations and electronic band structure computations in the LDA and GGA frameworks)
and norm-conserving pseudopotentials (applied when using hybrid functionals and the LDA screened exchange
approach) for the description of the core electrons, with

valence electronic configurations for the oxygen and aluminium atoms being 2s2 2p2 and 3s2 3p1 , in this order. A
Monkhorst-Pack 3 3 2 (4 4 4) grid was adopted
to evaluate integrals in reciprocal space for the hexagonal (rhombohedral) -Al2 O3 unit cell, while the plane
wave cutoff energy was set to 610 eV (750 eV) when
using the ultrasoft (norm-conserving) pseudopotentials,
ensuring a very good quality for the electronic structure it is about twice the cutoff energy values used
in the previous works on the DFT-calculated -Al2 O3
structural and electronic properties. The geometry optimization convergence thresholds were: energy variation smaller than 5.0 106 eV/atom, maximum force
smaller than 0.01 eV/
A, maximum stress smaller than
0.02 GPa and maximum atomic displacement smaller
than 5.0 104
A. Both the internal atomic coordinates
and the unit cell lattice parameters were relaxed, with
quality of the plane wave basis set being kept fixed. The
self-consistent field energy convergence tolerance was set
to 5.0 107 eV/atom with a convergence window of 3
iterations. The pure DFT calculations were carried out
using the larger hexagonal unit cell of alumina, while
the hybrid, sX-LDA calculations were performed using
the smaller rhombohedral unit cell to decrease the computational cost (both cells and the corresponding first
Brillouin zones are shown in Figure 1).
The lattice parameters resulting from the DFT geometry optimization procedures are shown in Table I, where
both the hexagonal and rhombohedral unit cell data are
presented. For the hexagonal structure, the ah (ch ) LDA
lattice parameter is 0.98% (1.0%) smaller than the experimental value, while the volume Vh is 3.0% smaller. This
is not surprising, as the LDA exchange-correlation energy
tends to overestimate the strength of the atomic binding. The GGA results, on the other hand, exhibit lattice
parameters larger than the measured ones (GGA functionals in general underestimates the binding strength
in solids), with the worst figures being obtained for the
RPBE unit cell: 1.3% larger for ah and 1.4% larger for
ch . In the PBE case, ah and ch are, respectively, 0.59%
and 0.66% larger than the X-ray data. Indeed, the best
agreement with the experimental parameters is found for
the PBESOL converged unit cell, with the theoretical
values ah and ch being only 0.11% and 0.15% smaller,
respectively (in this case, the PBESOL seems to slightly
overestimate the binding of the atoms, against typical
GGA behaviour). The angle for the rhombohedral
unit cell, in contrast, is overestimated by both the LDA
(0.05%) and PBESOL (0.03%) approaches, and underestimated when using the PBE (-0.06%) and RPBE (-0.1%)
exchange-correlation energies.
In Figure 2, we have a close-up of the Kohn-Sham band
structure near the main band gap of -Al2 O3 for the
hexagonal unit cell. One can see that the uppermost
valence band has its maximum at the point with very
small dispersion except along the -A line, which means a
large and anisotropic hole effective mass for this material.
The valence band electronic states are mainly originated

3
from O 2p orbitals, with very small contributions from
Al 3p and 3s states. The bottom of the conduction band,
on the other hand, has a much larger curvature, with a
nearly parabolic minimum at the point, indicating an
almost isotropic electron effective mass. These results are
in nice agreement with the paper published by Perevalov
et al. [6], which have predicted a perpendicular hole effective mass of 6.3 (in units of the free electron mass) and
an isotropic electron effective mass of approximately 0.4
using the GGA-PBE functional. The GGA-PBE band
gap we obtained was 6.045 eV, a value a bit smaller than
the value found in Perevalovs study (6.26 eV), probably due to pseudopotential differences. The conduction
band minimum has a very small density of states (about
1 electron/eV), with dominant contribution from Al 3s
states, followed by O 2s and a very tiny amount of Al
3p character. A secondary conduction band at the A
point with energy 6.7 eV has a much larger DOS (about
6 electrons/eV), with strong Al 3s contribution. Above
8.5 eV, the PDOS originates mainly from Al 3p and O
2p orbitals.
As we switch from one exchange-correlation functional
to another, the band gap changes, as shown in Table
2. In all cases, except for the sX-LDA calculation, the Al2 O3 crystal has a direct gap at the point. The RPBE
and PBESOL computations predict the same gap value,
of 5.881 eV, while the LDA result is 6.594 eV. For the
hybrid functionals, the closest match to experiment was
found for B3LYP, 8.674 eV (even better than the previous
calculation by Muscat et al. [27], which have obtained
8.5 eV), followed by PBE0 (8.554 eV), and HSE06 (8.088
eV, about 1 eV smaller than the HSE estimate of Choi
et al.[26]). The screened exchange LDA approach, in
contrast, predicts an indirect gap which is very close to
the experimental data, 8.826 eV, with the maximum of
the valence band along the -F line. The - gap, by
the way, is just 9 meV bigger.
The -sol scheme is a generalization to the solid state
of the self-consistent field approach used in molecular
systems , being based on the variation of unit cell total
energies with the unit cell charge [36]. Its main advantage is the much smaller computational cost in comparison with hybrid functional calculations and other more
advanced techniques for band gap correction. The -sol
corrected gaps for the PBE exchange-correlation family
of functionals are, in increasing order, 7.703 eV (RPBE),
7.885 eV (PBE), and 8.065 (PBESOL) eV (see Table 2).
For the LDA, this method improves the gap to 8.530 eV,
just about 0.3 eV below the measured gap.
From the top (bottom) of Figure 3, one can perform
a comparison between the PBE and PBE0 (LDA and
sX-LDA) electronic band structures in the first Brillouin
zone for the rhombohedral unit cell of alumina. For the
valence band, one can see that the PBE0 hybrid functional shifts the band curves to lower energies relative to
the PBE data, while the conduction bands are energetically shifted upwards. The shapes of the band curves,
however, do not change significantly. The same can be

said for the comparison between the LDA and sX-LDA


outputs, with the difference that the sX-LDA valence
band curves tend to be shifted to higher energy values
relative to the corresponding LDA bands.
Aiming to investigate the dependence of the nature of
the band gap with the unit cell lattice parameters, we
have performed a series of calculations using the LDA
and GGA-PBE functionals starting from the previously
optimized structures. We started by defining a lattice
parameter ratio = a/a0 = c/c0 for the hexagonal unit
cell, where a0 and c0 are the optimized values shown
in Table 1. This parameter was varied from 0.9 to
1.1 (shrinking or enlarging the crystal, as > or < 1,
respectively) and the internal atomic coordinates were
relaxed for each unit cell volume thus obtained. We have
plotted the unit cell total energy and the main energy
band gap as a function of at the top and bottom parts
of Figure 4. From it, one can see that the LDA total
energy is larger than the GGA-PBE value for a given ,
with both displaying a nearly parabolic shape. The main
energy band gap Eg , on the other hand, decreases almost
linearly with as it increases from 0.9 to 1.1. For the
LDA gap curve, Eg decreases from 10.435 eV ( = 0.9)
to 3.285 eV ( = 1.1), with an average rate of decrease of
-3.575 eV per 0.1 , while for the GGA-PBE case we have
a maximum gap of 10.325 eV ( = 0.9) and a minimum
of 2.857 eV( = 1.1), with average rate of decrease 3.734 eV per 0.1 . One can note, for the LDA Eg ()
curve, that the first point, corresponding to = 0.9,
has an energy gap a bit smaller than the expected from
linear extrapolation. It is also distinct due to the fact
that its minimum band gap is indirect, with the valence
band maximum at the point and the conduction band
minimum at M. Indeed, for both LDA and GGA-PBE
band gap curves, one can see that a slight increase of at
least 2% in the lattice parameter suffices to turn alumina
into an indirect gap insulator (see red circles and squares
at the right side of each plot in Fig. 4).
Figure 5 details the direct-indirect gap transition for
unit cell optimized employing the GGA-PBE functional.
At = 1.000 one can see the valence band maximum
at and a small shoulder at approximately one third of
the -K line. For = 1.010 (an increase of 1% in the
lattice parameters), the shoulder turns into a secondary
maximum just 1 meV below the main maximum at , and
for = 1.015, the indirect gap is greater than the direct
value by about 6 meV. Finally, for = 1.020, the indirect
gap is 12 meV above the direct gap at the point. The
valence band curves below the uppermost valence band
are energetically shifted upwards by the increase of . If
one considers the results of the measurements carried out
by Lucht et al. [13] of the lattice parameters of -Al2 O3
as a function of temperature, it is expected an increase
of more than 1.5% percent of a and b (using the T =
4.5 K data as reference) when the temperature is close
to 400 K, so one must hope to observe the direct-indirect
transition of aluminas main band gap as the temperature
increases to nearly that value.

4
In summary, we have presented the result of a series of DFT calculations to estimate more accurately
the main bad gap of -Al2 O3 and to elucidate its character (direct or indirect) as a function of the unit cell
lattice parameter. After geometry optimization, the
PBESOL exchange-correlation functional has predicted
the best lattice parameters in comparison with precise
X-ray diffraction data for alumina at T =4.5 K. For the
band gap, the most accurate results versus the experimental gap Eg = 8.8 eV were obtained by using the
screened-exchange LDA approach, which predicted an indirect band gap of 8.826 eV and a very close direct gap of
8.835 eV, followed by the B3LYP hybrid functional, with
Eg = 8.674 eV (direct). The PBE0 gap was 8.554 eV,
while the HSE06 functional had the worst figure among
the hybrid functionals, 8.088 eV. -sol corrected gaps

were also calculated, with the LDA--sol corrected gap


reaching 8.53 eV, as good as the PBE0 value, notwithstanding the much cheaper computational cost of this
approach. Direct to indirect gap transitions were observed for unit cells optimized using both the LDA and
GGA-PBE functionals as the unit cell parameters are increased, with a direct-indirect transition being found in
the GGA-PBE case for an (a, c) increase of just 1.5% relative to the optimized lattice parameters. This suggests
that a subtle gap type transition as a function of temperature must occur for T 400 K as the lattice parameters
of the crystal increase due to thermal expansion.
Acknowledgments V.N.F. is a senior researcher from
the Brazilian National Research Council (CNPq). E.
W. S. C. received financial support from CNPq project
307843/2013-0.

[1] International Technology Roadmap for Semiconductors:


2012 update review (Semiconductor Industry Association, San Josa, CA). See http://public.itrs.net/.
[2] M. L. Bortz and R. H. French, Appl. Phys. Lett. 55, 1955
(1989).
[3] M. E. Innocenzi, R. T. Swimm, M. Bass, R. H. French,
A. B. Villaverde, M. R. Kokta, J. Appl. Phys. 67, 7542
(1990).
[4] C. B. Samantaray, H. Sim, H. Hwang, App. Surf. Sci.
242, 121 (2005).
[5] S. M. Hosseini, H. A. R. Aliabad, A. Kompany, Eur.
Phys. J. B 43, 439 (2005).
[6] T. V. Perevalov, A. V. Shaposhnikov, V. A. Gritsenko,
H. Wong, J. H. Han, and C. W. Kim, JETP Lett. 85,
165 (2007).
[7] R. Ludeke, M. T. Cuberes, and E. Cartier, Appl. Phys.
Lett. 76, 2886 (2000).
[8] J. Robertson, Eur. Phys. J. Appl. Phys. 28, 265 (2004).
[9] A. F. Lima, J. M. Dantas, and M. V. Valic, J. Appl.
Phys. 112, 093709 (2012).
[10] I. P. Batra, J. Phys. C 18, 5399 (1982).
[11] F. S. Galasso, Structure and Properties of Inorganic
Solids (Pergamon, New York, 1970).
[12] ICSD 2003 Collection, Entry # 88029.
[13] M. Lucht, M. Lerche, H.-C. Wille, Y. V. Shvydko, H. D.
R
uter, E. Gerdau, P. Becker, J. Appl. Cryst. 36, 1075
(2003).
[14] Y.-N. Xu and W. Y. Ching, Phys. Rev. B 43, 4461 (1991).
[15] W. Y. Ching, Y.-N. Xu, J. Am. Cer. Soc. 77, 404 (1994).
[16] R. H. French, J. Am. Ceram. Soc. 73, 477 (1990).
[17] P. W. Peacock and J. Robertson, J. Appl. Phys. 92, 4712
(2002).
[18] W. Y. Ching, L. Ouyang, P. Rulis, H. Yao, Phys. Rev.
B 78, 014106 (2008).
[19] H. A. R. Aliabad, M. R. Benam, H. Arabshahi, Int. J.
Phys. Sci. 4, 437 (2009).
[20] T. V. Perevalov, V. A. Gritsenko, Physics - Uspekhi 53,
561 (2010).
[21] T. V. Perevalov, V. A. Gritsenko, V. V. Kaichev, Eur.
Phys. J. Appl. Phys. 52, 30501 (2010).
[22] S. J. Mousavi, M. R. Abolhassani, S. M. Hosseini, S. A.
Sebt, Chin. J. Phys. 47, 862 (2009).

[23] K. Shiiki, M. Igarashi, H. Kaijyu, Jpn. J. App. Phys. 42,


5185 (2003).
[24] H. Chang, Y. Choi, K. Kong, B.-H. Ryu, Chem. Phys.
Lett. 391, 293 (2004).
[25] A. G. Marinopoulos, M. Gr
uning, Phys. Rev. B 83,
195129 (2011).
[26] M. Choi, A. Janotti, C. G. Van de Walle, J. Appl. Phys.
113, 044501 (2013).
[27] J. Muscat, A. Wander, N. M. Harrison, Chem. Phys.
Lett. 342, 397 (2001).
[28] D. M. Ceperley, B. J. Alder, Phys. Rev. Lett. 45, 566
(1980).
[29] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett.
77, 3865 (1996).
[30] B. Hammer, L. B. Hansen, J. K. Norskov, Phys. Rev. B
59, 7413 (1999).
[31] J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov,
G. E. Scuseria, L. A. Constantin, X. Zhou, K. Burke,
Phys. Rev. Lett. 100, 136406 (2008).
[32] A. Seidl, A. G
orling, P. Vogl, J. A. Majewski, M. Levy,
Phys. Rev. B 53, 3764 (1996).
[33] C. Adamo, V. Barony, J. Chem. Phys. 110, 6158 (1998).
[34] A. V. Krukau, O. A. Vydrov, A. F. Izmaylov, G. E. Scuseria, J. Chem. Phys. 125, 224106 (2006).
[35] A. D. Becke, J. Chem. Phys. 98, 5648 (1993).
[36] M. K. Y. Chan, G. Ceder, Phys. Rev. Lett. 105, 196403
(2010).
[37] S. J. Clark, M. D. Segall, C. J. Pickard, P. J. Hasnip, M.
J. Probert, K. Refson, M. C. Payne, Z. Kristall. 220, 567
(2005).

5
TABLE I: Calculated lattice parameters for -Al2 O3 using
different DFT exchange-correlation functionals versus experiment. Hexagonal (h subscript) and rhombohedral (r subscript) unit cell data are shown. V is the unit cell volume.

ah (
A)
ch (
A)
Vh (
A3 )
ar (
A)
(deg)
Vr (
A3 )

LDA
4.70950
12.8458
246.740
5.07227
55.3220
82.2466

PBE
4.78447
13.0684
259.072
5.15812
55.2626
86.3573

RPBE PBESOL EXP [13]


4.81762 4.75095
4.75630
13.1692 12.9626
12.9820
264.701 253.387
254.338
5.19676 5.11799
5.12510
55.2292 55.3094
55.2937
88.2337 84.4624
84.7792

TABLE II: Calculated energy band gaps for -Al2 O3 using


different DFT exchange-correlation functionals in comparison
with the experimental data. Gap types (direct/indirect) are
also shown
Eg (eV)
LDA
PBE
RPBE
PBESOL
sX-LDA
PBE0
HSE06
B3LYP
EXP[2, 6]

6.594
6.045
5.881
5.881
8.826
( : 8.835)
8.554
8.088
8.674
8.8

Eg (eV) Transition
(-sol) (VBCB)
8.530

7.885

7.703

8.065

FIG. 1: Hexagonal (top) and rhombohedral (bottom) unit cell


representations of -Al2 O3 are shown at the left side, while
the corresponding first Brillouin zones are shown at the right
side. High symmetry points and reciprocal lattice axis are
depicted as well.

FIG. 2: Close-up of the GGA-PBE Kohn-Sham band structure (left) and respective partial density of states (right) near
the main band gap of -Al2 O3 (hexagonal unit cell).

FIG. 3: Top: PBE and PBE0 band structures near the main
band gap of -Al2 O3 (rhombohedral unit cell). Bottom: the
same for the LDA and sX-LDA band structures.

FIG. 4: Unit cell total energy (top) and main band gap (bottom) of -Al2 O3 as a function of the lattice parameter ratio
= a/a0 = b/b0 . Black (red) squares and circles indicate that
the main band gap for the corresponding is direct (indirect).

FIG. 5: Top of the GGA-PBE valence band of alumina for


distinct values of the lattice parameter ratio revealing a
direct to indirect gap transition (the bottom of the conduction
band, not shown here, is at the point for all values).

Vous aimerez peut-être aussi