Carbon 141 (2019) 274e282
lable at ScienceDirectContents lists avaiCarbon
journal homepage: www.elsevier .com/locate /carbonMapping the stacking interaction of triphenyl vinylene oligomers with
graphene and carbon nanotubes
A. Yaya a, b, A. Impellizzeri b, F. Massuyeau b, J.L. Duvail b, P. Briddon c, C.P. Ewels b, *
a Department of Materials Science & Engineering, University of Ghana, Ghana
b Institut des Mat
eriaux Jean Rouxel (IMN), UMR 6502, CNRS, Universit
e de Nantes, 2rue de La Houssiniere, 44322, Nantes, Cedex 3, France
c Department of Electrical Engineering, Newcastle University, Newcastle-Upon-Tyne, UKa r t i c l e i n f o
Article history:
Received 20 July 2018
Received in revised form
17 September 2018
Accepted 24 September 2018
Available online 25 September 2018
Keywords:
DFT
p-p stacking
SWCNTs
Nanotubes
Graphene
PPV
Triphenyl vinylene* Corresponding author.
E-mail address: chris.ewels@cnrs-imn.fr (C.P. Ewe
https://doi.org/10.1016/j.carbon.2018.09.071
0008-6223/© 2018 Elsevier Ltd. All rights reserved.a b s t r a c t
Using density functional calculations, we explore p-p and p-H stacking interactions in conjugated
polymer e carbon nanomaterial composites, through a detailed surface energy mapping procedure.
Taking the triphenyl vinylene oligomer as a structural model for poly para-phenylene vinylene (PPV), we
map intermolecular stacking configurations between PPV pairs. We then map PPV translation over the
surface of graphene, and explore orientation dependence in PPV-nanotube interaction with tubes of
different chirality and diameter. Preferential stacking orientations are shown, and commensurability
between the polymer and carbon substrate is demonstrated. Conjugated polymers such as PPV lie at the
crossroads between small organic molecules such as benzene and extended conjugated carbon systems
such as graphene. With the current calculations we are able to reconcile the range of stacking behaviours
seen in these diverse materials, with literature experimental polarized Raman spectroscopy results.
© 2018 Elsevier Ltd. All rights reserved.1. Introduction
Interactions involving p-p and p-H bonding play a major role in
all branches of organic chemistry, biochemistry and materials sci-
ence, from the stabilization of host-guess complexes and the in-
teractions of certain drugs to DNA [1] to the recognition of
molecular species (self-assembly) [2e4]. In particular, they are a
driving force in determining crystal stacking, from solids of small
organic molecules such as benzene, through polymers containing
aromatic rings, to large conjugated systems such as graphene
stacking in graphite. Such systems often exhibit a subtle competi-
tion between p-p and p-H bonding.
In hybrid composite systems, conjugated polymer composites
with graphene and single-walled carbon nanotubes (SWCNTs) are
believed to exhibit strong binding through non-covalent p-stacking
[5] and are promising for mechanical and optoelectronic applica-
tions [6e10]. Several theoretical and experimental polymer-
SWCNTs studies have demonstrated polymer wrapping of
SWCNTs [11e14]. Even though these studies do not alwaysls).represent bulk nanocomposite properties, they give useful infor-
mation on the nature of the polymer-nanotube interface. Molecular
dynamics studies of the interaction between poly (styrene), poly
(phenylacetylene), poly (para phenylene vinylene) (PPV) and
SWCNTs [15] have shown that monomer structure plays a vital role
in determining the strength of the interaction between the SWCNTs
and polymer, suggesting that the presence of aromatic rings in the
polymer backbone improve noncovalent binding of CNTs into
composite structures. There are fewer structural studies of
graphene-polymer composites and most use graphene oxide
because of its relative availability in quantities suitable for com-
posite production.
The enormous structural variety of monomeric and polymeric
aromatic compounds coupled with the lack of a thorough under-
standing of p-p and p-H bonding, makes it difficult to investigate
their interactions. These difficulties come from the treatment of the
electron correlation, dispersion, polarization and solvation, in a
manner that is appropriate to the context of the applications and
within the confines of available computer resources [16]. Experi-
mental methods such as NMR have been used to study the nature of
p-p interactions, giving partial information about their energetic
and substituent effects. However, the interpretation of these
experimental results is complex as solvation effects and secondary
A. Yaya et al. / Carbon 141 (2019) 274e282 275interactions complicate the situation [1].
In the light of this, quantum mechanical studies have become
key in the study of interactions involving p-p bonding. Higher or-
der wave function methods such a couple-cluster theory through
the use of perturbative triples, CCSD(T), with large basis sets
[1,17,18] can reach a good level of accuracy, but the computational
cost is too demanding for all except small systems: computational
time scales as ~O(N7), where N is the number of atoms in the sys-
tem. Less expensive methods such as Møller-Plesset perturbation
(MP2) can also become prohibitive when a large basis set is taken
into consideration. Also, they lead to overestimation of the electron
correlation effects that are inherent in pure p-p interactions [16].
These then also leave problems for the theoretical understanding of
larger systems containing p-p stacking interactions [19,20].
Previous groups have modelled PPV-PPV interaction at various
levels of theory [21]. Other high levels of electronic structure theory
have been applied to give an accurate model of p-stacking inter-
action in small organic systems [17,22,23]. However the computa-
tional cost means none of thesemethods can be applied to the large
systems we are considering here, such as PPV-nanotube in-
teractions. Therefore a simpler and computationally less
demanding approach is needed.
To address these large p-p systems we apply here density
functional techniques (DFT) using the local density approximation
(LDA), which scale much less critically with basis set size than the
advanced wave functional methods discussed above. We initially
benchmark this approach against these more complex techniques
using smaller molecules. Our approach is pragmatic, more akin to
numerical experimentation: The question we are posing is, to what
extent can simpler exchange-correlation description such as LDA
successfully reproduce p-p interaction for extremely large sys-
tems? Having benchmarked the approach we then explore the
energy surface as PPV moves over the surface of PPV, graphene and
carbon nanotubes. These computationally demanding mappings
allow us to develop a comprehensive model of p-p stacking
interaction in 0, 1 and 2D conjugated carbon materials.
2. Computational method
Density functional calculations are performed with the AIMPRO
code [24e26].We used the local spin density approximation (LSDA)
and the spin generalized gradient approximation (SGGA) parame-
trized by Perdew, Burke, and Ernzerholf (PBE) [27]. To describe the
PPV-graphene interaction exactly van der Waals forces are added
by using DFT-D3 [28] with Becke-Johnson (BJ) damping [29].
Relativistic pseudopotentials are included via the Hartwigsen-
Goedecker-Hütter scheme [30]. The basis consists of Gaussian
function sets multiplied by polynomial functions including all
angular momenta up to maxima p (l¼ 0, 1), d (l¼ 0, 1, 2) and f
(l¼0e3) [31]. For carbon a pdddp basis set was used resulting in 38
independent functions, hydrogen was treated using a ppp basis
with 12 independent functions. A system-dependent plane wave
energy cutoff of 150 Ha (Ha: Hartree energy) was taken, and a non-
zero electron temperature of kT¼ 0.04 eV for electronic level
occupation.
Periodic boundary conditions are used. Initial graphene and
nanotube geometry (including lattice vectors) are relaxed using
8 8 1 and 1 1 4 k-point grids adopting periodic boundary
conditions, respectively (previously tested for convergence by
changing the number of k-points in the periodic direction, as shown
in Supplementary Materials, Figure S1). Cell sizes were large
enough (e.g. at least 15.0 Å between nanotubes) to avoid interaction
between neighbouring species. After forming the hexagonal
supercells containing nanotubes with lengths in the range
24.0e24.3Å (168e440 atoms), we added the triphenyl vinyleneoligomer chain (C22H18), referred to hereafter as PPV for simplicity.
The nanotube atoms were fixed and only PPV atoms allowed to
fully relax with no constraints. Test calculations relaxing all atoms
including the nanotubes with LDA and PBE-D3 (BJ) indicated that
changes in the positions of the C atoms are minor upon insertion of
the PPV molecule and binding energy difference changes by
1e2meV. The nanotube length allows a separation distance be-
tween PPV molecules of more than 9 Å when parallel and 20 Å
when circumferential, giving negligible intramolecular in-
teractions. Graphene was modelled using an 8 8 graphene
supercell (C128). Previous studies of PPV-PPV derivative interactions
have shown that the LDA is a good approximation for equilibrium
calculations [32].
Atomic positions were geometrically optimised until the
maximum atomic position change in a given iteration dropped
below 105 a0 (a0: Bohr radius) and the total energy change has a
tolerance of 105 Ha. Binding energies are calculated in comparison
to the isolated separate components, e.g. for PPV binding to a
nanotube,
EB ¼ E(CNT ePPV) - ECNT - EPPV (1)
where EPPV, ECNT, and E(CNT ePPV) are the total energies of an isolated
molecule, the isolated nanotube supercell and the PPV-nanotube
system, respectively. Inter-PPV interaction was calculated with
two PPV chains in a large hexagonal supercell with full atomic
relaxation of both chains.
Contour map plots of PPV over SWCNTs, graphene and PPVwere
obtained by applying diffusion constraints to the PPV. It was first
translated in the xy plane to a given position in the mapping.
During subsequent geometric optimisation, two carbon atoms in
the same vinylene bond had their xy coordinate fixed and were
only allowed to vary their z height freely and independently. All
other atoms in the oligomer were allowed to vary their x, y and z
coordinates during the geometric optimisation. For PPV-PPV
interaction, the same vinylene constraint was applied to the un-
derlying ‘substrate’ PPV molecule, for the other substrates (gra-
phene or nanotube) these were kept frozen during the
optimisation. Constraining in this way allows the system as much
freedom as possible while still restricting sufficiently to avoid
twisting and torsional restacking. The approach is described further
in a previous study of benzene-benzene interaction [33]. Related
mapping approaches have also been used previously to explore
graphene platelet motion on the surface of pristine and edge-
terminated graphene, although with stricter optimisation con-
straints (fixed xy coordinates for all atoms) [34].
3. Results and discussion
3.1. Isolated PPV (triphenyl vinylene oligomer unit)
The isolated PPV chain studied in this work is made up of a
triphenyl vinylene repeat oligomer unit. This is an attractive poly-
mer for the theoretical study of properties of conjugated polymers
because it takes the form of relatively simple quasi-one-
dimensional molecules arranged in a 3D crystal structure. These
arrangements therefore make it possible to study the conforma-
tional and electronic characteristics of both the isolated and crystal
chains of the PPV polymers.
Comparing flat and bent configurations shown in Fig. 1a, the flat
conformation is 0.30 eV more stable in agreement with literature
[35], where the isolated monomer unit was found to be planar. We
find no energy difference within the precision of the calculations
between the cis- and trans-configurations. For extended p-conju-
gation the PPV chain prefers planarity, but steric hindrance
276 A. Yaya et al. / Carbon 141 (2019) 274e282
Fig. 1. (a) Front- and side-views of triphenyl repeat oligomer units of PPV with (left) stable flat geometry and (right) meta-stable bent configuration (0.30 eV less stable). The slight
torsional twist of 8 per PPV monomer is visible. (centre) Calculated LDA Kohn-Sham eigenvalues in the two configurations. In both cases the gap remains around 2.15eV, but once
bent the levels are shifted upwards in energy. Black (white) squares indicate filled (empty) states, only marked around the gap for clarity. Square of the wavefunction isosurfaces for
(b) highest occupied and (c) lowest unoccupied molecular orbitals for the flat configuration. (A colour version of this figure can be viewed online.)between hydrogen atoms on the benzenoid ring and the vinyl
linkage favours non-planar torsion. Indeedwe find a small torsional
angle of 8 per monomer in the flat configuration, although the
energy surface between 0 and 8 is essentially flat. This agrees well
with experimental x-ray studies which find a torsional angle of
10± 3 for PPV [36], and from 2 to 12 for (PPV)5 oligomer crystals
[37], and similar to previous AM1 studies of PPV oligomers which
found ~15 torsion [38].
Our calculated LDA band gap is largely invariant to bending, at
2.17 eV for the flat and 2.13 eV for the bent configuration (Fig. 1a).
The experimental band gap is 2.5 eV [39] so our value is quite close
but slightly underestimated as expected for the LDA. Previous cal-
culations find a band gap for the infinite PPV chain of 1.3 eV [35],
showing our short chain introduces confinement effects which
widen the gap. The top of the valence band in PPV is constructed
from four carbon p states per monomer: three from the benzenoid
ring and one from the vinylene bond. Fig. 1b shows the highest
occupied molecular orbital (HOMO) is delocalised over the vinyl-
ene, with some character on the central benzenoid ring furthest
from the molecule ends. The lowest unoccupied molecular orbital
(LUMO) sits in the corresponding anti-bonding location (Fig. 1c). In
the bent configuration the energy levels are shifted up by about
0.9 eV with respect to the flat configuration, due to increased steric
repulsion between neighbouring atoms and slightly increased
confinement induced by bending.
Having established that the flat triphenyl vinylene PPV isolated
chain with 8 torsional twist per monomer is the most stable form,we next examine its interaction with other species.3.2. PPV-PPV interaction
We study here the interaction between two PPV oligomers in
parallel and perpendicular T-shape configurations. This is done by
varying the relative positions of the two oligomers on a 16 16 grid
of points in an xy plane as described in the method, in order to
determine the most stable conformation at each configuration. In
this case we have used the trans-configuration.
Fig. 2 shows the resultant energy surfaces. The PS configuration
has three global minima (Fig. 2b). The two most stable correspond
to an axial shift resulting in AB-graphite type stacking configuration
(ii) and a staggered configuration (i) in which both p-p and H-p
interaction is maximised. The third slightly weaker minimum (iii)
also maximises H-p but with reduced p-p stacking as compared to
(ii). We have also included the final structure for the overlaid
molecules, (iv) to demonstrate the diffusion constraint applied. In
this case it can be seen that the constrained vinyl bond remains
overlaid and the rest of the twomolecules have distorted into order
to try tomove away fromAA-stacking. The orthogonal arrangement
(Fig. 2c) has two global minima which are related via symmetry (i
and ii). These correspond to the secondmolecule aligning along the
backbone of the first, with an offset of the two molecules along
their axis bringing the benzenoid hydrogen atoms into closer
proximity to the vinyl bonds and ring centres of the neighbour.
These show that the orthogonal (T-shaped) configuration is very
A. Yaya et al. / Carbon 141 (2019) 274e282 277
Fig. 2. Contoured energy surfaces showing binding energies (eV) between two PPV molecules as a function of their centre-centre offset in an xy plane (Å). (a) Schematic showing
how xy offset is defined (underlying PPV molecule shaded), (b) parallel-stacked (PS) geometries and (c) edge-on-face (T-shaped). The geometries of key structures are shown, most
stable structures are marked in blue. Black dots indicate structures that were optimised in order to generate the plot. (A colour version of this figure can be viewed online.)slightly less stable than the parallel stacked (PS) configuration, but
the energy difference between the most stable structures when
fully relaxed without constraints is only 0.08eV. This close relative
energy is similar to benzene-benzene where the energy of the T-
shaped and PS structures are very similar [22,32,33,40] (a differ-
ence in energy of only 0.0009eV using the CCSD(T) method [41]). In
the PPV-PPV case, parallel stacked is slightly more energetically
favoured due to the increased p-p conjugation. Binding energies
are roughly three times those of benzene; this is consistent because
the molecule is roughly three and half times bigger. We note that in
bulk crystalline form, PPV has a T-shaped (herringbone) packing
arrangement [35], reflecting a small shift in the relative stability of
the T-shape and PS stacking when going from our isolated dimer to
a fully surrounded 3D packing.
The results are consistent with previous literature calculations
for PPV-PPV oligomer interaction. Early quantum chemical calcu-
lations at MP2/3e21þþG level also show an offset parallel stackedglobal minimumwith in-plane lateral shear of 0.98 and 0.7 Å, close
to our values [21]. Their quoted intermolecule spacing of 3.5 Å is
larger than our value of 3.17 Å, but they also observe that repeating
the calculations with LDA leads to a decrease in intermolecular
spacing. We note that larger 3.3 Å (half-unit cell) translations were
commonly assumed in the early literature, which correspond to the
secondary local minima observed in Fig. 2b-ii and 2b-iii [21].3.3. PPV-graphene interaction
The calculated contour plots for the interaction of graphene
with PPV are shown in Fig. 3 (T-shaped) and Fig. 4 (Parallel stacked).
Thesewere obtained using the procedure described before inwhich
the graphene sheet was kept frozen during the optimisation run
with constraint applied in the PPV by constructing a grid of points
in the xy plane and then allowing it to move freely, with two atoms
in the same vinyl bond constrained to move only in the z direction
278 A. Yaya et al. / Carbon 141 (2019) 274e282
Fig. 3. Contour plot showing binding energies (eV) interactions between Graphene-PPV in the T-shaped conformation as a function of their centre-centre offset in a grid of xy plane
(Å). The geometries for the highest binding energy state, favourable state (colour blue) and that of the lowest unfavourable (colour red) are also indicated. Black dots indicate
structures optimised in order to generate the contour plot. (A colour version of this figure can be viewed online.)during relaxation.
Compared to PPV-PPV stacking the energy surfaces are rela-
tively flat, notably in the vertical T-shaped configuration where
energy variation is less than 0.05eV. Binding is much stronger than
PPV-PPV interaction, confirming there will be preferential inter-
action with graphene when it is inserted into a PPV matrix. Most
importantly, unlike PPV-PPV interaction where the energy differ-
ence between parallel and perpendicular stacking is very close, in
this case there is a strong (nearly 0.5eV) preference for parallel (PS)
stacking.
The global maximum (least stable) corresponds to the PS posi-
tions inwhich the atoms in neighbouring layers primarily face each
other, corresponding roughly to an “AA-stacking” type conforma-
tionwith an interlayer spacing of 3.51 Å between themolecules and
the surface with binding energy of 1.08 eV. The global minimum
(most stable) corresponds to that of PS adopting the parallel dis-
placed (PD) conformation (graphite AB-stacked conformation) with
a binding 0.14 eV higher and an interlayer spacing of 3.22 Å (see
Fig. 4aei). We see a small effect due to the k-point grid since if we
fully relax the PPV increasing from 22 1 to a 42 1 k-point grid
our calculated binding energies in the parallel and perpendicular
orientations are 1.07eV and 0.81eV respectively (unfortunately
such grids are not computationally practical for the entire mapping
calculations).
The contour plot shows that the AA-stacked PPV is a maximum
between AB-stacked geometries, and represents the largest barrier
for migration of PPV lying parallel on graphene sheet in the AB-
stacked configuration. We note that our calculated maximum
(0.14eV) and minimum (0.02eV) energy barriers on this landscape
are almost identical to those observed previously in similar energymapping calculations for graphene platelet motion over graphene
[34].
Next, we calculate the force involve in sliding the PPV over the
graphene from the global minimum position, AB-stacked, to the
global maximum, AA-stacking. This is the differential of the energy
plot in Fig. 4a, shown in Fig. 4b.
As expected, at the minimum energy AB-stacking the force is
zero (0 N). A force of 0.33 nN will be required to slide the PPV
directly from this stable position to the energy maximum AA-
stacking conformation which is improbable. However it will be
much easier to slide the PPV on the graphene between AB and AC-
stacking configurations which require a smaller force (0.07 nN) see
Fig. 4b. Henc!e this suggests preferential surface glide along the
armchair [10 1 0] direction of the graphene, at ~60 to the PPV axis,
wi!th!maximum resistance orthogonal to this along the graphene
[2 1 1 0] zigzag direction, ~-30 to the PPV axis.
This is important for composite application and this can be
compared to experiments where synergistic effect on composite of
this nature will have a higher mechanical strength in which the
interface between the PPV and the graphene form an AB-stacking
pattern.
These results are in broad agreement with experimental studies
on the frictional mechanisms between small graphene flakes and
graphites using a frictional force microscope that achieves a reso-
lution in the lateral forces down to 15 pN. It was found that there is
a strong dependence of friction on the orientation of the molecules
[42]. The frictional force was maximal when the orientation angle,
which defines the lattice mismatch between the flake and sub-
strate, was zero or 60 meaning the flakes slide over the graphite
surface in commensurate contact. Ultra-low friction behaviour and
A. Yaya et al. / Carbon 141 (2019) 274e282 279
Fig. 4. Contour plot showing binding energies (eV) interactions between Graphene-PPV in the parallel-stacked (PS) (a) conformation as a function of their centre-centre offset in a
grid of xy plane (Å). The geometries for the highest binding energy state, favourable state (colour blue) and that of the lowest unfavourable (colour red) are also indicated. Black dots
indicate structures optimised in order to generate the contour plot. (b) Energy gradient (force) plot of (a). (A colour version of this figure can be viewed online.)enhanced slipperiness was observed when the flake slides over
graphite surface in an incommensurate contact. Our modelling
approaches the commensurate case and shows there will be pref-
erential glide directions in this case.
A theoretical study on the adsorption of poly-aromatic hydro-
carbons (PAHs) on hydrogen terminated graphite shows that there
is a large force in moving from an AB-stacking orientation of these
PAHs to AA-stacking, in agreement with our studies [43].3.4. PPV-nanotube interaction
We next examine the interaction between the PPV oligomer and
a range of metallic and semi-conducting nanotubes with different
diameters (Table 1). In all cases PPV-CNT binding is significantly
stronger than PPV-PPV (0.30 eV), i.e. there will be strong and
preferable binding between the nanotubes and the polymer, in
agreement with prior theory and experiment [44,45].
The most stable orientation is always with the PPV parallel tothe nanotube axis, lying flat on the tube surface at a distance of
~3.33 Å (Fig. 5a), as outlined in the binding energy trend between
(4,4) nanotube and the molecule at different distances between
them (Supplementary Materials, Figure S2). The PPV chain remains
relatively flat, with small displacements of the hydrogen atoms.
There is no significant difference in binding energy between
metallic and semiconducting tubes, and instead a weak depen-
dence on tube diameter is seen favouring larger tubes.
Placing the PPV along the axis but perpendicular to the nano-
tube wall (Fig. 5b), or curving the PPV chain circumferentially
around the nanotube (Fig. 5c) result in weaker binding (around
0.4 eV less), although still significant and still stronger than PPV-
PPV interaction. Upon reducing nanotube length in order to allow
PPV molecules to interact, the calculated circumferential PPV
binding energy to the (4,4) tube changed by less than 0.02 eV
(Supplementary Materials, Figure S3).
When the PPV is orthogonal to the tube surface the contact area
is very localised and is thus essentially independent of nanotube
280 A. Yaya et al. / Carbon 141 (2019) 274e282
Table 1
Calculated DFT-LDA binding energy (eV) of triphenyl PPV to different carbon nanotubes, with graphene for comparison.
Tube Binding (eV)
(n,m) Diameter (Å) Axial parallel to surface Axial perpendicular to surface Circumferential parallel to surface Difference (Axial- Circumferential) parallel (eV)
(4,4) 5.50 0.89 0.49 0.40 0.49
(6,6) 8.14 0.93 0.51 0.57 0.36
(8,8) 10.85 0.97 0.52 0.70 0.28
(10,10) 13.56 1.01 0.52 0.80 0.21
(7,0) 5.58 0.87 0.50 0.35 0.52
(9,0) 7.12 0.93 0.49 0.52 0.41
(11,0) 8.64 0.94 0.51 0.60 0.34
(13,0) 10.17 0.97 0.52 0.70 0.27
Graphene e 1.07 0.81 1.07 e
Fig. 5. Different arrangements of the PPV oligomer with carbon nanotubes (in this example a zig-zag (9,0) nanotube), (a) axial parallel, (b) axial perpendicular and (c) circum-
ferential parallel configurations.diameter (see Table 1). It is unclear why there is a small jump in
binding energy between the nanotubes and the graphene cases.
Using both the LDA and PBE-D3 approximation, we find the
same binding energy difference of the system (towithin a fewmeV)
when the PPV molecule is placed either circumferentially or axially
parallel to the nanotube surface (SupplementaryMaterials, Table S2
and Figure S4). This shows the oligomers-graphene interaction
phenomenon shown here is robust and independent of the DFT
method.Fig. 6. Binding energy between armchair and zigzag carbon nanotubes and triphenyl PPV
reciprocal of the nanotube diameter. Linear best-fits are included, the cut-off at x¼0 is equ
effect is strongest for the smaller diameter tubes. (A colour version of this figure can be viFig. 6 shows a plot of PPV binding energy in the two parallel
(axial and circumferential) configurations vs the reciprocal of the
nanotube diameter, showing a linear relationship in each case. The
crossing point for 1/d¼ 0 (i.e. infinite diameter nanotubes) is at the
same binding energy for both, and exactly matches our calculated
graphene binding energy (1.07eV). Hence orientation effects in
nanotube-PPV composites are expected to be strongest for the
smallest tubes.
This theoretical work is in good agreement with theparallel to the tube surface and either axially or circumferentially oriented, versus the
ivalent to graphene (infinite diameter tubes). This shows the orientation dependence
ewed online.)
A. Yaya et al. / Carbon 141 (2019) 274e282 281experimental work by Massuyeau et al. [45] who studied com-
posites of PPV with SWCNTs inside nanoporous alumina templates.
Polarized-resolved Raman spectra exhibit strong anisotropic
orientation behaviour for the SWCNT and also the conjugated
polymer chains after their conversion into PPV, which was seen to
be dependent on the polarization angle, q. The observed maximum
at qm¼ 0 or 180 for a polarization direction parallel to the pore
axis, is strong evidence for alignment of both species parallel to the
pore and nanofiber axis. This optical anisotropy Raman behaviour is
comparable to other aligned carbon nanotube systems which have
also been observed in the emission and absorption behaviour of
SWCNTs [46,47]. While macroscopic alignment and solubility
behaviour is likely to be different in bulk material or thin films, this
result may still apply at the local level, indicating preference for
local alignment of PPV chains around inserted nanotubes. Such
short-range crystallisation will have important impact on, e.g.
charge transfer processes.
The results of these simulations in terms of final geometry (for
example distance between PPV and carbon materials) and quanti-
tative interaction energy are largely independent of DFT method,
for both metallic and semiconducting tubes, as shown in Supple-
mentary Table S1 and S2. They can be used to correlate the role of
aromaticity and backbone stiffness to the interfacial behaviour
between a SWCNT and PPV polymer, and this information can be
used to optimize the desired properties of the bulk material.
4. Conclusions
In the current study we demonstrate a continuous transition in
stacking behaviour as the extended aromaticity of the species un-
der study increases. While benzene-benzene interaction shows
almost iso-electronic structures for parallel AB- and T-stacking,
extending this to PPV oligomers shows a weak preference for par-
allel AB-stacking (which nonetheless reverts to T-stacking upon 3D
crystal formation). When the PPV interacts with more extended p-
systems such as nanotubes and graphene there is a transition to
strong preference for parallel stacking.
In the case of carbon nanotubes where there is an axial direction
and a diameter, the PPV ismost stable lying parallel to the nanotube
axis with binding energy increasing with nanotube diameter, in
order to minimise out-of-plane distortion of the molecule and
maximise its conjugation. This behaviour is independent of nano-
tube chirality, and metallic/semiconducting behaviour. Thus these
results demonstrate a general transition away from T-shape
stacking for small molecules to parallel stacking for larger systems.
This transition in stacking behaviour may have important im-
plications in nanocarbon-conjugated polymer composite design.
While such systems are beyond the scope of the current study, the
interface between crystallised PPV and an underlying graphene or
nanotube substrate must necessarily involve a transition from
parallel-stacked to T-stacking configuration of the PPV. Indeed the
polymer may be expected to also show stacking differences at
graphene edges and defects which show more localisation, similar
to extended conjugated polymer states. Changes in stacking
behaviour at edges have indeed been observed for small graphene
platelets [34].
Acknowledgements
AY acknowledges support from Pays de la Loire Region, France
and the University of Ghana Building a New Generation of Aca-
demics in Africa (BANGA-Africa) Project, with funding from the
Carnegie Corporation of New York. AI and CE acknowledge the
French Agence Nationale de Recherche Project ANR-16-CE24-0008-
01 “EdgeFiller” for funding. We acknowledge the CCIPL “Centre deCalcul Intensif Pays de la Loire” and where many of the calculations
were performed. We thank the referees for their useful and
insightful contribution to this paper.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.carbon.2018.09.071.
References
[1] C.D. Sherrill, T. Takatani, E.G. Hohenstein, J. Phys. Chem. 113 (38) (2009)
10146e10159. https://doi.org/10.1021/jp9034375.
[2] F. Jacob, J. Monod, J. Mol. Biol. 3 (3) (1961) 318e356. https://doi.org/10.1016/
S0022-2836(61)80072-7.
[3] B.H. Zimm, J. Chem. Phys. 33 (5) (1960) 1349. https://doi.org/10.1063/1.
1731411.
[4] J.H. Parish, Biochem. Educ. 13 (2) (1985), 92-92, https://doi.org/10.1016/0307-
4412(85)90046-9.
[5] D. Tuncel, Nanoscale 3 (9) (2011) 3545e3554. https://doi.org/10.1039/
c1nr10338e.
[6] M.-Q. Zhao, X.-F. Liu, Q. Zhang, G.-L. Tian, J.-Q. Huang, W. Zhu, F. Wei, ACS
Nano 6 (12) (2012) 10759e10769. https://doi.org/10.1021/nn304037d.
[7] Z. Yin, S. Sun, T. Salim, S. Wu, X. Huang, Q. He, Y. Ming Lam, H. Zhang, ACS
Nano 4 (9) (2010) 5263e5268. https://doi.org/10.1021/nn1015874.
[8] M.S. Gordon, L. Slipchenko, H. Li, J.H. Jensen, Annu. Rep. Comput. Chem. 3
(2007) 177e193. https://doi.org/10.1016/S1574-1400(07)03010-1.
[9] A.L. Ringer, C.D. Sherrill, Chem. Eur J. 14 (2) (2008) 2542e2547. https://doi.
org/10.1002/chem.200701622.
[10] T. Tran-Duc, N. Thamwattana, B.J. Cox, J.M. Hill, Philos. Mag. A 90 (13) (2010)
1771e1785. https://doi.org/10.1080/14786430903476349.
[11] C.-D. Tran, S. Lucas, D.G. Phillips, L.K. Randeniya, R.H. Baughman, T. Tran Cong,
Nanotechnology 22 (14) (2011) 145302. https://doi.org/10.1088/0957-4484/
22/14/145302.
[12] W. Yi, A. Malkovskiy, Y. Xu, X.-Q. Wang, A.P. Sokolov, M. Lebron-Colon,
M.A. Meador, Y. Pang, Polymer 51 (2) (2010) 475e481. https://doi.org/10.
1016/j.polymer.2009.11.052.
[13] Z.H. Zhu, G.Q. Lu, Langmuir 20 (24) (2004) 10751e10755. https://doi.org/10.
1021/la040062t.
[14] V. Lordi, N. Yao, J. Mater. Res. 15 (12) (2000) 2770e2779. https://doi.org/10.
1557/JMR.2000.0396.
[15] M. Yang, V. Koutsos, M. Zaiser, J. Phys. Chem. B 109 (20) (2005) 10009e10014.
https://doi.org/10.1021/jp0442403.
[16] R. Peverati, K.K. Baldridge, J. Chem. Theor. Comput. 4 (12) (2008) 2030e2048.
https://doi.org/10.1021/ct800252z.
[17] S. Tsuzuki, K. Honda, T. Uchimaru, M. Mikami, K. Tanabe, J. Am. Chem. Soc. 124
(1) (2002) 104e112. https://doi.org/10.1021/ja0105212.
[18] W. Langel, J. Solid State Electron. 6 (5) (2001) 359e360. https://doi.org/10.
1007/s10008-001-0256-1.
[19] S. Grimme, Angew Chem. Int. Ed. Engl. 47 (18) (2008) 3430e3434. https://doi.
org/10.1002/anie.200705157.
[20] J.W.G. Bloom, S.E. Wheeler, Angew. Chem. Int. Ed. 50 (34) (2011) 7847e7849.
https://doi.org/10.1002/anie.201102982.
[21] B.G. Sumpter, P. Kumar, A. Mehta, M.D. Barnes, W.A. Shelton, R.J. Harrison,
J. Phys. Chem. B 109 (16) (2005) 7671e7685. https://doi.org/10.1021/
jp0446534.
[22] M.O. Sinnokrot, E.F. Valeev, C.D. Sherrill, J. Am. Chem. Soc. 124 (36) (2002)
10887e10893. https://doi.org/10.1021/ja025896h.
[23] A. Vazquez-Mayagoitia, C.D. Sherrill, E. Apra`, B.G. Sumpter, J. Chem. Theor.
Comput. 6 (3) (2010) 727e734. https://doi.org/10.1021/ct900551z.
[24] M.J. Rayson, P. Briddon, Comput. Phys. Commun. 178 (2) (2008) 128e134.
https://doi.org/10.1016/j.cpc.2007.08.007.
[25] M.J. Rayson, P. Briddon, Phys. Rev. B 80 (20) (2009) 205104. https://doi.org/10.
1103/PhysRevB.80.205104.
[26] P. Briddon, M.J. Rayson, Phys. Status Solidi B 248 (6) (2011) 1309e1318.
https://doi.org/10.1002/pssb.201046147.
[27] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (18) (1996) 3865e3868.
https://doi.org/10.1103/PhysRevLett.77.3865.
[28] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 132 (15) (2010)
154104. https://doi.org/10.1063/1.3382344.
[29] S. Grimme, S. Ehrlich, L. Goergik, J. Comput. Chem. 32 (7) (2011) 1456e1465.
https://doi.org/10.1002/jcc.21759.
[30] C. Hartwigsen, S. Goedecker, Hutter, J. Phys. Rev. B 58 (7) (1998) 3641e3662.
https://doi.org/10.1103/PhysRevB.58.3641.
[31] J.P. Goss, M.J. Shaw, P. Briddon, Top. Appl. Phys. 104 (2007) 69e94. https://doi.
org/10.1007/11690320_4.
[32] B.G. Sumpter, M.L. Drummond, W.A. Shelton, E.F. Valeev, M.D. Barnes, Com-
put. Sci. Discov. 1 (1) (2008) 19, 015006, https://doi.org/10.1088/1749-4699/
1/1/015006.
[33] A. Yaya, E.K. Tiburu, B. Onwona-Agyeman, D. Dodoo-Arhin, J.K. Efavi, J. Com. &
modelling 3 (4) (2014) 27e42.
282 A. Yaya et al. / Carbon 141 (2019) 274e282[34] E. Cruz-Silva, X. Jia, H. Terrones, B.G. Sumpter, M. Terrones, M.S. Dresselhaus,
V. Meunier, ACS Nano 3 (7) (2013) 2834e2841. https://doi.org/10.1021/
nn4004204.
[35] G. Zheng, S.J. Clark, S. Brand, R.A. Abram, J. Phys. Condens. Matter 16 (47)
(2004) 8609e8620. https://doi.org/10.1088/0953-8984/16/47/013.
[36] D. Chen, M.J. Winokur, M.A. Masse, F.E. Karasz, Polymer 33 (15) (1992), 3316-
3122, https://doi.org/10.1016/0032-3861(92)90223-J.
[37] P.F. Van Hutten, J. Wildeman, A. Meetsma, G. Hadziioannou, J. Am. Chem. Soc.
121 (15) (1999) 5910e5918. https://doi.org/10.1021/ja990934r.
[38] S. Tretiak, A. Saxena, R.L. Martin, A.R. Bishop, Phys. Rev. Lett. 89 (9) (2002)
097402. https://doi.org/10.1103/PhysRevLett.89.097402.
[39] K.F. Voss, C.M. Foster, L. Smilowitz, D. Mihailovi
c, S. Askari, Z. Ni, S. Shi,
A.J. Heeger, F. Wudl, Phys. Rev. B 43 (6) (1991) 5109e5118. https://doi.org/10.
1103/PhysRevB.43.5109.
[40] F. Tournus, S. Latil, M.I. Heggie, J.-C. Charlier, Phys. Rev. B 72 (7) (2005)
075431. https://doi.org/10.1103/PhysRevB.72.075431.
[41] A.L. Ringer, C.D. Sherrill, Chem. Eur J. 14 (8) (2008) 2542e2547. https://doi.org/10.1002/chem.200701622.
[42] M. Dienwiebel, G.S. Verhoeven, N. Pradeep, J.W.M. Frenken, J.A. Heimberg,
W. Zandbergen, Phys. Rev. Lett. 92 (12) (2004) 126101. https://doi.org/10.
1103/PhysRevLett.92.126101.
[43] O.V. Ershova, T.C. Lillestolen, E. Bichoutskaia, Phys. Chem. Chem. Phys. 12 (24)
(2010) 6483e6491. https://doi.org/10.1039/c000370k.
[44] F. Massuyeau, Y. Zhao, A.A. El Mel, A. Yaya, F. Geschier, E. Gautron, S. Lefrant,
J.-V. Mevellec, C.P. Ewels, C.S. Hsu, E. Faulques, J. We
ry, J.L. Duvail, Nano
Research 6 (2) (2013) 149e158. https://doi.org/10.1007/s12274-013-0290-1.
[45] F. Massuyeau, J. We
ry, J.-L. Duvail, S. Lefrant, A. Yaya, C. Ewels, E. Faulques,
Beilstein J. Nanotechnol. 6 (2015) 1138e1144. https://doi.org/10.3762/bjnano.
6.115.
[46] G.S. Duesberg, I. Loa, M. Burghard, K. Syassen, S. Roth, Phys. Rev. Lett. 85 (25)
(2000) 5436e5439. https://doi.org/10.1103/PhysRevLett.85.5436.
[47] A.M. Rao, A. Jorio, M.A. Pimenta, M.S.S. Dantas, R. Saito, G. Dresselhaus,
M.S. Dresselhaus, Phys. Rev. Lett. 84 (8) (2000) 1820e1823. https://doi.org/10.
1103/PhysRevLett.84.1820.