Composites Part B 111 (2017) 261e269

Contents lists available at ScienceDirect

Composites Part B
journal homepage: www.elsevier.com/locate/compositesb

Temperature dependent mechanical properties of graphene reinforced

polymer nanocomposites e A molecular dynamics simulation
Feng Lin a, Y. Xiang a, b, Hui-Shen Shen c, *
a
School of Computing, Engineering and Mathematics, Western Sydney University, Locked Bag 1797, Penrith, NSW 2751, Australia
b
Centre for Infrastructure Engineering, Western Sydney University, Locked Bag 1797, Penrith, NSW 2751, Australia
c
School of Aeronautics and Astronautics, Shanghai Jiao Tong University, Shanghai 200240, People's Republic of China

a r t i c l e i n f o a b s t r a c t

Article history: This paper investigates the mechanical properties of graphene/PMMA nanocomposite system by using
Received 6 October 2016 the molecular dynamics simulations. The graphene nanoplates are assumed to be fully exfoliated in the
Received in revised form PMMA matrix and are all planar orientated, which are similar to the ones assembled using layer-by-layer
23 October 2016
technique. The Young's modulus and shear modulus of the composites with different graphene volume
Accepted 4 December 2016
Available online 9 December 2016
fractions under different temperatures are simulated and discussed. The results show that the Young's
and shear moduli increase with the increase of graphene volume fraction and decrease as the temper-
ature rises from 300 K to 500 K, while the efficiency of the reinforcement is reduced as the graphene
Keywords:
Nano-structures
content becomes higher. Simulations of single layer graphene under uniaxial tension, in-plane pure shear
Mechanical properties and uniformly distributed transverse load are performed and the effective thickness and the elastic
Computational modelling moduli of graphene are subsequently determined uniquely. The obtained stiffnesses of graphene are then
Graphene substituted into the simple rule of mixture to predict the overall mechanical properties of the composite.
Large discrepancies between the results from the MD simulations and the rule of mixture are observed.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction nanocomposites often exhibit enhanced thermal conductivity and

dimensional stability due to the excellent thermal properties of its
Graphene is one type of carbon allotropes which can also be graphene filler [4]. Besides structural applications, there is a
seen as a two dimensional hexagonal carbon sheet unzipped from a broader range of industries where graphene-polymer based
single walled carbon nanotube (CNT). Graphene has attracted nanocomposites can find their unique applications [5,6].
extensive attentions since it was first exfoliated in 2004 via me- Although the random distribution of graphene in polymer ma-
chanical cleavage [1]. Owing to its exceptional thermal, mechanical trix can enhance the thermo-mechanical properties of graphene-
and electrical properties, graphene and its derivatives have been based composites [7e14], the dispersion are still far from perfect
envisioned as the key ingredients of many futuristic applications. due to the poor miscibility between graphene and polymer matrix
Various attempts have been made to exfoliate or grow graphene in [15,16]. Li et al. [17] recently achieved planar orientation of the
large scale. In particular, efforts have been made trying to incor- graphene platelet in the PMMA matrix through forced assembly
porate them into polymer matrix in order to produce nano- and multilayer coextrusion. It demonstrated that in the direction
composites with exotic properties [2]. In general, the mechanical along graphene planar orientation, significant mechanical rein-
properties of the nanocomposites have certain degree of im- forcement can be achieved, and the reinforcing effect is higher than
provements over those of the matrix materials in terms of elastic random distributed graphene/PMMA composites. With the help of
modulus, tensile strength and toughness, which make the nano- in situ micro-Raman spectroscopy, Wang et al. [18] studied biaxial
composites ideal for fabricating lightweight gasoline tanks, compressive behaviour of monolayer graphene embedded in
strong wind turbines, medical implants, sports equipment, etc. [3]. PMMA matrix. It was found that the embedded graphene could
In addition to the mechanical reinforcement, graphene based continuously carry compressive loads after it reaches critical
buckling strain of 0.8%. At lower strain the graphene/PMMA system
shows mechanical recoverability, but at compressive strain higher
* Corresponding author. than 2.3%, interfacial debonding occurs at some areas. The effect of
E-mail address: hsshen@mail.sjtu.edu.cn (H.-S. Shen).

http://dx.doi.org/10.1016/j.compositesb.2016.12.004
1359-8368/© 2016 Elsevier Ltd. All rights reserved.
262 F. Lin et al. / Composites Part B 111 (2017) 261e269

lateral dimension of graphene on the properties of PMMA nano- infiltrated between graphene surfaces, which reduce certain poly-
composite was studied by Valle s et al. [19]. Compared with smaller mer chain mobility due to the confinement of graphene sheets.
flakes, the graphene with larger lateral dimensions provides better
interfacial stress transfer, thus increases the dynamic moduli as 2. Simulation methodology
well as the viscosity of the PMMA melt. The modulus of the com-
posite with larger graphene flake diameters is increased with the 2.1. Simulation method for graphene/PMMA composites
filler content loading without aggregation effects. Wang & Jayatissa
[20] studied the electrical conductivity of graphene/PMMA com- The Software used for simulation is LAMMPS (Large-scale
posite film. The sample being investigated was made by solution Atomic/Molecular Massively Parallel Simulator) [29]. COMPASS
phase mixing and spin coating technique, and had relatively large forcefield (i.e. Class2 in LAMMPS) is used as it was proven to be
weight fraction of graphene up to 20-40 wt%. Recently, the layer- effective and accurate in handling polymer subjects in condensed
by-layer (LbL) assembly technique has been employed to manner [30]. The energy terms of COMPASS can be expressed as
assemble structural layers made of polymer and graphene oxide in follows [30].
diverse dimensions [21,22]. The reduction of graphene oxide sheets
in LbL film can be later attained to restore the unique properties of Etotal ¼ Ediagonal þ Ecross þ Eintermol (1)
graphene, while keeping the structure of polymer counterparts
intact. Xh i
It is well perceived that the microstructure and material con- Ediagonal ¼ k2 ðb  b0 Þ2 þ k3 ðb  b0 Þ3 þ k4 ðb  b0 Þ4
b
stitutes determine the physical properties of the composite mate- Xh i
rial. For nanocomposite, due to its nanosized filler, molecular þ k2 ðq  q0 Þ2 þ k3 ðq  q0 Þ3 þ k4 ðq  q0 Þ4
dynamics (MD) simulation has been proven to be a valuable tool to q
X
investigate the interactions between the nanofiller and the sur- þ ½k1 ð1  cos fÞ þ k2 ð1  cos 2 fÞ
rounding matrix molecules. Previously, the effects of single walled f
CNT (SWCNT) on the mechanical properties of polymer nano- X
þ k3 ð1  cos 3 fÞ þ k2 c2 (2)
composite were studied by MD simulation. The simulated SWCNT/
c
PMMA and the SWCNT/PmPV system showed significant rein-
forcement in the SWCNT longitudinal direction [23]. Rissanou and X   X
Harmandaris [24] studied graphene reinforced PMMA composites Ecross ¼ kðb  b0 Þ b0  b0 0 þ kðb  b0 Þðq  q0 Þ
by MD simulation, in which the interfacial stress transfer and b;b0 b;q
X
deformation mechanism have been investigated. Simulation of the þ ðb  b0 Þ½k1 cos f þ k2 cos 2 f þ k3 cos 3 f
PMMA chains in the vicinity of graphene surface showed strong b;f
effect of graphene layers on the density, structure conformation as X
þ ðq  q0 Þ½k1 cos f þ k2 cos 2 f þ k3 cos 3 f
well as the dynamics of PMMA. It was found that PMMA chains
q;f
close to the graphene layer are practically frozen and trapped in a X   X  
metastable condition. Unlike CNTs, graphene exhibited a wrinkled þ kðq  q0 Þ q0  q0 0 þ kðq  q0 Þ q0  q0 0 cos f
morphology due to the thermal fluctuation and long-range q;q0 q;q0 ;f
attraction among atoms. Liu et al. [25] studied the effect of wrin- (3)
kled surface of graphene on the interfacial mechanical property of
2 !9 !6 3
nanocomposites by MD simulation. It was found that a wrinkled
X qi qj X rijo rijo
graphene surface can effectively enhance the interfacial mechanical Eintermol ¼ þ εij 42 3 5 (4)
property compared with a flat one and the wrinkled graphene/ i;j
rij i;j
rij rij
PMMA system has even stronger interaction than the graphene/PE
system. in which the valence terms represent internal coordinates of bond
Despite much simulation efforts being made to study the me- (b), angle (q), torsion angle (f), and out-of-plane angle (c), and the
chanical properties of graphene based polymer nanocomposites, to cross-coupling terms include combinations of two or three internal
the authors' knowledge, the previous studies mostly focused on the coordinates. The LJ-9-6 parameters ri;j o and ε follow the 6th order
ij
small graphene/graphene oxide platelet reinforcement, localized combination law as follows [30].
interface study, or random distribution of the nanofiller [7e14].
However, there are no comprehensive studies on the temperature 2   6 31=6
6
rio þ rjo
dependent mechanical properties of the graphene reinforced o 6 7
ri;j ¼4 5 (5)
nanocomposites. Therefore the current work aims to provide a set 2
of mechanical properties of the nanocomposites which can be
directly used for the mechanical analysis of nanocomposite struc-
2    3 3
tures. The pristine graphene sheets are used in the current model o 3 ro
because untreated graphene are able to offer better mechanical  1=2 6 ri j 7
εij ¼ 2 εi εj 4 6  6 5 (6)
properties than those functionalized graphene or defective gra- rio þ rjo
phene which are subjected to chemical treatment [26e28]. Inspired
by the composites made via layer by layer assembly, the single layer
and the partial charge q for atom i is the sum of all charge bond
graphenes are assumed to have large dimensions as well as planar
increments dij
orientation which are fully exfoliated in the PMMA matrix.
Compared with CNT which has hollow inaccessible interior surface, X
qi ¼ dij (7)
the wrinkled graphene surface interacts with polymer matrix on
j
both sides of the surface once embedded. High graphene weight
fractions are attempted in the current model, provided that poly- In the current simulation, the cut-off radius of Columbic in-
mer chains with lower molecular weight and radius of gyration are teractions and the L-J potential are both set at 12 Å.
 F. Lin et al. / Composites Part B 111 (2017) 261e269 263

Atactic PMMA polymer chain with 30 monomers is first built

using Materials Studio 7.0 polymer builder, as shown in Fig. 1. The
polymer chains are then filled to the top and bottom of the one
atom thin graphene in the simulation box in a Monte Carlo manner
to a predetermined density using the Amorphous Cell module of
Materials Studio 7.0. Periodical boundary conditions are applied in
all three dimensions. For clarity, graphene embedded in the PMMA
matrix is highlighted in golden colour (refer to web version of the
paper for pictures with colours), as shown in Fig. 2. The volume
fraction of graphene in nanocomposite is calculated by using the
relationship between weight fractions and volume fractions [31].

w
VG ¼   G    (8)
wG þ rG rm  rG rm wG

where wG is the mass fraction of graphene, and rG and rm are the

densities of monolayer graphene and matrix, respectively. For
example, for composite with the total number of atoms being
59524 and graphene weight fraction 8.8%, taking rG ¼ 4118 kg/m3
(see Section 3.2) and rm ¼ 1150 kg/m3 [23], from Eq. (8) the gra-
phene volume fraction is calculated to be VG ¼ 3%. After the
simulation cells with different graphene weight fractions are con-
structed, the molar mass of the graphene as well as the PMMA
molecules in the simulation cell is calculated and the correspond-
ing graphene weight fractions can be readily obtained. The calcu-
lated graphene volume fractions with respect to the graphene
weight fractions in the composite are listed in Table 1.
After constructing the initial model, energy minimization is first
performed using a combination of conjugate gradient and steepest
descent methods. In order to alleviate the close contact of atoms,
the model is first subjected to canonical ensemble (NVT) dynamics
runs under different time steps until the temperature reaches the
value which is close to but around 100 Ke200 K higher than the
desired temperature. The model is then undergone isothermal-
isobaric (NPT) ensemble dynamics runs until the pressure of
1 atm in all three dimensions of the simulation cell is achieved. It
has been reported that the melting temperature is increased when
reinforcing graphene is added in PMMA matrix [32]. Therefore, the
sampling temperature in this study is assumed to be 300 K, 400 K
and 500 K. The Nose -Hoover thermostat and barostat are used for
temperature and pressure control, and the Verlet algorithm is used
for time integration. It is noted that because of the pressure vari- Fig. 2. Graphene/PMMA system: (a) 3% graphene volume fraction; (b) 11% graphene
ation, there is a slight height change of the simulation cell, which in volume fraction.

turn changes the graphene volume fraction. However the height

variation is within 0.1% and can be neglected.
To calculate the tensile Young's modulus of the composites in
the three orthogonal directions, the pressure is set to be ramped
incrementally from 1 atm to a maximum value in the direction
which the Young's modulus needs to be calculated. The rate of
ramping is 1 MPa/ps. In the meantime, the compressive pressure
values of the other two directions are maintained as equal to the
pressure in the tensile direction multiplied by the Poisson's ratio of

Table 1
Graphene weight fractions and corresponding volume fractions.

Graphene weight Corresponding graphene volume

fraction (wt%) fraction ±0.05 (vl%)

8.8 3
14.9 5
Fig. 1. PMMA polymer chain with 30 monomers as matrix material colored by ele-
20.4 7
ments. i.e. atoms with red, light grey and dark grey colours are oxygen atoms,
25.7 9
hydrogen atoms and carbon atoms, respectively. (For interpretation of the references
30.9 11
to colour in this figure legend, the reader is referred to the web version of this article.)
264 F. Lin et al. / Composites Part B 111 (2017) 261e269

PMMA. In order to consider the interfacial stress transfer between found that the effective thickness of graphene is less than 0.34 nm.
the graphene and the matrix, at each simulation timestep during The Young's modulus of graphene decreases with increase of
pressure increment, a two-step time integration is adopted. Firstly temperature whereas the shear modulus depends weakly on
the PMMA molecules are subject to isothermal isobaric (NPT) dy- temperature variation. Min and Aluru [38] studied the shear
namics runs and only the atoms in the PMMA matrix are set to deformation of graphene using MD simulation. With the increase of
update positions which match the simulation box change, and then temperature, it was found that the shear modulus increases first up
the graphene molecule is subjected to a separated time integration to 800 K then decreases with further increase of temperature. As for
under microcanonical ensemble (NVE). As reported by Liu et al. coefficient of thermal expansion, the simulation results are rather
[25], the interfacial shear stress for a wrinkled-graphene/PE system dispersed. It can be positive, negative or change sign depending on
is much higher than a flat-graphene/PE system (i.e. 274 MPa versus the potentials which are used for computation [39]. It is also very
112 MPa). They also observed that wrinkled-graphene/PMMA sys- sensitive to the interaction between substrate and graphene [40].
tem has better mechanical properties than its graphene/PE coun- The current procedure for calculating the material properties of
terpart. We also observed that wrinkles exist in the graphene- single layer graphene is based on the work by Shen et al. [37]. There
PMMA system in the current study as shown in Fig. 3. We choose are, however, some notable differences in the model building. We
that the maximum value for the composites with 3% and 5% gra- use LCBOP (long-range carbon bond order potential) instead of
phene volume fractions is 150 MPa, and for the composites with 7%, AIREBO as demonstrated by Los and Fasolino that it can give more
9%, 11% graphene volume fractions the maximum value is 200 MPa. accurate results [41]. In order to minimize the size effect, the cur-
This is to ensure that there is no significant sliding between the rent graphene model is in near square shape of 147.6  147.7 (Å)2.
graphene and polymer matrix under the chosen maximum pres- The four edges of the graphene are distinguished in Fig. 4 as grey
sure on the system. As aforementioned, the pressure on the system area while the graphene proper is in golden colour (refer to web
is increased by 1 MPa/ps and sample stress-strain points are version of the paper for pictures with colours). For edge atoms, only
recorded before the system reaches the maximum pressure. Under the out of plane displacements are restrained and the in-plane
such pressure ramping setting, the corresponding strain rate is displacements are not restrained. Such choice is made as it is
estimated to be around 1.5%e2% per nanosecond. On average, about found that the choice of boundary conditions can influence the
1000 sample stress-strain points are used in the regression process calculated results of CTE, which is in consistence with the findings
to obtain the Young's modulus. After obtaining the Young's in Ref. [40]. The results of shear rigidity can also be influenced as
modulus, similar procedure is used to calculate the shear modulus, well. For calculating the in-plane stretching rigidity C11 and C22,
and the maximum value of the imposed shear pressure is 62 MPa. where 11 denotes the zigzag direction and 22 denotes the armchair
direction, uniaxial tensile forces are applied on each atom on one
pair of opposite edges. For in-plane shear rigidity C66 calculation,
2.2. Simulation for single layer graphene
forces are applied on four edges so that the graphene is essentially
in pure shear state. For bending calculation, the edge area is simply
Understanding the mechanical properties of single layer gra-
supported and uniformly distributed transverse load is applied on
phene itself is crucial for evaluating the effectiveness of rule of
the graphene proper, and the corresponding central deflection of
mixture for the graphene/PMMA composites. The elastic properties
the graphene sheet W is recorded. After performing the tension,
and intrinsic breaking strength of graphene have been measured by
shearing and bending simulations, the effective thickness, the
atomic force microscope [33]. It was estimated that the Young's
Young's moduli and shear modulus of the graphene sheet can be
modulus of graphene is of the order of 1 TPa, given the effective
calculated using the approach given in Ref. [37]. The CTE is
thickness of 0.34 nm. The in plane shear modulus of graphene film
grown by chemical vapor deposition on copper foil has been esti-
mated to be 280 ± 36 GPa with the effective thickness of 0.34 nm
[34]. The coefficient of thermal expansion (CTE) has been measured
using Raman spectroscopy in the temperature range between
200 K and 400 K. It was found that the thermal expansion coeffi-
cient remains negative in the whole temperature range [35].
Beside experimental work, theoretical and numerical studies on
graphene have also been performed. Ni et al. [36] studied the
anisotropic mechanical properties of graphene from MD simula-
tion. It was found that under the same uniaxial tensile load, the
fracture occurs sooner in the armchair direction instead of in the
zigzag direction. Shen et al. [37] studied the elastic properties of
graphene using MD simulation and classical plate theory. It was

Fig. 4. Single layer graphene MD model with different areas highlighted for material
Fig. 3. Wrinkled graphene surface under tension and thermal conditions. properties simulation.
 F. Lin et al. / Composites Part B 111 (2017) 261e269 265

calculated by tracking the displacement variation DL of the edge Table 2

area with respect to its original length L0 under the variation of Temperature dependent material properties for aligned graphene/PMMA compos-
ites in zigzag direction E11, armchair direction E22 and thickness direction E33.
temperature DT. After each variation of temperature DT , the
temperature is maintained and DL is calculated as the average Young's moduli Graphene volume 300 K (GPa) 400 K (GPa) 500 K (GPa)
length change, i.e. taking the average value of the 8000 steps of the fraction (%)

total 10000 steps. The CTE is then calculated using the equation E11 3 36.538 32.639 31.926
a ¼ DL=ðL0 DTÞ. 5 59.544 53.462 52.171
7 80.096 71.698 69.960
9 90.023 81.035 79.218
3. Results and discussion 11 96.388 88.557 87.039
E22 3 35.613 31.750 31.854
5 57.479 51.661 50.225
3.1. Graphene/PMMA composites 7 78.843 70.928 67.453
9 88.750 78.091 76.019
Before any deformation is applied, the structural properties of 11 94.265 86.063 84.743
E33 3 3.401 2.634 1.872
the composite have been studied. Assuming that each atom mass is
5 3.663 2.783 1.966
concentrated only at the centroid of the atom sphere, Fig. 5 illus- 7 4.380 3.407 3.160
trates the molar mass distribution against the composite height 9 5.182 4.271 3.226
profile for composite with 9% graphene volume fraction under 11 5.817 4.802 3.321
300 K and 1 atm. The two largest peaks indicate the position of
graphene while the lowest peaks with no mass indicate vdW gaps
between graphene and PMMA chains. In the vicinity of graphene, With the graphene volume fraction increases from 3% to 11%, the
due to the nonbond attraction forces, four interphase regions with Young's moduli become more sensitive to temperature variation.
an average thickness around 0.35 nm have been observed. The Compared with 3% graphene content, the composite with 11%
density of PMMA in the interphase region is higher than the ones in graphene has the largest reduction of moduli when temperature
the region away from the graphene surface, and the PMMA mole- rises from 300 K to 500 K. This is due to the fact that the material
cules are “arrested” in the region by wrinkled graphene surface. It properties of graphene are temperature dependent. It is also
was reported that the existence of interphase region has also been observed that for composites under the same temperature, the rate
found for graphene oxide-silk fibroin nanocomposite membrane, of improvement decreases with the increase of graphene volume
and the strong interactions between nanofiller and matrix in the fraction, which suggests that composites with higher graphene
interphase region are thought to be the reason for the measured content may not have equivalent improvement as the composites
ultra high tensile modulus up to 145 GPa [42]. As discussed previ- with lower graphene content.
ously, the wrinkled graphene will also provide additional interfacial Table 3 presents the temperature dependent shear modulus G12
strength for the graphene/PMMA system [25]. of composites in the graphene plane with different graphene vol-
Table 2 presents the temperature dependent Young's moduli of ume fractions. Similar influences of temperature and graphene
composites with different graphene volume fractions. E11 is the volume fractions on shear modulus can also be observed. Table 4 is
Young's modulus along the zigzag direction of graphene, E22 is the the results of G13 and G23 for the composites at 300 K. Overall the
Young's modulus along the armchair direction and E33 is in the results indicate that the shear moduli in the transverse plane of the
thickness direction. From the results it can be seen that due to the composite are lower than or similar to that of the PMMA matrix.
planar orientation of graphene, the mechanical properties of the The transverse shear moduli of the composite are not improved by
composite exhibit orthotropic characteristics. Generally speaking, increasing the graphene content. It may be explained that under
the Young's moduli increase with the increase of graphene volume transverse shear, the perpendicular graphene surface further dis-
fraction and decrease as the temperature rises from 300 K to 500 K. rupts the intermolecular grip forces between PMMA molecules

Fig. 5. Molar mass distribution for composite with 9% graphene volume fraction under 300 K and 1 atm.
266 F. Lin et al. / Composites Part B 111 (2017) 261e269

Table 3 Table 5
Temperature dependent material properties G12 for aligned graphene/PMMA com- Temperature-dependent inplane rigidities for monolayer graphene (L11 ¼ 14.76 nm,
posites in plane direction. L22 ¼ 14.77 nm).

Graphene volume 300 K (GPa) 400 K (GPa) 500 K (GPa) Temperature (K) C11 (GPa nm) C22 (GPa nm) C66 (GPa nm)
fraction (%)
1 351.76 350.88 126.14
3 11.388 11.572 11.700 100 345.18 344.64 132.83
5 15.655 12.919 11.450 200 342.52 341.79 127.24
7 23.644 19.574 17.106 300 340.65 339.74 128.38
9 33.635 25.566 22.202 400 332.56 331.39 129.90
11 34.713 26.735 23.478 500 328.65 326.18 131.55
700 326.50 323.63 127.06
1000 312.04 309.36 121.17
Table 4
Temperature dependent material properties G13 and G23 for aligned graphene/
PMMA composites in thickness direction at 300 K.
Note that these equations are for the plate with movable in-
Graphene volume G13 (GPa) G23 (GPa) plane displacements instead of immovable in-plane boundary
fraction (%)
conditions as used in Ref. [37]. Under 300 K, the Poisson's ratio vG
12
3 0.401 0.306 is calculated to be 0.177, the central deflection W under uniformly
5 0.301 0.055
distributed transverse load q of 6 MPa is calculated to be 0.387 nm.
7 0.081 0.260
9 0.138 0.271 The effective length in the zigzag direction L11 and armchair di-
11 0.179 0.062 rection L22 is measured to be 14.259 nm and 14.190 nm, respec-
tively. The calculated effective thickness of the graphene sheet
using Eq. (9) is hG ¼ 0.188 nm, and rG ¼ 4118 kg/m3.
which usually exist in pure PMMA matrix, so that the reinforcing Fig. 7 shows the results of CTE under the temperature variation
effect would be weakened. Furthermore, we cannot obtain G13 and in the current study. It can be seen that at lower temperature the
G23 results under 400 K or 500 K due to that the higher thermal CTE remains negative and it turns to positive when the temperature
kinetic energies cause further oscillations of atomic displacements is higher than around 520 K. This trend agrees with the studies in
which further diverge the already weak results of the shear moduli. Refs. [39] and [43]. However the present values of the CTE for
temperature being lower than 300 K are closer to the ones in
3.2. Single layer graphene Ref. [39] rather than [43]. For CTE in the temperature range less
than 300 K, whether it will monotonically increase or attain a
Table 5 presents the temperature dependent results of in-plane minimal value between 100 K and 300 K remains unclear and
stretching rigidity C11 and C22 as well as the in-plane shear rigidity further investigation is needed.
C66 for the simulated graphene sheet. Fig. 6 shows the effect of Table 6 summaries the material properties of the monolayer
temperature changes on the surface moduli C11, C22 and C66 varying graphene simulated in this study. It is observed that the Young's
from 1 K to 1000 K. It can be seen that the surface moduli C11 and moduli E11 G and EG decrease with the increase of temperature and
22
C22 are decreased with increase in temperature, and the surface the Young's moduli in zigzag direction are slightly higher than the
shear modulus C66 weakly depends on the temperature variation. corresponding ones in the armchair direction. The shear modulus
To simplify calculation, the effective thickness hG is assumed to GG12 increases first then decreases as temperature rises. It is, how-
be invariable with temperature. The equations for calculating the ever, in general not very sensitive to temperature variation at low
effective thickness hG are expressed as [37]. temperatures.
212 31=2
ð1vG12 vG21 ÞqL411 ð3Þ
 Aw W 3
hG ¼ 4 5
C11 3.3. Comparison with the conventional rule of mixture
ð1Þ
(9)
Aw W
A number of micromechanics models have been proposed for
where the determination of effective properties of carbon nanotube-
reinforced and/or graphene-reinforced composites, where the
ð1Þ p6  2 4
 rule of mixture is commonly used. A comparison study of the cur-
AW ¼ 1 þ 2c4 b þ c3 b (10)
16 rent results against the equations of rule of mixture has been car-
ried out. The effective Young's moduli and shear modulus can be
" ! #
c3 b     expressed as [44].
4
ð3Þ 3p6 1 2 4 4
Aw ¼ þ 1 þ 2c4 b þ c3 b þ 1 þ c3 b
64 g13 g31 G
  E11 ¼ VG E11 þ Vm E m (13)
1  vG G
12 v21
1 V Vm
(11) ¼ GG þ m (14)
E22 E22 E
and
  1 V Vm
2 4 ¼ GG þ m (15)
g13 ¼ 1 þ 18c4 b þ 18c3 b ; E33 E33 E
 
2 4
g31 ¼ 81 þ 18c4 b þ c3 b ;
b ¼ L11 =L22 ; (12) 1 V Vm
¼ G þ m (16)
c3 ¼ C22 =C11 ; G12 GG G
  12
c4 ¼ v21 þ 2ðC66 =C11 Þ 1  vG G
12 v21
where E11, E22 and E33 are nanocomposite Young's moduli in the
 F. Lin et al. / Composites Part B 111 (2017) 261e269 267

Fig. 6. Temperature-dependent inplane rigidities for monolayer graphene.

Fig. 7. CTE of monolayer graphene under the temperature variation.

graphene zigzag, armchair and transverse directions, G12 is the composites cannot be employed to predict the stiffnesses of the
shear modulus of the nanocomposite, E11 G , E G and E G are the graphene reinforced nanocomposites. Further investigations on the
22 33
Young's moduli of the graphene in the zigzag, armchair, transverse correct form of equations are needed in order to determine the
m m
directions, GG
12 is the shear modulus of the graphene, E and G are mechanical properties of the graphene/PMMA composites.
the Young's modulus and shear modulus of the matrix, and Vm and
VG are the volume fractions of matrix and graphene, respectively.
Here we take E33G ¼ 39.511 GPa [45]. Em ¼ 2.5 GPa, Gm ¼ 0.912 GPa. 4. Conclusions
Using the MD results of graphene in Table 6 and the equations of
the rule of mixture of Eqs. (13e16), the stiffnesses of the composite Molecular dynamics simulations have been performed to study
under T ¼ 300 K for different graphene volume fractions are ob- the temperature dependent mechanical properties of graphene
tained and listed in Table 7. It can be seen that E22 is much smaller reinforced PMMA nanocomposites with certain material consti-
than E11 using the rule of mixture, whereas the MD results show tutes arrangement. By inspecting the simulated model it is found
that E11 and E22 are almost the same. The results are not consistent that the graphene has wrinkled surface under thermal conditions
with the MD simulation results of the nanocomposite in the pre- and a PMMA interphase region exists in the vicinity of the graphene
vious sections and it is due to the nanosize effect, surface effect and surface. The mechanical properties of the nanocomposites exhibit
other factors. Similar to CNT reinforced nanocomposites [23], the orthotropic characteristics due to the planar orientated graphene
simple rule of mixture used in the conventional fibre reinforced fillers. The Young's and shear moduli of the nanocomposites in-
crease with the increase of graphene volume fraction and decrease
268 F. Lin et al. / Composites Part B 111 (2017) 261e269

Table 6 strength and thermal stability. Carbon 2011;49:198e205.

Temperature-dependent material properties for monolayer graphene [9] Jang JY, Jeong HM, Kim BK. Compatibilizing effect of graphite oxide in gra-
(L11 ¼ 14.76 nm, L22 ¼ 14.77 nm, thickness hG ¼ 0.188 nm, vG phene/PMMA nanocomposites. Macromol Res 2009;17:626e8.
12 ¼ 0.177,
rG ¼ 4118 kg/m3). [10] Khezerlou H, Tahmasebipour M. Molecular dynamic simulation of graphene-
poly methyl methacrylate nano-composite. J Nanoelectron Optoelectron
Temperature (K) G (GPa)
E11 G (GPa)
E22 GG
12 (GPa) aG11 (106/K) aG22 (106/K) 2014;9:580e3.
[11] Roussou R-E, Karatasos K. Graphene/poly(ethylene glycol) nanocomposites as
300 1812 1807 683 0.90 0.95 studied by molecular dynamics simulations. Mater Des 2016;97:163e74.
400 1769 1763 691 0.35 0.40 [12] Rahman R, Haque A. Molecular modeling of crosslinked grapheneeepoxy
500 1748 1735 700 0.08 0.08 nanocomposites for characterization of elastic constants and interfacial
700 1737 1721 676 0.25 0.30 properties. Compos B-Eng 2013;54:353e64.
1000 1660 1646 645 0.32 0.32 [13] Shiu S-C, Tsai J-L. Characterizing thermal and mechanical properties of gra-
phene/epoxy nanocomposites. Compos B-Eng 2014;56:691e7.
[14] Zhang Y, Zhuang X, Muthu J, Mabrouki T, Fontaine M, Gong Y. Timon Rabczuk
T. Load transfer of graphene/carbon nanotube/polyethylene hybrid nano-
Table 7 composite by molecular dynamics simulation. Compos B-Eng 2014;63:27e33.
Stiffness of graphene/PMMA composite under T ¼ 300 K for different graphene [15] Saravanan N, Rajasekar R, Mahalakshmi S, Sathishkumar TP, Sasikumar KSK,
volume fractions using the rule of mixture equations in Eqs. (13)e(16). Sahoo S. Graphene and modified graphene-based polymer nanocomposites - a
review. J Reinf Plast Compos 2014;33:1158e80.
Graphene volume E11 (GPa) E22 (GPa) E33 (GPa) G12 (GPa) [16] Hu K, Kulkarni DD, Choi I, Tsukruk VV. Graphene-polymer nanocomposites for
fraction (%) structural and functional applications. Prog Polym Sci 2014;39:1934e72.
[17] Li XG, McKenna GB, Miquelard-Garnier G, Guinault A, Sollogoub C, Regnier G,
3 56.78 (55%)a 2.58 (93%) 2.57 (24%) 0.94 (92%) et al. Forced assembly by multilayer coextrusion to create oriented graphene
5 92.97 (56%) 2.63 (95%) 2.62 (28%) 0.96 (94%) reinforced polymer nanocomposites. Polymer 2014;55:248e57.
7 129.16 (61%) 2.69 (97%) 2.68 (39%) 0.98 (96%) [18] Wang G, Liu L, Dai Z, Liu Q, Miao H, Zhang Z. Biaxial compressive behavior of
9 165.35 (84%) 2.75 (97%) 2.73 (47%) 1.00 (97%) embedded monolayer graphene inside flexible poly (methyl methacrylate)
11 201.54 (109%) 2.81 (97%) 2.79 (52%) 1.03 (97%) matrix. Carbon 2015;86:69e77.
a [19] Valles C, Abdelkader AM, Young RJ, Kinloch IA. The effect of flake diameter on
Difference ¼ 100%  [(Eij)-(Eij)MD]/(Eij)MD (i,j ¼ 11, 22, 33 and 12).
the reinforcement of few-layer graphene-PMMA composites. Compos Sci
Technol 2015;111:17e22.
[20] Wang W, Jayatissa AH. Comparison study of graphene based conductive
as the temperature rises from 300 K to 500 K. The Young's and nanocomposites using poly(methyl methacrylate) and polypyrrole as matrix
materials. J Mater Sci-Mater Electron 2015;26:7780e3.
shear moduli become more sensitive to temperature variation for
[21] Lee T, Min SH, Gu M, Jung YK, Lee W, Lee JU, et al. Layer-by-layer assembly for
the composites with higher graphene volume fractions. Increasing graphene-based multi layer nanocomposites: synthesis and applications.
graphene contents may not necessarily results in proportional Chem Mater 2015;27:3785e96.
increasing of the composite moduli. Simulations of single atom thin [22] Sheng K, Bai H, Sun Y, Li C, Shi G. Layer-by-layer assembly of graphene/pol-
yaniline multilayer films and their application for electrochromic devices.
graphene are also performed to determine its mechanical proper- Polymer 2011;52:5567e72.
ties. The comparison between the MD simulation results and the [23] Han Y, Elliott J. Molecular dynamics simulations of the elastic properties of
simple rule of mixture reveals that the conventional equations of polymer/carbon nanotube composites. Comput Mater Sci 2007;39:315e23.
[24] Rissanou AN, Harmandaris V. Structure and dynamics of poly(-
rule of mixture for microfiber reinforced composite may not be methylmethacrylate)/graphene systems through atomistic molecular dy-
suitable for the graphene reinforced nanocomposites. The current namics simulations. J Nanoparticle Res 2013;15:1589.
study may provide valuable insight on the material properties [25] Liu F, Hu N, Ning H, Liu Y, Li Y, Wu L. Molecular dynamics simulation on
interfacial mechanical properties of polymer nanocomposites with wrinkled
which are currently missing for subsequent analysis on the me- graphene. Comput Mater Sci 2015;108:160e7.
chanical behaviors of graphene reinforced nanocomposite struc- [26] Ansari R, Ajori S, Motevalli B. Mechanical properties of defective single-
tures with such material constitutes arrangement. layered graphene sheets via molecular dynamics simulation. Superlattices
Microstruct 2012;51:274e89.
[27] Ansari R, Motevalli B, Montazeri A, Ajori S. Fracture analysis of monolayer
Acknowledgment graphene sheets with double vacancy defects via MD simulation. Solid State
Commun 2011;151:1141e6.
[28] Ansari R, Mirnezhad M, Rouhi H. Mechanical properties of fully hydrogenated
This study was supported by the Australian Research Council graphene sheets. Solid State Commun 2015;201:1e4.
Grant DP140104156 and the National Natural Science Foundation of [29] Plimpton S. Fast parallel algorithms for short-range molecular dynamics.
J Comput Phys 1995;117:1e19.
China under Grant 51279103. The authors are very grateful for these [30] Sun H. COMPASS: an ab initio force-field optimized for condensed-phase
financial supports. The authors also greatly appreciate the applications - overview with details on alkane and benzene compounds.
computational resources provided to this project by the School of J Phys Chem B 1998;102:7338e64.
[31] Rafiee MA, Rafiee J, Wang Z, Song H, Yu Z-Z, Koratkar N. Enhanced mechanical
Computing, Engineering and Mathematics at Western Sydney
properties of nanocomposites at low graphene content. ACS Nano 2009;3:
University. 3884e90.
[32] Sheng X, Xie D, Cai W, Zhang X, Zhong L, Zhang H. In situ thermal reduction of
graphene nanosheets based poly(methyl methacrylate) nanocomposites with
References effective reinforcements. Ind Eng Chem Res 2015;54:649e58.
[33] Lee C, Wei XD, Kysar JW, Hone J. Measurement of the elastic properties and
[1] Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, et al. intrinsic strength of monolayer graphene. Science 2008;321:385e8.
Electric filed effect in atomically thin carbon films. Science 2004;306:666e9. [34] Liu X, Metcalf TH, Robinson JT, Houston BH, Scarpa F. Shear modulus of
[2] Mukhopadhyay P, Gupta RK. Graphite, graphene, and their polymer nano- monolayer graphene prepared by chemical vapor deposition. Nano Lett
composites. Hoboken: Taylor and Francis; 2012. 2012;12:1013e7.
[3] Du J, Cheng H-M. The Fabrication, Properties, and uses of graphene/polymer [35] Yoon D, Son YW, Cheong H. Negative Thermal expansion coefficient of gra-
composites. Macromol Chem Phys 2012;213:1060e77. phene measured by raman spectroscopy. Nano Lett 2011;11:3227e31.
[4] Zhang Y-F, Zhao Y-H, Bai S-L, Yuan X. Numerical simulation of thermal con- [36] Ni Z, Bu H, Zou M, Yi H, Bi K, Chen Y. Anisotropic mechanical properties of
ductivity of graphene filled polymer composites. Compos B-Eng 2016;106: graphene sheets from molecular dynamics. Phys B-Conden Matt 2010;405:
324e31. 1301e6.
[5] Das TK, Prusty S. Graphene-based polymer composites and their applications. [37] Shen L, Shen H-S, Zhang C-L. Temperature-dependent elastic properties of
Polymer-Plast Technol Eng 2013;52:319e31. single layer graphene sheets. Mater Des 2010;31:4445e9.
[6] Potts JR, Dreyer DR, Bielawski CW, Ruoff RS. Graphene-based polymer nano- [38] Min K, Aluru NR. Mechanical properties of graphene under shear deformation.
composites. Polymer 2011;52:5e25. Appl Phys Lett 2011;98:013113.
[7] Zhao X, Zhang QH, Chen DJ, Lu P. Enhanced mechanical properties of [39] Magnin Y, Forster GD, Rabilloud F, Calvo F, Zappelli A, Bichara C. Thermal
graphene-based poly(vinyl alcohol) composites. Macromolecules 2010;43: expansion of free-standing graphene: benchmarking semi-empirical poten-
2357e63. tials. J Phys Conden Matt 2014;26:185401.
[8] Vadukumpully S, Paul J, Mahanta N, Valiyaveettil S. Flexible conductive gra- [40] Jiang JW, Wang JS, Li BW. Thermal expansion in single-walled carbon nano-
phene/poly(vinyl chloride) composite thin films with high mechanical tubes and graphene: nonequilibrium Green's function approach. Phys Rev B
 F. Lin et al. / Composites Part B 111 (2017) 261e269 269

2009;80:205429. order Cauchy-Born rule. Comput Mater Sci 2012;51:445e54.

[41] Los JH, Fasolino A. Intrinsic long-range bond-order potential for carbon: [44] Frankland SJV, Harik VM, Odegard GM, Brenner DW, Gates TS. The stresse-
performance in Monte Carlo simulations of graphitization. Phys Rev B strain behavior of polymerenanotube composites from molecular dynamics
2003;68:024107. simulation. Compos Sci Technol 2003;63:1655e61.
[42] Hu K, Gupta MK, Kulkarni DD, Tsukruk VV. Ultra-robust graphene oxide-silk [45] Cho J, Luo JJ, Daniel IM. Mechanical characterization of graphite/epoxy
fibroin nanocomposite membranes. Adv Mater 2013;25:2301e7. nanocomposites by multi-scale analysis. Compos Sci Technol 2007;67:
[43] Guo X, Liao JB, Wang XY. Investigation of the thermo-mechanical properties of 2399e407.
single-walled carbon nanotubes based on the temperature-related higher