www.advmat.
de
COMMUNICATION
Synthesis, Structure, and Properties of Boron- and
Nitrogen-Doped Graphene
By L. S. Panchakarla, K. S. Subrahmanyam, S. K. Saha,
Achutharao Govindaraj, H. R. Krishnamurthy, U. V. Waghmare,* and
C. N. R. Rao*
Graphene has emerged as an exciting material because of the
novel properties associated with its two-dimensional structure.[1,2]
Single-layer graphene is a one-atom thick sheet of carbon atoms
densely packed into a two-dimensional honeycomb lattice. It is
the mother of all graphitic forms of carbon, including
zero-dimensional fullerenes and one-dimensional carbon nano-
tubes.[1] The remarkable feature of graphene is that it is a Dirac
solid, with the electron energy being linearly dependent on the
wave vector near the vertices of the hexagonal Brillouin zone. It
exhibits a room-temperature fractional quantum Hall effect[3]
and an ambipolar electric field effect along with ballistic
conduction of charge carriers.[4] It has been reported recently
that a top-gated single-layer graphene transistor is able to reach
electron- or hole-doping levels of upto 5
1013 cm2. The doping
effects are ideally monitored by Raman spectroscopy.[5–10] Thus,
the G-band in the Raman spectrum stiffens for both electron- and
hole-doping, and the ratio of the intensities of the 2D- and G-band
varies sensitively with doping. Doping graphene through
molecular charge-transfer caused by electron-donor and -acceptor
molecules also gives rise to significant changes in the electronic
structure of graphenes composed of a few layers, as evidenced
by changes in the Raman and photoelectron spectra.[6,7]
Charge-transfer by donor and acceptor molecules soften and
stiffen the G-band, respectively. The difference between electro-
chemical doping and doping through molecular charge-transfer
is noteworthy. It is of fundamental interest to investigate how
these effects compare with the effects of doping graphene by
substitution with boron and nitrogen and to understand
dopant-induced perturbations of the properties of graphene.
Secondly, opening the bandgap in graphene is essential
[*] Prof. C. N. R. Rao, L. S. Panchakarla, K. S. Subrahmanyam,
Dr. A. Govindaraj
International Centre for Materials Science
Chemistry and Physics of Materials Unit
New Chemistry Unit and CSIR Centre of Excellence in Chemistry
Jawaharlal Nehru Centre for Advanced Scientific Research
Jakkur P.O., Bangalore 560064 (India)
E-mail: cnrrao@jncasr.ac.in
Prof. U. V. Waghmare
Theoretical Sciences Unit
Jawaharlal Nehru Centre for Advanced Scientific Research,
Bangalore 560064 (India)
E-mail: waghmare@jncasr.ac.in
S. K. Saha, Prof. H. R. Krishnamurthy
Department of Physics, Indian Institute of Science
Bangalore 560012 (India)
DOI: 10.1002/adma.200901285
4726 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
for facilitating its applications in electronics, and graphene
bilayers[11] are an attractive option for this. With this motivation,
we prepared, for the first time, B- and N-doped graphene (BG
and NG) bilayer samples by employing different strategies and
investigated their structure and properties. We also carried out
first-principles density functional theory (DFT) calculations to
understand the effect of substitutional doping on the structure of
graphene as well as its electronic and vibrational properties.
To prepare BGs and NGs, we exploited our recent result in
which it was determined that arc discharge between carbon
electrodes in a hydrogen atmosphere yields graphenes (HG)
composed of two to three layers.[12] The method makes use of the
fact that in the presence of hydrogen, graphene sheets do not
readily roll into nanotubes. In the case of BG, we carried out
the arc discharge using graphite electrodes in the presence
H2 þ B2H6 (BG1) or using boron-stuffed graphite electrodes
(BG2). We prepared NG by carrying out the arc discharge in
the presence of H2 þ pyridine (NG1) or H2 þ ammonia (NG2).
We also performed the transformation of nanodiamond in the
presence of pyridine to obtain a third NG (NG3). All doped
graphene samples were characterized by a variety of physical
methods along with an undoped sample prepared by arc
discharge in hydrogen (HG).
In Figure 1a, we show the typical core-level X-ray photoelectron
spectra (XPS) of BG2 along with the elemental mapping from
electron energy loss spectroscopy (EELS) carried out on
transmission electron microscope (TEM) images. The C 1s
signal is at approximately 285 eV, while the B 1s peak is around
189 eV. The position of this feature indicates that boron is bonded
to carbon in the sp2 carbon network.[13] Based on the XPS data, we
found that BG1 and BG2 contain 1.2 and 3.1 at.% of boron,
respectively. EELS measurements performed on the graphene
nanostructures in the electron microscope indicated the content
of boron in these samples to be 1.0 and 2.4 at.%, respectively. In
Figure 1b, we show the typical XPS data for the NG2 sample along
with the EELS elemental mapping. The asymmetric shape of the
N 1s peak indicates the existence of at least two components. On
deconvolution, we found peaks at 398.3 and 400 eV, the first one
being characteristic of pyridinic nitrogen (sp2 hybridization) and
the second due to nitrogen in the graphene sheets.[13,14] XPS data
indicated that NG1, NG2, and NG3 contain 0.6, 1, and 1.4 at.% of
nitrogen, respectively. EELS results were comparable, indicating
that NG2 contained 0.9 at.% nitrogen.
TEM images clearly show the presence of two to three layers in
the BG and NG samples. In Figure 2a and b, we show the
high-resolution electron microscopy (HREM) images of the BG1
and NG1 samples, respectively. In Figure 3, we show typical
Adv. Mater. 2009, 21, 4726–4730
www.advmat.de
Figure 1. a) C 1s and B 1s XPS signals of B-doped graphene (BG2), EELS
elemental mapping of C and B of BG2. b) C 1s and N 1s XPS signals of
N-doped graphene (NG2), EELS elemental mapping of C and N of NG2.
atomic force microscopy (AFM) images of BG1 and NG1. The
images correspond to two-layer graphenes. The AFM images
generally show the presence of two to three layers in the various
BG and NG samples, with the occasional presence of single
layers. Thermogravimetric analysis indicated that the BG and NG
samples undergo combustion at temperatures only slightly lower
than the parent graphene (580 8C). The BG and NG samples
exhibit higher electrical conductivity than the undoped graphene
(Fig. 4).
Figure 2. TEM images of a) B-doped graphene (BG1), b) N-doped graphene (NG1). Calculated
scanning tunneling microscopy (STM) images of c) B- and d) N-doped bilayers. B- and N-doping which dominates its energetics. The interpla-
results in depletion or addition of electronic charge on carbon atoms on the sublattice of the nar separation is reduced by almost 2.7% in
substituted dopant, which is evident in the weaker black (B) and white colors (N), respectively. B-doped bilayer graphene while it remains
Adv. Mater. 2009, 21, 4726–4730 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
We examined the Raman spectra of all BG and NG samples in
COMMUNICATION
comparison with that of HG, i.e., pure graphene, prepared by the
H2-discharge method (Figure 5a). The Raman spectra (with
632.8 nm excitation) of these samples show three main features
in the 1000–3000 cm1 region,[5,6,9,15] the G-band (1570 cm1),
the defect-related D-band (1320 cm1), and the 2D-band
(2640 cm1). It is noteworthy that the G-band stiffens both
with boron and nitrogen doping as can be seen in Figure 5b. This
is similar to what happens with electrochemical doping[5] but
differs from what occurs when the doping is done by molecular
charge-transfer.[6,7] The blue-shift in the case of BG2, which has a
higher B-content, is larger than that of BG1. In the case of NG, the
sample with the highest N content, NG3, shows the largest shift.
Stiffening of the G-band arises from the nonadiabatic removal of
the Kohn anomaly at the G-point, and its broadening is due to the
absence of blockage of the decay channels of the phonons into
electron–hole pairs.[15] The intensity of the D-band is higher with
respect to that of the G-band in all doped samples. On doping, the
relative intensity of the 2D-band generally decreases with respect
to that of the G-band. It is known that the 2D-band is sensitive to
doping.[5–7] The 2D-band can also be deconvoluted to show the
presence of more than one layer (2 and/or 3). The changes in
the G-band and the other bands in the Raman spectra along
with the XPS and EELS results mentioned earlier suggest that
the dopant B- and N-atoms occur substitutionally. We calculated
the in-plane crystallite sizes (La) of the undoped and doped
graphene samples based on Raman data by the formula:[16]
La ðnmÞ ¼ ð2:4
1010 Þl4 ðID =IG Þ1 (1)
Here, l is the wavelength used for Raman measurements while
ID and IG represent the intensity of the D- and G-bands,
respectively. The crystallite sizes of the HG, BG1, BG2, NG1,
NG2, and NG3 samples are estimated to be 64, 30, 26, 43, 41, and
19 nm, respectively. The doped graphene samples exhibit smaller
crystallite sizes compared to undoped graphene.
We considered two configurations of doped bilayer graphene
in our simulation (4
4 supercell), where the substituted atoms
(3.125% B or 3.125% N) in the two layers are a)
close to and b) far from each other. The
configurations with dopant atoms separated by
larger distances are more energetically favor-
able. While the configuration with well-
separated N-atoms is lower in energy by
25 meV than that with N at the nearest
sites in the two planes, the corresponding
energy difference for B-substitution is
rather small (4 meV). This implies that homo-
geneous B-substitution may be easier than
N-substitution. The origin of this difference
can be traced to the structure. The BC bond
is about 0.5% longer than the CC bond while
the NC bond length is about the same as that
of the CC bond; this results in significant
structural relaxation of the B-doped bilayer,
4727
 www.advmat.de

almost unchanged in N-doped bilayers. We

COMMUNICATION

estimate the energy of formation of doped

Figure 3. Typical AFM images of flakes of B- and N-substituted graphenes: a) BG1, b) NG1.
Figure 4. Electrical conducting data of NG1, BG1, and pristine grapene
(HG).
Figure 5. a) Raman spectra of undoped (HG) and doped (BG and NG) graphene samples. corrections as in the monolayer. This assump-
b) Shifts of the G-band caused by electron (N) doping (with pyridine) and hole (B) doping.
graphene from graphene and gaseous dopant
atoms to be 5.6 and 8.0 eV atom1 for B and N
doping, respectively, suggesting that the
synthesis of BG and NG is indeed possible.
Our calculations reveal that the linearity in
the dispersion of the electronic bands within
1 eV of the Fermi energy is almost unchanged
with B- and N-doping (see Fig. 6). This means
that the doped graphenes have the potential to
exhibit the interesting properties of pristine
graphene. The Fermi energy, which is at the
apex of the conical band-structure near the
K-point of the Brillouin zone of graphene, is
shifted by 0.65 and 0.59 eV in the case of 2
at.% B and N-substitutions. These shifts are of
1.0 and 0.9 eV in 3.125 at.% B- and N-doped
bilayers, respectively. This results in p-type and
n-type semiconducting behavior of graphene
(Fig. 6), as expected. We note that a large amount (96%) of the
shift in Fermi energy arises from substitution with dopants, and
the remaining 4% arises from lattice relaxation. This is also
reflected in the almost symmetric shift in Fermi energy exhibited
by B- and N-substitution, which were seen to be associated with
large and small structural relaxations, respectively. This is indeed
encouraging because this means that the effects of doping should
scale reasonably well to lower concentrations.
As Raman spectroscopy is ideally suited for the characteriza-
tion of doped graphene,[5–10] it is important to know how it
correlates with the concentration of carriers or dopants. The shift
in the G-band frequency measured by Raman spectroscopy has
many physical contributions, and we considered calculations to
uncover their relative magnitudes. Unexpectedly, on the basis of
the lengths of BC and NC bonds, Yang et al.[17] and Coville
et al.[18,19] have reported interesting findings, in which the G-band
shifts are in the same direction as that for carbon nanotubes. We
demonstrate (see Table 1) that the shifts in vibrational frequencies
of graphene with B- and N-doping indeed have opposite signs if
one takes into account only the changes in bond length obtained
at a fixed lattice constant. Relaxation or the change of lattice
constant is highly asymmetric: the lattice constant increases by
0.32% with 2% boron substitution in single-
layer graphene, while it decreases slightly with
N-substitution, resulting in a large decrease
and a slight increase in frequency, respectively.
With these two mechanisms alone, the shift in
G-band frequency with either B- or
N-substitution is negative, contrary to our
measured trends. After adding the dynamic
corrections,[20–22] however, the frequency
shifts become positive, in agreement with
experimental observations reported here.
Similar changes in the G-bands of bilayer
graphene (see Table 1) are estimated from our
calculations with B- and N-substitution,
assuming the same magnitude of dynamic
tion is reasonable for shifts in Fermi energy

4728 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2009, 21, 4726–4730
www.advmat.de

from the introduction of carriers. In fact, the

Figure 6. Electronic structure of substitutionally doped graphene monolayers (a,b), and bilayers
(c,d), with boron and nitrogen, respectively. as is the in-plane lattice constant of the supercell. It is
remarkable that doping results in a small gap opening up in the monolayer and a weak quadratic the electronic structure of graphene or of bilayer
dispersion in the bilayer. This is ideal for applications of graphene in electronic devices.
above 0.39 eV and less than 1 eV (the second band is populated by
electrons or holes, as seen in Fig. 2) because the density of states of
the bilayer matches with that of the monolayer, and hence the
dynamic corrections are expected to be similar (see Fig. 3 and 4 of
Ref. 19). Thus, the comparison between experiment and
calculations shown here can be used to conclude that dynamic
corrections are important in understanding the changes
in the phonon frequencies of graphene and of bilayers arising
Table 1. Various contributions to the shifts in phonon frequencies (cm )
of graphene and bilayers resulting from B- or N-substitution.
G-band Fixed Relaxed
lattice lattice
COMMUNICATION
overall neutrality of substitutionally doped
graphene allows a clean and accurate calculation
and further testing of how dynamic corrections
arising from doping can affect the G-band.
Our observation of the nonsegregating ten-
dency or the homogeneity in the distribution of
boron implies that the disorder or the number of
possible configurations will increase with the
concentration of dopant atoms and result in
more prominent peaks of the D-band, as
evidenced in the measured Raman spectra
(Fig. 5). Secondly, the changes in frequency of
the Raman 2D-band can be understood from our
results for the phonons at K þ DK accessible in
the spectrum of G-point phonons, determined
for the 4
4 supercell. We found that the shifts
in the frequency of phonons at K þ DK are
comparable to the observed values (see Table 1,
lower panel) and much weaker in magnitude
than those at K, re-affirming the strong
Kohn-anomaly in the phonons at the K-point.
We note that the eigenmodes of the G- and
D-band (Fig. 7) also change with doping; features
in the atomic displacements develop near the
dopant atoms, providing a physical picture as to
why these modes could be used to characterize
the nature of the dopant atom and its effects on
graphene.
Having demonstrated that different routes
are possible for the synthesis of BG and NG, and based on a
consistent agreement between experiments and calculations, and
based on derived com-plementary information with respect to the
electronic structure, we conclude that BGs and NGs can be
synthesized to exhibit p- and n-type semiconducting electronic
properties that can be systematically tuned with the concentration
of B and N, and that they can be characterized with Raman
spectroscopy. Interestingly, elemental doping and electrochemical
doping produce similar shifts in the Raman G-band, but molecular
1 charge-transfer gives rise to different effects. Realization of such p-
and n-type conducting graphene bilayers should be usable in a
variety of devices similar to those based on semiconductors.
With dynamic
correction

Pristine monolayer graphene 1570.7 1570.7 1570.7 Experimental

2% Boron-doped monolayer 1579.4 1561.2 1584.6
Synthesis of Boron- and Nitrogen-Doped Graphene: BG1 was prepared by
2% Nitrogen-doped monolayer 1546.5 1553.4 1574.4
performing the arc discharge of graphite electrodes in the presence of
Pristine bilayer graphene 1603.8 1603.8 1603.8
hydrogen, helium, and diborane (B2H6). B2H6 vapor was generated by the
3.125% Boron-doped bilayer 1609.7 1582.4 1617.8
addition of BF3-diethyl etherate to sodium borohydride in tetraglyme. B2H6
3.125% Nitrogen-doped bilayer 1566.1 1577.7 1609.9
vapor was carried to the arc chamber by flowing hydrogen (200 Torr, 1
D-bands Frequency D-bandshift 2D-bandshift Torr ¼ 133.32 Pa) and subsequently He (500 Torr) through the B2H6
Pristine bilayer graphene 1404.1 0.0 0.0 generator. BG2 was prepared by carrying out the arc discharge using
3.125% Boron-doped bilayer 1387.2 16.9 33.2 boron-packed graphite electrodes (3 at.% B) in the presence of H2 (200
3.125% Nitrogen-doped bilayer 1400.0 4.1 8.2 Torr) and He (500 Torr).
NG1 was prepared by performing the arc discharge of graphite
‘‘Fixed lattice’’ refers to an internally relaxed structure keeping the lattice constant the electrodes in the presence of H2, He, and pyridine vapor. Pyridine vapor
same as that of an undoped system; ‘‘Relaxed lattice’’ refers to a system where the was carried to the arc chamber by flowing hydrogen (200 Torr) and
lattice constant is also relaxed for doped configurations. Dynamic corrections are subsequently He (500 Torr) through a pyridine bubbler. NG2 was prepared
obtained using the formalism in Refs. 20 and 21. by carrying out arc discharge of graphite electrodes in the presence of

Adv. Mater. 2009, 21, 4726–4730 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 4729

COMMUNICATION
Figure 7. Atomic displacements of the G-band and D-band modes calculated a,d) for pristine
bilayers, b,e) for B-doped bilayers, and c,f) for N-doped bilayers. The extended modes evolve to
exhibit localized features upon doping: boron is displaced more than carbon because it is lighter,
and the displacement of nitrogen is much weaker than that of carbon. Dynamical corrections
have not yet been added to the provided frequencies.
H2 (200 Torr), He (200 Torr), and NH3 (300 Torr). The transformation
of nanodiamond was carried out in the presence of He and pyridine
vapor at 1650 8C to obtain NG3. We also carried out the transformation
of nanodiamond in the presence of NH3 to obtain an additional NG
sample.
Sample Characterization: Raman spectra were recorded with LabRAM
HR high resolution Raman spectrometer (Horiba Jobin Yvon) using a
He–Ne Laser (l ¼ 632.8 nm). TEM images were obtained with a JEOL JEM
3010 instrument. AFM measurements were performed using a NanoMan
instrument. XPS spectra were recorded using a VG Scientific ESCA LAB V
spectrometer. EELS were recorded using the high-resolution images
obtained with a transmision electron microscope (FEI, TECNAI F30)
equipped with an energy filter for EELS operating at 200 kV. Thermogravi-
metric analysis (TGA) was carried out using Mettler Toledo TGA 850
instrument. DC resistance measurements were carried out by the standard
four-probe method using the resistivity option in the Physical Property
Measurement System (PPMS), Quantum Design, USA.
Computational Methods: Our first-principles calculations are based on
ultra-soft pseudopotentials and density functional theory as implemented
in Quantum Espresso [23]. The electron exchange correlation energy was
approximated with the PBE [24] form of the generalized gradient
approximation functional for single-layer graphene and the PZ [25]
form of the local density approximation functional for bilayer graphene
because the latter describes the binding between the two graphene planes
more realistically. Doping with B and N was simulated by replacing one
carbon atom in a 5
5 (4
4) supercell, amounting to 2 (3.125) at.%
substitution in the graphene monolayer (bilayer). In each case, the
structure was fully relaxed to the lowest energy. Kohn–Sham wave functions
were expressed in the plane-wave basis with an energy cutoff of 40 Ry, and
integrations over the Brillouin zones were sampled with a 9
9
1
(6
6
1 for bilayer) mesh of uniformly spaced K-points. Vibrational
properties were determined using density functional theory linear response
formalism within the adiabatic approximation, and further dynamic
corrections were obtained using the time-dependent perturbation theory
[21] and the linear form of electronic band dispersion at the Fermi level.
Our methodology and calculational parameters were previously tested for
convergence [26,27].
Acknowledgements
L.S.P. and K.S.S. gratefully acknowledge CSIR, New Delhi for senior
research fellowships, and S.K.S. acknowledges financial support from the
4730 ß 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advmat.de
JNCASR. We would also like to acknowledge use of
the central computing facility of the Centre for
Computational Materials Science at JNCASR.
Received: April 16, 2009
Published online: August 15, 2009
[1] A. K. Geim, K. S. Novoselov, Nat. Mater. 2007, 6,
183.
[2] M. I. Katsnelson, Mater. Today 2007, 10, 20.
[3] K. S. Novoselov, Z. Jiang, Y. Zhang, S. V. Morozov,
H. L. Stormer, U. Zeitler, J. C. Maan, G. S. Boebinger,
P. Kim, A. K. Geim, Science 2007, 315, 1379.
[4] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang,
Y. Zhang, S. V. Dubonos, I. V. Grigorieva, A. A.
Firsov, Science 2004, 306, 666.
[5] A. Das, S. Pisana, B. Chakraborty, S. Piscanec, S. K.
Saha, U. V. Waghmare, K. S. Novoselov, H. R.
Krishnamurthy, A. K. Geim, A. C. Ferrari, A. K. Sood,
Nat. Nanotechnol. 2008, 3, 210.
[6] B. Das, R. Voggu, C. S. Rout, C. N. R. Rao, Chem.
Commun. 2008, 5, 155.
[7] a) R. Voggu, B. Das, C. S. Rout, C. N. R. Rao, J. Phys. Condens. Matter 2008,
20, 472204. b) W. Chen, S. Chen, D. C. Qi, X. Y. Gao, A. T. S. Wee, J. Am.
Chem. Soc. 2007, 129, 10418.
[8] A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S.
Piscanec, D. Jiang, K. S. Novoselov, S. Roth, A. K. Geim, Phys. Rev. Lett.
2006, 97, 187401.
[9] A. C. Ferrari, Solid State Commun. 2007, 143, 47.
[10] K. S. Subrahmanyam, S. R. C. Vivekchand, A. Govindaraj, C. N. R. Rao, J.
Mater. Chem. 2008, 13, 1517.
[11] T. Ohta, A. Bostwick, T. Seyller, K. Horn, E. Rotenberg, Science 2006, 313,
951.
[12] K. S. Subrahmanyam, L. S. Panchakarla, A. Govindaraj, C. N. R. Rao, J. Phys.
Chem. C 2009, 113, 4257.
[13] a) L. S. Panchakarla, A. Govindraj, C. N. R. Rao, ACS Nano 2007, 1, 494.
b) K. Raidongia, D. Jagadeesan, M. Upadhyay-Kahaly, U. V. Waghmare,
S. K. Pati, M. Eswaramoorthy, C. N. R. Rao, J. Mater. Chem. 2008, 18, 83.
[14] R. Sen, B. C. Satishkumar, A. Govindaraj, K. R. Harikumar, G. Raina,
J.-P. Zhang, A. K. Cheetham, C. N. R. Rao, Chem. Phys. Lett. 1998, 287, 671.
[15] S. Pisana, M. Lazzeri, C. Casiraghi, K. S. Novoselov, A. K. Geim, A. C.
Ferrari, F. Mauri, Nat. Mater. 2007, 6, 198.
[16] L. G. Cançado, K. Takai, T. Enoki, M. Endo, Y. A. Kim, H. Mizusaki, A. Jorio,
L. N. Coelho, R. M. Paniago, M. A. Pimenta, Appl. Phys. Lett. 2006, 88,
163106.
[17] Q. Yang, P. Hou, M. Unno, S. Yamauchi, R. Saito, T. Kyotani, Nano Lett.
2005, 5, 2465.
[18] K. C. Mondal, A. M. Strydom, R. M. Erasmus, J. M. Keartland, N. J. Coville,
Mater. Chem. Phys. 2008, 111, 386.
[19] E. N. Nxumalo, V. O. Nyamori, N. J. Coville, J. Organomet. Chem. 2008,
692, 2942.
[20] T. Ando, J. Phys. Soc. Jpn. 2006, 75, 124701.
[21] M. Lazzeri, F. Mauri, Phys. Rev. Lett. 2006, 97, 266407.
[22] T. Ando, J. Phys. Soc. Jpn. 2007, 76, 104711.
[23] S. Baroni, S. de Gironcoli, A. Dal Corso, P. Giannozzi, http://
www.pwscf.org (last accessed May 2008).
[24] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865.
[25] J. P. Perdew, A. Zunger, Phys. Rev. B 1981, 23, 5048.
[26] S. K. Saha, U. V. Waghmare, H. R. Krishnamurthy, A. K. Sood, Phys. Rev. B
2007, 76, 201404.
[27] S. K. Saha, U. V. Waghmare, H. R. Krishnamurthy, A. K. Sood, Phys. Rev. B
2008, 78, 165421.
Adv. Mater. 2009, 21, 4726–4730