PCCP

View Article Online

PAPER View Journal | View Issue

Computer simulations of 3C-SiC under

hydrostatic and non-hydrostatic stresses
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.

Cite this: Phys. Chem. Chem. Phys.,

2016, 18, 8132
H. Z. Guedda,a T. Ouahrani,ab A. Morales-Garcı́a,c R. Franco,d M. A. Salvadó,d
P. Pertierrad and J. M. Recio*d

Received 5th January 2016,
Accepted 5th February 2016
DOI: 10.1039/c6cp00081a
www.rsc.org/pccp
The response of 3C-SiC to hydrostatic pressure and to several uni- and bi-axial stress conditions is
thoroughly investigated using first principles calculations. A topological interpretation of the chemical
bonding reveals that the so-called non-covalent interactions enhance only at high pressure while the
nature of the covalent Si–C bonding network keeps essentially with the same pattern. The calculated
low compressibility agrees well with experimental values and is in concordance with the high structural
stability of this polymorph under hydrostatic pressure. Under uniaxial [001] stress, the c/a ratio shows
a noticeable drop inducing a closure of the band gap and the emergence of a metallic state around
40 GPa. This behavior correlates with a plateau of the electron localization function exhibiting a roughly
constant and non-negligible value surrounding CSi4 and SiC4 covalent bonded units.

1 Introduction undergo different structural phase transformations whose actual

The marked importance of 3C-SiC (also called b-SiC) demands
extensive studies concerning its response to hydrostatic and
non-hydrostatic stresses. At the same time, 3C-SiC constitutes a
paradigmatic system to check how computational simulations
perform under these general stress conditions. In fact, as a
covalent compound, silicon carbide has unique mechanical,
chemical, electrical and thermal properties, which have led to a
number of technological applications.1,2 Among them, SiC has
revealed to be a new outstanding material that can replace
silicon in the fabrication of power electronics semiconductor
devices.2 Under normal conditions, SiC may adopt a large variety
of semiconducting polytypes, which are all very close in energy.3,4
They consist of different stacking of hexagonal Si–C bilayers; the
most common of which are the cubic zinc blende structure with a
repeated unit of three bilayers (3C-SiC), and various hexagonal
structures with repeated units of two (2H-SiC, wurtzite), four
(4H-SiC), and six (6H-SiC) bilayers. Interestingly enough, all SiC
polytypes present the same nearest neighbor tetrahedral coordina-
tion, whereas the main differences between polytypes arise due to
secondary interactions between next nearest neighbors.
By combining high temperature and hydrostatic pressure
(using, for example, diamond anvil cell-type devices), SiC can
a
Laboratoire de Physique Théorique, Université de Tlemcen, 13000 Tlemcen, Algeria
b
École Préparatoire en Sciences et Techniques, 13000 Tlemcen, Algeria
c
MALTA Team and Department of Physical and Macromolecular Chemistry,
Charles University in Prague, Hlavova 2030, 128 40 Prague 2, Czech Republic
d
MALTA Team and Departamento de Quı́mica Fı́sica y Analı́tica,
Universidad de Oviedo, E-33006 Oviedo, Spain. E-mail: jmrecio@uniovi.es
thermodynamic boundaries are not well defined and depend on
several factors such as compression/decompression rates. This is
due to the large energetic barriers associated with the creation/
rupture of strong Si–C covalent bonds and the subsequent hysteresis
involved in transitions.5–7 High pressure behavior of 3C-SiC has
been extensively studied using first principles methodologies since
the pioneering work of Chang and Cohen.8 Besides phase stability,
these authors determined an indirect electronic gap for cubic SiC
and calculated its evolution under applied hydrostatic pressure.
A better agreement with the experimental zero pressure bulk
modulus of this polytype was found by Karch et al.9 who also
reported for the first time phonon related properties of 3C-SiC
after a thorough theoretical investigation. In contrast with the
number of studies dealing with the hydrostatic regime, there
are rare explicit calculations of 3C-SiC under non-hydrostatic
conditions. We would like to notice, however, the extensive work
of Umeno and co-workers as regards the theoretical calculation
of ideal critical strengths of SiC polytypes.10,11 Using methodol-
ogies based on the density functional theory (DFT), these authors
were able to characterize the (anisotropic) mechanical behavior
of 3C-SiC by computing a number of tensile and shear stress–
strain curves involving the effect of multiaxial stress. In parti-
cular, the direction normal to (100) was determined as the one
showing the highest ideal strength for this polytype.
Pressure, as well as temperature, is a factor that determines
the state of materials. As hydrostatic pressure and/or generalized
stresses are applied, solids become denser and can experience
dramatic modifications in their physical properties. Atomic dis-
tances usually shrink, thus influencing the electron distributions
and causing appreciable changes in the conducting/insulator

8132 | Phys. Chem. Chem. Phys., 2016, 18, 8132--8139 This journal is © the Owner Societies 2016

Paper
character of the compound (see, for example, ref. 12 and
references therein). The particular structural and chemical
features of the bonding network in the 3C-SiC compound
provide a paradigmatic case to investigate the effect of these
stresses. More than 250 identified polymorphs of silicon carbide13
are very well known. This rich polymorphism being out of the
scope of this work, our main aim is addressing the under-
standing of how hydrostatic pressure and non-hydrostatic
stresses affect the structure and the chemical bonding network
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.
of the 3C-SiC polytype, with a special emphasis on the implica-
tion of the induced changes in the potential apparition of new
phases. Knowledge of how the basic interactions within the
system evolve under extreme conditions is fundamental for
atomistic models dealing with the mechanical stability of
3C-SiC (see ref. 11), and also for helping in the understanding
of its technological applications. Atomic-level information is
the key in this regard and can be provided by means of the
topological analysis of the electron density, the electron locali-
zation function (ELF) and the reduced density gradient (RDG).
The first two scalar fields are useful for a general description of
the chemical bonding network of 3C-SiC that can be extrapo-
lated to the rest of the SiC polytypes. RDG is introduced to
check if this system responds or not as a layered compound, and
provides also information on dispersion interactions between next
nearest neighbors. Local pressures14,15 and atomic equations of
state are computed to interpret the mechanism of response of
3C-SiC under these stress conditions. We observe that struc-
tural parameters and chemical interactions provide a clear
correlation with the emergence of a metallic state as [001]
uniaxial stress is applied.
The rest of the paper is divided into three sections. Next, we
present computational details of total energy calculations
under generalized stress conditions and a brief introduction
of the topological approaches used in our work. Our results on the
response of 3C-SiC to hydrostatic and non-hydrostatic conditions
are discussed in three separate subsections: structural parameters,
chemical interactions, and electronic structure. Main conclusions
are summarized in the last section.
2 Computational details
2.1 Total energy calculations under generalized stress
conditions
Total energy ab initio calculations have been performed under
the framework of the density functional theory (DFT) as imple-
mented in the abinit code.16 A standard scheme has been
followed using norm-conserving non-local Troullier–Martins
pseudo-potentials17 and the Perdew–Burke–Ernzerhof exchange–
correlation functional18 within the generalized gradient approxi-
mation (PBE-GGA). Some calculations at the LDA level have also
been carried out for the sake of comparison with GGA results.
The set of plane waves was extended up to a kinetic-energy cutoff
of 290 Hartree, and Monkhorst–Pack grids (up to 12  12  12)
appropriate to each structure were used to perform integrations
in the reciprocal space. In fact, convergence in the calculations
This journal is © the Owner Societies 2016
View Article Online
PCCP
was studied by looking at the energy and the optimized lattice
parameter as the grid becomes denser, and sufficient accurate
values (within 105 Hartree and 104 Å, respectively) were
already reached for 10  10  10 grids. To simulate the
mechanically stressed state of the 3C-SiC crystal lattice we
employed the abinit code to perform unit cell optimizations
for a given stress condition. At each stress value, the structures
were fully relaxed using a quasi-Newton scheme combined
with the BFGS algorithm for Hessian updating. Optimiza-
tion was ended when the forces on atoms were lower than
5  105 Hartree bohr1 and a deviation of the stress tensor not
greater than 0.1 GPa was found.
The unit cell of 3C-SiC can be described in terms of Si atoms
in the vertices and centers of the faces whereas four C atoms
lie in the four diagonals of the cube at 1/4 or 3/4 of the
corresponding ends. Therefore, just one structural parameter
needs to be determined in this cubic symmetry. However,
when non-hydrostatic stress is applied, symmetry is lowered
(see below). In all these optimizations symmetry constraints
were not used.
The mechanical state of the crystal was fixed using the value
of the six independent components of the symmetrical stress
tensor:
0 1
s11 s12 s13
B C
B C
s ¼ B s12 s22 s23 C (1)
@ A
s13 s23 s33
Calculations were made at different compressive stress values
of s ranging from 5 to 55 GPa. Four types of stress conditions
were explored: (i) hydrostatic, where the off-diagonal compo-
nents vanish and the three diagonal components are equal and
correspond to the external pressure, s11 = s22 = s33 = s = p,
(ii) uniaxial stress along the cubic [001] direction, where the only
non-zero element is s33 = s, (iii) uniform biaxial stress in
the plane (001) where s11 = s22 = s, and (iv) uniaxial stress
along [111], where the six components were set to the same
value, s/3. Notice that although the stress components are
usually negative for compression, we define s as the negative of
the non-zero components. Therefore, positive values of s will be
used in the presentation and discussion of our results.
2.2 Topological approaches and local pressure
Topological analysis of scalar fields related to the electron density
constitute practical tools to characterize the nature of chemical
interactions in crystalline solids, and allow us to follow their
changes under hydrostatic pressure (see, for example, ref. 19 and
references therein). The topological analysis relies on a partition of
the molecular space achieved in the framework of the theory of
-
gradient dynamics applied to a scalar potential function, say V(r),
which contains the physical or chemical information that we shall
call the chemical potential function. In our work, we have specifi-
cally followed formalisms based on the topological partition
induced by the electron density,20 the electron localization function
(ELF),21 and the so-called reduced density gradient (RDG).22
Phys. Chem. Chem. Phys., 2016, 18, 8132--8139 | 8133

PCCP
In the Atoms In Molecules Theory of Bader (QTAIM), the
scalar function is the electron density distribution (r), whose
basins have their attractors most commonly located on the
nuclei and which are therefore associated with the atoms that
constitute the molecule or the condensed system. Under the
QTAIM the bulk properties of the solids are partitioned into
additive atomic contributions (local properties). This partition
allows us to investigate how the bulk compressibility splits into
local contributions associated with the atoms (Si and C) con-
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.
tained in the crystal and to understand the response of 3C-SiC
to external stresses at a microscopic level. QTAIM analysis is
performed using the CRITIC2 code.23 Since our first principles
calculations are performed using pseudopotentials, the charge
density only contains contributions from the valence electrons.
In order to perform an accurate analysis a total charge density
was built adding a core charge to the valence one. A fine FFT
grid was used to reproduce the correct total charge.
In order to recover a chemist’s representation of a molecule
consistent with Lewis’ valence picture, another local function
that is able to describe the electron pair regions is used. As
interpreted by Savin et al.,24 the ELF core can be understood as a
local measure of the excess of local kinetic energy of electrons due to
the Pauli principle. The analysis of ELF surfaces provides a partition
of the three-dimensional space into non-overlapping basins, which
can be thought of as electronic basins corresponding to bonds (B),
lone pairs (LP), and atomic core shells (C). The ELF has been
defined to have values between 0 and 1 (1 corresponding to
perfect localization) to ease the visualization of its isosurfaces.
All electron wave functions are preferable for a quantitative
analysis of the ELF topology in a crystalline phase. Thus the 1D
ELF profile along the Si–C bond is calculated using ELK,25
a code implementing the all-electron full-potential linearized
augmented-plane wave (FP-LAPW) method.
The total electron density is also the input to carry out the
analysis of non-covalent interactions (NCIs), a visualization index
which relies on the electron density and the reduced-density
gradient (RDG or s).22 The combination of s and r allows a rough
partition of the real space into NCI domains: high-s and low-r
correspond to non-interacting tails, low-s and high-r correspond
to covalent bonds, and low-s and low-r to non-covalent inter-
actions, and more precisely with dispersion-like interactions.
The sign of the second density Hessian eigenvalue (l2) gives us
the classification of interactions as attractive or repulsive. NCI
analysis is also implemented in the CRITIC2 code.23
3 Response of 3C-SiC to hydrostatic
pressure and non-hydrostatic stress
conditions
3.1 Structural properties and the equation of state
As a starting point, we compare our calculated lattice parameter
(a) and equation of state (EOS) parameters, the bulk modulus
(B0) and its first pressure derivative (B0 0 ), both evaluated at zero
pressure, with experimental data26,27 in Table 1. LDA results
tend to underestimate a and overestimate B0 with respect to the
8134 | Phys. Chem. Chem. Phys., 2016, 18, 8132--8139
View Article Online
Paper
Table 1 Calculated lattice parameter (a), bulk modulus (B0) and its
pressure derivative (B0 0 ) of 3C-SiC
a (Å) B0 (GPa) B0 0
LDA 4.307 229 3.83
GGA 4.370 211 3.94
Expt. 4.359a 227b 4.1b
a
From ref. 26. b
From ref. 27.
experimental measurements. Contrary to that, the GGA approach
overestimates a and underestimates B0. This is the usual behavior
and within the expected deviations (in absolute value): 1.2 and
0.2% for a according to LDA and GGA levels of calculation,
respectively, and 1.5% (LDA) and 6.0% (GGA) for B0. The
calculated values of B0 0 are also in good agreement with experi-
mental data. These results agree very well with many other
previous calculations too.28 The low compressibility (k = 1/B0)
of this compound suggests that a high hydrostatic pressure
would be necessary to promote a phase transition. In fact,
this polymorph has been found to be stable up to 100 GPa at
room temperature.29
More interesting is to evaluate the strains on the unit cell under
non-hydrostatic conditions. Both, uniaxial [001] and uniform
biaxial (001) stresses, yield to a tetragonal distortion of the cubic
cell. Whereas the uniaxial stress induces a significant decrease in
the c/a ratio, the shift is opposite when the biaxial stress along the
(001) plane is applied. These are the expected effects considering
the directions of the stresses and the bonding network. Fig. 1(a)
illustrates the behavior of c/a under these non-hydrostatic condi-
tions. Obviously, hydrostatic pressure keeps the unit cell with the
original cubic symmetry of 3C-SiC. As regards the application of
uniaxial [111] stress, the unit cell distorts towards a trigonal
symmetry in which the cell angle increases linearly, less than
5% in 20 GPa (see Fig. 1(b)). In fact, the effect of the [111] stress
on the cubic unit cell of 3C-SiC is weak because the strong
covalent Si–C bonds lie along this direction and are hardly
affected by hydrostatic and non-hydrostatic stresses as we will
detail in the next subsection.
Fig. 1 (a) c/a and (b) cell angle versus hydrostatic and some uniaxial and
biaxial stress conditions.
This journal is © the Owner Societies 2016

Paper
3.2 Chemical interactions and local pressures
By means of the topological partition induced by the electron
density, we evaluate the charges and volumes associated with
C and Si as hydrostatic pressure is applied. The whole picture
results in an ionic compound with a total charge transferred
from Si to C of 2.67 electrons at zero pressure and 2.78 electrons at
60 GPa. Therefore, the most electronegative element, carbon,
plays the role of the anion in 3C-SiC. Consequently, the fractional
occupation of C (75%) in the unit cell is greater than that of Si
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.
(25%). This result does not change very much in the 0–60 GPa
pressure range. The computed evolution of the atomic volumes
allows us to evaluate the atomic equation of state parameters
for C and Si. Although we obtain a greater bulk modulus for
C (B0(C) = 288 GPa, B0(Si) = 138 GPa), once the fractional
occupations are taken into account the anion is the one that
contributes the most to the compressibility of the compound.
Overall, QTAIM provides a consistent description of 3C-SiC in
terms of ions whose charges and volumes are affected by
hydrostatic pressure similar to what happens in many other
compounds analyzed with this formalism.30,31
In the case of the ELF topological analysis, the 1-D profile
along the Si–C bond path clearly reveals the existence of one
bond basin V(C–Si) between the core of C (K-shell) and the core
and outer-core of Si (K and L shells) (see Fig. 2). RB, r1 and r2 are
appropriate parameters to characterize this covalent bond.32 RB
provides a measure of the bond size. Typical values for covalent
bonds are around 1 Å.32 r1 and r2 are the distances from Si
and C, respectively, to the position along the bond path where
the maximum of the ELF in the bond basin is located, and have
different values for heteronuclear bonds. We obtain values of
RB = 0.970 Å, r1 = 1.074 Å, and r2 = 0.810 Å. Notice that although
the ELF maximum is displaced towards the position of Si due
to the greater electronegativity of C, r1 4 r2 because Si has an
outer-core L shell.
In Fig. 3, we plot the 1-D ELF profile along the [111] direction
of the unit cell, thus including C at (0,0,0), Si at (0.25,0.25,0.25)
and C at (1,1,1). A normalized distance is used to put in the
Fig. 2 One dimensional representation of the electron localization
function along the Si–C covalent bond. RB, r1 and r2 stand for the bond
basin size and distances, respectively, from Si and C to the maximum of
the bond basin.
This journal is © the Owner Societies 2016
View Article Online
PCCP
Fig. 3 One dimensional representation of the electron localization func-
tion along the [111] direction of 3C-SiC under hydrostatic pressure (upper
panel) and [001] stress (lower panel).
same figure curves from hydrostatic (upper panel) and non-
hydrostatic (lower panel) conditions with different stresses.
Clearly, the left part of the plots (up to 0.25) contains the same
information as in Fig. 2. Going from 0.25 to 1 along the
normalized distance coordinate, we observe the K and L shells
of Si followed by a region of very low ELF values with a small hill
around 0.4. Approaching the ending C atom, another hill with a
maximum around 0.9 and an ELF value of 0.6 is displayed
before the K-shell of C. It should be pointed out that these hills
are projections of the ELF from other C–Si bond basins.
It is worth indicating the values of the two C–Si distances
involved in these plots. At 0 GPa, the shorter C–Si distance (ds)
is 1.881 Å, whereas the larger one (dl) is 5.644 Å. At 40 GPa,
ds = 1.741 Å and dl = 5.224 Å. Under uniaxial [001] stress of
54 GPa, the reduction of ds and dl is smaller, 1.863 and 5.581 Å,
respectively. The whole ELF profile is very similar regardless of
the hydrostatic pressure. What we compress when pressure is
applied on 3C-SiC are the strong covalent Si–C bonds. This fact
explains the high bulk modulus of this polymorph (227 GPa),
greater than other binary ionic compounds such as MgO with
B0 = 160 GPa. Taking into account the results from both QTAIM
and ELF analysis, we can conclude that the bonding network of
3C-SiC is covalently polarized.
Which is more striking here is to notice that the ELF profile
shows a meaningful enhancement between the no-bonded
Si and C atoms when the uniaxial [001] stress increases up to
54 GPa. Under this uniaxial condition, ELF values around
0.2 dominate the profile in a wide region along this [111]
direction and are indicative of an ELF plateau surrounding
the Si–C covalent bonding network, similar to the ELF picture
between cores in metals. We associate the progressive apparition
of the enhanced ELF profile as the uniaxial [001] stress is applied
(not shown in Fig. 3 for the sake of clarity) with the noticeable
decrease in the c/a ratio induced by the same uniaxial [001] stress
previously depicted in Fig. 1. This correlation is worth to be
emphasized since it is related to the changes in the electronic
structure that will be discussed in the last subsection.
Phys. Chem. Chem. Phys., 2016, 18, 8132--8139 | 8135

PCCP
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.
Fig. 4 RDG versus sign(l2)  r plot for 3C-SiC as a function of uniaxial
[001] stress.
Fig. 5 RDG-based NCI isosurfaces (in green) of the 3C-SiC structure as a
function of [001] stress.
The emergence of non-covalent interactions displaying disper-
sion interactions as the uniaxial [001] stress increases can be
detected in the plots of the RDG versus the density multiplied by
the sign of l2 (see Fig. 4). The peak located at around 0.04
identifies the C–Si covalent bonding and remains constant under
this uniaxial stress indicating that this bond does not change its
character. On the other hand, the non-covalent interactions are
located in the range of 0–0.02. In Fig. 4, we see that the position
of the peaks slightly shifts towards higher density values by the
effect of the uniaxial stress. These peaks are associated with NCI
domains between the atomic positions and Si–C covalent bonds
shown in the 3-D representations of Fig. 5. The observed shifts of
the peaks are clearly illustrated by the increasing sizes and shapes
of the NCI isosurfaces. Overall, NCIs do not play any meaningful
role in the cohesion of 3C-SiC, which is dominated by the covalent
polarized network of bonds. However, these weak interactions
might be responsible for the relative stability of different polytypes
under applied hydrostatic or non-hydrostatic stress.
Once the topological tools have been applied to characterize the
chemical bonding of 3C-SiC and its changes under hydrostatic and
non-hydrostatic conditions, we can analyze at an atomic level
the main mechanism of volume reduction. The definition of local
pressure presented in our previous studies of II–VI polymorphs,14,15
 
@E
pi ¼  ; (2)
@Vi T¼0
8136 | Phys. Chem. Chem. Phys., 2016, 18, 8132--8139
View Article Online
Paper
Fig. 6 Calculated total energy versus volumes of Si and C basins, and the
SiC unit formula in the B3 phase.
where E is the formula unit energy of 3C-SiC and Vi the volume of a
given topological basin, provides a means to evaluate the resistance
of chemically meaningful regions of the unit cell to the applied
stress. Curves in Fig. 6 contain the information to evaluate the local
pressures of the atomic basins of C and Si calculated from QTAIM.
Given a pressure-induced increase in energy from the equilibrium
value (DE), the reduction of the unit cell volume per formula unit
(DV(SiC)) can be understood as the sum DV(C) + DV(Si). The lower
the reduction of the volume the greater the resistance of the atom
to be compressed. Therefore, according to our calculations (see
Fig. 6), the Si atom is the chemical species showing the greatest
resistance. Considering the ionic partition induced by the electron
density, we conclude that it is the space assigned to the anion (C)
the one that experiences a greater reduction as pressure is applied.
This result is also in agreement with our previous analysis in terms
of ionic contributions to the bulk compressibility.
3.3 Electronic structure
Using standard LDA or GGA functionals, it is well known that
band gaps of semiconductors are often underestimated.33 In
an attempt to overcome this shortcoming, Engel and Vosko
constructed a new functional (EV-GGA) that is able to better
reproduce energy gaps at the expense of less accuracy in
structural optimizations.34 To see how hydrostatic and non-
hydrostatic stresses affect the electronic properties of 3C-SiC,
we evaluate the corresponding band structures using the
EV-GGA functional implemented in the WIEN2k code35 for
the optimized strained structures computed with abinit. As
a good compromise between computational efficiency and
accuracy for WIEN2k calculations, we have considered a num-
ber of basis functions up to RMT  Kmax = 9, where RMT is the
minimum radius of the muffin-tin spheres and Kmax gives
-
the magnitude of the largest k vector in the plane wave basis.
The dependence of the energy on the number of k-points in the
irreducible wedge of the Brillouin zone has been examined,
and a 12  12  12 mesh (the same as in the abinit calcula-
tions) has been used.
The calculated band gaps for the optimized strained struc-
tures computed under different hydrostatic, uniaxial and biaxial
conditions are depicted in Fig. 7. At zero pressure, we observe
This journal is © the Owner Societies 2016

Paper
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.
Fig. 7 Energy gap, Eg, of 3C-SiC versus hydrostatic pressure, uniaxial
[001] and [111] and biaxial (001) stresses.
that the gap between the valence band and the conduction gap
is 2.00 eV, in good agreement with the experimental value of
Goldberg et al.36 (2.36 eV). The calculated value from our PBE-
GGA level of calculation is only 1.37 eV, within the expectation
from standard DFT calculations. We have found an interesting
result: the compression along the c-axis ([001] uniaxial stress)
leads to a vanishing band gap in the 3C-SiC polytype at around
40 GPa. Explicitly, in Fig. 8, the band structures of 3C-SiC at
Fig. 8 Band structures of 3C-SiC at zero pressure (upper panel) and at an
uniaxial [001] stress of 40 GPa (lower panel). Symmetry points along the BZ
of a common body centered tetragonal cell are used.
This journal is © the Owner Societies 2016
View Article Online
PCCP
0 GPa (upper panel) and at 40 GPa under [001] uniaxial stress
(lower panel) are displayed. For the sake of comparison, both
band structures are plotted using points in the Brillouin zone
(BZ) of a body centered tetragonal cell. These plots clearly reveal
that the indirect gap at zero pressure is closed at this specific
uniaxial stress.
We have observed that the closure of the gap progresses
as the uniaxial stress is increased. We have shown previously
that the c/a ratio also decreases continuously under the same
stress. Also, as we have noticed before, values of ELF as high as
0.2 were found in the 1-D profile conecting CSi4 and SiC4
tetrahedra along the [111] direction under this [001] uniaxial
stress. Although we have not shown ELF curves for lower [001]
uniaxial stresses, our results indicate a clear progression in the
ELF values as this uniaxial stress increases that has not been
found under the other non-hydrostatic or hydrostatic conditions
studied in this work. This interesting structural–chemical–
electronic correlation is worth to be emphasized. Given the
relevance of this result, we have checked that the optimized
3C-SiC structure under this [001] uniaxial stress (40 GPa) is
dynamically stable: computed phonon dispersion curves show
that all frequencies are positive.
A more detailed analysis in terms of the partial density of the
electronic states associated with each energy interval is pre-
sented in Fig. 9. The zero of energy is chosen to coincide with
the top of the valence band. C(s) orbitals present the highest
contribution in the energy range from 15 to 11 eV for
all uniaxial stresses explored. Between 10 eV and the Fermi
level the dominating C(p) and the Si(p) orbitals show higher
contributions than the couple Si(s)/C(s) with Si(d) states con-
tributing more than s states near the Fermi level. The overall
picture we obtain in the valence region is similar to that found
in ref. 37. In the lower conduction band, p states also show a
higher contribution than s states but now the differences are
smaller and Si(d) contribution is of comparable magnitude.
Compared with ref. 37, our calculation gives an inverted C(p)/C(s)
ratio and a smaller Si(d) contribution. Contributions from
Fig. 9 Partial density of electronic states at zero pressure (upper panel)
and at an uniaxial [001] stress of 42 GPa (lower panel).
Phys. Chem. Chem. Phys., 2016, 18, 8132--8139 | 8137

PCCP
s states become more pronounced with increasing stress leading to s
and p couples with similar profiles at 42 GPa, where the uniaxial
[001] stress induces the closure of the band-gap through a long flat
tail in the conduction band (see Fig. 9, lower panel). It is to be noted
that a semiconductor–metal transition has also been favoured in
other families of materials, such as transition metal dichalcogenide
monolayers (MX2), by the application of non-hydrostatic stresses. It
is worth noting the case of MoS2, where this transition has been
recently observed successfully under hydrostatic and uniaxial stress
conditions. It takes place at around 68 GPa38 under hydrostatic
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.
conditions, and, with the application of small out-of-plane axial
pressure, it happens only between 2 and 3 GPa,39 proving the
efficacy of the uniaxial stress, as in our 3C-SiC polymorph.
4 Summary and conclusions
First principles electron structure calculations complemented
with the topological analysis of r-based scalar fields provide
detailed information on the response of the 3C-SiC polytype to
hydrostatic and non-hydrostatic conditions. We have contrib-
uted to increase the knowledge on the structural, chemical and
electronic changes induced in this compound by hydrostatic
pressure and several uniaxial and biaxial stresses. The covalent
polarized bonding network shows a high stability under all the
conditions studied, whereas chemical interactions among SiC4 and
CSi4 tetrahedra are meaningfully enhanced only under the applica-
tion of uniaxial [001] stress, as it is illustrated by the topological
analysis of the electron localization function and the reduced density
gradient. This effect is originated by the large reduction of the c/a
ratio evaluated under this same stress. Interestingly enough, we have
found that these structural and chemical changes correlate with the
computed drop in the band gap that motivates an electronic
transition from the semiconducting state under ambient conditions
to a metallic behavior of 3C-SiC when the uniaxial [001] stress
increases up to 40 GPa. A more in-depth study would be neccesary
in order to provide a microscopic description of the transition
mechanism and to confirm the structural–chemical–electronic rela-
tionships. Overall, our investigation clearly reveals that theoretical
simulations constitute a helpful strategy to explore different non-
hydrostatic conditions that can disclose otherwise hidden properties
of materials as the metallic behavior of the 3C-SiC polytype.
Acknowledgements
Financial support from the Asturias FICYT Agency (Project
GRUPIN 14-049), the Spanish MICINN and FEDER programs
under project CTQ2012-35899-C02-01, MALTA-Consolider
Ingenio-2010 program under project CSD2007-00045, and
MAT2015-71070-REDC project is gratefully acknowledged.
References
1 A. Addamino, in Silicon Carbide 1973, ed. R. C. Marshall,
J. W. Faust Jr and C. E. Ryan, University of South Carolina
Press, 1974, p. 179.
8138 | Phys. Chem. Chem. Phys., 2016, 18, 8132--8139
View Article Online
Paper
2 Silicon Carbide. Recent Major Advances, ed. W. J. Choyke,
H. Matsunami and G. Pensl, Springer-Verlag, Berlin, 2004.
3 P. Käckell, B. Wenzien and F. Bechstedt, Phys. Rev. B:
Condens. Matter Mater. Phys., 1994, 50, 17037.
4 P. Käckell, J. Furthmüller and F. Bechstedt, Phys. Rev. B:
Condens. Matter Mater. Phys., 1999, 60, 13261.
5 M. Catti, Phys. Rev. Lett., 2001, 87, 035504.
6 M. S. Miao and W. R. L. Lambrecht, Phys. Rev. Lett., 2005,
94, 255501.
7 H. Y. Xiao, F. Gao, X. T. Zu and W. J. Weber, J. Phys.:
Condens. Matter, 2009, 21, 245801.
8 K. J. Chang and M. L. Cohen, Phys. Rev. B: Condens. Matter
Mater. Phys., 1987, 35, 8196.
9 K. Karch, P. Pavone, W. Windl, O. Schütt and D. Strauch,
Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17054.
10 Y. Umeno, Y. Kinoshita and T. Kitamura, Strength Mater.,
2008, 40, 2.
11 Y. Umeno, A. Kubo and S. Nagao, Comput. Mater. Sci., 2015,
109, 105 and references therein.
12 An Introduction to High-Pressure Science and Technology, ed.
J. M. Recio, J. M. Menéndez and A. Otero-de-la-Roza, CRC
Press, Boca Raton, Florida, 2015.
13 R. Cheung, Silicon carbide microelectromechanical systems for
harsh environments, Imperial College Press, 2006, p. 3.
14 T. Ouahrani, J. M. Menéndez, M. Marqués, J. Contreras-
Garca, V. G. Baonza and J. M. Recio, EPL, 2012, 98, 56002.
15 T. Ouahrani, R. Franco, J. M. Menéndez, M. Marqués and
J. M. Recio, J. Alloys Compd., 2014, 592, 150.
16 X. Gonze, G.-M. Rignanese, M. Verstraete, J.-M. Beuken,
Y. Pouillon, R. Caracas, F. Jollet, M. Torrent, G. Zerah,
M. Mikami, Ph. Ghosez, M. Veithen, J.-Y. Raty, V. Olevano,
F. Bruneval, L. Reining, R. Godby, G. Onida, D. R. Hamann
and D. C. Allan, Z. Kristallogr., 2005, 220, 558.
17 N. Troullier and J. L. Martins, Phys. Rev. B: Condens. Matter
Mater. Phys., 1991, 43, 1993.
18 J. P. Perdew, K. Burke and M. Ernezerhof, Phys. Rev. Lett.,
1996, 77, 3865.
19 J. Contreras-Garcı́a, M. Calatayud, J.-P. Piquemal and
J. M. Recio, Comput. Theor. Chem., 2012, 998, 193.
20 R. F. W. Bader, Atoms in Molecules: A Quantum Theory,
Oxford University Press, Oxford, 1990.
21 B. Silvi and A. Savin, Nature, 1994, 371, 683.
22 E. R. Johnson, S. Keinan, P. Mori-Sánchez, J. Contreras-
Garcı́a, A. J. Cohen and W. Yang, J. Am. Chem. Soc., 2010,
132, 6498.
23 A. Otero-de-la-Roza, E. R. Johnson and V. Luaña, Comput.
Phys. Commun., 2014, 185, 1007.
24 A. Savin, O. Jepsen, J. Flad, L. K. Andersen, H. Preuss and
H. G. von Schnering, Angew. Chem., Int. Ed. Engl., 1992,
31, 187.
25 http://elk.sourceforge.net.
26 R. D. Carnahan, J. Am. Ceram. Soc., 1968, 51, 223.
27 I. V. Aleksandrov, A. F. Goncharov, S. M. Stishov and
E. V. Yakovenko, Pis’ma Zh. Eksp. Teor. Fiz., 1989, 50, 116
(JETP Lett., 1989, 50, 127).
28 W. H. Lee and X. H. Yao, Comput. Mater. Sci., 2015, 106, 76.
This journal is © the Owner Societies 2016

Paper
29 M. Yoshida, A. Onodera, M. Ueno, K. Takemura and
O. Shimomura, Phys. Rev. B: Condens. Matter Mater. Phys.,
1993, 48, 10587.
30 V. Luaña, A. Costales and A. Martı́n Pendás, Phys. Rev. B:
Condens. Matter Mater. Phys., 1997, 55, 4285.
31 J. M. Recio, R. Franco, A. Martı́n Pendás, M. A. Blanco,
L. Pueyo and R. Pandey, Phys. Rev. B: Condens. Matter Mater.
Phys., 2001, 63, 184101.
32 J. Contreras-Garca, M. Marqués, J. M. Menéndez and
J. M. Recio, Int. J. Mol. Sci., 2015, 16, 8151.
Published on 12 February 2016. Downloaded by Universite De Liege on 11/3/2020 7:46:01 AM.
33 P. Dufek, P. Blaha and K. Schwarz, Phys. Rev. B: Condens.
Matter Mater. Phys., 1994, 50, 7279.
34 E. Engel and S. H. Vosko, Phys. Rev. B: Condens. Matter
Mater. Phys., 1993, 47, 13164.
This journal is © the Owner Societies 2016
View Article Online
PCCP
35 P. Blaha, K. Schwarz, G. K. H. Madsen, D. Kvasnicka and
J. Luitz, WIEN2k, An Augmented Plane Wave Plus Local
Orbitals Program for Calculating Crystal Properties, Vienna
University of Technology, Vienna, 2001.
36 Y. Goldberg, M. E. Levinshtein and S. L. Rumyantsev, in
Silicon Carbide (SiC), ed. M. E. Levinshtein, S. L. Rumyantsev
and M. S. Shur, New York, 2001, pp. 93–148.
37 G. L. Zhao and D. Bagayoko, New J. Phys., 2000, 2, 16.
38 A. P. Nayak, T. Pandey, D. Voiry, J. Liu, S. Moran,
A. Sharma, C. Tan, C.-H. Chen, L.-J. Lee, M. Chhowalla,
J.-F. Lin, A. K. Singh and D. Akinwande, Nano Lett., 2015,
15, 346.
39 M. Peña-Álvarez, E. del Corro, A. Morales-Garcı́a, L. Kavan,
M. Kalbac and O. Frank, Nano Lett., 2015, 15, 3139.
Phys. Chem. Chem. Phys., 2016, 18, 8132--8139 | 8139