You are on page 1of 12

Computational and Theoretical Chemistry 1151 (2019) 12–23

Contents lists available at ScienceDirect

Computational and Theoretical Chemistry


journal homepage: www.elsevier.com/locate/comptc

Electronic structure of boron and nitrogen doped isomeric graphene T


nanoflakes

Cesar Gabriel Vera de la Garza, Esau Martínez Olmedo, Serguei Fomine
Instituto de Investigaciones en Materiales, Universidad Nacional Autónoma de México, Apartado Postal 70-360, CU, Coyoacán, México, DF 04510, Mexico

A R T I C LE I N FO A B S T R A C T

Keywords: Electronic properties of nitrogen and boron doped isomeric graphene nanoflakes have been explored using
Haeckelite dispersion corrected B3LYP functional and CASSCF methods. The most thermodynamically stable isomers of
Ionization energy nitrogen and boron doped systems contain phenalene and azulene motifs substituted in positions 7 and 9, re-
Doping with boron spectively. Nitrogen doping promotes nanoflake planarity, increases singlet-triplet gap and a band gap, while
Doping with nitrogen
boron doping promotes dome shaped nanoflake geometry, polyradicalic ground state, reduces singlet-triplet gap.
Electron affinity
Isomeric nanoflakes are less sensitive to doping compared to graphene nanoflakes. Nitrogen is a weak n type
dopant for isomeric nanoflakes, while boron cannot be considered as a p type dopant. This effect is explained by
non-uniform electron density distribution in pristine isomeric nanoflakes as compared to graphene nanoflakes.

1. Introduction metal oxides have been discovered long before graphene [12] and
graphene discovery promoted detailed study of their properties.
Two dimensional (2D) structures have recently been recognized as Others, like 2D boron nitride (BN) have been prepared shortly after
potentially very important materials for multiple applications in op- graphene discovery [13]. These 2D layered materials demonstrate in-
toelectronic, chemical sensing, battery fabrication, supercapacitor and teresting electronic properties including superconductivity [14,15] and
transparent semiconductor manufacturing [1–3]. The most remarkable have been considered promising in battery fabrication, catalysis, as
and the most studied 2D material today is definitely graphene; one sensors and radiation detectors [16–20].
atom thick carbon material discovered in 2004 [4,5]. Graphene is made Carbon is a unique element. The carbon chemistry is the richest
of fused benzene rings, benzene itself is a very stable molecule im- chemistry out of all elements in the Periodic Table. This is due to very
parting to graphene extraordinary structural and mechanical stability. special combination of carbon atom properties; small size and large
Moreover, graphene shows prominent electronic and magnetic prop- number of valence electrons allowing formation of strong hetero and
erties [6]. Among them are: very high charge carrier mobility, room homo nuclear bonds. As a consequence graphene is not the one and the
temperature Hall and ambipolar effects to name a few. Delouse et al. only possible 2D structure made of carbon atoms exclusively.
explored theoretically graphene nanoribbons using the formalism of Nevertheless, the ability of carbon atoms to form different 2D allo-
crystalline orbitals [7]. It has been found out, that symmetry-breakings tropies has been relatively overlooked. The number of studies related to
in spin-densities for singlet states, is the consequence of an approximate the detailed search of 2D carbon allotropies is rather limited [21–23].
treatment of electron correlation in single-determinant approaches, The most complete study explores possible 2D carbon allotropies using
and, therefore this is a methodological artefact. Rayne and Forest [8,9] density functional tight binding approach (DFTB) calculations [21]. All
studied graphene nanoribbons assuming a closed shell singlet ground found allotropies are less stable than graphene, but the two most stable
state. They concluded that graphene nanoribbons have closed shell of them called haeckelites are very interesting because of their stability
ground state with vanishing degenerated singlet and triplet states at and chemical relevancy. Haeckelites are made of fused pentagons and
polymeric limit. In a recent research [10,32] a conclusion has been heptagons, producing 2D structures with azulene motif (Fig. 1).
reached that graphene nanoribbons, present strong multiradicalic We have demonstrated in our recent work using hybrid density
character [11]. functional (DFT) methods [24] that haeckelites are structurally stable
The outstanding properties of graphene have stimulated intense systems representing a minimum at the potential energy surface. This
search for other 2D materials. Some of the 2D materials like MoS2 and is, however, no surprise since the haeckelite structures can be found in


Corresponding author.
E-mail address: fomine@unam.mx (S. Fomine).

https://doi.org/10.1016/j.comptc.2019.01.022
Received 16 October 2018; Received in revised form 17 January 2019; Accepted 27 January 2019
Available online 29 January 2019
2210-271X/ © 2019 Published by Elsevier B.V.
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Fig. 1. D3bj dispersion corrected B3LYP/def2-SVP optimized ground states of H1 graphene allotropy NFs doped with boron and nitrogen. The bond lengths increase
from red to green. The shortest and the longest bond lengths are of 1.340 and 1.613 Å, respectively. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)

the real world as topological defects in graphene [23]. These defects are reduction [24]. Haeckelite nanoflakes (NFs) have a finite bandgap (Eg)
called Stone-Wales-type defects. Haeckelites are metallic [21] however, which decreases with the NF size [24]. We have discovered that unlike
their electronic properties can be modified by the dimensionality graphene, graphene NFs, and 2D haeckelites, which are plane,

13
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

haeckelite NFs are often not plane and planar structures are actually very different from that of the inner atoms.
transition states, not a minimum. Another important difference be- Doped isomeric graphene NFs are denoted as X-Y-mnK where X
tween graphene and haeckelite NFs is that the latter have single re- stands for the dopant type, N or B, Y – for the NF type, H1 or H2, and
ference ground state which facilitates their description within DFT mnK is the dopant position code where the first two numbers denote
frame. Another approach to the properties modification apart from the the position of dopant atoms and the last letter-the distance between
dimensionality reduction is the introduction of chemical defects either them, L (large) or S (short). Graphene NFs are denoted as X-G-K, where
by doping or by chemical modification. This approach has successfully X stands for the dopant type, G – for graphene and K is the distance
been applied to graphene, graphene nanotubes and graphene nano- between dopant atoms, similar to isomeric graphene NFs. Cation or
flakes [25,26]. The present work explores properties of haeckelite NFs anion radicals are referred to with + and − signs, respectively.
doped with boron and nitrogen atoms. Unlike graphene where all car-
bons are equivalent, thus reducing the number of possible structures for 3. Results and discussion
doped systems, in haeckelites and especially haeckelite NFs, there are
many nonequivalent positions increasing number of possible isomers. 3.1. Relative stability and geometry
Therefore, this study focuses only on two variables; the distance be-
tween chemical defects and the nature of these defects. Boron and ni- Table 1 shows the relative stability of doped haeckelite and gra-
trogen atoms have been selected as a dopants since they belongs to the phene NFs. As seen from Table 1 the most stable isomers are always H2
same period as carbon ensuring strong electron interaction with carbon doped systems. Actually, pristine H2 NF is 43.1 kcal/mol more stable
atoms, but they differ in electronegativity and the number of valence than H1. The doped NFs continue this trend. However, the most stable
electrons. NFs have different doped positions for B- and N-doped species. As seen,
the most stable N-doped system is N-H2-11L, while the most stable B-
2. Computational details doped NF is B-H2-22S. Nature of the dopant atom strongly affects the
stability of the NF. While N-H2-11L is the most stable among N-doped
All geometry optimizations were carried out using TURBOMOLE 7.2 systems, B-H2-11L has the highest total energy out of all B-doped NFs.
code [27]. We applied D3bj dispersion corrected B3LYP functional [28] The position of dopant atom also impacts the stability of a NF. Thus, the
in combination with def2-SVP basis set for all geometry optimizations. relative energies of doped NFs of the same type, which differ only by
B3LYP functional has been shown to be a reliable tool for predicting the the position of doping atom can vary by as much as 40 kcal/mol.
geometry of conjugated polymers [29]. Although functionals with Figs. 1 and 2 depicture optimized geometries of NFs where the bond
higher fraction of Hartree-Fock exchange like BHandHLYP and M062x lengths are color coded; the shortest are red, the longest are green.
predict the bond length alternation (BLA) pattern slightly better in When comparing the relative energies of doped NFs with different do-
conjugated systems [30], the high fraction of HF exchange in these pant positions it is clearly seen, that the relative stability of isomers
functionals may lead to the artificial overstabilization of the high spin depends not on the distance between the dopant atoms but rather on
states due to spin contamination. High spin contamination is a result of their position within a NF. These “low energy” locations are different
the multireference character of the ground state which is not un- for N- and B-doped systems. The lowest energy isomers for H1 and H2
common in 2D conjugated systems [11]. In some cases where the ex- N-doped NFs are 11L and 11S, respectively, while for B-doped NFs the
pectation values of 〈S2〉 exceed 10% of theoretical value, a complete most stable isomers are 22L and 22S, respectively. As seen from Figs. 1
active space calculations (CASSCF) in combination with 6-31G* basis and 2, in all the most stable nitrogen doped NFs nitrogen atom replaces
set with active space consisting of 14 electrons and 14 orbitals were carbon atoms in the position 7 of phenalene moiety of the NF, while in
performed to decrease the effect of spin contamination on the relative the case of B-doped NFs, boron replaces position 9 of azulene motifs.
energies of the different spin states. Gaussian 16 code has been used for Unlike graphene, H1 and H2 NFs have strong BLA [24] which can
CASSCF calculations [31]. also be appreciated in Figs. 1 and 2. Doping causes local deformation of
Triplet stability analysis calculations have been carried out for all the NF geometries. As seen, CeB bonds are shorter compared to the
closed shell singlet wavefunctions. If the instability was detected, the original CeC bonds, while CeN bonds are larger. This is in line with an
molecular geometry had been reoptimized using unrestricted “open increase of the bond energies from CeN to CeC and further to CeB
shell singlet” broken symmetry (BS) wavefunction. For Eg estimations [32]. Doping not only causes local geometry changes, but also global
two different methods have been used, the first one is the HOMO-LUMO ones. Fig. 3 shows the ground state geometries of pristine H1 and H2
energy difference extracted directly from DFT calculations. It has been NFs and some of the N- and B-doped NFs.
shown that simple B3LYP HOMO-LUMO difference reproduces sur- As it has been shown earlier [24]. H1 NF is a dome shaped while H2
prisingly well Eg of conjugated polymers [28]. The second is the time isomer is totally plane. Doping with N and B atoms promotes different
dependent (TD) implementation of MN15 functional in combination NF shapes. Thus, all H2 NFs doped with N are plane in their ground
with def2-SVP basis set. This model has previously been tested to show state according to the frequency calculations. On the other hand, all H1
excellent results for prediction of the lowest excitation energies of na- NFs doped with both N and B atoms are dome shaped. However, in the
nocene [24]. case of B-doped H2 NFs the situation is different. While pristine H2 NF
Frequency calculations have been performed for all studied NFs to is plane, all B-doped are dome shaped. To quantify this shape change on
ensure that the structures are the true minima having no imaginary doping we calculated the root mean square (RMS) deviation from plane
frequencies. for each NF, and the results are shown in Table 1. The results show that
Two isomeric haeckelite NFs have been selected as a model for the nature and the position of dopants tune the NF shape. We could
doping, H1 and H2 (Figs. 1 and 2). They contained two dopant atoms, summarize various trends inspecting Table 1.
which corresponds to 2.1 atomic %. Even number of dopant atoms is
important to maintain formal closed shell electronic structure in doped 1. Nitrogen doping promotes planarity of a NF while doping with
molecules, thus allowing a valid comparison between pristine and boron promotes non planarity. As seen, all N-doped H1 NFs have
doped systems. The positions of the doping centers have been selected smaller deviation from plane compared to pristine H1 NF. In case of
to explore different types of the location sites and different distances H2 NFs, all N-doped and pristine NF’s are plane. On the other hand,
between doping centers, thus avoiding heteroatom-heteroatom bonds all boron doped H1 NFs show larger deviation from plane in com-
formation and edge atoms replacement. NeN and BeB bonds are no- parison with pristine H2. Moreover, all B-doped H2 NFs except for
tably weaker compared to CeN and CeB ones [32], thus decreasing the B-H2-11L one are dome shaped unlike plane H2 NF.
stability of doped NFs, while chemical environment of edge atoms is 2. The effect of doping on the NF shape is the most notable for closely

14
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Fig. 2. D3bj dispersion corrected B3LYP/def2-SVP optimized ground states of H2 graphene allotropy NFs doped with boron and nitrogen. The bond lengths increase
from red to green. The shortest and the longest bond lengths are of 1.300 and 1.585 Å, respectively. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)

located dopant atoms in both cases nitrogen and boron. As seen from to boron doped NFs where dopant atoms are well separated.
Table 1, N-doped H1 type NFs show significantly lower deviation
from plane for closely positioned nitrogen atoms compared to the Generally speaking, closely located dopant atoms affect the NF
more separated locations. In the case of boron doping, closely placed shape in greater extent than separated dopant atoms.
boron atoms promote non planarity in much larger extent compared There are two different factors of doping that could affect the NF

15
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Table 1 factor. As seen from Table 1, the deviation from plane for charged H1
Relative energies of doped isomeric graphene nanoflakes and deviations from species are very similar to that for neutral H1 and very different from
plane of these NFs. that for doped NFs, suggesting that it is the structural factor (difference
Nanoflake Erel, kcal/mol Dev, Å in the bond lengths) that tunes shapes of doped NFs. This hypothesis is
also confirmed by the analysis of the molecular shape of dications,
N-H1-11L 36.4 36.6 dianions, tetraanions and tetracations of doped NFs. Dications and
N-H1-11S 34.4 15.2
tetracations of N-doped NFs are isoelectronic to neutral pristine NFs,
N-H1-12L 55.6 31.6
N-H1-12S 56.6 9.93 and neutral B-doped NFs, respectively. Dianions and tetraanions of B-
N-H1-22L 74.5 34.8 doped NFs are isoelectronic to neutral pristine and neutral N-doped
N-H1-22S 73.0 12.0 (9.3)a (3.02)c NFs, respectively. As seen from Table 1 the deviation from plane is
B-H1-11L 95.4 62.3
defined mainly by the nature of the dopant atom and not by not by the
B-H1-11S 75.6 79.1
B-H1-12L 52.9 60.0
number of electrons.
B-H1-12S 46.9 75.0
B-H1-22L 44.7 59.3
B-H1-22S 38.8 85.5 (80.0)b (78.0)d
3.2. Electronic nature of the ground states
H1 – 43.5 (39.6)a (41.1)b
N-H2-11L 0 0.0 The doping of the NFs affects the relative energy and the nature of
N-H2-11S 2.14 0.0 the ground states. It has been shown earlier, that pristine H1 and H2
N-H2-12L 22.4 0.0
NFs have single reference singlet ground states unlike the corre-
N-H2-12S 22.2 0.0
N-H2-22L 40.9 0.0 sponding graphene NFs [11] which often show multiconfigurational
N-H2-22S 32.9 0.0 ground states. Table 2 shows the relative singlet-triplet energies
B-H2-11L 44.9 0.0 (singlet-triplet gaps) and the corresponding 〈S2〉 expectation values for
B-H2-11S 39.5 24.3 states with different multiplicity. Notable spin contamination is an in-
B-H2-12L 28.6 2.20
B-H2-12S 12.5 28.1
dicator of possible multireference character of the electronic state
B-H2-22L 5.07 5.10 which may lead to unreliable relative DFT energies of the electronic
B-H2-22S 0 31.1 states with different multiplicities [11].
H2 – 0.0 Both pristine H1 and H2 have singlet ground state, but the triplet-
N-G-L 0 0
singlet gap is much larger for H2, moreover, the closed shell singlet
N-G-S 12.6 0
B-G-L 0 0 solution is unstable for H1 and stable solution corresponds to BS wa-
B-G-S 7.79 0 vefunction. Doping with nitrogen and boron has different impacts on
G – 0 both, singlet-triplet gaps and the nature of the ground states. As seen
a
from Table 2, nitrogen doping opens up singlet triplet gap for all NF
Dicaton.
b types, while B-doping mostly decreases it. The effect of N-doping is the
Dianion.
c most notable for N-H1-11L NF where singlet-triplet gap increases from
Tetracation.
d
Tetraanion. 0.46 kcal/mol for pristine H1 to 13.2 kcal/mol for N-doped system. For
N-doped NFs of H2 series N-doping also results in an increase of the
shape, the electronic and the structural. On the one hand, carbon and singlet-triplet gap, although this effect is less important. According to
dopant atoms have different number of electrons and on the other hand, Table 2 the distance between N atoms does not affect the singlet-triplet
CeC, CeB and CeN bond lengths are also different. To distinguish gap in a systematic way.
between these 2 factors we calculated the deviation from plane for B-doping clearly affects the singlet – triplet band gap. The effect
dicationic, tetracationic, dianionic and tetraanionic species of some of depends on the distance between dopant atoms. In the majority of
the doped and pristine NFs. The results are shown in Table 1. Dication cases, well separated dopants reduce singlet triplet gap while closely
and dianion of H1 are isoelectronic to B- and N-doped species, re- positioned dopants increase it slightly. The most important finding,
spectively, but they contain no heteroatoms, thus excluding structural however, is that triplet ground states have been detected at DFT level
for 4B-doped nanoflakes B-H1-11L, B-H1-22L, B-H2-11S and B-H2-

Fig. 3. D3bj dispersion corrected B3LYP/def2-SVP optimized ground states of pristine and some of the doped NFs.

16
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Table 2
HOMO-LUMO (EH-L), singlet-triplet (ES-T) gaps, 〈S2〉 expectation values for open shell singlet (OSS) and triplet states, ionization energies (IP) and electron affinities
(EA). Lowest singlet excitation energies (S0 → S1) of studied nanoflakes.
Nanoflake EH-L, eV ES-T, kcal/mol 〈S2〉 OSS 〈S2〉 Triplet IP, eV EA, eV S0 → S1, eV

N-H1-11L 1.21 13.2 – 2.02 5.00 2.35 1.10


N-H1-11S 1.08 10.7 – 2.03 5.07 2.57 1.03
N-H1-12L 1.04 9.80 – 2.01 4.95 2.47 0.70
N-H1-12S 1.12 11.1 – 2.12 5.09 2.54 1.00
N-H1-22L 1.05 10.0 – 2.04 4.99 2.50 0.81
N-H1-22S 1.02 9.74 – 2.03 5.07 2.62 0.94
B-H1-11L 0.67 −2.08 (28.2) 1.92 3.05 4.99 2.94 0.04
B-H1-11S 0.76 2.32 – 2.04 5.00 2.83 0.24
B-H1-12L 0.72 2.52 – 2.05 4.95 2.84 0.51
B-H1-12S 0.81 4.71 – 2.03 5.08 2.83 0.59
B-H1-22L 0.72 −0.53 (0.14) 1.04 2.05 5.14 2.98 0.59
B-H1-22S 0.88 7.51 – 2.04 5.10 2.70 0.64
H1 0.642 0.46 1.13 2.02 5.06 2.98 0.296
N-H2-11L 1.10 11.4 – 2.06 4.98 2.42 0.491
N-H2-11S 1.17 12.1 – 2.05 5.01 2.39 0.47
N-H2-12L 1.03 8.76 – 2.09 4.99 2.54 0.39
N-H2-12S 1.29 12.8 – 2.06 5.12 2.40 0.59
N-H2-22L 0.96 6.50 – 2.06 5.05 2.66 0.13
N-H2-22S 1.35 15.5 – 2.05 5.15 2.38 0.68
B-H2-11L 0.7 1.36 0.28 2.10 5.09 2.94 0.30
B-H2-11S 0.71 −0.24 (17.2)a 1.02 2.04 5.09 2.94 0.50
B-H2-12L 0.73 −0.91 (4.72)a 1.04 2.11 5.34 3.14 0.69
B-H2-12S 0.99 7.36 – 2.06 5.22 2.81 0.80
B-H2-22L 0.638 0.67 0.60 2.05 5.29 3.17 0.33
B-H2-22S 0.952 5.97 – 2.03 5.28 2.90 0.53
H2 0.95 7.66 – 2.06 5.32 2.92 0.34
N-G-L 0.56 −0.37 (−7.99)a 1.25 2.15 4.52 2.56 0.45
N-G-S 0.76 0.02 2.10 3.09 4.63 2.24 0.41
B-G-L 0.70 −0.10 (−12.39)a 1.48 2.44 5.08 2.96 0.35
B-G-S 0.98 −1.15 (9.39)a 2.00 3.32 5.26 2.93 0.35
G 1.12 −5.74 (12.62)a 2.05 3.13 4.80 2.75 1.48

a
CASSCF/6-31G*//B3LYP/def2-SVP.

Fig. 4. Superposition of active orbitals with populations close to 1 from CASSCF/6-31G* single point calculations (a) and unpaired spin density obtained from BS
B3LYP/def2-SVP calculations (b).

12L. When revising 〈S2〉 expectation values for open shell singlet states M

of these systems, one can notice strong spin contamination, while spin Nu = ∑ n2i (2 − ni )2
i=1 (1)
contamination of triplets is much lower. This may result in an artificial
overstabilization of the triplet states. Therefore, we have performed where Nu is the number of effectively unpaired electrons, ni is the oc-
single point CASSCF calculations using DFT optimized geometry for all cupation of i-th natural orbital and M is a total number of active orbi-
singlet and triplet states for these 4 cases where DFT calculations in- tals. We have shown previously [11] that unlike graphene NFs where
dicate triplet ground states to eliminate all possible artefacts. Since the number of effectively unpaired electrons increases rapidly with size,
CASSCF method lacks spin contamination problem, there would be no in case of H1 and H2 NFs the number of effectively unpaired electrons
artificial overstabilization of high spin states. does not change with the NF size remaining about 0.2. We have esti-
As seen from Table 2, for all four boron doped NFs CASSCF calcu- mated this number using CASSCF wavefunction for 4 boron doped NFs
lations confirm singlet to be ground state. showing high spin contamination of singlet states. The results were of
CASSCF calculation also allows us to estimate the number of ef- 0.2, 2.0, 0.1, 0.1 effectively unpaired electrons for B-H1-11L, B-H1-
fectively unpaired electrons in the molecule, using Head-Gordon for- 22L, B-H2-11S and B-H2-1-2L NFs, respectively. As seen, strong spin
mula (1) [33] contamination is not always an indication of the multireference

17
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Fig. 5. Total (TDOS) and projected (PDOS) density of states for selected doped and pristine NFs. PDOS1 and PDOS2 are the contributions from the NF and dopant
atoms, respectively.

18
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Fig. 6. Electrostatic potential mapped onto total electron density for pristine and doped graphene and isomeric graphene NFs, obtained from B3LYP/def2-SVP
calculations.

character of the electronic state. However, it also can be noted, that B- calculations the number of effectively unpaired electrons for singlet
doping may lead to the polyradicalic ground state of the NFs. Fig. 4 states of G, B-G-S, B-G-L, N-G-S and N-G-L NFs are of 4.1, 4.0, 3.0, 2.1
visualises unpaired spin density obtained from BS B3LYP calculations, and 3.1, respectively. Therefore DFT energies are not reliable for cor-
and active orbitals with population close to 1 from CASSCF calcula- rect assignment of their ground states. The ground state of graphene NF
tions. As seen, both methods give qualitatively similar picture of spin must be singlet according to Ovchinnikov’s rule [34], and this is con-
distribution in biradicalic state of B-H1-22L. It is interesting to note firmed by both, CASSCF and DFT calculations. Doped graphene NFs,
that although boron doping induces polyradicalic character in the NFs, however, not necessarily must have singlet ground state since Ovchin-
boron atoms do not have any appreciable spin density on them. nikov rule does not apply to them anymore. Both nitrogen and boron
Unlike H1 and H2, NFs doping can change the multiplicity of the dopant promote triplet ground states in graphene NFs when dopant
ground state in graphene NFs as seen from Table 2. Spin contamination atoms are well separated, while graphene NFs with closely positioned
is very strong in graphene NFs due to strong multireference character of dopant atoms maintain singlet ground state of graphene NFs (Table 2).
the singlet and even triplet states [11]. Thus, according to CASSCF Fig. 5 shows total (TDOS) and projected (PDOS) density of states for

19
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Fig. 7. Spin density distribution in polaron cations and polaron anions of some N doped H1 and H2 NFs estimated at B3LYP/def2-SVP level.

pristine and the most the stable B- and N-doped NFs. TDOS of graphene the Fermi level.
NF is also shown for comparison. Graphene is semimetal while allo-
tropies H1 and H2 systems are metallic [21]. The corresponding NFs
possess a band gap as seen from Fig. 5 in accordance with more ela- 3.3. Ionization energies, electron affinities and the band gaps
borated TD-DFT calculations. The Fermi energy (Ef) is the lowest for H2
and the highest for H1 NF, however the difference is very small as seen Ionization potentials (IP), electron affinities (EA) and Eg's of NFs are
from the TDOS spectra. N-doping slightly increases Ef, while B-doping listed in Table 2. As seen, doping with nitrogen changes IPs very little,
decreases it slightly. All doped NFs show a bandgap in agreement with reducing them in most of the cases. The reduction of IPs is more notable
TD-DFT calculations. TDOS of doped and pristine NFs are very similar, for H2 NFs compared to H1. Nitrogen doping also reduces slightly EAs
the contribution of heteroatoms to TDOS is rather small especially near for both NF types. It is interesting to note that IPs and EAs depend on
the distance between doping centers. Smaller distance between dopant

20
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

Fig. 8. Spin density distribution in polaron cations and polaron anions of some B-doped H1 and H2 NFs estimated at B3LYP/def2-SVP level.

centers results in smaller changes in both IPs and EAs. The difference is are also shown for the comparison purpose. When comparing nitrogen
not large (about 0.1 eV) but consistent for all doped NFs. The behavior doped NFs with these of pristine ones, one can note that a part of the
of H1 and H2 NFs on doping with nitrogen atoms resembles that of electron density of nitrogen atoms is delocalized over the entire NFs. In
graphene NFs as seen from Table 2. However, in the case of graphene this particular case nitrogen doping causes similar effects in isomeric
the results of IP and EA reduction are more notable. In all cases N graphene and graphene NFs, however, to a smaller extent.
dopants belong to “graphitic” type of nitrogen [36] where nitrogen In case of doping with boron atoms, IPs of doped H1 and H2 NFs
atom shares one electron with π electron cloud of the NF, thus being n remain practically unaffected as can be seen from Table 2. EAs are also
type of dopant [15]. This effect can be appreciated in Fig. 6 where the affected very little by B-doping for H1 NF type, slightly increasing for
electrostatic potential of nitrogen and boron doped H1, H2 and gra- doped H2 NFs. This behavior differs from that of graphene NF where B-
phene NFs are mapped over their total electron densities. Pristine NFs doping causes notable increase of both IP and EA. Boron is a p dopant

21
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

atom for graphene NF [35], but as a dopant, boron atoms behaves electronic structure of H1 and H2 NFs. Nitrogen atoms promote single
differently in H1 and H2 NFs. As seen from Fig. 6 showing the elec- reference singlet ground state, they increase singlet – triplet gap, and
trostatic potential mapped over the total electron densities for boron Eg, boron decreases singlet – triplet energy gap and in some cases
doped H1, H2 and graphene NFs, there is a notable difference between promotes polyradicalic ground states.
boron-doped graphene NFs from the one side, and H1, H2 NFs from the IP and EA of H1 and H2 NFs are less affected by doping compared to
other. The electrostatic potential maps are very similar for H1 and H2 graphene NF. Nitrogen can still be considered as n dopant for H1 and
NFs. It is clearly seen that in the case of graphene NFs, boron is a p H2 NFs, while boron cannot be considered as p type of dopant for
dopant, receiving electron density from the rest of π-electron cloud as isomeric graphene NFs. The probable reason for this behavior is an
an electron deficient atom. In the case of H1 and H2 NFs, however, “internal” doping of isomeric graphene NFs; non uniform density dis-
there is no significant change in the electrostatic potential of the NFs on tribution in these NF decreases the effect of heteroatom doping on the
doping. These results demonstrate that nitrogen atoms act as n dopant electronic properties.
in H1 and H2 NFs although the doping effect is weaker compared to
that of graphene NF. On the other hand, boron cannot be considered as Acknowledgments
a p type dopant for H1 and H2 contrary to graphene NFs. The reason
why isomeric graphene NFs are less sensible to doping compared to We acknowledge the financial support from CONACyT (Grant
graphene NFs can be understood revising Figs. 6–8. Figs. 6 and 7 vi- 251684) and from supercomputing facilities of National Autonomous
sualize spin density delocalization pattern in polaron cations and po- University of Mexico.
laron anions of isomeric NFs corresponding to the largest and the lowest
IPs and EAs. As seen from the figures, the doping centers do not par- Appendix A. Supplementary material
ticipate in stabilization neither polaron cations nor polaron anions.
In case of graphene NFs the situation is quite different, boron and Supplementary data to this article can be found online at https://
nitrogen atoms do participate in stabilization of polaron cations and doi.org/10.1016/j.comptc.2019.01.022.
polaron anions, respectively [11,36]. Another point is related with
“internal” doping existing in isomeric graphene H1 and H2 systems. As References
seen from Fig. 6, graphene NF shows a uniform electron density dis-
tribution across the NF. Contrary to that, H1 and H2 NFs show highly [1] M. Zeng, Y. Xiao, J. Liu, K. Yang, L. Fu, Exploring two-dimensional materials toward
non uniform electron density distribution where some of carbon atoms the next-generation circuits: from monomer design to assembly control, Chem. Rev.
(2018).
are electron deficient and another are electron enriched due to none- [2] X. Song, J. Hu, H. Zeng, Two-dimensional semiconductors: recent progress and
quivalent chemical environments. This can be considered as “internal” future perspectives, J. Mater. Chem. C (2013).
doping and new external doping center causes smaller impact on the [3] M. Chhowalla, H.S. Shin, G. Eda, L.J. Li, K.P. Loh, H. Zhang, The chemistry of two-
dimensional layered transition metal dichalcogenide nanosheets, Nat. Chem.
electronic structure of the H1 and H2 compared to graphene NFs, since (2013).
H1 and H2 are already “doped”. When it comes down to charges, both [4] K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, Y. Zhang, S.V. Dubonos,
doping centers nitrogen and boron bear small negative charges, ac- I.V. Grigorieva, A.A. Firsov, Electric field effect in atomically thin carbon films,
Science (2004).
cording to Mulliken population analysis, boron (−0.3) being slightly [5] A.H. Castro Neto, N.M.R. Peres, K.S. Novoselov, A.K. Geim, F. Guinea, The elec-
more negative than nitrogen (−0.1). Charges are rather similar for all tronic properties of graphene, Rev. Mod. Phys. (2009).
positions and all NF types. More negative charges of boron atoms make [6] Y.S. Dedkov, M. Fonin, Electronic and magnetic properties of the graphene-ferro-
magnet interface, New J. Phys. (2010).
sense since boron accepts π electron density from carbon atoms which
[7] M. Huzak, M.S. Deleuze, B. Hajgat, Half-metallicity and spin-contamination of the
is in total agreement with the electrostatic potential map. electronic ground state of graphene nanoribbons and related systems: an impossible
Unlike IPs and EAs of the NFs, the doping impacts notably both, compromise, J. Chem. Phys. (2011), https://doi.org/10.1063/1.3626554.
Eg’s, calculated as HOMO-LUMO gap and as the lowest excitation en- [8] J.P. Malrieu, G. Trinquier, A recipe for geometry optimization of diradicalar singlet
states from broken-symmetry calculations, J. Phys. Chem. A (2012), https://doi.
ergy using TD DFT framework. N-doping increases Eg from a fraction of org/10.1021/jp303825x.
eV to more than 1 eV and the results are consistent for all methods used [9] S. Rayne, K. Forest, A comparison of density functional theory (DFT) methods for
for the calculation of Eg. This increase of Eg is more notable for doped estimating the singlet-triplet (S0–T1) excitation energies of benzene and poly-
acenes, Comput. Theor. Chem. (2011), https://doi.org/10.1016/j.comptc.2011.08.
H1 NFs compared to doped H2 ones. As seen from Table 2 there is no 010.
visible correlation between separation distance for nitrogen dopants [10] A.E. Torres, S. Fomine, Electronic structure of graphene nanoribbons doped with
and Eg. The effect of B-doping on the Eg is much less and more erratic. nitrogen atoms: a theoretical insight, Phys. Chem. Chem. Phys. (2015), https://doi.
org/10.1039/c5cp00227c.
The Eǵs evolution of H1 and H2 on doping differs from that of graphene [11] A.E. Torres, P. Guadarrama, S. Fomine, Multiconfigurational character of the
NFs (Table 2). As seen both, boron and nitrogen doping reduce Eg's of ground states of polycyclic aromatic hydrocarbons. A systematic study, J. Mol.
graphene NFs. This behavior is similar for all methods used for Eg Model. (2014), https://doi.org/10.1007/s00894-014-2208-6.
[12] A. Koma, K. Yoshimura, Ultrasharp interfaces grown with Van der Waals epitaxy,
calculation. Surf. Sci. (1986).
[13] A.B. Preobrajenski, A.S. Vinogradov, N. Mårtensson, Monolayer of h-BN chemi-
4. Conclusions sorbed on Cu(111) and Ni(111): the role of the transition metal 3d states, Surf. Sci.
(2005).
[14] K. Taniguchi, A. Matsumoto, H. Shimotani, H. Takagi, Electric-field-induced su-
The relative stability of doped isomeric graphene NFs depends perconductivity at 9.4 K in a layered transition metal disulphide MoS2, Appl. Phys.
greatly on both the dopant nature and the position. The difference in Lett. (2012).
the relative energies can be as high as 40 kcal/mol. The most stable N- [15] E. Coronado, C. Martí-Gastaldo, E. Navarro-Moratalla, E. Burzurí, A. Camón, F. Luis,
Hybrid magnetic/superconducting materials obtained by insertion of a single‐mo-
and B-doped NFs have common substitution patterns: nitrogen doped lecule magnet into TaS2 layers, Adv. Mater. (2011).
NFs have a phenalene fragment substituted in position 7 and boron [16] C. Feng, J. Ma, H. Li, R. Zeng, Z. Guo, H. Liu, Synthesis of molybdenum disulfide
doped NFs – azulene moiety with boron atom in position 9. (MoS2) for lithium ion battery applications, Mater. Res. Bull. (2009).
[17] J.V. Lauritsen, M. Nyberg, J.K. Nørskov, B.S. Clausen, H. Topsøe, E. Lægsgaard,
Unlike graphene NFs, the doping of isomeric graphene NFs with F. Besenbacher, Hydrodesulfurization reaction pathways on MoS2 nanoclusters
boron or nitrogen noticeably affect their shape. Boron doping promotes revealed by scanning tunneling microscopy, J. Catal. (2004).
dome shaped geometry, while nitrogen doping favors planar geometry. [18] H.S. Lee, S.W. Min, Y.G. Chang, M.K. Park, T. Nam, H. Kim, J.H. Kim, S. Ryu, S. Im,
MoS 2 nanosheet phototransistors with thickness-modulated optical energy gap
This effect depends on the distance between dopant atoms. Closely lo- (Supplementary), Nano Lett. (2012).
cated dopant atoms have stronger effect on the NF shape compared to [19] W. Choi, M.Y. Cho, A. Konar, J.H. Lee, G.B. Cha, S.C. Hong, S. Kim, J. Kim, D. Jena,
separated dopant atoms. J. Joo, S. Kim, High-detectivity multilayer MoS2 phototransistors with spectral
response from ultraviolet to infrared, Adv. Mater. (2012).
Boron and nitrogen doping produce different effects on the

22
C.G. Vera de la Garza et al. Computational and Theoretical Chemistry 1151 (2019) 12–23

[20] H. Li, Z. Yin, Q. He, H. Li, X. Huang, G. Lu, D.W.H. Fam, A.I.Y. Tok, Q. Zhang, elements H-Pu, J. Chem. Phys. 132 (2010), https://doi.org/10.1063/1.3382344.
H. Zhang, Fabrication of single- and multilayer MoS2 film-based field-effect tran- [29] T.M. McCormick, C.R. Bridges, E.I. Carrera, P.M. Dicarmine, G.L. Gibson,
sistors for sensing NO at room temperature, Small (2012). J. Hollinger, L.M. Kozycz, D.S. Seferos, Conjugated polymers: evaluating DFT
[21] A.N. Enyashin, A.L. Ivanovskii, Graphene allotropes, Phys. Status Solidi Basic Res. methods for more accurate orbital energy modeling, Macromolecules (2013).
(2011). [30] S. Yang, M. Kertesz, Bond length alternation and energy band gap of polyyne, J.
[22] X. Rocquefelte, G.-M. Rignanese, V. Meunier, H. Terrones, M. Terrones, J.- Phys. Chem. A (2006).
C. Charlier, How to identify haeckelite structures: a theoretical study of their [31] M.J.G. Frisch, W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,
electronic and vibrational properties, Nano Lett. (2004). G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato, X.
[23] A. Chuvilin, J.C. Meyer, G. Algara-Siller, U. Kaiser, From graphene constrictions to Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, Gaussian
single carbon chains, New J. Phys. (2009). 16, Rev. A.03, 2016, doi:111.
[24] C.G. Vera de la Garza, G.L. García, E.M. Olmedo, E.R. Peña, S. Fomine, Electronic [32] T.L. Cottrell, The Strenght of Chemical Bonds, Butterworths, London., 1958.
structure of isomeric graphene nanoflakes, Comput. Theor. Chem. 1140 (2018) [33] M. Head-Gordon, Characterizing unpaired electrons from the one-particle density
125–133. matrix, Chem. Phys. Lett. (2003).
[25] A. Govindaraj, C.N.R. Rao, Doping of graphene by nitrogen, boron, and other ele- [34] A.A. Ovchinnikov, Multiplicity of the ground state of large alternant organic mo-
ments, Functionalization of Graphene, (2014). lecules with conjugated bonds - (Do Organic Ferromagnetics Exist?), Theor. Chim.
[26] Y. Liu, Y. Shen, L. Sun, J. Li, C. Liu, W. Ren, F. Li, L. Gao, J. Chen, F. Liu, Y. Sun, Acta (1978).
N. Tang, H.M. Cheng, Y. Du, Elemental superdoping of graphene and carbon na- [35] L. Zhao, R. He, K.T. Rim, T. Schiros, K.S. Kim, H. Zhou, C. Gutiérrez,
notubes, Nat. Commun. (2016). S.P. Chockalingam, C.J. Arguello, L. Pálová, D. Nordlund, M.S. Hybertsen,
[27] TURBOMOLE V7.2 2017, a development of University of Karlsruhe and D.R. Reichman, T.F. Heinz, P. Kim, A. Pinczuk, G.W. Flynn, A.N. Pasupathy,
Forschungszentrum Karlsruhe GmbH, 1989–2007, TURBOMOLE GmbH, since Visualizing individual nitrogen dopants in monolayer graphene, Science 80 (2011).
2007. Available from < http://www.turbomole.com > . [36] A.E. Torres, R. Flores, L. Fomina, S. Fomine, Electronic structure of boron-doped
[28] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, A consistent and accurate ab initio finite graphene sheets: unrestricted DFT and complete active space calculations,
parametrization of density functional dispersion correction (DFT-D) for the 94 Mol. Simul. (2016).

23

You might also like