MULTIFUNCTIONAL NANOCOMPOSITE REINFORCED BY
GRAPHITE NANOPLATELETS
By
Huang Wu

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILSOPHY
Chemical Engineering
2011

ABSTRACT
MULTIFUNCTIONAL NANOCOMPOSITE REINFORCED BY GRAPHITE
NANOPLATELETS
By
Huang Wu
Nanoparticles have the potential to impart multiple desirable properties to polymers
and composites due to their theoretical high properties and easier processing.
Compared with other nanoparticles, graphite or graphene nanoplatelets are drawing
more and more attention because of their combination of exceptional mechanical,
electrical, thermal properties that are equal to carbon nanotubes but with the same
two-dimensional platelet morphology as nanoclay, which also makes them capable of
providing improved barrier properties in polymers. However, the 2D structure of
graphite nanoplatelets also requires additional attention to be directed at the
processing of these nanoparticles in polymers. Unlike symmetric nanofillers such as
carbon black, the dispersion of 2D platelet fillers in a polymer matrix is more difficult
due to the strong action between their surfaces. Besides getting good dispersion, the
orientation of 2D particles also needs to be taken into account due to their
anisotropic properties. However, if done properly, both dispersion and orientation
can be controlled through nano-structuring to create novel materials with unique
morphologies and properties.
In this research, graphite nanoplatelets were fabricated through exfoliation route to
preserve the exceptional electrical and thermal properties. A novel polishing and

plasma etching process for scanning electron microscopic observation of these
nanocomposite samples has been developed to study the relationships among
processing, morphology and performance. As a result, the dispersion and orientation
of GNP could be used to explain the mechanical and electrical properties of the
composites. It was discovered that orientation of GNP particles is the main factor in
achieving high mechanical properties. Orientation that causes high mechanical
properties and low electrical conductivity can be controlled by annealing the
composite under high temperature.
Based on these principles, methods for nanostructuring of the exfoliated graphite
nanoplatelets were identified and investigated to optimize desired properties. First,
composites with uniformly aligned filler particles can be produced by cast film
extrusion. Improvement of both tensile modulus and strength along with barrier
properties was observed compared with injection and compression molded samples.
Second, pre-coating of polymer powder and solid state ball milling (SSBM) were used
to create percolated networks at low GNP concentrations to enhance electrical
conductivity. A combination of melt-extrusion and SSBM resulted in the composites
with both high mechanical properties and high electrical conductivity. Third, GNP
particles were assembled into a thin paper-like film to create a flexible, durable and
light-weighted material with extremely high electrical conductivity (~2200 S/cm) and
thermal conductivity (~313W/m/K), low permeability to small molecules and the
engineering possibility to be incorporated into fiber reinforced composites.

Copyright by
Huang Wu
2010

To my wife and my parents

v

ACHNOWLEDGEMENTS
Firstly, I’d like to thank The Boeing Company for sponsoring this project. I’d also like
to express my personal gratitude to Mr. Gregg Bogucki from Boeing for his help.
Then, I’d like to express my sincere appreciation to my advisor, Prof. Lawrence T. Drzal,
for taking me as his student and helping me throughout my PhD research project. I
am also thankful to my committee members, Prof. Greg Baker, Krishnamurthy
Jayaraman, Ilsoon Lee and Ramani Narayan for their comments on my research.
Furthermore, I’d like to thank department of Chemical Engineering and Material
Science.
I am grateful to every CMSC staff, Per Askeland, Inhwan Do, Ed Drown, Hiroyuki
Fukushima, Hazel-Ann Hosein, Karen Lillis, Wanjun Liu, Hwan-Man Park, Mike Rich,
Brian Rook and Inger Weitlauf. Just like good neighbors, they are always there when I
need help.
My awesome officemates, Pat, Xian, Anchita, Deb, Tao, Jinglei, Dee, Yan and Sanjib,
inspired me a lot through our chatting in the office, lab and during coffee time. I
enjoyed staying and working in EB1105 thanks to the friendly and joyful environment
created by them.
Last but not least, I want to thank my wife and my parents for their unconditional
love and infinite support in my life, especially their encouragement during my
studies.

vi

Table of Contents
List of Tables ……………………………………………………………………………………………………………xi

List of Figures ............................................................................................................... xii

Chapter 1 Introduction .............................................................................................. 1
Background and Literature Review ............................................................................ 1
Multifunctional Polymer Composite .............................................................. 1
Nanocomposite .............................................................................................. 2
Processing of Nanocomposites ...................................................................... 5
Paper Form Nanocomposite .......................................................................... 7
Dissertation Objective ................................................................................................ 8
REFERENCES ............................................................................................................. 10

Chapter 2 A novel method of revealing the morphology of nanocomposites by
Scanning Electron Microscope ..................................................................................... 16
Abstract .................................................................................................................... 16
Introduction ............................................................................................................. 16
Experimental Details and Discussion ....................................................................... 18
SEM Sample Preparation ............................................................................. 18
Image Processing ......................................................................................... 22
Fluorescent Imaging ..................................................................................... 25
Conclusions .............................................................................................................. 27
REFERENCES ............................................................................................................. 28

Chapter 3 Effect of Molding and Annealing on Dispersion and Orientation on the
Mechanical Properties and Electrical Conductivity of Graphene Nanoplatelet
-Polyetherimide Composites ........................................................................................ 30
Abstract .................................................................................................................... 30
Introduction ............................................................................................................. 31
Experimentation....................................................................................................... 34
Materials ...................................................................................................... 34
Melt-extrusion ............................................................................................. 35
Molding ........................................................................................................ 35
Characterization ........................................................................................... 36
Results and Discussions ........................................................................................... 37
Compression molding .................................................................................. 37
vii

Injection molding ......................................................................................... 38
Model prediction of modulus ...................................................................... 41
Effect of annealing ....................................................................................... 47
Conclusions .............................................................................................................. 51
REFERENCES ............................................................................................................. 52

Chapter 4 Effect of Graphite Nanoplatelets on Coefficient of Thermal Expansion of
Polyetherimide Composite........................................................................................... 55
Abstract .................................................................................................................... 55
Introduction ............................................................................................................. 56
Experimentation....................................................................................................... 58
Materials ...................................................................................................... 58
Melt-extrusion ............................................................................................. 58
Injection Molding ......................................................................................... 59
Coefficient of Thermal Expansion ................................................................ 59
Annealing ..................................................................................................... 59
Results and Discussions ........................................................................................... 60
Linear Thermal Expansion ............................................................................ 60
Annealing Effect ........................................................................................... 61
Modeling for Coefficient of Thermal Expansion .......................................... 67
Conclusions .............................................................................................................. 78
REFERENCES ............................................................................................................. 80

Chapter 5 Highly Aligned Multifunctional Graphite Nanoplatelet-Polyether imide
Composite in Film Form ............................................................................................... 84
Abstract .................................................................................................................... 84
Introduction ............................................................................................................. 85
Experimentation....................................................................................................... 86
Materials ...................................................................................................... 86
Melt-extrusion and film casting ................................................................... 87
Annealing of Extrusion Cast Film ................................................................. 88
Melt-extrusion and Compression Molding .................................................. 90
Electrical Conductivity Measurement .......................................................... 90
Thermal Diffusivity and Thermal Conductivity ............................................ 91
Scanning Electron Microscope (SEM) Observation...................................... 91
Tensile Properties ......................................................................................... 92
O 2 Permeability ............................................................................................ 93
Results and Discussions ........................................................................................... 94
Morphology of Extrusion Cast Films ............................................................ 94
Tensile Properties ......................................................................................... 98
O 2 Permeability .......................................................................................... 105
viii

Effect of Annealing ..................................................................................... 108
Conclusions ............................................................................................................ 110
REFERENCES ........................................................................................................... 112

Chapter 6 Dispersion Optimization of Exfoliated Graphene Nanoplatelet in
Polyetherimide Nanocomposites: Extrusion vs. Pre-Coating vs. Solid State Ball
Milling
115
Abstract .................................................................................................................. 115
Introduction ........................................................................................................... 116
Experimentation..................................................................................................... 119
Materials .................................................................................................... 119
Melt-extrusion ........................................................................................... 119
Pre-coating ................................................................................................. 120
Solid State Ball Milling (SSBM) ................................................................... 120
Compression Molding ................................................................................ 121
Electrical Resistivity Measurement ............................................................ 121
Flexural Strength and Modulus Testing...................................................... 122
Scanning Electron Microscope (SEM) Observation.................................... 122
Results and Discussions ......................................................................................... 122
Electrical Conductivity................................................................................ 122
Flexural Properties ..................................................................................... 126
Combining Melt-Extrusion with SSBM ....................................................... 128
Conclusions ............................................................................................................ 132
REFERENCES ........................................................................................................... 134

Chapter 7 Effect of Particle size on the Thermal and Electrical Conductivity of
Graphite Nanoplatelet/Polyetherimide Composites ................................................. 137
Abstract .................................................................................................................. 137
Introduction ........................................................................................................... 138
Experimental Materials and Methods ................................................................... 141
Materials .................................................................................................... 141
Melt-extrusion ........................................................................................... 142
Pre-coating ................................................................................................. 142
Compression Molding ................................................................................ 142
Electrical Resistivity Measurement ............................................................ 143
Thermal Diffusivity and Thermal Conductivity .......................................... 143
Results and Discussion ........................................................................................... 143
Electrical conductivity ................................................................................ 143
Thermal Conductivity ................................................................................. 145
Conclusions ............................................................................................................ 149
References .............................................................................................................. 150
ix

Chapter 8 GNP Paper: Light weight Composite having Electrical, Thermal
Conductivity and Gas Barrier Properties ................................................................... 154
Abstract .................................................................................................................. 154
Introduction ........................................................................................................... 155
Experimentation..................................................................................................... 157
Materials .................................................................................................... 157
GNP Paper Fabrication ............................................................................... 158
Impregnating GNP Paper ........................................................................... 162
Characterization ......................................................................................... 163
Results and discussions .......................................................................................... 165
Effect of Annealing and Pressing................................................................ 165
Properties of GNP Paper ............................................................................ 166
Impregnation of GNP Paper ....................................................................... 169
Conclusions ............................................................................................................ 176
References .............................................................................................................. 178

Chapter 9 Summary and Conclusions .................................................................... 183
Future Investigations and Opportunities ............................................................... 187

x

List of Tables
Table 3.1, Parameters used in Tandon-Weng equation ....................................... 43
Table 3.2, Fit values of the volume ratio of unidirectional region in injection
molded coupons .......................................................................................... 44
Table 4.1, Properties of GNP and PEId ................................................................. 68
Table 5.1, Dimension and stretching ratio of extrusion cast films ...................... 88
Table 5.2, Thickness change after annealing for GNP/PEId extrusion cast films . 89
Table 7.1, Thermal conductivity, coefficient of thermal expansion and density of
fillers[18] .................................................................................................... 138
Table 8.1, Density and porosity of GNP papers, data are actual value of the
particular sample, not an average ............................................................. 159
Table 8.2, Actual composition in GNP-PEId ....................................................... 174
Table 9.1, Summary of properties of GNP/PEId composites and GNP paper
composites ................................................................................................. 186

xi

List of Figures
Figure 1.1, Proposed model for the tortuous zigzag diffusion path in a composite
filled by aligned 2D particles (adapted from literature [4]) ........................... 3
Figure 1.2, Molecular structure of PEId ................................................................. 9
Figure 2.1, Sectioning injection molded coupon. Dash lines indicate the
directions of sectioning (For interpretation of the references to color in this
and all other figures, the reader is referred to the electronic version of this
dissertation). ................................................................................................ 18
Figure 2.2, Neat polyetherimide surface under plasma etching ......................... 20
Figure 2.3, Comparison between fractured, polished, and plasma etched
surfaces. Images were taken from 5wt% GNP/PEId composites. ................ 21
Figure 2.4, Comparison between fractured surface and plasma etched surface
for pre-coated composite (a) fractured surface of GNP/PP, adapted from
literature [6] and (b) plasma etched surface of GNP/PEId........................... 22
Figure 2.5, Schematics of image processing, white objects stand for the
projection of GNP particles on the observed plane, θ is the angle between
the length direction of the projection and the reference direction, d is the
larger dimension of the projection .............................................................. 23
Figure 2.6, Degree of Alignment on the entire thickness-with plane
(cross-section of 10wt% GNP-5/PEId injection molded coupon) ................ 25
Figure 2.7, Comparison between SEM and fluorescent images. Images were
taken from carbon fiber (CF)/GNP/epoxy composite with one layer of GNP
in between of two layers of carbon fiber stacks (details available in Chapter
8) .................................................................................................................. 26
Figure 3.1, Comparison between GNP-5/PEId composites made from injection
molding and compression molding.............................................................. 37
Figure 3.2, Orientation of GNPGNP particles observed on: (a) cross-section of a
compression molded coupon; (b) the cross-section of an injection molded
coupon perpendicular to the flow direction & (c) along the flow direction
(the scale bar represents 20 micron) ........................................................... 40
xii

Figure 3.3, Schematics of alignment situation of GNP particles in PEId matrix: (a)
random orientation after compression molding, (b) aligned orientation
after injection molding................................................................................. 41
Figure 3.4, Comparison between experimental data (tensile modulus of
injection molded samples and flexural modulus of compression molded
samples) and Tandon-Weng modeling ........................................................ 42
Figure 3.5, (a) TEM image of GNP particle in PEId matrix after melt-extrusion,
showing stacking of layers & (b) SEM image of 1wt% GNP-5/PEId composite
made from melt-extrusion, showing reduced particle size (original average
size: 3.9 micron in diameter) ....................................................................... 45

Figure 3.6, 𝑎11 along the thickness direction on the cross-section plane of flow
direction (diamond series are 𝑎11 values and dash lines indicate the
distinguish of two regions) and the representative morphology ................ 46

Figure 3.7, Effect of annealing on electrical resistivity ........................................ 49
Figure 3.8, 5wt% GNP-5/PEId composite (a) before annealing and (b) after
annealing. Images were taken along the flow direction (as during injection
molding), showing the morphology of the near-surface part of the tensile
bar coupon ................................................................................................... 50
Figure 3.9, Time dependence of the annealing effect ......................................... 50
Figure 4.1, Schematics of TMA sample preparation ............................................ 59
Figure 4.2, LTE of injection molded GNP/PEId composites.................................. 60
Figure 4.3, LTE of annealed GNP/PEId composites .............................................. 62
Figure 4.4, Schematics of confinement to GNP particle rotation ........................ 62
Figure 4.5, Comparison of LTE: before annealing and after annealing................ 64
Figure 4.6, CTE of GNP/PEId composites before and after annealing ................. 65
Figure 4.7, SEM images of GNP-5/PEId composites: 5wt% injection molded
sample, (a) before annealing, (b) after annealing, and 10wt% injection
molded sample, (c) before annealing, (d) after annealing. Images are taken
at the near-surface part of the tensile bar coupon (see Chapter 3 for
details).......................................................................................................... 66
xiii

Figure 4.8, Model prediction for CTE of GNP/PEId composites........................... 68
Figure 4.9, SEM images showing orientation of GNP particles observed on the
cross-section of an injection molded coupon perpendicular to the flow
direction ....................................................................................................... 71
Figure 4.10, Configuration for measuring LTE in Z direction ............................... 71
Figure 4.11, Model prediction of LTE in (a) 0 degree direction, (b) 90 degree
direction, (c) Z direction ............................................................................... 72
Figure 4.12, alignment of GNP particles on thickness-width plane of injection
molded coupon ............................................................................................ 74

Figure 4.13, Morphology for different 𝑎11 values, where white lines stand for
GNP particles and black part stands for the matrix. The images are taken
from 10wt% GNP-5/PEId composite injection molded coupon on the
cross-section of flow direction (width-thickness plane) .............................. 75
Figure 4.14, Alignment in 90 degree direction .................................................... 76
Figure 5.1, Film extrusion, showing stretching ratio ........................................... 87
Figure 5.2, Experiment setup of annealing extrusion cast film ........................... 89
Figure 5.3, Schematic of electrical resistance measurement .............................. 91
Figure 5.4, Cross-sections that were observed by SEM. Dash-line indicates the
direction in which the film was cut. ............................................................. 92
Figure 5.5, Sample preparation for tensile modulus and strength testing .......... 93
Figure 5.6, Measured permeation vs. testing time for compression molded neat
PEId film, showing the plateau after about 12 hours .................................. 94
Figure 5.7, SEM images of cross-sections of extrusion cast film, showing GNP
particle alignment. The bubbles in the images are on the epoxy putty
region, not a feature of the film sample. ..................................................... 96
Figure 5.8, SEM images of cross-sections of annealed extrusion cast film,
showing the change in GNP particle alignment ........................................... 97
Figure 5.9, Degree of alignment of GNP particles in extrusion cast films ........... 98
xiv

Figure 5.10, Typical stress-strain curve for (a) neat PEId film (b) 10wt%
GNP-5/PEId composite film with SR=11 (c) 10wt% GNP-5/EPId composite
film with SR=34 .......................................................................................... 100
Figure 5.11, (a) Tensile modulus (b) tensile strength of GNP-5/PEId composites
films. According to Figure 10, the strength value shown here is the strength
as yield ....................................................................................................... 102
Figure 5.12, Elongation of film during tensile testing (a) neat PEId film (b)
GNP-5/PEId composite film ....................................................................... 103
Figure 5.13, Normalized tensile modulus and tensile strength, showing
comparison between injection molded coupon and films made from
various stretching ratios (a) modulus (b) Strength .................................... 104
Figure 5.14, Relative O 2 Permeability of compression molded GNP-5/PEId film.
Series “model” was calculated by assuming perfect alignment 𝑎 = 1.
Series “model with alignment” was calculated using the 𝑎 value shown in
Figure 5.15. ................................................................................................ 106
Figure 5.15, SEM images of compression molded GNP-5/PEId film, showing
2

alignment of GNP particles. The <cos ɵ> values shown in the images are
averages from images taken at different position of the sample .............. 107
Figure 5.16, Relative O 2 permeability of GNP-5/PEId extrusion cast film ........ 108
Figure 5.17, Effect of annealing on electrical resistivity of GNP-5/PEId extrusion
cast films (a) through-plane direction (b) in-plane direction..................... 109
Figure 5.18, Effect of annealing on thermal conductivity of GNP-5/PEId extrusion
cast film ...................................................................................................... 110
Figure 6.1, SEM image of GNP-PEId composite after SSBM, showing geometry of
the particles ............................................................................................... 121
Figure 6.2, Comparison of electrical resistivity of GNP/PEId composites from
different compounding methods: (a) in-plane direction (b) through-plane
direction. .................................................................................................... 124
Figure 6.3, SEM images of GNP/PEId composites (a) polished surface of the cross
section of 3wt% GNP-1/PEId composite from pre-coating compounding,
xv

showing orientation and long continuous graphite phase (white lines) (b)
3wt% GNP-15/PEId composite from melt-extrusion compounding, showing
size reduction and distribution of GNP particles (c) polymer powder surface
after pre-coating with 3wt% GNP-1 (d) polymer powder surface after SSBM
with 3wt% GNP-1. ...................................................................................... 125
Figure 6.4, Flexural modulus of GNP/PEId composites...................................... 128
Figure 6.5, Flexural Strength of GNP/PEId composites ...................................... 128
Figure 6.6, Properties of GNP/PEId composites from two-step compounding:
melt-extrusion with GNP-15 then SSBM with GNP-1: (a) in-plane electrical
resistivity (b) through-plane electrical resistivity (c) flexural modulus (d)
flexural strength ......................................................................................... 130
Figure 6.7, SEM image of 3wt% GNP/PEId extrusion compounded + 1wt% GNP-1
in SSBM step, showing two kinds of morphology: extrusion compounded
GNP particles embedded in polymer matrix and SSBM coated GNP particles
lying on the surface: (a) liquid nitrogen fractured cross-section, left side
showing coating morphology while right side showing extrusion
morphology (b) polished cross-section, discrete particles showing extrusion
morphology while continues line showing coating morphology ............... 132
Figure 6.8, Schematics of the effect of SSBM on the composite powder made
from melt-extrusion ................................................................................... 132
Figure 7.1, Electrical conductivity of GNP/PEI composites ................................ 144
Figure 7.2, SEM image of pre-coated 3wt% GNP-1/PEI composite showing
continuous phase created by pre-coating. The white lines indicate GNP
phase while the dark part indicates the PEI matrix. The observed surface is
prepared by polishing and plasma etching as explained elsewhere. ........ 144
Figure 7.3, Thermal conductivity of GNP/PEI composites ................................. 145
Figure 7.4, Schematics of particle size and connectivity: (a) although the
particles are well connected, the contacting resistance is huge; (b) large
particles which are barely touching each other can still provide higher
thermal conductivity than (a) .................................................................... 147
Figure 7.5, Thermal conductivity of pre-coated GNP-15/PEI composite .......... 148
Figure 8.1, TGA result of GNP and PEIn ............................................................. 158
xvi

Figure 8.2, SEM images showing cross-section of the GNP paper .................... 160
Figure 8.3, GNP paper and its flexibility (a) GNP paper (b) paper plane folded
from GNP paper (c) morphology of the surface of the GNP paper (d)
morphology of the folding edge of the paper plane ................................. 161
Figure 8.4, Structure and nomination of CF/GNP paper composites ................ 163
Figure 8.5, Stress-strain curve of GNP paper ..................................................... 164
Figure 8.6, Effect of Annealing and Compression on the density and electrical
conductivity of GNP-W paper. “as-made” stands for filtrated paper without
annealing or pressing. Pressure is the nominal value calculated from
press/sample area ...................................................................................... 165
Figure 8.7, Comparison between carbon-based papers (a) tensile modulus (b)
electrical conductivity (c) thermal conductivity ........................................ 167
Figure 8.8, SEM images showing cross-section morphology of (a) GRAFOIL® and
(b) CNT Bucky paper. Samples were prepared by polishing-O 2 plasma
etching process .......................................................................................... 169
Figure 8.9, Confocal microscopic images of GNP papers impregnated by (a) 0.1
g/mL PEId/dichloromethane solution (b) 0.2 g/mL PEId/dichloromethane
solution (c) pressed GNP- PEId paper (d) GNP-E paper. Green indicates
fluorescence signal, black indicates GNP ................................................... 170
Figure 8.10, SEM images showing cross-section of (a) pressed GNP-PEId (b)
CGC-PEId .................................................................................................... 171
Figure 8.11, Properties of GNP paper composites from soaking ...................... 172
Figure 8.12, Properties of GNP paper/CF composites ....................................... 173
Figure 8.13, Schematics of surface resistance measurement for CF/GNP paper
composite................................................................................................... 176

xvii

Chapter 1 Introduction
Background and Literature Review
Multifunctional Polymer Composite
Polymer and polymer composites have been widely used to replace metal and metal
alloys because of their light weight, good chemical resistance and lower
manufacturing costs in automobile, aerospace, recreational goods, infrastructure,
electronic devices, etc.. Although the modulus and strength may be the principal
structural requirement for polymer composites, many applications would also
benefit greatly if other properties could be added thereby increasing the
functionality of the same material. For example, one inherent difference between
polymer and metal is the electrical conductivity: most polymers are insulators while
metals are conductors. Incorporating adequate electrical conductivity into polymers
and polymer composites has been a major goal recently due to the increasing
demand of lightning strike protection in polymer composites constructed aircraft.
Similarly, any desirable property is thermal conductivity which is very low in polymer
composites. Materials used in electronic packaging require high thermal conductivity
to dissipate heat. Another example of the need for multifunctional composites is the
material used in containers such as fuel tanks, which requires mechanical strength
for structural purpose, electrical conductivity for static charge hazard prevention, and
barrier properties to the fuel molecules.
1

Nanocomposite
One promising way to provide multifunctionality to a polymer composite is to add
nanofillers. Polymer nanocomposites usually refer to a polymer matrix filled by
high-surface-area reinforcing fillers [1] including carbon nanotubes (CNTs), carbon
nanofibers, nanoclay, carbon black (CB), graphene platelets and so on. Since their
discovery, nanoparticles were expected to function as substitute reinforcements for
conventional reinforcement fibers because of the theoretically high particle
properties and the significant improvement in nanocomposite properties attained at
low loadings.

Furthermore significant advantages were expected from gains in

processing by eliminating the need for layer-by-layer assembly or autoclave
processing. The functionality of the nanofillers is highly dependent on their
morphology. For example, as conductive fillers, the zero-dimensional CB can improve
composite electrical conductivity as a result of electrical percolation at low (~5%)
loading for non-polar, low crystallinity thermoplastics [2], while high aspect ratio CNT
filled composites with a similar matrix have a percolation threshold of 0.1vol% [3].
The high aspect ratio and linear shape of CNTs make it easier to form a percolated
network with more contacts being formed between filler particles. Two-dimensional
(nano-platelet) fillers such as alumino-silicate nanoclay can also produce composites
with low gas permeability which cannot be achieved by CNTs or CB because of the
highly tortuous path formed by their impenetrable nature. If properly aligned, the 2D
particles will force the small molecules to travel through a tortuous zigzag path
2

instead of diffusing freely, resulting in much lower permeability [4-6] (Figure 1.1).
Research also suggests that this kind of filler can also be used to improve the fire
retardancy of its composite [7].

Figure 1.1, Proposed model for the tortuous zigzag diffusion path in a composite
filled by aligned 2D particles (adapted from literature [4])
Despite the promising properties of CNTs, it is not advantageous to use them in
commercial composites because of their expensive nature and limited availability.
CB/rubber composites are the standard in the tire industry, and colloidal silica as a
rubber additive is making inroads into the tire industry [8]. However, these two fillers
do not possess the excellent electrical properties found in CNTs.

Exfoliated Graphite Nanoplatelet and its Composites
Graphene based nanoplatelet particles, which combine good electrical and thermal
properties similar to CNTs and the two-dimensional morphology of alumino-silicate
clays, are considered to be an excellent potential additive in terms of achieving
multifunctionality for the composites. Graphite or graphene nanoplatelets have
received growing interest because their high mechanical properties (~1TB for either
3

single layer or multi layer graphene theoretically [9, 10] and experimentally [11-13]),
-6

excellent electrical (~1.0×10

ohm/cm resistivity for pure graphene [14]), and

thermal (~5000 W/m/K [15-18]) conductivity. The platelet shape of graphite also
benefits the composites by enhancing barrier properties and coupled with the high
thermal oxidative resistance of graphene, fire retardancy can be achieved as well as
demonstrated in Figure 1.1. Such properties are considered to be essential for
multifunctional composites.
The process of generating thin graphite platelets consisting of several layers of
graphene sheets is relatively simple and cost-effective. Usually, three steps are
involved in fabricating these nanofillers from natural graphite: (i) intercalation, where
both proton-donor agents and electron acceptors can be intercalated in between
graphene layers; (ii) exfoliation, where a large amount of heat is applied to expand
the intercalated graphite through vaporization of the intercalant; (iii) pulverization,
where the exfoliated graphite is reduced in size to smaller dimensions. As
demonstrated in Drzal’s group [19], exfoliated graphite nanoplatelets (GNP)
consisting of small stacks of graphene can be fabricated by acid intercalation,
microwave heating and ultrasonication pulverization and be scaled inexpensively into
quantity production by this route. The size and thickness of GNP is controllable
during fabrication. Typical sizes of the platelets range from submicron to 25 microns
in diameter usually notated as GNP-1, GNP-5 and GNP-15 where the number refers
to the average statistical diameter with thicknesses in the 5-10 nm range. Compared
with the other popular fabrication route requiring formation of graphene oxide
4

followed by reduction of graphite oxide [20] back to graphene, the
2

intercalation-expanding route avoids the disruption of sp bonding of the graphite
which is the key to its superior thermal and electrical properties as well as its
economical price.

Processing of Nanocomposites
Because of the two dimensional nature of GNP and its high aspect ratio, final
properties of GNP based nanocomposites can vary due to the processing methods
selected. For example, the molding processes are important for their ability to make
complex shaped integrated composites. Kim et al [21] have shown that differences in
electrical conductivity of polycarbonate/graphene composites can be attained
depending on whether they were made by injection molding or compression molding.
Compression molded samples showed much better electrical conductivity than
injection molded ones. Mohanty et al [22] found that post-molding treatment, such
as annealing after injection molding can provide better flexural modulus and strength
and alter the morphology of nanocomposites and thus their properties.
Typical compounding methods for thermoplastic matrix includes: (1) melt
compounding, such as combining high density polyethylene (HDPE) and exfoliated
graphite mixed in a Brabender mixer [23, 24] or twin-screw extruder [25]; (2) solution
mixing, where a solvent is used to dissolve polymer and then disperse graphite [26,
27]; (3) in-situ polymerization, where the polymer is polymerized with presence of
graphite [28, 29].
5

While being the most widely used method in industrial applications, melt-extrusion
falls short in the electrical conductivity and percolation threshold achieved in the
resulting composite [30, 31]. Solution mixing is capable of separating and dispersing
nano particles which results in lower percolation threshold [32], but the presence of
solvent may cause problems while processing the composite and increase the cost of
manufacture. In-situ polymerization only applies to limited category of polymers, and
the energy cost for exfoliating graphite during the process is a large concern. Efforts
have been made to find a practical and efficient way to disperse the nano graphite
particles in order to achieve lower percolation. Some researchers used
melt-extrusion of unsaturated polyester resin (UPR) to modify expanded graphite [33]
which resulted in a percolation threshold at 5.7wt%, instead of 22.2wt% for
composites with conventional graphite. Recently, Drzal’s group [30] developed a GNP
pre-coating method to achieve low percolation threshold where a non-solvent of the
polymer is used to disperse GNP particles. While adding polymer powder into the
suspension, the GNP particles are coated onto the polymer surface. The GNP forms a
complete continuous coating on the surface of the polymer powder. Subsequent
compression molding results in nanocomposites with the continuous GNP phase
intact after the molding process.
Dynamic percolation behavior of CB composites has been studied by Wu et al [34]
who found that due to Brownian motion, CB particles are able to reposition
themselves locally under high temperature during which the viscosity of the polymer
matrix is low. Particles of platelet shape such as alumino-silicates and graphene
6

might not be affected to the same extent by Brownian motion because of their larger
dimensions. Researchers have shown faster particle re-organization in layered
alumino-silicate/polymer composites than expected compared to Brownian motion
theory [35-37]. The real driving force might be the elastic restoring force of the
polymeric matrix to its stress free state generated during processing though no
widely-accepted conclusion has been drawn yet. Kim [21] proposed a disorientation
mechanism for graphene composites based on X-ray scattering and rheology data,
pointing out that at low loadings (<3wt%) of graphite filler, the rotation experienced
by the particles caused by annealing still cannot increase particle contact with each
other, while at high loadings (>12wt%), the movement of filler particles is restricted
due to excluded volume interactions between particles.

Paper Form Nanocomposite
Reassembly of nano material into macro-scale structures has become one interesting
approach to utilize the nanofillers. The reassembled structure can also be
incorporated into a composite. Carbon nanotubes (CNT) Bucky paper [38-48] and
graphite oxide (GO) paper [20, 49, 50] have been widely studied. The paper form of
these materials can be self-standing and show good flexibility [20, 46]. Electrical
conductivity is reported to be 200 S/cm for Bucky paper with aligned CNT [46] and
350 S/cm for reduced GO paper [49]. The paper fabricated from ref. [46] and [49] are
claimed to be binder-free but other types of paper made with a polymeric

binder

would have lower values. Compared with natural graphite which has a conductivity
7

of ~2000 S/cm, these values are still one order of magnitude lower. The reason is, for
CNT, the contact resistance between each tube is high due to the curvature of the
contacting surfaces resulting in point-to-point contacts; for GO paper, the chemically
reduced graphite oxide is not as conductive as pristine graphite. Li et al [51] and
Biswas et al [52] both fabricated Langmuir-Blodgett thin films from exfoliated
graphite nanoplatelets. The electrical and thermal conductivity were reported to be
over 1000 S/cm [52] and 380 W/m/K [53], respectively.

Dissertation Objective
The major objective of this research is to investigate the feasibility of using GNP
particles to impart the polymer composite with multifunctionality, and find the
optimum way to achieve these desired multifunctional properties. The major
polymer matrix used during this research is the polyetherimide (PEId, Ultem®)
(Figure 1.2). PEId is a high temperature amorphous engineering thermalplastic. The
glass transition temperature for PEId is 216℃. PEId has excellent chemical resistance
and thermal stability. Moreover, it is inherently flame resistant and only generates a
very small amount of low-density gas when burned, making it an ideal material for
applications in advanced structures such as submarine, airplane and high-floor
offices [54].

8

O

O

*

N

N
O

CH3

*

n

O
O

O
CH3

Figure 1.2, Molecular structure of PEId
In order to establish a viable method for producing multifunctional GNP composites
for different applications, the relationships among common processing methods, the
morphology and the performance of the GNP composite are first studied. A novel
approach of SEM sample preparation has been developed for revealing the internal
polymer nanocomposite morphology. The effect of injection and compression
molding on the mechanical properties, electrical conductivity and coefficient of
thermal expansion is investigated and explained based on the morphology. The
remaining part of this dissertation is directed at how to design and nanostructure the
GNP in the composites for desired properties based on what has been learned.
Eventually, a nanostructured film or paper form of GNP composite is shown to the
best way to achieve the full potential of the GNP nanoparticle.

9

REFERENCES

10

REFERENCES
1.

Vaia, R.A. and E.P. Giannelis, Polymer nanocomposites: Status and
opportunities. Mrs Bulletin, 2001. 26(5): p. 394-401.

2.

Huang, J.C., Carbon black filled conducting polymers and polymer blends.
Advances in Polymer Technology, 2002. 21(4): p. 299-313.

3.

Bauhofer, W. and J.Z. Kovacs, A review and analysis of electrical percolation in
carbon nanotube polymer composites. Composites Science and Technology,
2009. 69(10): p. 1486-1498.

4.

LeBaron, P.C., Z. Wang, and T.J. Pinnavaia, Polymer-layered silicate
nanocomposites: an overview. Applied Clay Science, 1999. 15(1-2): p. 11-29.

5.

Kim, J.K., et al., Moisture barrier characteristics of organoclay-epoxy
nanocomposites. Composites Science and Technology, 2005. 65(5): p.
805-813.

6.

Yano, K., et al., SYNTHESIS AND PROPERTIES OF POLYIMIDE CLAY HYBRID.
Journal of Polymer Science Part a-Polymer Chemistry, 1993. 31(10): p.
2493-2498.

7.

Porter, D., E. Metcalfe, and M.J.K. Thomas, Nanocomposite fire retardants - A
review. Fire and Materials, 2000. 24(1): p. 45-52.

8.

Schaefer, D.W. and R.S. Justice, How nano are nanocomposites?
Macromolecules, 2007. 40(24): p. 8501-8517.

9.

Chang, T.C. and H.J. Gao, Size-dependent elastic properties of a single-walled
carbon nanotube via a molecular mechanics model. Journal of the Mechanics
and Physics of Solids, 2003. 51(6): p. 1059-1074.

10.

Meo, M. and M. Rossi, Prediction of Young's modulus of single wall carbon
nanotubes by molecular-mechanics based finite element modelling.
Composites Science and Technology, 2006. 66(11-12): p. 1597-1605.

11.

Lee, C., et al., Measurement of the elastic properties and intrinsic strength of
monolayer graphene. Science, 2008. 321(5887): p. 385-388.

12.

Zhao, H., K. Min, and N.R. Aluru, Size and Chirality Dependent Elastic
Properties of Graphene Nanoribbons under Uniaxial Tension. Nano Letters,
2009. 9(8): p. 3012-3015.
11

13.

Wong, C.L., et al., Characterization of nanomechanical graphene drum
structures. Journal of Micromechanics and Microengineering, 2010. 20(11).

14.

Tune, L., Physicists Show Electrons Can Travel More Than 100 Times Faster in
Graphene (https://newsdesk.umd.edu/scitech/release.cfm?ArticleID=1621).
2008.

15.

Watanabe, E., et al., Ballistic thermal conductance of electrons in graphene
ribbons. Physical Review B, 2009. 80(8).

16.

Jiang, J.W., J.S. Wang, and B.W. Li, Thermal conductance of graphene and
dimerite. Physical Review B, 2009. 79(20).

17.

Saito, K., J. Nakamura, and A. Natori, Ballistic thermal conductance of a
graphene sheet. Physical Review B, 2007. 76(11).

18.

Balandin, A.A., et al., Superior thermal conductivity of single-layer graphene.
Nano Letters, 2008. 8(3): p. 902-907.

19.

Fukushima, H., Graphite Nanoreinforcements in Polymer Nanocomposites, in
Department of Chemical Engineering and Materials Science. 2003, Michigan
State University: East Lansing.

20.

Stankovich, S., et al., Synthesis of graphene-based nanosheets via chemical
reduction of exfoliated graphite oxide. Carbon, 2007. 45(7): p. 1558-1565.

21.

Kim, H. and C.W. Macosko, Processing-property relationships of
polycarbonate/graphene composites. Polymer, 2009. 50(15): p. 3797-3809.

22.

Mohanty, A.K., et al., Effect of process engineering on the performance of
natural fiber reinforced cellulose acetate biocomposites. Composites Part
a-Applied Science and Manufacturing, 2004. 35(3): p. 363-370.

23.

Wang, Y.S., M.A. Ogurkis, and J.T. Lindt, ELECTRICAL-PROPERTIES OF
EXFOLIATED-GRAPHITE FILLED POLYETHYLENE COMPOSITES. Polymer
Composites, 1986. 7(5): p. 349-354.

24.

Krupa, I. and I. Chodak, Physical properties of thermoplastic/graphite
composites. European Polymer Journal, 2001. 37(11): p. 2159-2168.

25.

Zheng, W., X.H. Lu, and S.C. Wong, Electrical and mechanical properties of
expanded graphite-reinforced high-density polyethylene. Journal of Applied
Polymer Science, 2004. 91(5): p. 2781-2788.
12

26.

Zheng, W.G., S.C. Wong, and H.J. Sue, Transport behavior of PMMA/expanded
graphite nanocomposites. Polymer, 2002. 43(25): p. 6767-6773.

27.

Zheng, W. and S.C. Wong, Electrical conductivity and dielectric properties of
PMMA/expanded graphite composites. Composites Science and Technology,
2003. 63(2): p. 225-235.

28.

Pan, Y.X., et al., A new process of fabricating electrically conducting nylon
6/graphite nanocomposites via intercalation polymerization. Journal of
Polymer Science Part B-Polymer Physics, 2000. 38(12): p. 1626-1633.

29.

Zou, J.F., et al., Conductive mechanism of polymer/graphite conducting
composites with low percolation threshold. Journal of Polymer Science Part
B-Polymer Physics, 2002. 40(10): p. 954-963.

30.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, A new compounding method for
exfoliated graphite-polypropylene nanocomposites with enhanced flexural
properties and lower percolation threshold. Composites Science and
Technology, 2007. 67(10): p. 2045-2051.

31.

Kim, S., I. Do, and L.T. Drzal, Multifunctional xGnP/LLDPE Nanocomposites
Prepared by Solution Compounding Using Various Screw Rotating Systems.
Macromolecular Materials and Engineering, 2009. 294(3): p. 196-205.

32.

Ansari, S. and E.P. Giannelis, Functionalized Graphene Sheet-Poly(vinylidene
fluoride) Conductive Nanocomposites. Journal of Polymer Science Part
B-Polymer Physics, 2009. 47(9): p. 888-897.

33.

She, Y.H., G.H. Chen, and D.J. Wu, Fabrication of polyethylene/graphite
nanocomposite from modified expanded graphite. Polymer International,
2007. 56(5): p. 679-685.

34.

Wu, G.Z., et al., A delay of percolation time in carbon-black-filled conductive
polymer composites. Journal of Applied Physics, 2000. 88(3): p. 1480-1487.

35.

Schmidt, G., et al., Shear orientation of viscoelastic polymer-clay solutions
probed by flow birefringence and SANS. Macromolecules, 2000. 33(20): p.
7219-7222.

36.

Solomon, M.J., et al., Rheology of polypropylene/clay hybrid materials.
Macromolecules, 2001. 34(6): p. 1864-1872.

37.

Ren, J.X., et al., Disorientation kinetics of aligned polymer layered silicate
13

nanocomposites. Macromolecules, 2003. 36(11): p. 4188-4194.
38.

Liu, J., et al., Fullerene pipes. Science, 1998. 280(5367): p. 1253-1256.

39.

Cooper, S.M., et al., Gas permeability of a buckypaper membrane. Nano
Letters, 2003. 3(2): p. 189-192.

40.

Wang, Z., et al., Processing and property investigation of single-walled carbon
nanotube (SWNT) buckypaper/epoxy resin matrix nanocomposites.
Composites Part a-Applied Science and Manufacturing, 2004. 35(10): p.
1225-1232.

41.

Endo, M., et al., 'Buckypaper' from coaxial nanotubes. Nature, 2005.
433(7025): p. 476-476.

42.

Gonnet, P., et al., Thermal conductivity of magnetically aligned carbon
nanotube buckypapers and nanocomposites. Current Applied Physics, 2006.
6(1): p. 119-122.

43.

Gou, J.H., Single-walled nanotube bucky paper and nanocomposite. Polymer
International, 2006. 55(11): p. 1283-1288.

44.

Kim, Y.A., et al., Fabrication of high-purity, double-walled carbon nanotube
buckypaper. Chemical Vapor Deposition, 2006. 12(6): p. 327-+.

45.

Zheng, F., et al., Single-walled carbon nanotube paper as a sorbent for organic
vapor preconcentration. Analytical Chemistry, 2006. 78(7): p. 2442-2446.

46.

Wang, D., et al., Highly oriented carbon nanotube papers made of aligned
carbon nanotubes. Nanotechnology, 2008. 19(7).

47.

Xu, G.H., et al., The feasibility of producing MWCNT paper and strong MWCNT
film from VACNT array. Applied Physics a-Materials Science & Processing,
2008. 92(3): p. 531-539.

48.

Dumee, L.F., et al., Characterization and evaluation of carbon nanotube
Bucky-Paper membranes for direct contact membrane distillation. Journal of
Membrane Science, 2010. 351(1-2): p. 36-43.

49.

Chen, H., et al., Mechanically strong, electrically conductive, and
biocompatible graphene paper. Advanced Materials, 2008. 20(18): p. 3557-+.

50.

Park, S., et al., Graphene oxide papers modified by divalent ions - Enhancing
14

mechanical properties via chemical cross-linking. Acs Nano, 2008. 2(3): p.
572-578.
51.

Li, X.L., et al., Highly conducting graphene sheets and Langmuir-Blodgett films.
Nature Nanotechnology, 2008. 3(9): p. 538-542.

52.

Biswas, S. and L.T. Drzal, A Novel Approach to Create a Highly Ordered
Monolayer Film of Graphene Nanosheets at the Liquid-Liquid Interface. Nano
Letters, 2009. 9(1): p. 167-172.

53.

Xiang, J.L. and L.T. Drzal, Thermal Conductivity of a Monolayer of Exfoliated
Graphite Nanoplatelets Prepared by Liquid-Liquid Interfacial Self-Assembly.
Journal of Nanomaterials, 2010.

54.

http://www.sabic-ip.com. Last visited: March 2011

15

Chapter 2 A novel method of
revealing the morphology of
nanocomposites by Scanning Electron
Microscope
Abstract
Revealing the morphology of the nanocomposites explicitly helps to understand the
relationship between particle structure, method of processing and performance. A
novel SEM sample preparation process involving epoxy fixation of the sample,
mechanical polishing and O2 plasma etching is developed, leading to images with
excellent morphological information as a result of good contrast between GNP
particles and the matrix. Such good contrast enables important information to be
collected for further analysis such as particle size and orientation by the
image-processing software. For samples with a porous structure or cracks,
fluorescent images can be used to locate the epoxy fixation material and reveal the
real structure of the material without the complicating effects of the fixation agent..

Introduction
“Seeing Is Believing.” Since the natural state of the fillers such as their distribution,
dispersion and orientation is the bridge between the processing route and the
16

performance of the polymer nanocomposites, it is critical to be able to actually see
the internal morphology, at a micro scale. Modern microscopic techniques have
given researchers various ways to observe the morphology for micro-scale devices
and materials. Scanning electron microscope (SEM) and transmission electron
microscope (TEM) are the most widely used techniques. SEM is useful for checking
the fracture surface of a composite. However, when the size of the filler particle is in
hundreds of nanometers to a few microns range, it is hard to distinguish the filler
morphology from the features generated by the polymer matrix [1, 2]. TEM is
extremely useful to investigate the dispersion state of nanofillers such as
nano-silicate [3, 4], carbon nanotubes (CNT) [2, 5]. However, only a small portion of
the sample can be observed at one time, making it difficult to understand the overall
condition of the fillers throughout the entire sample.
Recently, graphite nanoplatelet (GNP)/polymer composites have drawn a lot of
attention. To get a clear picture of the dispersion, distribution and orientation of the
GNP particles in GNP/polymer composite, a new time-efficient approach for making
SEM samples has been developed. After adequate polishing and oxygen plasma
etching, the prepared cross-section of the composite sample shows excellent
contrast between the nanofillers and the matrix. The images can be processed in
software to gather information such as particle size, number density and degree of
alignment, which is crucial to understanding the performance of the composite and
guiding the nanostructuring for desired properties.

17

Experimental Details and Discussion
SEM Sample Preparation
For a composite that is reinforced by anisotropic fillers, 2D images of merely
cross-section of the sample cannot tell everything. Especially in injection molded
coupons, particles are forced to align in the flow direction while still confined by the
sample geometry. In this research, an injection molded sample was prepared by
sectioning in three directions: along the flow (0 degree), width (90 degree) and
thickness (z) directions (Figure 2.1).

Figure 2.1, Sectioning injection molded coupon. Dash lines indicate the directions of
sectioning (For interpretation of the references to color in this and all other figures,
the reader is referred to the electronic version of this dissertation).
After sectioning, the to-be-observed surface was attached to an aluminum substrate
by double-sided tape. A plastic ring holder was placed around the sample. An epoxy
fixative agent was added into the holder and cured under room temperature. The
sample-embedded epoxy puck was then polished by SiC polishing paper with finish
18

up to #4000 grit. This was followed by polishing with 1 micron and 0.05 micron
alumina powder/water slurry.
The alumina powder/water polished sample has a very smooth surface with few
polishing artifacts. However, it is not suitable for SEM observation since the contrast
under SEM relies most on topographical differences i.e. height difference. In order to
create height difference between the GNP particles and polymer matrix, O 2 plasma
was applied to the polished surface. The sample was placed in a chamber with the
polished surface facing up. The chamber was evacuated and then purged by O 2 ,
which was ionized under a power source of 400 watts. The total time used for
GNP/polymer composite was 1 hour. The polymer materials on the polished sample
including the polymer matrix and the epoxy fixative are slowly removed as a result of
the plasma etching while the GNP particles are mostly unaffected by the plasma due
2

to its stable sp structure. The effect of plasma etching on polished polyetherimide
(PEId) surface is shown in Figure 2.2.

19

Figure 2.2, Neat polyetherimide surface under plasma etching: (a) no plasma, (b) 30
min plasma, (c) 90 min plasma.
If no plasma is applied, the polished surface shows a smooth morphology. After 30
min plasma treatment, the surface becomes rougher. After 90 min, many pores with
features of a few hundred of nanometers appear on the surface. This indicates that
the O 2 plasma is very effective at etching away polymer.
The comparison between SEM images taken from a fractured surface, a polished
surface and a plasma etched surface is shown in Figure 2.3. Compared with
un-etched surfaces, the plasma treated one shows the best contrast between GNP
particles and the matrix. In Figure 2.3(c), the white lines stand for the cross-section
of the GNP particles. The dark part is occupied by the matrix. On the fractured
surface (Figure 2.3(a)), it is hard to distinguish GNP particles from the matrix due to
the roughness of sample. Although polished surface can show some contrast, it is not
as clear as the etched one, and a lot of smaller particles are invisible due to the
20

entrapment of polymer matrix (Figure 2.3(b)).

Figure 2.3, Comparison between fractured, polished, and plasma etched surfaces.
Images were taken from 5wt% GNP/PEId composites: (a) fractured surface, (b)
polished surface, (c) plasma etched surface
Kalaitzidou et al [6] developed a novel approach to fabricate GNP nanocomposites by
coating the GNP particles onto a polymer powder surface (pre-coating process). After
compression molding, the coated structure forms a continuous GNP phase and helps
the electrical percolation. In their work a fractured surface was used to demonstrate
the percolated structure. However, it is hard to interpret due to surface roughness.
Figure 2.4 shows the comparison between the fractured surface adapted from
literature [6] and the plasma etched surface for the sample made from the same
process. While Figure 2.4(a) suffers from a mixed morphology of GNP particles and
the roughness of the matrix, Figure 2.4(b) explicitly shows the continuous phase of
the GNP (white lines).
21

Figure 2.4, Comparison between fractured surface and plasma etched surface for
pre-coated composite (a) fractured surface of GNP/PP, adapted from literature [6]
and (b) plasma etched surface of GNP/PEId (scale bars: (a)150 micron, (b)20 micron)

Image Processing
Image-Pro Plus software (Media Cybernetics, Inc.) was used to gather information
from the SEM images. This software is capable of distinguishing one phase from
another based on the contrast of the image. In our image, the GNP phase always
stands out from the matrix, resulting in bright (high intensity) spots surrounded by
dark phase. Two properties of the particles are of extreme interest: the degree of
alignment and the size. Figure 2.5 summarizes the type of information about the GNP
particles obtained by the Image-Pro software.

22

Figure 2.5, Schematics of image processing, white objects stand for the projection of
GNP particles on the observed plane, θ is the angle between the length direction of
the projection and the reference direction, d is the larger dimension of the projection.
(For interpretation of the references to color in this and all other figures, the reader
is referred to the electronic version of this dissertation)
Note that the thickness of GNP particles cannot be obtained from this kind of SEM
images, because the projected thickness of the white object is actually the projection
of part of the GNP basal plane, which is much larger than the real thickness of the
platelet.
Getting good contrast makes it possible to process the images in the software. Figure
2.5 shows the schematics of gathering information from the SEM image. In this thesis,
two properties are majorly of interest: the average size of the particle in the
composite, and the degree of alignment. The size of an individual particle, d, can be
calculated by the Image-Pro software by counting the pixels between the two ends of
the largest dimension of the particle. The length that a pixel represents can be
obtained by comparing it with the scale bar. Assuming every particle has the same
thickness, the volume-weighted average size of the particles can be expressed as:

23

∑ 𝑑3
𝑖�
< 𝑑 >=
∑ 𝑑2
𝑖

(1)

The direction of the projection of GNP particle on the observed plane can be
represented by the angle (θ) between the projected radius and the reference

𝑎 =< 𝑐𝑜𝑠 2 𝜃 >

direction. The degree of alignment is defined as:

(2)

Since the sample size for SEM can be relatively large, it is possible to survey the
whole cross-section of a composite sample. An automated navigation stage can be
used to move the SEM sample a few microns in a step and SEM images can be taken
at each location. The information of interest, such as degree of orientation, can be

𝑎 =< 𝑐𝑜𝑠 2 𝜃 > )

plotted as a function of position for the entire sample (Figure 2.6).
Figure 2.6 shows the alignment situation (

on the entire

cross-section of an injection molded coupon. In practice, over 900 images were taken
from one sample but not all the images were processed to get the 𝑎 value. Due to

its symmetric nature, only one quarter of the surface needs to be surveyed, e.g. from
(0, 0) to (1500, 6300) in Figure 2.6, while the other portion can be mirrored.

24

Figure 2.6, Degree of Alignment on the entire thickness-with plane (cross-section of
10wt% GNP-5/PEId injection molded coupon)

Fluorescent Imaging
Epoxy fixative is useful as a support during polishing. However, when a porous or
crack-containing sample is to be observed, the epoxy can fill into the pores and
compromise the results such as porosity one can get from image analysis. To solve
this problem, a complementary fluorescent imaging approach is developed. During
SEM sample preparation, the epoxy fixative was mixed with a fluorescent dye. After
25

polishing and plasma etching, the sample was first observed with a confocal
microscope with a laser source. The well dispersed fluorescent dye molecules are
excited by the laser beam and emit a fluorescence signal, which can be detected by
the confocal microscope. By comparing with the SEM images, one can tell the
original polymer matrix phase from the epoxy fixative phase.

Figure 2.7, Comparison between SEM and fluorescent images. Images were taken
from carbon fiber (CF)/GNP/epoxy composite with one layer of GNP in between of
two layers of carbon fiber stacks (details available in Chapter 8)
In Figure 2.7(a), a polymer-rich region presents between the GNP layer and CF layer.
However, in the corresponding fluorescent image (Figure 2.7(b)), no fluorescence
signal is present at the same position. This indicates that the polymer is the matrix of
the composite, not the epoxy fixative, so the sample is fully consolidated with no
delamination. Meanwhile, Figure 2.7(c) shows a polymer-rich region inside the GNP
26

layer. The corresponding fluorescent image (Figure 2.7(d)) confirms that the material
there is from the epoxy fixative because of the strong fluorescent signal.

Conclusions
The sectioning-polishing-O 2 plasma etching process is found to be very effective in
revealing the morphology of the GNP/polymer nanocomposites. Artifact free images
can be obtained by SEM imaging of the sample prepared in this way. As a result of
the good contrast between nanofillers and the matrix, the image can be processed
by Image-Pro software and information such as particle size and orientation can be
gathered for further analysis. The complementary fluorescent image can identify the
epoxy fixative material used in the sample preparation process and helps interpret
the morphology of samples with internal porosity.

27

REFERENCES

28

REFERENCES
1.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, Mechanical properties and
morphological characterization of exfoliated graphite-polypropylene
nanocomposites. Composites Part a-Applied Science and Manufacturing,
2007. 38(7): p. 1675-1682.

2.

Tang, W.Z., M.H. Santare, and S.G. Advani, Melt processing and mechanical
property characterization of multi-walled carbon nanotube/high density
polyethylene (MWNT/HDPE) composite films. Carbon, 2003. 41(14): p.
2779-2785.

3.

Giannelis, E.P., Polymer-layered silicate nanocomposites: Synthesis, properties
and applications. Applied Organometallic Chemistry, 1998. 12(10-11): p.
675-680.

4.

Lan, T. and T.J. Pinnavaia, CLAY-REINFORCED EPOXY NANOCOMPOSITES.
Chemistry of Materials, 1994. 6(12): p. 2216-2219.

5.

Ginic-Markovic, M., et al., Synthesis of new polyaniline/nanotube composites
using ultrasonically initiated emulsion polymerization. Chemistry of Materials,
2006. 18(26): p. 6258-6265.

6.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, A new compounding method for
exfoliated graphite-polypropylene nanocomposites with enhanced flexural
properties and lower percolation threshold. Composites Science and
Technology, 2007. 67(10): p. 2045-2051.

29

Chapter 3 Effect of Molding and
Annealing on Dispersion and
Orientation on the Mechanical
Properties and Electrical Conductivity
of Graphene Nanoplatelet
-Polyetherimide Composites
Abstract
In this research, exfoliated graphene nanoplatelets (GNP, ~10 nm thick and ~5
microns in diameter)) and Polyetherimide (PEId) were compounded via
melt-extrusion in a Leistritz twin-screw extruder. Both injection molding and
compression molding were used to fabricate GNP-PEId nanocomposites at various
concentrations for measurement of their mechanical properties and electrical
conductivity. It was found that the electrical percolation threshold of the composites
processed by compression molding (<3wt%) was much lower than those made by
injection molding (>10wt%). However, the composites that were made by injection
molding showed higher modulus and strength. Scanning electron microscopy (SEM)
investigation of the morphology found that injection molding led to higher GNP
orientation, although the degree of orientation varied from the surface to the center
30

of the sample. Modification of the Annealing of the

samples made by injection

molding under the same temperature as during compression molding was found to
increase the electrical conductivity with the annealing time. The largest
improvement in electrical conductivity with annealing was in the through-plane
direction (perpendicular to flow direction in injection molding). SEM images
confirmed an increase in the random orientation of the GNP after annealing.
Tandon-Wang equation with respect to the nanoparticle orientation resulted in good
agreement with the experimental results.

Introduction
Polymer nanocomposites usually refer to a polymer matrix filled by high-surface-area
reinforcing fillers [1] including carbon nanotubes (CNTs), carbon nanofibers, nanoclay,
carbon black (CB), graphene platelets and so on. Since their discovery, nanoparticles
were expected to function as substitute reinforcements for conventional
reinforcement fibers because of the theoretically high particle properties and the
significant improvement in nanocomposite properties attained at low loadings.
Furthermore significant advantages were expected from gains in processing by
eliminating the need for layer-by-layer assembly or autoclave processing. The
functionality of the nanofillers is highly dependent on their morphology. For example,
as conductive fillers, the zero-dimensional CB can improve composite electrical
conductivity as a result of electrical percolation at low (~5%) loading for non-polar,
low crystallinity thermoplastics [2], while high aspect ratio CNT filled composites with
31

a similar matrix have a percolation threshold of 0.1vol% [3]. The high aspect ratio and
linear shape of CNTs make it easier to form a percolated network with more contacts
being formed between filler particles. Two-dimensional (nano-platelet) fillers such as
alumino-silicate nanoclay can also produce composites with low gas permeability
which cannot be achieved by CNTs or CB because of the highly tortuous path formed
by their two-dimensional nature [4, 5]. Research also suggests that this kind of filler
can also be used to improve the fire retardancy of its composite [6].
Despite of the promising properties of CNTs, it is not advantageous to use them in
commercial composites because of their expensive nature and limited availability.
CB/rubber composites reached the expectation of the tire industry, and colloidal
silica is showing a promising future in the same area [7]. However, these two fillers
do not possess the excellent electrical properties as found in CNTs.
Graphene based nanoplatelet particles, which combine good electrical and thermal
properties similar to CNTs and the two-dimensional morphology as alumino-silicate
clays, are considered to be the excellent filler in terms of multifunctionality. The
process of generating thin graphite platelets consisting of several layers of graphene
sheets is relatively simple and cost-effective. Usually, three steps are involved in
fabricating these nanofillers from natural graphite: (i) intercalation, where both
electron-donor agents and electron acceptors can be intercalated in between
graphene layers; (ii) exfoliation, where a large amount of heat is applied to expand
the intercalated graphite through vaporization of the intercalant; (iii) pulverization,
where the exfoliated graphite is reduced in size to smaller dimensions. As
32

demonstrated in Drzal’s group [8], exfoliated graphite nanoplatelets (GNP) consisting
of small stacks of graphene can be fabricated by acid intercalation, microwave
heating and ultrasonication pulverization and be scaled inexpensively into quantity
production by this route. The size and thickness of GNP is controllable during
fabrication. Typical sizes of the platelets range from submicron to 25 microns in
diameter usually notated as GNP-1, GNP-5 and GNP-15 where the number refers to
the average statistical diameter with thicknesses in the 5-10 nm range.
Because of the two dimensional nature of GNP and its high aspect ratio, final
properties of GNP based nanocomposites can vary due to the processing methods
selected. For example, the molding processes are important for their ability to make
complex shaped integrated composites. Kim et al [9] have shown that differences in
electrical conductivity of polycarbonate/graphene composites can be attained
depending on whether they were made by injection molding or compression molding.
Compression molded samples showed much better electrical conductivity than
injection molded ones. Mohanty et al [10] found that post-molding treatment, such
as annealing after injection molding can provide better flexural modulus and strength
and alter the morphology of nanocomposites and thus their properties. Dynamic
percolation behavior of CB composites has been studied by Wu et al [11] who found
that due to Brownian motion, CB particles are able to reposition themselves locally
under high temperature during which the viscosity of the polymer matrix is low.
Particles of platelet shape such as alumino-silicates and graphene, might not be
subject to Brownian motion because of their larger dimensions. Researchers have
33

shown faster particle re-organization in layered alumino-silicate/polymer composites
than expected compared to Brownian motion theory [12-14]. The real driving force
might be the elastic restoring force of the polymeric matrix generated during
processing though no widely-accepted conclusion has been drawn yet. Kim proposed
a disorientation mechanism for graphene composites based on X-ray scattering and
rheology data, pointing out that at low loadings (<3wt%) of graphite filler, the
rotation experienced by the particles caused by annealing still cannot increase
particle contact with each other, while at high loadings (>12wt%), the movement of
filler particles is restricted due to excluded volume interactions between particles.
The purpose of this research is to understand the relationship between processing,
morphology and properties of the GNP/polyetherimide (PEId) nanocomposites. PEId
has a high glass transition temperature, extremely good chemical resistance and
excellent thermal stability and is an amorphous thermoplastic material.
Melt-extrusion in a twin-screw extruder was used to compound GNP and PEId pellets
which were either injection molded or compression molded. Annealing under
confinement at the molding temperature (340℃) was investigated for its effect on
mechanical and electrical properties.

Experimentation
Materials
Polyetherimide (PEId, Ultem 1010) was provided by Sabic Americas, Inc. (Houston,
34

TX). GNP-5 was provided by XG Sciences, Inc. (East Lansing, MI) with an average
diameter of 3.9 micron and a specific surface area of 40m2/g.

Melt-extrusion
Melt-extrusion was carried on a Leistritz co-rotating twin-screw extruder
(MIC27/6L-48D). Barrel temperatures were set to 310C with a melt temperature at
340C. The die pressure was ~5.5MPa. Screw speed was set at 150 rpm.

Molding
After extrusion compounding, part of the composite material was compression
molded in a CARVER Laboratory Press (Model 2731, Fred S. Carver Inc.). A
picture-frame mold was used with a dimension of 76mm × 76mm × 3.3mm. The
material was heated up to 340C and held for 15 min. A compression pressure of
20MPa was applied to the material and was held for 15 min. After that, the press was
cooled with water. The other part of the composite material was injection molded
with a Milacron VSX 85-4.44 (Cincinnati, OH) injection molder with a Master
Precision tensile coupon die (Series No. 24157, Greenville, MI). The melt pressure
during injection molding was 90MPa. The central part of the injection molded tensile
bars were cut into a 12.7mm by 12.7mm square and annealed in a mold of the same
size under a pressure of 23MPa at 340C for 1 hour.

35

Characterization
The resistance of GNP/PEId composites was measured in two directions: the
‘through-plane direction’ which through the coupon thickness and was the same
direction as the applied pressure in compression molding while

the ‘in-plane

direction’ which in the direction of flow in the injection molded coupon and was
perpendicular to the pressure in the compression molded coupon. The resistance
value was then converted into resistivity by taking into account the sample
dimensions. Flexural strength and modulus were tested under ASTM 790 standard on
a UTS SFM-20 machine (United Calibration Corp.) at room temperature. The test was
performed at a flexural rate of 0.03 in/min. according to ASTM 790 standard, the
sample strain should not exceed 5% during flexural test, so the strength value was
calculated from the force at 5% strain if the sample hadn’t failed before that. Tensile
strength and modulus were tested under ASTM D638 with a straining rate of 0.02
in/min.
The morphology of the nanocomposites was observed by Scanning Electron
Microscope (JSM-6400, Jeol, Tokyo, Japan). In order to investigate the distribution
and orientation of the fillers, the tensile bar coupons were cut along as well as
perpendicular to the material flow direction as in injection molding and put into an
epoxy holder to undergo a series of polishing steps. Oxygen plasma was used after
the final polishing to etch away a layer of polymer in order to expose the GNP
particle. The polished surface was then sputter coated with gold.
36

Results and Discussions
Compression molding
(b) Electrical Resistivity - inplane direction
15
10
inj.
5
0
comp.

Electrical Resistivity through-plane direction
15
10
inj.
5
comp.
0

log(r/ohm-cm)

log(r/ohm-cm)

(a)

0 1 3 5 10
GNP-5 loading wt%

(c)

3

inj.

0

strength/MPa

modulus/GPa

6

comp.
1

3

GNP-5 loading wt%

(d)

Flexural modulus

0

0 1 3 5 10

5 10

Flexural strength

150
100

0

strength/MPa

modulus/GPa

(f)

6
inj.

0
1

2

3

4

1

3

5 10

GNP-5 loading wt%

Tensile modulus

3

comp.

0

GNP-5 loading wt%

(e)

inj.

50

5

GNP-5 loading wt%

Tensile strength

150
100
inj.

50
0
1

2

3

4

5

GNP-5 loading wt%

Figure 3.1, Comparison between GNP-5/PEId composites made from injection
molding and compression molding
The flexural modulus and strength and electrical resistivity of the GNP-5/PEId
composites made from compression molding are shown in Figure 3.1. It is found that
37

the electrical percolation threshold is around 3wt% for the through-plane direction
and 1 wt% for the in-plane direction. This is the results of the GNP particles having a
slightly preferential alignment direction because of the compressive force during
molding as shown in Figure 3.2(a). The flexural modulus of the composites is
increased with an increase of GNP content. At 10wt% loading, a 70% improvement of
modulus is observed. For flexural strength, an increase is seen with the GNP content
up to 3wt%, while a decrease is recorded afterwards. Interestingly, 3wt% loading is
around the electrical percolation threshold of the composites, suggesting the
forming of a continuous filler phase is producing strength reducing defects in the
composite.

Injection molding
The orientation of the GNP particles in the injection molded coupons is much more
complicated because of the material flow and the walls of the mold. The injection
molded coupons show anisotropic behavior due to material flow. GNP orientation
should be present at in two cross-sections: one surface perpendicular to the flow
direction and the other one along the flow direction (Figures 3.2 (b) and (c)). In
Figure 3.2(b), different orientations of the GNP particles can be seen at different
position of the cross-section. The portion of the samplethat is close to the wall has
almost perfect orientation parallel to it, while a change of orientation can be
observed at the corner. As the distance between the particle and the wall increase,
the GNP particle undergoes less orientation. This is because viscous melt of the
38

composite has the highest shear rate at the wall and the lowest at the center of the
coupon.
The orientation of the filler particles leads to differences in the mechanical behavior
between the compression and injection molded samples. In Figure 3.1 (c), the
modulus of the GNP-5/PEId composites made from injection molding is higher than
the ones made from compression molding. At 10wt%, the modulus is improved by
100% as compared to the neat polymer. The strength of the injection molded
coupons increases steadily with the increase of GNP loading up to 5wt% while at
10wt% it remains constant at the same level as the neat polymer. For tensile
modulus, the same trend is observed as flexural modulus, while tensile strength
drops steadily with the increase of the filler loading.
The electrical resistivity of the injection molded samples is much higher than the
compression molded ones. It is found that in the through-plane direction, the GNP
particles do not percolate even at 10wt% loading. For the in-plane direction, the
10wt% sample shows percolated behavior, but the resistivity is two orders of
magnitude higher than its compression molded counterpart.

39

(a)

(b)

(c)
Figure 3.2, Orientation of GNPGNP particles observed on: (a) cross-section of a
compression molded coupon; (b) the cross-section of an injection molded coupon
perpendicular to the flow direction & (c) along the flow direction (the scale bars
represent 20 micron)
40

(a)

(b)

Figure 3.3, Schematics of alignment situation of GNP particles in PEId matrix: (a)
random orientation after compression molding, (b) aligned orientation after injection
molding
As shown in Figure 3.3(a), the particles are randomly oriented. The connection
between particles is good. However, in Figure 3.3(b), most of the particles are aligned.
The alignment not only prevents the particles from connecting with each other in the
through-plane direction, but also places the particles apart in the in-plane direction.

Model prediction of modulus
Because of the complicated orientation situation of GNP/PEId composites, it’s not
easy to predict their mechanical behavior. Figure 3.4 shows the calculated modulus

𝐸11

=

1
1+𝜙(−2𝜐 𝑚 𝐴3 +(1−𝜐 𝑚 ) 𝐴4 +(1+𝜐 𝑚 ) 𝐴5 𝐴)/2𝐴

value of the composites using the Tandon-Weng equation [15].

𝐸𝑚

(1)

In equation (1), E 11 , E m are the Young’s modulus of the composite and neat matrix

polymer, respectively. ϕ is the volume fraction of the filler. υm is the Poisson’s
ratio of the matrix. A and Ai are functions of ϕ,

υm

and Eshelby’s tensor.

Tandon and Weng gave formulas for both uni-directional and randomly oriented [16]
41

situations. The values of the parameters used in the equation are listed in Table 3.1.

Figure 3.4, Comparison between experimental data (tensile modulus of injection
molded samples and flexural modulus of compression molded samples) and
Tandon-Weng modeling
Although the reported Poisson’s Ratio in literature varies from -0.171 to 4.958, the
predicted modulus does not vary significantly using different values due to the small
thickness of GNP particles. 0.006 was chosen in model calculation. The aspect ratio
of the GNP particles after processing can be determined by the following equation
[9]:

Where

ϕsphere = 0.29

𝐴𝑓 =

3𝜙 𝑠𝑝ℎ𝑒𝑟𝑒

2𝜙 𝑝𝑙𝑎𝑡𝑒𝑙𝑒𝑡

(2)

which is the percolation threshold of spherical-shaped

conductive fillers [17]. From Figure 3.1 (a) (b), the electrical percolation takes place
around 1wt% for the in-plane direction and 3wt% for the through-plane direction.
From equation (2), the actual aspect ratio of the GNP particles should be between 23
and 71. The average value 47 is used in the model.
42

Interestingly, if we consider the surface area of the GNP prior to processing and the
average size of the particles, the expected aspect ratio should be around 300.
However, during melt-extrusion, these platelet-shaped particles are reduced in
lateral dimension because of the mechanical shearing of the twin-screws,. (Figure
3.4)
From TEM images, a stack of GNP usually contains ~3-5 layers (each layer is a GNP
particle consisting of a few graphene sheets with a thickness of ~10nm) with a
thickness varying from 30 to 80 nm. The SEM images show an average diameter of
GNP particles after processing as 2 micron, which means the aspect ratio of the
particles is between 25 and 67. This result is consistent with the result calculated
from equation (2).
In Figure 3.4, the random platelet orientation Tandon-Weng equation predictions of
the compression molded GNP/PEId composites agree very well at loadings up to
5wt%. However, at 10wt%, the experimental data is higher than predicted. This is
because at 10wt%, the distance between particles is very small, resulting in a
reduction in the ability of the platelet particles to freely rotate during molding.
Table 3.1, Parameters used in Tandon-Weng equation
Component

Modulus (GPa)

Poisson’s Ratio

Density (g/cm3)

GNP

1060

-0.171~4.958[18]

2.1

PEId

3.3

0.36

1.29

43

However, for injection molded samples, neither random nor unidirectional model
accurately predicts the observed behavior, indicating that a complicated orientation
state exists in these samples. To simplify the problem, we only analyze the
orientation along the flow direction which is also the direction of tensile testing.
From Figure 3.2(c), the particles close to the surface of the sample tend to orient in
the flow direction while the ones at the center of the coupon tend to be randomly
oriented. This phenomenon suggests a combination of the unidirectional model and
the random model. If we divide the sample coupon into two regions, one considered
as unidirectional and the other random, then the modulus of the sample can be
expressed as:

𝐸11
𝐸𝑚

= 𝑥

𝐸11,𝑢𝑛𝑖𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛𝑎𝑙
𝐸𝑚

+ (1 − 𝑥)

𝐸11,𝑟𝑎𝑛𝑑𝑜𝑚
𝐸𝑚

(3)

where x stands for the volume ratio of unidirectional region. By fitting equation (3)
with the experimental data, the values of x at different loadings can be found. (Table
3.2)
Table 3.2, Fit values of the volume ratio of unidirectional region in
injection molded coupons
GNP loading

1wt%

3wt^

5wt%

10wt%

x

1

0.62

0.46

0.41

44

(a)

(b)

Figure 3.5, (a) TEM image of GNP particle in PEId matrix after melt-extrusion,
showing stacking of layers & (b) SEM image of 1wt% GNP-5/PEId composite made
from melt-extrusion, showing reduced particle size (original average size: 3.9 micron
in diameter). (scale bars: (a) 50 nm, (b) 5 micron)
To verify this approach, SEM images were taken at the cross-section of flow direction
from edge to center (sample thickness: 3.3mm). The SEM images were processed in
Image-Pro Plus software (Media Cybernetics, Inc., MD) to calculate the orientation
information, a 11 of the particles:

𝑎11 =< 𝑐𝑜𝑠 2 𝜃 >

where 𝜃 is the angle between the GNP basal plane and the flow direction.

45

(4)

(a)

(b)

(c)
Figure 3.6, a11 along the thickness direction on the cross-section plane of flow
direction (diamond series are a11 values and dash lines indicate the distinguish of
two regions) and the representative morphology
46

Figure 3.6 shows the change of degree of orientation from the surface to the center
unidirectional (a11

= 1) or completely random (a11 = 1/3), it is clear that the

of the sample. Although none of the points reaches the extreme value for either

sample can be split into two parts with different levels of orientation. Moreover, for

the 3wt% sample, the transition from high to low orientation occurs between 1mm
and 1.2mm, indicating the volume ratio of the unidirectional region to be 0.61~0.73.
The same procedure can be applied to 5wt% and 10wt% samples, where the volume
ratios are 0.36~0.48 and 0.24~0.48 respectively. This result correlates well with the
computed values as shown in Table 3.1.

Effect of annealing
It has been shown that the percolation behavior of nanocomposites filled with
conductive nanoparticles is influenced by annealing under high temperature (above
the melt temperature of the matrix polymer). Similarly during annealing, GNP
particles are able to reorient due to polymer chain relaxation and elastic restoring
force [12].
From Figure 3.7, the electrical resistivity in the through-plane direction decreases
significantly after annealing. The 5wt% GNP-5/PEId composite displays percolation.
In the in-plane direction, the decrease in electrical resistivity is less than in the
through-plane direction, but is significant. A 1.5 order of magnitude decrease is
attained for the 5wt% loading and a 2.5 order of magnitude decrease for the 10wt%
loading.
47

SEM images of the composites taken both before and after annealing document the
GNP re-orientation. Figure 3.8 shows the morphology comparison between the
composites before and after annealing. It is found that the area that is close to the
surface of the tensile bar has a higher degree of orientation (Figure 3.8(a)). However,
after annealing, the highly aligned orientation is destroyed as shown in Figure 3.8 (b).
As discussed earlier, less oriented particles have a better chance to connect with
each other (Figure 3.3) and therefore provide path for electrical conductivity. The fact
that the decrease of resistivity in the through-plane direction is larger than that in
the in-plane direction indicates that the movement of the particles does not result in
their aggregation in the flow direction. The particles rotate at the original position
rather than move along the flow direction.

48

(a)

(b)
Figure 3.7, Effect of annealing on electrical resistivity
Figure 3.9 shows the change of electrical resistivity of GNP/PEId composites along
with the annealing time. For the 5wt% sample, the minimum time for the effect of
annealing to affect the percolation point is between 40 and 60 minutes. For the 10wt%
sample, the time is around 10 minutes. With higher GNP loadings, the annealing
percolation time decreases. This is due to the smaller distance between GNP
49

particles in the composites as the concentration increases. Composites with higher
loadings of GNP have a shorter distance between particles, requiring less movement
of the particles to make contact with each other.

(a)

(b)

Figure 3.8, 5wt% GNP-5/PEId composite (a) before annealing and (b) after annealing.
Images were taken along the flow direction (as during injection molding), showing
the morphology of the near-surface part of the tensile bar coupon (scale bars
represent 20 micron)

Figure 3.9, Time dependence of the annealing effect

50

Conclusions
GNP and PEId were compounded by melt-extrusion followed by either injection
molding or compression molding. It was found that injection molding leads to better
mechanical properties while compression molding results in better electrical
conductivity. When the Tandon-Weng equation was applied to predict the modulus
of the samples, it was found that the random particle orientation Tandon-Weng
equation fits compression molded samples well while a hybrid combination of
unidirectional and random model best explains the behavior of injection molded
samples. Annealing is found to be effective in overcoming the orientation effects
caused by molding and leads to higher electrical conductivity of the GNP/PEId
composites as a result of the rotation of GNP particles during the high temperature
annealing treatment. SEM images of GNP morphology in the molded coupons
confirm the nanoplatelet orientation and annealing effects and support these
conclusions.

51

REFERENCES

52

REFERENCES
1.

Vaia, R.A. and E.P. Giannelis, Polymer nanocomposites: Status and
opportunities. Mrs Bulletin, 2001. 26(5): p. 394-401.

2.

Huang, J.C., Carbon black filled conducting polymers and polymer blends.
Advances in Polymer Technology, 2002. 21(4): p. 299-313.

3.

Bauhofer, W. and J.Z. Kovacs, A review and analysis of electrical percolation in
carbon nanotube polymer composites. Composites Science and Technology,
2009. 69(10): p. 1486-1498.

4.

LeBaron, P.C., Z. Wang, and T.J. Pinnavaia, Polymer-layered silicate
nanocomposites: an overview. Applied Clay Science, 1999. 15(1-2): p. 11-29.

5.

Kim, J.K., et al., Moisture barrier characteristics of organoclay-epoxy
nanocomposites. Composites Science and Technology, 2005. 65(5): p.
805-813.

6.

Porter, D., E. Metcalfe, and M.J.K. Thomas, Nanocomposite fire retardants - A
review. Fire and Materials, 2000. 24(1): p. 45-52.

7.

Schaefer, D.W. and R.S. Justice, How nano are nanocomposites?
Macromolecules, 2007. 40(24): p. 8501-8517.

8.

Fukushima, H., Graphite Nanoreinforcements in Polymer Nanocomposites, in
Department of Chemical Engineering and Materials Science. 2003, Michigan
State University: East Lansing.

9.

Kim, H. and C.W. Macosko, Processing-property relationships of
polycarbonate/graphene composites. Polymer, 2009. 50(15): p. 3797-3809.

10.

Mohanty, A.K., et al., Effect of process engineering on the performance of
natural fiber reinforced cellulose acetate biocomposites. Composites Part
a-Applied Science and Manufacturing, 2004. 35(3): p. 363-370.

11.

Wu, G.Z., et al., A delay of percolation time in carbon-black-filled conductive
polymer composites. Journal of Applied Physics, 2000. 88(3): p. 1480-1487.

12.

Schmidt, G., et al., Shear orientation of viscoelastic polymer-clay solutions
probed by flow birefringence and SANS. Macromolecules, 2000. 33(20): p.
7219-7222.
53

13.

Solomon, M.J., et al., Rheology of polypropylene/clay hybrid materials.
Macromolecules, 2001. 34(6): p. 1864-1872.

14.

Ren, J.X., et al., Disorientation kinetics of aligned polymer layered silicate
nanocomposites. Macromolecules, 2003. 36(11): p. 4188-4194.

15.

Tandon, G.P. and G.J. Weng, THE EFFECT OF ASPECT RATIO OF INCLUSIONS ON
THE ELASTIC PROPERTIES OF UNIDIRECTIONALLY ALIGNED COMPOSITES.
Polymer Composites, 1984. 5(4): p. 327-333.

16.

Tandon, G.P. and G.J. Weng, AVERAGE STRESS IN THE MATRIX AND EFFECTIVE
MODULI OF RANDOMLY ORIENTED COMPOSITES. Composites Science and
Technology, 1986. 27(2): p. 111-132.

17.

Shante, V.K.S. and Kirkpatr.S, INTRODUCTION TO PERCOLATION THEORY.
Advances in Physics, 1971. 20(85): p. 325-&.

18.

Cho, J., J.J. Luo, and I.M. Daniel, Mechanical characterization of
graphite/epoxy nanocomposites by multi-scale analysis. Composites Science
and Technology, 2007. 67(11-12): p. 2399-2407.

54

Chapter 4 Effect of Graphite
Nanoplatelets on Coefficient of
Thermal Expansion of Polyetherimide
Composite
Abstract
Thermal expansion is one of the major concerns for polymer composites. In this
research, graphite nanoplatelets (GNPs) was added to polyetherimide (PEId)
thermoplastic polymer in order to reduce the coefficient of thermal expansion (CTE)
of the injection molded composite. First, the coefficient of linear thermal expansion
(LTE) was measured in three directions in the anisotropic coupon: 0 degree, 90
degree and the out of plane Z direction. It is found that the GNP particles are very
effective in terms of reducing the LTE in 0 degree direction due to high degree of
alignment. After annealing above glass transition temperature, significant increase of
0 degree LTE and decrease of Z degree LTE were observed. The bulk CTE was
calculated by adding up the LTEs in all three directions and is found to be
independent of annealing. Second, several models were applied to predict both CTE
and LTE. It is found that Schapery’s lower limit model fits the experimental CTE very
well. Chow’s model was applied for LTEs in three directions. The behavior of
GNP-5/PEId composites is explained by the combination of Chow’s model and
55

morphology obtained by scanning electron microscope (SEM).

Introduction
Thermoplastic materials are receiving more and more attention as the matrix to host
either continuous fibers or nano-scale particles. Compared to thermoset polymers,
thermoplastics are generally tougher and more resistant to chemical decomposition
[1]. Moreover, thermoplastics can be processed much faster than thermosets which
require a dedicated curing cycle. The thermoplastics also have a much longer shelf
life. However, a key factor that has been bothering the thermoset matrices is still not
solved by switching to thermoplastics: the expansion and contraction of the polymer
due to change in temperature. If used as a structural material, the dimensional
change of the polymer due to cyclical (season or location) changes may cause serious
deformation. In a composite consisting of continuous fibers, the difference in the
coefficient of thermal expansion (CTE) between fiber and matrix would result in
significant residual stress in the composite [2-5] which will generate defects [6, 7]
and affect matrix-dominated properties such as moisture absorption [8] and glass
transition temperature [9].
Many efforts have been made to reduce the CTE of a thermoplastic system. Nano
fillers such as carbon nanotubes (CNTs) [10-13], nano clay [14-17] and nano graphite
[18] are mostly used in such applications. Although the difference in CTE still exists
for nano fillers as compared with their matrices [19], the extremely small size of the
filler minimizes the damage especially when the filler is homogeneously dispersed.
56

Warrier et al [10] incorporated CNTs into carbon fiber (CF)/epoxy composite and
succeeded in increasing the glass transition temperature and reducing the CTE,
indicating the effectiveness of using nano fillers to minimize the residual stress in a
fiber reinforced composite. Similar results are also seen in the work in the Drzal
group [Toshiya Kamae, unpublished results].
Nano clay is the most popular filler for reducing CTE of the composites. Gu et al [16]
reported a 65% reduction in CTE by adding 3wt% organomontmorillonite in
polyimide by a special processing route. This reduction is huge compared to other
results [15, 17] where a reduction around 10% is normally found for the same
loading. However, the major contributor in Gu’s work is the thermal imidization of
the matrix instead of the clay. For CNT, the influence is smaller. A 2% reduction is
found in 3wt% CNT/epoxy system [11].
A newly-developed nano filler, graphite nanoplatelet (GNP) has already shown great
potential to reduce the CTE in the composites. With the same 2D geometry, the GNP
can provide a reduction in the CTE of its composite comparable to nano clay.
Kalaitzidou et al [18] reported a 25% reduction for 3vol% GNP filled polypropylene.
Moreover, the GNP also possesses high thermal and electrical conductivity and can
be produced by a scalable, cost-effective process [20], making it the ideal filler for
multifunctional composites.
PEId is an engineering thermoplastic that has unique flame-resistance properties and
would only generate a very small amount of low density gas when forced to burn,
making it well suited and widely used in airplane, aerospace, submarine and
57

high-floor building applications. However, these applications may also involve
dramatic ambient temperature change, which would benefit from a lower CTE. The
need of using continuous CF in PEId matrix also would benefit from a reduction in the
transverse CTE as a result of adding GNP to the GNP/PEId composite. In this research,
the linear coefficient of thermal expansion (LTE) is investigated in injection molded
GNP/PEId composites in three dimensions: 0 degree (flow direction), 90 degree and
the out-of-plane Z direction. The CTE of the thermally annealed samples are also
measured. In addition, models for both LTE and CTE are applied according to the
morphology of the GNP/PEId composites.

Experimentation
Materials
Polyetherimide (PEI) was obtained from Sabic-IP (Ultem 1010). GNP-5 (5 micron
average diameter, 40m2/g surface area) was provided by XG-Science Inc. (Lansing,
MI).

Melt-extrusion
GNP and PEId were blended by melt-extrusion in a Leistritz twin-screw extruder
(MIC27/6L-48D). The screws were operated in the co-rotation setup. Barrel
temperatures were set to 310℃, with a melt temperature at 340℃. The die pressure
was at around 5.5MPa. Screw speed was set at 150 rpm.

58

Injection Molding
The composite material was injection molded by a Milacron VSX 85-4.44 (Cincinnati,
OH) injection molder with a Master Precision tensile coupon (Type I dog bone) die
(Series No. 24157, Greenville, MI). The melt pressure during injection molding was
90MPa.

Coefficient of Thermal Expansion
The CTE of the GNP/PEId composites was measured by Thermal Mechanical Analyzer
(TMA, TA Instruments). The central portion was cut from the injection molded tensile
coupon. The LTEs were measured under a 10℃/min heating rate on TMA. Three
dimensions including 0 degree, 90 degree and Z direction (Figure 4.1) were measured.
The CTE was calculated by combining the three LTEs for an individual sample.

Figure 4.1, Schematics of TMA sample preparation

Annealing
The injection molded samples were annealed in a CARVER Laboratory Press (Model
2731, Fred S. Carver INC.) under 340℃ and 23 MPa. A steel picture-frame mold of
59

the same size as the dog bone coupon is used to confine the dimensions of the
sample. The annealed dog bone coupons were cut and measured by TMA in the
same manner as the un-annealed samples.

Results and Discussions
Linear Thermal Expansion
The melt flow during injection molding leads to alignment of GNP particles which
causes the composite to behave anisotropically. Figure 4.2 shows the LTEs in 0 degree,
90 degree and Z directions for injection molded coupons of neat PEId and GNP/PEId
composites up to 10wt% filler loading.

Figure 4.2, LTE of injection molded GNP/PEId composites
With 3wt% GNP particles, the composite shows a reduction of 33% in 0 degree LTE.
Additional GNP particles cannot further reduce the LTE in the same direction. Instead,
the 90 degree LTE starts to decrease when the GNP loading reaches 5wt% and a
60

reduction of 26% is recorded at 10wt% loading. Meanwhile, the Z direction LTE
remains at the same level regardless of GNP loading.
Interestingly, even the injection molded neat PEId sample shows anisotropic LTEs.
This could be due to the fact that the polymer chains are also aligned along the flow
direction during injection molding. Generally, higher bond energy leads to lower
thermal expansion. Since the 0 degree has more covalent bonds due to chain
alignment, its LTE is the smallest among the three directions.

Annealing Effect
In a GNP/PEId composite, the anisotropic behavior in LTEs is significantly larger than
in neat PEId sample because of particle alignment. In Chapter 3, annealing is found to
allow GNP particle rotation and re-orientation. Although the polymer chain
relaxation should be the driving force underneath this phenomenon, no proof is
provided.
Now by comparing the LTEs of GNP/PEId samples before and after annealing, it is
clear that the polymer chain itself also re-orients during annealing. In Figure 4.3, the
annealed neat PEId samples show isotropic LTEs in all three directions, indicating that
there is no significant preferential chain alignment.

61

Figure 4.3, LTE of annealed GNP/PEId composites
GNP/PEId composites with up to 3wt% filler loading show isotropic LTEs as well.
However, with 5wt% or more filler, anisotropic behavior still exists. This can be
regarded to the confinement to GNP particle rotation caused by a dense population
of filler particles (Figure 4.4).

Figure 4.4, Schematics of confinement to GNP particle rotation
The comparison of LTEs before and after annealing is shown in Figure 4.5. In 0 degree
62

direction, the LTE generally increases significantly after annealing, except for 10wt%
loading (Figure 4.5(a)). Below 5wt%, the difference in LTE consists of both polymer
chain (16% increase as indicated by neat PEId group) and particle re-orientation (14%
increase for 1wt% GNP/PEId, 47% increase for 3wt% GNP/PEId). The fact that the
increase of LTE caused by annealing starts to drop from 5wt% loading (20% increase)
indicates that the confinement begins to take place. At 10wt%, although the polymer
chain re-orientation still exists, the population of the particles is large enough to
generate significant interference between particles, resulting in much less rotation of
the GNP particles (Figure 4.4). The almost no change in 10wt% sample indicates that
the apparent LTE in 0 degree direction is predominated by the GNP particles at this
high loading.
After annealing, no significant change of LTE in 90 degree direction can be observed
(Figure 4.5(b)). For this specific direction, the two factors, chain and GNP particle
re-orientation, actually have opposite effect on LTE. The chain re-orientation will
result in more polymer chains aligned in 90 degree compared with before annealing
while the rotation of two-dimensional GNP particles results in less orientation. These
two opposite effects cancel each other and keep the LTE in 90 degree direction
unchanged.
In the Z direction, annealing can reduce LTE (Figure 4.5(c)). Both chain and GNP
particle re-orientation benefit the reduction. The fact that there is a significant
decrease for 10wt% sample indicates that the polymer chain re-orientation still
happens during annealing.
63

(a)

(b)

(c)
Figure 4.5, Comparison of LTE: before annealing and after annealing
64

Bulk thermal expansion (CTE) can be calculated by combining the LTEs in all three
directions (Figure 4.6). It can be seen that the CTE of the composites decreases with
the increase of GNP-5 loading. Besides, after annealing, most samples retain the
same CTE except for 10wt%. This indicates that for most cases, the rotation of the
GNP particles and the re-orientation of the polymer chains do not affect the bulk
thermal expansion, unless the particle rotation is interfered by the interaction
between particles due to high enough GNP concentration (Figure 4.4).

Figure 4.6, CTE of GNP/PEId composites before and after annealing

65

(a)

(b)

(c)

(d)

Figure 4.7, SEM images of GNP-5/PEId composites: 5wt% injection molded sample, (a)
before annealing, (b) after annealing, and 10wt% injection molded sample, (c) before
annealing, (d) after annealing. Images are taken at the near-surface part of the
tensile bar coupon (see Chapter 3 for details) (scale bars represent 20 micron)
Scanning electron microscope (SEM) images can reveal the rotation of GNP particles
after annealing (Figure 4.7). For 5wt%, the rotation is significant. Before annealing,
almost all the particles are aligned perfectly while they are randomly oriented after
annealing. However, there is not too much difference before and after annealing for
10wt% sample.

66

Modeling for Coefficient of Thermal Expansion
Bulk Thermal Expansion
Several models have been developed to predict the bulk CTE of the composites [21].

𝛾 𝑐 = 𝛾 𝑓 𝜙 + 𝛾 𝑚 (1 − 𝜙)

Rule of Mixture [22] gives a first order approximate:

Where

𝛾𝑐 , 𝛾 𝑓 , 𝛾 𝑚

(1)

are the bulk CTE of the composite, filler and matrix

respectively and 𝜙 is the volume fraction of the filler and can be calculated from

weight fraction 𝑤:

Where

𝑑

𝑚,

𝑑𝑓

𝜙=

𝑤

𝑤+(1−𝑤)

𝑑𝑓
𝑑 𝑚

(2)

are the density of the matrix and the filler.

A more sophisticated model proposed by Turner [23] takes into account the
interaction between filler and matrix. The equation can be expressed as:

Where

𝑘

𝑚,

𝑘𝑓

𝛾𝑐 =

(1−𝜙)𝑘 𝑚 𝛾 𝑚 /𝑑
(1−𝜙)𝑘 𝑚 /𝑑

𝑚 +𝜙𝑘 𝑓

𝛾 𝑓 /𝑑 𝑓

𝑚 +𝜙𝑘 𝑓 /𝑑 𝑓

(3)

are the bulk modulus of the matrix and the filler respectively and

Schapery [24] gives out the upper and lower limit of the CTE of composites:

𝐾 𝑐𝑢 , 𝐾 𝑐𝑙

𝛾 𝑐𝑙 = 𝛼 𝑚 +

𝛾 𝑐𝑢 = 𝛼 𝑚 +

𝑘 𝑓 (𝑘 𝑚 −𝑘 𝑐𝑢 )(𝛾 𝑓 −𝛾 𝑚 )
𝑘 𝑐𝑢
(𝑘 𝑚 −𝑘 𝑓 )
𝑘 𝑓 �𝑘 𝑚 −𝑘 𝑙𝑐 �(𝛾 𝑓 −𝛾 𝑚 )
(𝑘 𝑚 −𝑘 𝑓 )
𝑘 𝑙𝑐

(4)

(5)

Where the superscript l and u stand for the lower and upper limit, respectively, and
are the upper and lower limits of the bulk modulus of the composite and
67

𝑘 𝑐𝑢 = 𝑘 𝑓 +

1− 𝜙
�

are provided by Hashin and Shtrikman [25, 26]:

Where

𝜇

𝜙
𝑘 𝑙𝑐 = 𝑘 𝑚 + �

𝑚 and

𝜇𝑓

1

𝑘 𝑚 −𝑘 𝑓

1

𝑘 𝑓 −𝑘 𝑚

+

+

3𝜙

(3𝑘 𝑓 +4𝜇 𝑓 )

3(1−𝜙)

(3𝑘 𝑚 +4𝜇 𝑚 )

(6)

(7)

are the shear modulus of the matrix and the filler, respectively.

𝑘 and 𝑔 can be calculated from Young’s modulus 𝐸 and Poisson ratio 𝜈:

𝑘=

𝐸

2(1+𝜈)

and

All known parameters are listed in Table 4.1.

𝜇=

𝐸

(3−2𝜈)

Table 4.1, Properties of GNP and PEId
PEId
167.5
CTE (micron/m/℃)
Young’s Modulus (GPa)
3.13
Poisson Ratio
0.36
3
Density (g/cm )
1.29

GNP
23.5 [27]
1060
0.006 [28]
2.1

Figure 4.8, Model prediction for CTE of GNP/PEId composites
68

(8)

Among the four models, Schapery’s lower limit prediction fits the experimental data
best (Figure 4.8). Turner’s equation and Schapery’s upper limit prediction largely
overvalue the effect of GNP particles on the CTE of the composite while Rule of
Mixture underestimated the effect.

Linear Thermal Expansion
Based on the Eshelby tensor, Chow et al developed an analytical model for
calculating the LTE of the composite filled by perfectly aligned inclusions that can be

𝛼𝐿 = 𝛼 𝑚 +

considered as ellipsoids [29]:

where

𝛼𝐿

𝛼𝑇 = 𝛼 𝑚+

𝑘 𝑓 (𝛾 𝑓 −𝛾 𝑚 )𝐺1 𝜙
𝑘 𝑚 2𝐾1 𝐺3 +𝐺1 𝐾3

(9)

𝑘 𝑓 (𝛾 𝑓 −𝛾 𝑚 )𝐺3 𝜙
𝑘 𝑚 2𝐾1 𝐺3 +𝐺1 𝐾3

(10)

parallel with the alignment of the major axis of the ellipsoid (length) and

𝐾𝑖

LTE of the transverse direction.

𝐾𝑖 = 1 + �

In equations (9) and (10),

where

𝛼𝑖

𝛼𝑇

is the longitudinal LTE of the composite which is in the direction that is

𝐺𝑖 = 1 + �

and

𝛽𝑖

𝑘𝑓

and

𝑘 𝑚
𝜇𝑓

𝜇 𝑚

𝐺𝑖

− 1� [(1 − 𝜙) 𝛼 𝑖 + 𝜙]

is the

(i=1, 3) are:

− 1� [(1 − 𝜙) 𝛽 𝑖 + 𝜙]

(11)

(12)

are functions of the filler aspect ratio 𝜌 = 𝑙𝑒𝑛𝑔𝑡ℎ/𝑑𝑖𝑎𝑚𝑒𝑡𝑒𝑟,

the aspect ratio of the filler 𝐴 and the Poisson ratio of the matrix

𝜈𝑚

[30].

Similar to nanoclay, GNP can also be considered as ellipsoids with a
69

thickness/Diameter ratio less than 1 [31]. Note that

𝛼𝑇

𝛼𝐿

is actually the LTE of the

normal direction of the GNP particle and should be applied for the Z direction, while
is the LTE of the radius direction of the GNP particle and should be applied for

𝛼0 = 𝛼

𝑚,0 +

the 0 degree direction. So equation (9) and (10) become:

𝛼𝑍 = 𝛼

𝑚,𝑍 +

𝑘 𝑓 (𝛾 𝑓 −𝛾 𝑚 )𝐺1 𝜙
𝑘 𝑚 2𝐾1 𝐺3 +𝐺1 𝐾3
𝑘 𝑓 (𝛾 𝑓 −𝛾 𝑚 )𝐺3 𝜙
𝑘 𝑚 2𝐾1 𝐺3 +𝐺1 𝐾3

(13)

(14)

For 90 degree direction, the situation is more complicated. Due to the geometry of
the mold used in injection molding, some of the GNP particles are aligned along the
thickness dimension of the coupon while majority of them are aligned along the

𝛼𝐿

𝛼 𝑇.

width dimension (Figure 4.9). This morphology indicates that the overall LTE in 90
degree direction should be a combination of

and

Note that the same

situation also can be applied to the Z direction. However, during TMA measurement,
the testing probe is placed at the center of the sample. Due to the testing sample
geometry, the probe cannot detect the dimension change of the edge area (Figure
4.10). On the other hand, for 90 degree direction, the probe is large enough to detect
the dimension change caused by the entire sample.

70

Figure 4.9, SEM images showing orientation of GNP particles observed on the
cross-section of an injection molded coupon perpendicular to the flow direction
(scale bars represent 20 micron)

Figure 4.10, Configuration for measuring LTE in Z direction

71

Figure 4.11, Model prediction of LTE in (a) 0 degree direction, (b) 90 degree direction,
(c) Z direction

72

Figure 4.11 shows the prediction of Chow’s LTE model calculated with different filler

aspect ratio 𝐴. For 0 degree direction, the experimental data follow the model with

A=47 for loadings not higher than 3wt%. This aspect ratio value is consistent with the
result from Chapter 3. From 5wt% and above, the experimental data shift to lower
aspect ratio region of the model predictions, indicating either agglomeration of the
particles or the decrease of alignment of the particles.
For 90 degree direction, it is clear that with 3wt% loading or less, the experimental
data follow the longitudinal model, indicating that the LTE is predominated by the
GNP particles with their normal aligned in 90 degree direction. That is to say, the
particles that are aligned along the thickness dimension are the major contributor.
However, with 5wt% or more GNP added, the LTE in 90 degree direction starts to
follow the transverse model, indicating the particles aligned along the width
dimension start to be predominant in the composite.
To observe the change of particle alignment that affects 90 degree LTE, SEM images
were taken over the entire width-thickness plane of the injection molded coupon.

𝑎11

The SEM images were processed in Image-Pro Plus software (Media Cybernetics, Inc.,

𝑎11 =< 𝑐𝑜𝑠 2 𝜃 >

MD) to pick up the orientation information that can calculate

of the particles:
(15)

where 𝜃 is the angle between a GNP particle and the width dimension of the

𝑎11

injection molded coupon.
Then, the

values can be plotted against the position.

73

(a)

(b)

(c)

(d)

Figure 4.12, alignment of GNP particles on thickness-width plane of injection molded
coupon
In Figure 4.12, the patterns are significantly different between 1wt% and other
loadings, indicating that more particles are aligned along the thickness dimension
74

𝑎11

other than the width dimension for 1wt% sample. SEM images of representative
morphology for different

values are shown in Figure 4.13.

(a)a11 = 0.8~1.0, unidirectionally
aligned in width (90 degree) direction

(b)a11 = 0.6~0.8, preferentially aligned
in width (90 degree) direction

(c)a11 = 0.4~0.6, transition region:
aligned in 45 degree direction

(d)a11 = 0.2~0.4, preferentially aligned
in thickness direction (Z direction)

(e)a11 = 0.0~0.2, unidirectionally aligned in thickness direction (Z direction)
Figure 4.13, Morphology for different a11 values, where white lines stand for GNP
particles and black part stands for the matrix. The images are taken from 10wt%
GNP-5/PEId composite injection molded coupon on the cross-section of flow
direction (width-thickness plane)
75

(a)

(b)
Figure 4.14, Alignment in 90 degree direction
76

(c)

(d)
Figure 4.14, Alignment in 90 degree (cont.)
77

Figure 4.14 shows more information of the change of alignment with increase of GNP
loading. The series that fall into the solid frame stand for the majority of the
thickness-width plane. From 1wt% to higher loadings, the size of the frame decrease
significantly, indicating that more and more particles are aligned in 90 degree
direction with higher GNP loadings. The value of the line which sits in the dash frame
(series 6300) keeps increasing with the increase of GNP loading. These two trends
are consistent with the experimental data of LTE in 90 degree direction.
Interestingly, the LTE in the Z direction does not follow Chow’s model at all (Figure
4.11(c)). While the predicted result indicates that the LTE in the normal direction of
platelet particle should increase with the loading of filler, the experimental data of
composite stay at the same level as neat polymer no matter how high the loading is.

Conclusions
LTEs of the injection molded GNP-5/PEId composites were measured in the 0 degree,
90 degree and Z direction. It is found that the 0 degree LTE decrease with the
increase of GNP-5 loading and can be predicted by Chow’s model. The LTE in 90
degree direction also decrease with GNP loading at above 5wt% while the LTE
remains the same at lower than 3wt% loadings. This behavior is explained by the
morphology of the GNP particles in the cross-section of injection molded coupons
which show a development of GNP alignment with an increase of loading. It is also
observed that the bulk CTE of GNP/PEId composites can be best predicted by
Schapery’s lower limit model. Annealing is effective in changing the alignment of
78

polymer chains and GNP particles, resulting in significant increase of LTE in 0 degree
direction and decrease in Z direction. However, the overall bulk CTE is not affected by
annealing.

79

REFERENCES

80

REFERENCES
1.

Johnston, N.J., T.W. Towell, and P.M. Hergenrother, Physical and mechanical
properties of high-performance thermoplastic polymers and their composites,
in Thermoplastic composite materials, C. LA, Editor. 1991, Elsevier Science
Publishers B.V.: Amsterdam. p. 27-71.

2.

Parlevliet, P.P., H.E.N. Bersee, and A. Beukers, Residual stresses in
thermoplastic composites - A study of the literature - Part I: Formation of
residual stresses. Composites Part a-Applied Science and Manufacturing,
2006. 37(11): p. 1847-1857.

3.

Nairn, J.A. and P. Zoller, MATRIX SOLIDIFICATION AND THE RESULTING
RESIDUAL THERMAL-STRESSES IN COMPOSITES. Journal of Materials Science,
1985. 20(1): p. 355-367.

4.

Barnes, J.A. and G.E. Byerly, THE FORMATION OF RESIDUAL-STRESSES IN
LAMINATED THERMOPLASTIC COMPOSITES. Composites Science and
Technology, 1994. 51(4): p. 479-494.

5.

Chapman, T.J., et al., PREDICTION OF PROCESS-INDUCED RESIDUAL-STRESSES
IN THERMOPLASTIC COMPOSITES. Journal of Composite Materials, 1990.
24(6): p. 616-643.

6.

Kugler, D. and T.J. Moon, Identification of the most significant processing
parameters on the development of fiber waviness in thin laminates. Journal of
Composite Materials, 2002. 36(12): p. 1451-1479.

7.

Abedian, A. and W. Szyszkowski, Influence of the free surface on the thermal
stresses in unidirectional composites. Composites Part a-Applied Science and
Manufacturing, 1997. 28(6): p. 573-579.

8.

Adams, R.D. and M.M. Singh, Low temperature transitions in fibre reinforced
polymers. Composites Part a-Applied Science and Manufacturing, 2001. 32(6):
p. 797-814.

9.

D'Amore, A., et al., Long-term behaviour of PEI and PEI-based composites
subjected to physical aging. Composites Science and Technology, 1999. 59(13):
p. 1993-2003.

10.

Warrier, A., et al., The effect of adding carbon nanotubes to glass/epoxy
composites in the fibre sizing and/or the matrix. Composites Part a-Applied
Science and Manufacturing, 2010. 41(4): p. 532-538.
81

11.

Yuen, S.M., et al., Preparation and thermal, electrical, and morphological
properties of multiwalled carbon nanotube and epoxy composites. Journal of
Applied Polymer Science, 2007. 103(2): p. 1272-1278.

12.

Zhang, Y.C. and X. Wang, Thermal effects on interfacial stress transfer
characteristics of carbon nanotubes/polymer composites. International
Journal of Solids and Structures, 2005. 42(20): p. 5399-5412.

13.

Wei, C.Y., D. Srivastava, and K.J. Cho, Thermal expansion and diffusion
coefficients of carbon nanotube-polymer composites. Nano Letters, 2002. 2(6):
p. 647-650.

14.

Agag, T., T. Koga, and T. Takeichi, Studies on thermal and mechanical
properties of polyimide-clay nanocomposites. Polymer, 2001. 42(8): p.
3399-3408.

15.

Yasmin, A., et al., Mechanical and thermal behavior of clay/epoxy
nanocomposites. Composites Science and Technology, 2006. 66(14): p.
2415-2422.

16.

Gu, A.J. and F.C. Chang, A novel preparation of polyimide/clay hybrid films
with low coefficient of thermal expansion. Journal of Applied Polymer Science,
2001. 79(2): p. 289-294.

17.

Asif, A.A., et al., Surface morphology, thermomechanical and barrier
properties of poly(ether sulfone)-toughened epoxy clay ternary
nanocomposites. Polymer International, 2010. 59(7): p. 986-997.

18.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, Multifunctional polypropylene
composites produced by incorporation of exfoliated graphite nanoplatelets.
Carbon, 2007. 45(7): p. 1446-1452.

19.

Liao, K. and S. Li, Interfacial characteristics of a carbon nanotube-polystyrene
composite system. Applied Physics Letters, 2001. 79(25): p. 4225-4227.

20.

Drzal, L.T. and H. Fukushima, Expanded graphite and products produced
therefrom. 2004: US.

21.

Wong, C.P. and R.S. Bollampally, Thermal conductivity, elastic modulus, and
coefficient of thermal expansion of polymer composites filled with ceramic
particles for electronic packaging. Journal of Applied Polymer Science, 1999.
74(14): p. 3396-3403.
82

22.

Orrhede, M., R. Tolani, and K. Salama, Elastic constants and thermal
expansion of aluminum-SiC metal-matrix composites. Research in
Nondestructive Evaluation, 1996. 8(1): p. 23-37.

23.

Turner, P.S., THERMAL-EXPANSION STRESSES IN REINFORCED PLASTICS.
Journal of Research of the National Bureau of Standards, 1946. 37(4): p.
239-250.

24.

Schapery, R.A., THERMAL EXPANSION COEFFICIENTS OF COMPOSITE
MATERIALS BASED ON ENERGY PRINCIPLES. Journal of Composite Materials,
1968. 2(3): p. 380-&.

25.

Hashin, Z. and S. Shtrikman, A VARIATIONAL APPROACH TO THE THEORY OF
THE ELASTIC BEHAVIOUR OF MULTIPHASE MATERIALS. Journal of the
Mechanics and Physics of Solids, 1963. 11(2): p. 127-140.

26.

Hashin, Z., ANALYSIS OF COMPOSITE-MATERIALS - A SURVEY. Journal of
Applied Mechanics-Transactions of the Asme, 1983. 50(3): p. 481-505.

27.

Kirby, R.K., T.A. Hahn, and B.D. Rothrock, in American Institute of Physics
handbook, D.E. Gray, Editor. 1972, McGraw-Hill: New York. p. 4-119 to 4-142.

28.

Cho, J., J.J. Luo, and I.M. Daniel, Mechanical characterization of
graphite/epoxy nanocomposites by multi-scale analysis. Composites Science
and Technology, 2007. 67(11-12): p. 2399-2407.

29.

Chow, T.S., EFFECT OF PARTICLE SHAPE AT FINITE CONCENTRATION ON
THERMAL-EXPANSION OF FILLED POLYMERS. Journal of Polymer Science Part
B-Polymer Physics, 1978. 16(6): p. 967-970.

30.

Chow, T.S., EFFECT OF PARTICLE SHAPE AT FINITE CONCENTRATION ON
ELASTIC-MODULI OF FILLED POLYMERS. Journal of Polymer Science Part
B-Polymer Physics, 1978. 16(6): p. 959-965.

31.

Yoon, P.J., T.D. Fornes, and D.R. Paul, Thermal expansion behavior of nylon 6
nanocomposites. Polymer, 2002. 43(25): p. 6727-6741.

83

Chapter 5 Highly Aligned
Multifunctional Graphite
Nanoplatelet-Polyether imide
Composite in Film Form
Abstract
The objective of this research is to investigate the effect of adding graphite
nanoplatelet (GNP) particles into polyether imide film. Extruded films were produced
with various amounts of GNP and different degrees of post-extrusion stretching. A
uniform high degree of alignment of the GNP particles along both 0 degree
(stretching) direction and 90 degree direction was obtained under high stretching
ratios. Tensile properties including tensile modulus, strength and elongation were
investigated. O 2 permeability was reduced with the increase of GNP loading and
stretching ratio of the film. Bharadwaj’s model was applied to predict the
permeability. Information of degree of alignment was obtained through analyzing
SEM images in image-processing software and used with Bharadwaj’s model,
resulting in good agreement with experimental data. Annealing of the stretched films
under 340℃showed a significant improvement of both electrical and thermal
conductivity in the film direction.

84

Introduction
Polymer nanocomposite films have drawn more and more attention because of their
unique multifunctional properties and the possibility to be incorporated into other
structural or functional materials. Polyether imide (PEId, Ultem resin, Sabic IP) is a
thermoplastic material which has high chemical and thermal resistance, good flame
resistance and low smoke generation [1]. Major applications of PEId are as a
structural material for confined space, including high-floor office, airplane and
submarines. Just like the other thermoplastic based material, PEId can offer lighter
and more durable solutions as alternatives to metal and alloys, but falls short in
properties like thermal and electrical conductivity and barrier properties to small
molecules. Meanwhile, a group of nanofillers such as carbon nanotubes (CNT) [2],
nanoclay [3] and nano graphite [4] are considered to be efficient in providing such
multifunctionality to their composites. Among those fillers, the graphite nanoplatelet
(GNP) combines electrical and thermal conductivity similar to CNT and
two-dimensional morphology similar to nanosilicates, resulting in the capability of
meeting most of the requirements if incorporated into composites.
In a composite reinforced by anisotropic fillers, the alignment of the filler is usually
the key to performance of the composite. Lan et al [5] designed a clay-epoxy
thermoset composite with sub-ambient Tg to allow alignment of clay particles under
strain. Analogous to this system would be stretching the thermoplastic
nanocomposite material above Tg to obtain high alignment such as by a blown film
85

process. Zenkiewicz et al [6] reported the dependence of blowing ratio on the barrier
properties of clay/PLA film. It is found that higher blowing ratio would lead to lower
permeability by oxygen, water vapor and carbon dioxide. TEM images shown by the
author verified that the particles tend to be more aligned under higher blowing ratio.
However, not all the properties can benefit from alignment. Kim et al [7] proposed a
schematic that describes electrical conductivity changes due to the transition of
graphite particles from the aligned state to the un-aligned state. Similar results are
also discussed in Chapter 3. Adequate post-treatment would be necessary to alter
the alignment of fillers in order to achieve optimized and balanced performance.
The goal of this research is to fabricate unidirectional-stretched GNP/PEId film
through extrusion and film casting. Stretching ratios from low to high were applied to
obtain composite film with different degrees of alignment of the GNP particles.
Tensile properties, electrical and thermal conductivity and barrier to oxygen
transport have been measured. Annealing under high temperature was used to
adjust the degree of alignment in order to enhance electrical conductivity.

Experimentation
Materials
Polyetherimide (PEI, Ultem 1010) was provided by Sabic Americas, Inc. (Houston, TX).
GNP-5 was provided by XG Sciences Inc. (East Lansing, MI). The GNP particles have an
2

average diameter of 3.9 micron with a specific surface area of 40m /g.
86

Melt-extrusion and film casting
Melt-extrusion was carried on Leistritz twin-screw extruder (MIC27/6L-48D). The
screws were in the co-rotation mode. The shearing provided by co-rotating screws
can help separate the filler particles and result in a better dispersion than
counter-rotation. Barrel temperatures were set to 310℃ with a melt temperature at
340℃. The die pressure was at around 5.5MPa with screw speed set at 150 rpm.
GNP was fed at a side feeder at 1/3 screw length away from the main feeder where
the polymer was fed. The melt was extruded through a film die and then stretched
and collected by a three-roll film collector, whose rolls were heated to 160℃ by oil
to get uniform, wrinkle-free film. The stretching ratio is defined as in Figure 5.1:

Figure 5.1, Film extrusion, showing stretching ratio
Between the film die opening and the first nip point of the three roll stack, the film is
stretched in the melt or rubbery state. The distance between the die and the rolls is
an important factor which can affect the shape of the product, such as thickness and
width [8]. Also, since the film is exposed in air, a temperature gradient is developed
87

along the stretching direction as well. For this research, the distance was kept to a
minimum in order to get as thin a film as possible. The polymer chains are able to
move before reaching the rolls. For a crystalline polymer, this effect will also result in
rearrangement of crystalline and amorphous domains resulting in higher crystallinity
[9]. For PEId, since it’s amorphous, the stretching would only result in the orientation
of both chains and GNP particles.
In the experiment setup, the stretching ratio was controlled by fixing the material
feeding speed and changing the speed of the three roll stack. The stretching ratio (SR)
Die Opening Thickness (DOT) × Die Opensing Width (DOW)
Film Thickness (FT) × Film Width (FW)

then is calculated as:
SR =

The die opening width was 152 mm and the films were collected at 1, 4, 8 and 16
fpm three-roll speed. Film dimension and SR value are listed in Table 5.1.
Table 5.1, Dimension and stretching ratio of extrusion cast films
Roll speed
1
4
8
16
(fpm)
Die Opening
0.58
0.91
0.91
0.91
(mm)
Thickness (mm)
0.094±10
0.052±8
0.030±4
0.28±4
Width (mm)
144
136
132
127
SR
2
11
20
34

Annealing of Extrusion Cast Film
The extrusion cast film was also annealed under 340℃ in a hot press. To keep the
dimension to the same as the un-annealed sample, a piece of aluminum foil was
88

used as the mold. A square hole with the same dimension as the GNP/PEId film was
cut on the aluminum foil. The foil, along with the sample in the hole, was
sandwiched by two sheets of PTFE film and two steel plates. The whole set-up was
then placed in the pre-heated 340℃ hot press (CARVER Laboratory Press, Model
2731, Fred S. Carver INC.). A pressure of 23 MPa was applied immediately and held
for 1 hour. After that, water cooling was applied to the hot press to cool the sample.
The setup of the experiment is shown in Figure 5.2.

Figure 5.2, Experiment setup of annealing extrusion cast film
The change in thickness dimension after 1 hour annealing is shown in Table 5.2. Note
that the samples used for annealing were the central part of the extrusion cast film
with all the edges trimmed off, so the thickness is slightly different from Table 5.1.
Table 5.2, Thickness change after annealing for GNP/PEId
extrusion cast films
Stretching ratio

11

20

34

After annealing (µm)

105±5

58±2

34±2

113±8

61±5

36±4

Before annealing (µm)

89

Melt-extrusion and Compression Molding
GNP-5 and PEId were also melt-blended and extruded as strands with the same
twin-screw extruder setup as used for film and then cut into pellets. An aluminum
foil with a round hole cut in the center was used as mold. The composite pellets were
piled at the center of the hole and then compressed into ~150 µm thick films.

Electrical Conductivity Measurement
The resistance of GNP/PEI composite films was measured in two directions: the
through-plane direction which is across the thickness of the sample and in-plane
direction which is the stretching direction for extrusion cast films and the radius
direction for compression molded films (Figure 5.3). The resistance value was then
converted into resistivity by taking into account the sample dimensions.

90

Figure 5.3, Schematic of electrical resistance measurement

Thermal Diffusivity and Thermal Conductivity
Thermal diffusivity of the GNP/PEId composite films was measured by laser flash
method (LFA 447 NanoFlash, Netzsch) in through-plane direction. The heat capacity
Cp was determined by Differential Scanning Calorimetry (DSC, TA Instruments) and
the density of the samples was calculated by the rule of mixtures. Thermal
conductivity was calculated by multiplying thermal diffusivity by Cp and density.

Scanning Electron Microscope (SEM) Observation
SEM was used to observe the morphology of the samples at high vacuum mode. The
composite film samples were cut and mounted in a cast epoxy holder (manufacture:
LECO, St. Joseph, MI). Then the surface was polished with #4000 grit finish, 1 micron
alumina powder/water slurry on a LECO LP20 polisher and then 0.05 micron alumina
powder/water slurry on a Buehler Vibromet polisher (Lake Bluff, IL) for 2h. The
91

polished surface was etched by plasma in a chamber filled with O 2 at room
temperature and 0.283 torr. A 10nm gold layer was also sputter coated to the surface
for observation. For extrusion cast films, cross-sections of both stretching direction (0
degree) and 90 degree direction were observed (Figure 5.4).

Figure 5.4, Cross-sections that were observed by SEM. Dash-line indicates the
direction in which the film was cut.

Tensile Properties
Tensile properties including modulus, strength and elongation were tested under
ASTM D638 with a stretching rate of 0.1 in/in/min. A UTS SFM-20 machine (United
Calibration Corp.) was used for the testing. The extrusion cast films were cut into
Type V specimens with the geometry dimensions shown in Figure 5.5. Samples were
cut along both 0 degree and 90 degree directions.

92

Figure 5.5, Sample preparation for tensile modulus and strength testing

O 2 Permeability
O 2 permeability was measured by MOCON OXTRAN 2/20 model (MOCON, Inc.). The
2

composite film was sandwiched by two pieces of aluminum masks with a 1 cm

opening at the center. Permeation of O 2 was measured under a pressure of 30 psi
2

and recorded with a unit of cc/m /day. A testing cycle included the following steps:
(1) the O 2 detector was re-zeroed at the beginning of each cycle for 30 min with
purge of H 2 /N 2 mixture, (2) the O 2 detector was connected with testing cell A for 30
min, (3) the O 2 detector was connected with testing cell B for 30 min. Two cells were
tested alternatively in a single experiment. Multiple cycles with a typically 12-hour
total testing time was applied to all the samples in order to saturate the sample and
reach equilibrium state. Typical testing curve is shown in Figure 5.6. The permeability
p was then calculated by:

p=

Permeation × Film Thickness
Pressure
2

The unit used in this article is cc-m/m /day/Pa.
93

Figure 5.6, Measured permeation vs. testing time for compression molded neat PEId
film, showing the plateau after about 12 hours

Results and Discussions
Morphology of Extrusion Cast Films
The properties of the nanocomposites are highly dependent on the morphology,
especially for the composites filled by anisotropic fillers. Tandon and Weng used an
analytical model to demonstrate the difference in elastic modulus between
composites with randomly-oriented fillers and uni-directionally-oriented fillers [10,
11]. In Chapter 3, it was discussed that the injection molded GNP-5/PEId coupon had
a tensile modulus in between of Tandon-Weng’s unidirectional model and the
random model and SEM images showed that the GNP particles were aligned at the
near surface region but randomly oriented at the center of the coupon. One major
goal of fabricating extrusion cast film is to achieve a more uniform alignment through
the entire thickness of the sample, as can be seen in GNP/PEId composite films by
94

SEM observation (Figure 5.7).
Figure 5.8 shows the change of the particle alignment after annealing for GNP/PEId
extrusion cast films. All the films showed significant change in degree of alignment
after annealing. Image-Pro Plus software (Media Cybernetics, Inc.) was used to

gather information of the angle θ between GNP particle and 0 or 90 degree
directions, depending on different cross-sections. For example, in the cross-section of
0 degree direction, the alignment along the 90 degree direction is revealed, with

θ

the angle between a GNP particle and the 90 degree direction. On the other hand, in

the cross-section of the 90 degree direction, the alignment along 0 degree direction

is revealed, with θ being the angle between a GNP particle and the 0 degree

𝑎11 =< 𝑐𝑜𝑠 2 𝜃 >

direction. The degree of alignment was characterized by:

Values of

𝑎11

along different directions are shown in Figure 8. It can be observed

that with the increase of stretching ratio, the GNP particles are more aligned in the
PEId matrix. However, the alignment appears to reach a maximum at SR=20. Higher
stretching ratio (SR=34) cannot further align the particles. After 1 hour annealing,
most of the samples showed the same degree of alignment in both 0 and 90 degree
directions.

95

Figure 5.7, SEM images of cross-sections of extrusion cast film, showing GNP particle
alignment. The bubbles in the images are on the epoxy putty region, not a feature of
the film sample. (scale bars: (a)(e)~(h): 10 micron, (b)~(d): 2 micron)
96

Figure 5.8, SEM images of cross-sections of annealed extrusion cast film, showing the
change in GNP particle alignment (scale bars represent 2 micron)
97

Figure 5.9, Degree of alignment of GNP particles in extrusion cast films
For the film with SR=11, the comparison between Figure 5.7(a)(b) and Figure 5.8(a)(b)
indicates that the re-orientation of the particles happens more significantly at close
to film surface than the center. This phenomenon implies that the particles close to
the surface can rotate more freely than the ones at the center of the film. It is
noticed that the re-oriented particles can connect with each other, although the
change in degree of alignment is not dramatic, from 0.9 to around 0.75 which can
still be considered as preferential alignment (Figure 5.9).

Tensile Properties
Polymer material can be categorized into two types: ductile polymer and brittle
polymer. The former would exhibit necking behavior after yielding while the latter
would fail before yielding. Neat PEId film shows typical ductile behavior with
generation of necking phenomenon and undergoes a long cold draw before failing
(Figure 5.10(a), Figure 5.12(a)). With the presence of GNP particles, the composite
98

film exhibits more like brittle material, which fails without necking. Interestingly, in
Figure 5.10(b), it seems that the sample reached yielding but failed immediately.
Other tested samples showed similar behavior for SR=11 and SR=20 composite films,
indicating that the yielding of the polymer results in the failure of the sample, and no
necking probably means that the confinement of the polymer chains by the GNP
restricts polymer chain re-arrangement. The failure of the film might be initiated by
the GNP particles that are not perfectly aligned or well bonded to the matrix.
However, when the stretching ratio is 34, the film performs as a ductile material in
the 0 degree direction. In Figure 5.12, the elongation at break of SR=34 film is
significantly higher than its yield elongation and the elongation of other samples. The
stress in Figure 5.10(c) reaches a maximum at around 3.7% strain and decreased
slightly afterwards, but remains close to the maximum until fracture. This indicates
that the presence of GNP particles results in enhanced strength of the composite
after the yielding of the matrix. However, such behavior was not observed in the 90
degree direction. The film broke at the same time of yielding, just like SR=11 and
SR=20 films (Figure 5.12(b)).

99

stress/MPa

Stress-Strain Curve of Neat PEId Film
SR=11
Yield point
100
50

Break
point

0
0

20

40

60

80

strain %
(a)

stress/MPa

Stress-Strain Curve of 10wt% GNP5/PEId Film SR=20
150

Yield point

100
Break
point

50
0
0

1

2

3

4

5

strain %
(b)

stress/MPa

Stress-Strain Curve of 10wt% GNP5/PEId Film SR=34
Yield point

150
100
50

Break point

0
0

2

4

6

8

strain %
(c)
Figure 5.10, Typical stress-strain curve for extrusion cast films
100

10

From the morphology study, it can be observed that the alignment of GNP particles is
very uniform in the composite film. Meanwhile, since the film was stretched under
relatively high temperature, the amorphous polymer chain can also align along the
stretching direction, making the neat polymer to perform like fiber with transverse
isotropic properties [12, 13].
Figure 5.11 shows the tensile modulus and strength of GNP-5/PEId films. Both
modulus and strength show anisotropic behavior in 0 and 90 degree directions. For
modulus, the difference is not obvious for neat PEId film but very significant for
composite films. For strength, the difference becomes more significant with the
increase in stretching ratio. The highest stretching ratio results in lower modulus and
strength, especially for SR=34, which indicates that the film samples might have been
over-stretched during the film casting process. Figure 5.12(b) also implies
over-stretching with high stretching ratio, showing that the elongation at yield
decreases with the increase of the stretching ratio.

101

(a)

(b)
Figure 5.11, (a) Tensile modulus (b) tensile strength of GNP-5/PEId composites films.
According to Figure 10, the strength value shown here is the strength as yield
For 10wt% GNP-5/PEId film with an SR=11, a 134% improvement in modulus and a 9%
increase in strength were recorded in the 0 degree direction while the improvement
in 90 degree direction were 79% and 3% respectively. Since the degree of alignment
for the GNP particles is the same in both directions (Figure 5.9), the cause of the
102

difference can be attributed to the polymer chain alignment. Considering that the
neat polymer did not exhibit such large difference in the two directions, the presence
of GNP might also increase the extent that the polymer chains get aligned.

(a)

(b)
Figure 5.12, Elongation of film during tensile testing (a) neat PEId film (b) GNP-5/PEId
composite film

103

(a)

(b)
Figure 5.13, Normalized tensile modulus and tensile strength, showing comparison
between injection molded coupon and films made from various stretching ratios (a)
modulus (b) Strength
When compared to injection molded composite coupons (Chapter 3), the stretched
films show higher modulus and strength due to better alignment. Due to the fact that
the injection molded coupon and the film were measured with different sample
dimensions, the values are normalized by the modulus and strength of the neat
samples that were fabricated under the same condition. An ascending trend can be
104

observed from injection molded samples to films with higher stretching ratio for both
modulus and strength, except for SR=34 films possibly due to severe overstretching
(Figure 5.13). This trend indicates the benefit of particle alignment on both tensile
modulus and tensile strength.

O 2 Permeability
Polymer and composite membranes have been widely used as either barrier to small
molecules or selective separator for gases [14-16]. The morphology of the filler is the
deciding factor for the change of the permeability. Unless specially modified to
enhance the bonding between filler and matrix, spherical particles like silica would
actually enlarge the permeability due to poor interface between filler and matrix and
can be used as gas separators due to selectivity to gas types [17-19]. However, Amino
silane-modified clay resulted in reduced permeability to gases for the same matrix
because of chemical coupling between filler and matrix [18, 20]. Nunes [20],
Takahashi [17, 18] and Kim [19] suggested that the dispersion state of the
nanoparticles is the key to the barrier properties. For platelet-shaped fillers such as
layered-silicate and graphite, the gas permeability of their composites is usually
reduced, due to the tortuous zigzag path forced by impenetrable platelet fillers [15,
21]. The alignment and aspect ratio of the filler are two major factors for the

𝑝𝑐

=

1−𝜙

permeability of the composites as indicated by Bharadwaj’s model [22]:

𝑝𝑚

𝜌
1+
2

𝜙<𝑐𝑜𝑠 2 𝜃>
105

(1)

Where

𝑝𝑐

and

𝑝𝑚

are the permeability of composite and neat matrix polymer

respectively. The ratio between is determined by filler volume fraction 𝜙, the aspect
ratio of the filler 𝜌 and the degree of alignment

𝑎 =< 𝑐𝑜𝑠 2 𝜃 >. It can be seen

from Equation (1) that the higher aspect ratio and alignment lead to lower
permeability of the composite.

“model” was calculated by assuming perfect alignment 𝑎 = 1. Series “model with
alignment” was calculated using the 𝑎 value shown in Figure 5.15.

Figure 5.14, Relative O 2 Permeability of compression molded GNP-5/PEId film. Series

106

Figure 5.15, SEM images of compression molded GNP-5/PEId film, showing
alignment of GNP particles. The a values shown in the images are averages from
images taken at different position of the sample (Scale bars represent 20 micron)
Figure 5.14 indicates that the O 2 permeability decreases with the increase of GNP-5
content. With 10wt% GNP-5 added, the permeability of the composite film is
reduced to about 50% of that of the neat polymer. The aspect ratio of the GNP-5
particles is set to 47 according to Chapter 3. In Figure 5.14, if perfect alignment is
assumed, the model prediction is over-estimating the effect of the filler. Figure 5.15
shows the SEM images of the cross-section of compression molded films. The degree
of alignment is used in equation (1) and the corresponding predicted value is shown
in Figure 5.14 as “model with alignment”. It can be seen that this model correlates
with the experimental data very well.

107

Figure 5.16, Relative O 2 permeability of GNP-5/PEId extrusion cast film
Figure 5.16 shows the relative O2 permeability of extrusion cast film with various
stretching ratios. With the increase of stretching ratio, the permeability decreases.
About 65% decrease is observed for the film with SR=34. The series “model” is
calculated from Equation (1) with an aspect ratio of 47 and 𝑎 values in Figure 5.9. It

can be observed that the model proposed by Bharadwaj predicts the behavior of the
stretched GNP/PEId films very well.

Effect of Annealing
Figure 5.8 and Figure 5.9 show the change of GNP particle alignment after 1 hour 340℃
annealing. The re-arrangement of the GNP orientation helps the connection between
particles thus enhance both electrical and thermal conductivity for the composite.

108

(a)

(b)
Figure 5.17, Effect of annealing on electrical resistivity of GNP-5/PEId extrusion cast
films (a) through-plane direction (b) in-plane direction
Figure 5.17 indicates that 1 hour annealing would reduce the resistivity of the
composite film significantly in both through-plane and in-plane directions. About 4
orders of magnitude decrease was observed for all samples.

109

Figure 5.18, Effect of annealing on thermal conductivity of GNP-5/PEId extrusion cast
film
The same effect can also be observed for thermal conductivity. Figure 5.18 shows a
increase of 173% in the through-plane thermal conductivity for 10wt% composite
films. The reason of the improvement is that the GNP particles re-oriented after
annealing (Figure 5.8, 5.9). Figure 5.8 clearly shows that the particles are less aligned
in the in-plane direction after annealing, acting as better heat conductors in the
through-plane direction for the film. Figure 5.9 shows significant decrease of degree
of alignment in the in-plane direction, which means an increase of that in the
through-plane direction. Since the GNP particle has very high anisotropic thermal
conductivity (high in-plane, low through-plane), the alignment change resulted in
significant improvement in through-plane conductivity after annealing.

Conclusions
Neat PEId and GNP-5/PEId films have been successfully fabricated via extrusion cast
110

and unidirectional stretching. The alignment of the GNP particles formed by
extrusion cast fabrication benefited the modulus and strength as compared with
injection molded coupons. The addition of GNP particles changed the polymer from
ductile to brittle type material. However, with SR=34, the composite exhibited ductile
behavior during tensile testing with less strength drop after yielding during necking
process. The unidirectional stretched film behaved anisotropically. For the 10wt%
GNP-5/PEId film with SR=11, a 134% improvement in modulus and a 9% increase in
strength were recorded in the 0 degree direction while the improvement in the 90
degree direction was 79% and 3% respectively. A 50% decrease of O2 permeability
has been observed for compression molded 10wt% GNP-5/PEId film while a 65%
decrease was recorded for extrusion cast film at the same loading. It was found that
Bharadwaj’s model can predict the permeability for GNP-5/PEId system very well if
incorporated with the alignment information of GNP particles obtained by SEM
observation. Annealing was effective to re-orient the GNP particles and provide
connection between them, resulting in improved electrical and thermal conductivity.
However, permeability may also be increased due to less aligned particles after
annealing as indicated by Bharadwaj’s model (Equation 1). SEM images showed that
the degree of alignment dropped from around 0.9 to around 0.76. Due to the good
barrier property and improved electrical and thermal conductivity, the GNP film has
high potential for use where multifunctional properties are required.

111

REFERENCES

112

REFERENCES
1.

http://www.sabic-ip.com. Last visited: March 2011

2.

Bauhofer, W. and J.Z. Kovacs, A review and analysis of electrical percolation in
carbon nanotube polymer composites. Composites Science and Technology,
2009. 69(10): p. 1486-1498.

3.

Choudalakis, G. and A.D. Gotsis, Permeability of polymer/clay nanocomposites:
A review. European Polymer Journal, 2009. 45(4): p. 967-984.

4.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, Multifunctional polypropylene
composites produced by incorporation of exfoliated graphite nanoplatelets.
Carbon, 2007. 45(7): p. 1446-1452.

5.

Lan, T. and T.J. Pinnavaia, CLAY-REINFORCED EPOXY NANOCOMPOSITES.
Chemistry of Materials, 1994. 6(12): p. 2216-2219.

6.

Zenkiewicz, M., J. Richert, and A. Rozanski, Effect of blow moulding ratio on
barrier properties of polylactide nanocomposite films. Polymer Testing, 2010.
29(2): p. 251-257.

7.

Kim, H. and C.W. Macosko, Processing-property relationships of
polycarbonate/graphene composites. Polymer, 2009. 50(15): p. 3797-3809.

8.

Shiromoto, S., et al., A Neck-in Model in Extrusion Lamination Process.
Polymer Engineering and Science, 2010. 50(1): p. 22-31.

9.

Lee, S.Y., S.Y. Park, and H.S. Song, Lamellar crystalline structure of hard elastic
HDPE films and its influence on microporous membrane formation. Polymer,
2006. 47(10): p. 3540-3547.

10.

Tandon, G.P. and G.J. Weng, THE EFFECT OF ASPECT RATIO OF INCLUSIONS ON
THE ELASTIC PROPERTIES OF UNIDIRECTIONALLY ALIGNED COMPOSITES.
Polymer Composites, 1984. 5(4): p. 327-333.

11.

Tandon, G.P. and G.J. Weng, AVERAGE STRESS IN THE MATRIX AND EFFECTIVE
MODULI OF RANDOMLY ORIENTED COMPOSITES. Composites Science and
Technology, 1986. 27(2): p. 111-132.

12.

Murakami, S., et al., A STUDY ON THE STRUCTURAL-CHANGES DURING
UNIAXIAL
DRAWING
AND
OR
HEATING
OF
POLY(ETHYLENE
NAPHTHALENE-2,6-DICARBOXYLATE) FILMS. Polymer, 1995. 36(2): p. 291-297.
113

13.

Matsuo, M., et al., Characteristics of ultradrawn polyethylene films as a
function of temperature estimated by the positron annihilation lifetime
method. Macromolecules, 2002. 35(8): p. 3059-3065.

14.

Stern, S.A., POLYMERS FOR GAS SEPARATIONS - THE NEXT DECADE. Journal of
Membrane Science, 1994. 94: p. 1-65.

15.

LeBaron, P.C., Z. Wang, and T.J. Pinnavaia, Polymer-layered silicate
nanocomposites: an overview. Applied Clay Science, 1999. 15(1-2): p. 11-29.

16.

Giannelis, E.P., Polymer-layered silicate nanocomposites: Synthesis, properties
and applications. Applied Organometallic Chemistry, 1998. 12(10-11): p.
675-680.

17.

Takahashi, S. and D.R. Paul, Gas permeation in poly(ether imide)
nanocomposite membranes based on surface-treated silica. Part 1: Without
chemical coupling to matrix. Polymer, 2006. 47(21): p. 7519-7534.

18.

Takahashi, S. and D.R. Paul, Gas permeation in poly(ether imide)
nanocomposite membranes based on surface-treated silica. Part 2: With
chemical coupling to matrix. Polymer, 2006. 47(21): p. 7535-7547.

19.

Kim, J.H. and Y.M. Lee, Gas permeation properties of poly(amide-6-b-ethylene
oxide)-silica hybrid membranes. Journal of Membrane Science, 2001. 193(2):
p. 209-225.

20.

Nunes, S.P., et al., Membranes of poly(ether imide) and nanodispersed silica.
Journal of Membrane Science, 1999. 157(2): p. 219-226.

21.

Yano, K., et al., SYNTHESIS AND PROPERTIES OF POLYIMIDE CLAY HYBRID.
Journal of Polymer Science Part a-Polymer Chemistry, 1993. 31(10): p.
2493-2498.

22.

Bharadwaj, R.K., Modeling the barrier properties of polymer-layered silicate
nanocomposites. Macromolecules, 2001. 34(26): p. 9189-9192.

114

Chapter 6 Dispersion Optimization of
Exfoliated Graphene Nanoplatelet in
Polyetherimide Nanocomposites:
Extrusion vs. Pre-Coating vs. Solid
State Ball Milling
Abstract
Polyetherimide (PEId) nanocomposites reinforced with high stiffness and
electrically conductive exfoliated graphite (graphene) nanoplatelets (GNP, ~10nm
thick and ~1-15 microns in diameter) were fabricated by various processing methods
to achieve good dispersion including: melt-extrusion, pre-coating of the polymer
powder with GNP , solid state ball milling as well as combinations of these methods.
GNP compounded with PEId by melt extrusion have a high percolation threshold
(~10wt%). In order to lower the percolation threshold, a pre-coating method using a
non-solvent was used, in which the GNP particles are coated onto the PEId polymer
powder surface by surface deposition from a non-solvent. As a result of this
processing approach, the electrical conductivity is greatly increased with a
percolation threshold as low as 2wt%. However, the electrical conductivity
improvement comes at the expense of only a very small increase in flexural modulus
115

and a reduction in flexural strength to a value lower than the neat polymer. Solid
state ball milling (SSBM) was investigated as an alternative process to enhance
dispersion, adhesion and to reduce GNP size. High electrical conductivity and
improved modulus were achieved by this approach along with better mechanical
properties. Further improvements to the mechanical properties could be made by
combining the extrusion and SSBM approaches. Examination of the nanocomposite
morphology explains the effect of these combined compounding approaches on GNP
particle dispersion and their relation to the improved GNP/PEId nanocomposite
performance.

Introduction
Nanocomposites are defined as composite materials with at least one Gibbsian solid
phase where at least one dimension is in the nanometer range (1-100 nm) [1].
Graphene based nanoparticles, which combine good electrical and thermal
properties similar to CNTs and the two-dimensional morphology similar to
silica-alumina

layered

clays,

are

considered

to

be

potentially

excellent

multifunctional nanoparticles. The process of generating thin graphene nanoplatelets
consisting of several layers of graphene sheets is relatively simple and cost-effective.
Usually, three steps are involved in fabricating these nanoparticles from natural
graphite: (i) intercalation, where both electron-donor agents and electron acceptors
can be intercalated in between graphene layers; (ii) exfoliation, where a large
amount of heat is applied to expand the intercalated agents; (iii) pulverization, where
116

the exfoliated graphite is down-sized to smaller platelets. As demonstrated by Drzal’s
group [2, 3], exfoliated graphite nanoplatelets (GNP) can be fabricated from acid
intercalation, microwave heating and ultra-sonic pulverization. Cost-effective
industrial scale manufacturing can be attained by this route (XG Sciences,
Inc., www.xgsciences,com, East Lansing, MI). The GNP nanoplatelet lateral width is
controllable during fabrication. Typical sizes of the nanoplatelets include 1 micron, 5
micron, and 15 micron in diameter and usually notated as GNP-1, GNP-5 and GNP-15
respectively.
The properties of a nanocomposite are highly affected by its compounding method
which may alter the distribution, separation and orientation of the nanoparticles.
Typical compounding methods for thermoplastic matrix include [4]: (1) melt
compounding, such as high density polyethylene (HDPE) and exfoliated graphite
mixed in a Brabender mixer [5, 6] or twin-screw extrusion [7]; (2) solution mixing,
where a solvent is used to dissolve the polymer and then the graphite is added and
dispersed [8, 9]; (3) in-situ polymerization, where monomers are usually intercalated
in between graphite layers followed by polymerization which at the same time
exfoliates the graphite [10, 11].
While it is convenient in industrial application, melt-extrusion falls short of attaining
the electrical conductivity at low percolation threshold of the resulting composite [4,
12]. Solution mixing is capable of separating and dispersing nanoparticles which
results in lower percolation threshold [13], but the use and recovery solvent is
problematic from an environmental and cost perspective. In-situ polymerization can
117

only be applied to a limited category of polymers. Efforts have been made to find a
practical and efficient way to disperse the graphene nanoplatelets in order to achieve
lower percolation threshold. Some researchers used unsaturated polyester resin
(UPR) to modify expanded graphite [14] followed by melt-extrusion which resulted in
a percolation threshold at 5.7wt%, instead of 22.2wt% for composites with
conventional graphite. Recently, Drzal’s group[4] developed a ‘pre-coating’ method
to achieve low percolation threshold where a non-solvent of the polymer is used to
disperse GNP particles. After adding polymer powder into this suspension, the GNP
particles can deposit on the polymer surface resulting in a controllable, continuous
coating on the surface of the polymer powder. Subsequent compression molding
results in deformation and consolidation of the polymer but the continuous GNP
phase remains intact after the molding process forming a low concentration
percolated network.
The main purpose of this research is to find an alternative compounding method for
adding GNP to amorphous high temperature thermoplastics, such as polyetherimide
(PEId) which has a high glass transition temperature, extremely good chemical
resistance and excellent thermal stability, in order to achieve better nanocomposite
mechanical properties and a low percolation threshold. Three compounding
methods were used and compared. First, the GNP/PEId nanocomposites were
fabricated from melt-extrusion. The flexural modulus and strength were measured
along with electrical resistivity and the percolation threshold. To further improve
electrical conductivity and lower the percolation threshold, the pre-coating method
118

using a non-solvent (acetone) was used. The third method, solid state ball milling
(SSBM) of the dry mixture of PEId and GNP was investigated, where the GNP and
PEId are combined in a mill along with metal balls in a vibrating chamber at room
temperature (far below the Tg of the PEId), as an alternative process to enhance the
adhesion between PEId and GNP particles and at the same time to reduce particle
size. This solvent-free method is believed to be a simple, cost-effective method
having advantages over the pre-coating method to fabricate composites with low
percolation threshold.

Experimentation
Materials
Polyetherimide (PEId) was obtained from GE Plastics (Ultem 1010, in two forms:
pellet and powder). GNP-1 (particles with a diameter of 1 micron) and GNP-15
(particles with a diameter of 15 micron) were prepared in our lab [2].

Melt-extrusion
Melt-extrusion was carried on Leistritz twin-screw extruder (MIC27/6L-48D). The
screws were operating in the counter-rotation mode. Barrel temperatures were set
to 330℃, with a melt temperature at 340℃. The die pressure was at around 5.5MPa.
Screw speed was set at 100 rpm. The pellet product obtained from melt-extrusion
was pulverized into powder by a cryogenic mill.

119

Pre-coating
GNP was first dispersed in acetone under ultrasonication. PEId powder (~300 microns
in diameter) was then added into the suspension while stirring. Acetone was then
evaporated. The mixture was further mixed in a Speed Mixer (Flextec Corporation).
The sample was dried in a vacuum oven at 120℃.

Solid State Ball Milling (SSBM)
SSBM was carried out in SPEX SamplePrep 80000 Dual Mixer/Mill. 10 gram sample
(PEId and GNP) was placed in a cylindrical hardened steel vial (5.7 cm diameter, 7.62
cm long) with four 6.35 mm and two 12.7 mm steel balls. During processing, two
vials were vibrated at high frequency in the device. The processing time was 200
minutes. Both neat polymer powder and extrusion-compounded composite powder
were used in the vial with or without additional GNP. The resulting material after
SSBM is in flake-like shape with a typical size of 150 micron in diameter and a few
microns in thickness (Figure 6.1). The particle shown in Figure 1 consists of several
smaller particles sitting on a larger plate, indicating a wide size distribution and some
extend of re-agglomeration of the particles.

120

Figure 6.1, SEM image of GNP-PEId composite after SSBM, showing geometry of the
particles (Scale bar represents 50 micron)

Compression Molding
After compounding, the composite material was compression molded in a CARVER
Laboratory Press (Model 2731, Fred S. Carver INC.). A picture-framed mold was used
with a dimension of 76mm × 76mm × 3mm. The material was heated to 340℃
and held for 15 min. A compression pressure of 20MPa was applied to the material
and was held for 45 min in order to achieve better consolidation. After that, the
press was cooled by water.

Electrical Resistivity Measurement
The resistance of GNP/PEId composites was measured in two directions: the
through-plane direction (thickness direction) which was the same direction as the
applied pressure in compression molding and the in-plane direction which was
perpendicular to pressure. The resistance value was then converted into resistivity by
taking into account the sample dimensions.
121

Flexural Strength and Modulus Testing
Flexural strength and modulus were tested under ASTM 790 standard on a UTS
SFM-20 machine (United Calibration Corp.) at room temperature. The test was
performed at a flexural rate of 0.03 in/min. Maximum force, instead of force at 5%
strain, was used to calculate the strength of the composite.

Scanning Electron Microscope (SEM) Observation
A JEOL JSM-6400 (Tokyo, Japan) SEM was used to observe the morphology of the
samples. The powder form samples were observed directly with a 10nm gold layer
that was sputter coated. The composite flex bar samples were cut and mounted in a
cast epoxy holder (manufacture: LECO, St. Joseph, MI). Then the surface was polished
with #4000 grit finish, 1 micron alumina powder/water slurry on a LECO LP20
polisher and then 0.05 micron alumina powder/water slurry on a Buehler Vibromet
polisher (Lake Bluff, IL) for 2h. The polished surface was etched by plasma in a
chamber filled with O 2 at room temperature and 0.283 torr. A 10nm gold layer was
also sputter coated to the surface for observation.

Results and Discussions
Electrical Conductivity
Figure 6.2 shows the electrical resistivity of the GNP/PEId composites. The resistivity
in the in-plane direction is always lower than that in the through-plane direction due
122

to the particle anisotropy and alignment during compression molding (Figure 6.3 (a)).
The percolation threshold of melt-extrusion processed nanocomposites is higher
than 10wt% (about 5 vol%). This is mainly because during melt-extrusion the high
polymer melt viscosity prevents the nanoparticles from forming a percolated
network (Figure 6.3 (b)). While the GNP particles were homogeneously distributed
into the polymer matrix during melt-extrusion, the stacked layers of graphene
formed by the platelet-shape nature could not be separated efficiently during
extrusion, reducing the number of effective nanoparticles. The size reduction during
melt-extrusion (Figure 6.3 (b)) also affects electrical conductivity because of the
decrease of aspect ratio. Encapsulation of the GNP by the polymer matrix prevents
the GNP particles from touching each other resulting in a lower conductivity [15].
Mixing the GNP and PEId in a solvent (dichloromethane) resulted in improved
conductivity and reduced the percolation threshold ~5wt%. One possible explanation
is that in a solvent system, the viscosity is much lower than in a melt system, so the
polymer chains move freely and are able to permeate into the graphite layers,
preventing the graphene sheets which are separated by ultrasonication from
re-stacking. The net result is that the concentration of effective particles would then
be much more than in melt-extrusion.

123

(a)

(b)
Figure 6.2, Comparison of electrical resistivity of GNP/PEId composites from different
compounding methods: (a) in-plane direction (b) through-plane direction.
Pre-coating compounding and SSBM both produce nanocomposites with a low
percolation threshold at lower than 3wt%. Composites from SSBM show the lowest
percolation threshold at 1wt% (the relatively larger error bar is due to that the 1wt%
loading is just around percolation threshold where a little variation inside the sample
124

can result in a large difference in electrical resistivity), the electrical conductivity of
which is greater that the nanocomposites produced by pre-coating in the
through-plane direction but falls short in the in-plane direction at higher loadings.

(a)

(b)

(c)
(d)
Figure 6.3, SEM images of GNP/PEId composites (a) polished surface of the cross
section of 3wt% GNP-1/PEId composite from pre-coating compounding, showing
orientation and long continuous graphite phase (white lines) (b) 3wt% GNP-15/PEId
composite from melt-extrusion compounding, showing size reduction and
distribution of GNP particles (c) polymer powder surface after pre-coating with 3wt%
GNP-1 (d) polymer powder surface after SSBM with 3wt% GNP-1. (scale bars: (a) 50
micron, (b)~(d): 2 micron)
This result indicates that the SSBM process has a similar effect on forming a
continuous conductive phase in the polymer matrix by forming a complete
125

continuous coating of GNP on the polymer powder surface (Figure 6.3 (c)(d)). A few
shiny edges in the SEM image indicate a thick and uniform coating of GNP particles
has been produced. Composites with GNP-15 show better conductivity after
percolation because the nanoparticle higher aspect ratio provides a continuous
conductive phase with fewer contact points and lower contact resistance.

Flexural Properties
The flexural modulus and strength of GNP/PEId nanocomposites made by pre-coating
and melt-extrusion compounding are shown in Figure 6.4 and Figure 6.5, respectively.
5wt% GNP-15/PEId composite made by melt-extrusion shows a 30% increase in
flexural modulus while the same loading composite made by pre-coating
compounding only shows an improvement of 13% in modulus. This indicates that
during pre-coating, GNP-15 particles wrapped around the polymer powder particles
form a thick layer of graphite. Due to the large aspect ratio, the particles tend to
overlap each other. The polymer cannot flow into the graphite layers in the following
compression molding process, resulting in fewer effective GNP particles available to
increase the matrix modulus. Composites with GNP-1 from the same pre-coating
process show slightly better modulus than the ones made with GNP-15 because of
less overlapping of the nanoparticle particles.
Flexural strength is highly dependent on the defects in the sample. Melt-extrusion
separates and surrounds GNP particles from each other with polymer and provides
the composite with the highest strength which reaches a maximum at 3wt%. The
126

strength starts to drop at higher than 3wt% because of the increase of the number
and the size of defects created by agglomerates of GNP. Also, due to the confinement
of GNP particles, the polymer matrix cannot elongate to the intrinsic strain limit of
the neat polymer. Note that the maximum force in flexural testing was used here to
calculate the strength, instead of the force at 5% strain which is suggested by ASTM
790. The reason is, composites made from pre-coating or SSBM are usually brittle
and could not reach 5% strain. Pre-coating results in very poor strength because of
the continuous phase of graphite acts as a weak point in the composite. The
graphite-rich area fails at low loads more easily than the polymer matrix.
With the same amount of GNP-1, SSBM composites always show a higher modulus
and strength than the pre-coated GNP nanocomposites. SEM images (Figure 6.3
(c)(d)) indicate that the powder from SSBM process has a surface morphology where
GNP particles are embedded into the polymer powder which results in better
adhesion between polymer matrix and GNP during the compression molding process.
This is due to the fierce high energy generated during SSBM from the collisions
resulting in partial melting of the polymer. SSBM is also very efficient in reducing the
size of the polymer powder. Smaller-sized polymer powder means a larger surface
area for the same amount of material, resulting in thinner graphite layers on each
polymer powder particle surface and better dispersion.

127

Figure 6.4, Flexural modulus of GNP/PEId composites

Figure 6.5, Flexural Strength of GNP/PEId composites

Combining Melt-Extrusion with SSBM
Melt-extrusion distributes the GNP particles very well in order to achieve high
flexural properties but it is not a sufficient way to prevent the particles from
re-stacking. However, very small amounts of GNP added in the SSBM process are able
to achieve electrical percolation in the nanocomposite. Figure 6.6 shows the
electrical resistance, flexural modulus and strength for the composites produced by
128

the two-step compounding approach, combining melt-extrusion and SSBM. The
x-axis in Figure 6.6 indicates the amount of GNP-15 used in melt-extrusion step,
while the three series with different patterns indicate the difference in the SSBM step.
With 1wt%

additional GNP-1 added in the SSBM step, all the composites produced

by melt-extrusion compounding all percolated with a higher modulus and a much
smaller drop of strength as compared to the neat polymer than the composites with
similar resistivity made from pre-coating or SSBM compounding approaches.

Figure 6.6, Properties of GNP/PEId composites from two-step compounding
129

(c)

(d)
Figure 6.7, Properties of GNP/PEId composites from two-step compounding (cont.):
melt-extrusion with GNP-15 then SSBM with GNP-1: (a) in-plane electrical resistivity
(b) through-plane electrical resistivity (c) flexural modulus (d) flexural strength
Figure 6.7 shows the combined morphology of this kind of composite. The right side
of the image is typical morphology of the extrusion compounded nanocomposite.
The GNP particles are distributed homogeneously in the bulk while separated by
polymer matrix. This morphology is beneficial for mechanical properties. The left side
of the image shows the coating morphology. The relatively smooth surface indicates
130

a uniform coating layer of GNP particles. The boundary between the left and right
side is a layer of graphite phase formed by coating. This kind of coating morphology
also indicates a continuous phase for the conductive nanoparticles formed during
SSSB process where the GNP particles were coated on the polymer powder surface.
The composite made from this process can attain percolation at lower concentrations
because of the continuous conductive phase.
Besides coating of higher concentrations of GNP-1 particles, GNP and PEId size
reduction during SSBM also aids melt-extrusion-compounding achieve lower
percolation in the nanocomposite. Areas containing higher amounts of fragile GNP
agglomerates can be broken down to individual particles during collision of polymer
powder particles in SSBM resulting in better separation and dispersion of GNP. Also,
with the size reduction, the GNP particles that are encapsulated in polymer matrix
are now exposed to the surface. (Figure 6.8) After SSBM, the anisotropic composite
particles usually have a diameter of around 150 microns with a thickness of about 10
microns (Figure 6.1) which is close to the size of GNP-15. The GNP exposed on the
surface during SSBM will form a continuous conductive phase, resulting in a
percolated GNP-PEId nanocomposite.

131

(a)
(b)
Figure 6.8, SEM image of 3wt% GNP/PEId extrusion compounded + 1wt% GNP-1 in
SSBM step, showing two kinds of morphology: extrusion compounded GNP particles
embedded in polymer matrix and SSBM coated GNP particles lying on the surface: (a)
liquid nitrogen fractured cross-section, left side showing coating morphology while
right side showing extrusion morphology (b) polished cross-section, discrete particles
showing extrusion morphology while continues line showing coating morphology
(Scale bars: (a) 100 micron, (b) 5 micron)

Figure 6.9, Schematics of the effect of SSBM on the composite powder made from
melt-extrusion

Conclusions
A new efficient, solventless compounding method, solid state ball milling, has been
identified as a viable new processing method for nanocomposites. This method
132

succeeds in simultaneously lowering the percolation threshold by forming a
continuous GNP coating on the surface of size-reduced polymer powder with good
adhesion and achieves excellent mechanical properties. Composites fabricated with
the SSBM approach showed percolation threshold similar to pre-coating but with
better mechanical properties and without using solvent. The morphology of the
GNP-PEId nanocomposites identified the results of different compounding methods.
While melt-extrusion tends to form discrete GNP phases by separating and reducing
the size of the GNP particles, pre-coating and SSBM succeed in forming a continuous
phase of the conductive GNP nanoparticles.

133

REFERENCES

134

REFERENCES
1.

Komarneni, S., NANOCOMPOSITES. Journal of Materials Chemistry, 1992.
2(12): p. 1219-1230.

2.

Fukushima, H., Graphite Nanoreinforcements in Polymer Nanocomposites, in
Department of Chemical Engineering and Materials Science. 2003, Michigan
State University: East Lansing.

3.

Kalaitzidou, K., Exfoliated Graphite Nanoplatelets as Reinforcement for
Multifunctional Polypropylene Nanocomposites, in Department of Chemical
Engineering and Materials Science. 2006, Michigan State University: East
Lansing.

4.

Wang, Y.S., M.A. Ogurkis, and J.T. Lindt, ELECTRICAL-PROPERTIES OF
EXFOLIATED-GRAPHITE FILLED POLYETHYLENE COMPOSITES. Polymer
Composites, 1986. 7(5): p. 349-354.

5.

Krupa, I. and I. Chodak, Physical properties of thermoplastic/graphite
composites. European Polymer Journal, 2001. 37(11): p. 2159-2168.

6.

Zheng, W., X.H. Lu, and S.C. Wong, Electrical and mechanical properties of
expanded graphite-reinforced high-density polyethylene. Journal of Applied
Polymer Science, 2004. 91(5): p. 2781-2788.

7.

Zheng, W.G., S.C. Wong, and H.J. Sue, Transport behavior of PMMA/expanded
graphite nanocomposites. Polymer, 2002. 43(25): p. 6767-6773.

8.

Zheng, W. and S.C. Wong, Electrical conductivity and dielectric properties of
PMMA/expanded graphite composites. Composites Science and Technology,
2003. 63(2): p. 225-235.

9.

Pan, Y.X., et al., A new process of fabricating electrically conducting nylon
6/graphite nanocomposites via intercalation polymerization. Journal of
Polymer Science Part B-Polymer Physics, 2000. 38(12): p. 1626-1633.

10.

Zou, J.F., et al., Conductive mechanism of polymer/graphite conducting
composites with low percolation threshold. Journal of Polymer Science Part
B-Polymer Physics, 2002. 40(10): p. 954-963.

11.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, A new compounding method for
exfoliated graphite-polypropylene nanocomposites with enhanced flexural
properties and lower percolation threshold. Composites Science and
135

Technology, 2007. 67(10): p. 2045-2051.
12.

Kim, S., I. Do, and L.T. Drzal, Multifunctional xGnP/LLDPE Nanocomposites
Prepared by Solution Compounding Using Various Screw Rotating Systems.
Macromolecular Materials and Engineering, 2009. 294(3): p. 196-205.

13.

Ansari, S. and E.P. Giannelis, Functionalized Graphene Sheet-Poly(vinylidene
fluoride) Conductive Nanocomposites. Journal of Polymer Science Part
B-Polymer Physics, 2009. 47(9): p. 888-897.

14.

She, Y.H., G.H. Chen, and D.J. Wu, Fabrication of polyethylene/graphite
nanocomposite from modified expanded graphite. Polymer International,
2007. 56(5): p. 679-685.

15.

Yu, J., et al., Conductivity of polyolefins filled with high-structure carbon black.
Journal of Applied Polymer Science, 2005. 98(4): p. 1799-1805.

136

Chapter 7 Effect of Particle size on
the Thermal and Electrical
Conductivity of Graphite
Nanoplatelet/Polyetherimide
Composites
Abstract
Graphite nanoplatelets (GNPs) with different sizes were compounded with
polyetherimide (PEId) by melt-blending and pre-coating in order to investigate their
effect on the electrical and thermal conductivity of the composites. Pre-coating
compounding successfully created a percolated GNP network with an electrical
percolation threshold at 1wt%, while melt-blended samples required concentrations
over 5wt% to achieve percolation. The pre-coated samples also show higher thermal
conductivity than the melt-blended ones. Smaller GNP particle size results in lower
thermal conductivity. An improvement of 1600% in thermal conductivity was
recorded for pre-coated 10wt% GNP-15/PEId composite as compared to the neat
polymer and a polynomial dependence on GNP concentration was found to predict
thermal conductivity.

137

Introduction
Polymer and polymer composite materials are considered to be a viable substitute
for metal alloys and are being widely used because of their light weight and good
chemical resistance. However, the lack of thermal conductivity of neat polymers or
polymer blends limits some of the potential applications of this lightweight material
in certain areas, such as bi-polar plates for fuel cells, engine shields, packaging of
electronic devices, and so on. To overcome this problem, conductive fillers such as
metal powder [1-5], ceramic particles [6-8], carbon nanotubes [9-13] and
graphene-based nanofillers [14-17] have been introduced into polymer matrices.
Table 7.1 gives the thermal conductivity, coefficient of thermal expansion (CTE) and
density of various materials that are commonly used as fillers in composites.
Table 7.1, Thermal conductivity, coefficient of thermal expansion
and density of fillers[18]
Thermal
Linear CTE
Density (g/cm3)
Material
Conductivity
(10-6/℃)
(W/moK)
Copper
398
17
8.9
Silver
429
18
10.49
Aluminum
247
23
2.7
Tungsten
155
4.5
19.3
Kovar
17
5.1
8.39
Beryllium oxide
260
6
3
Aluminum nitride
320
4.5
3.3
Silicon carbide
270
3.7
3.3
Graphite (in-plane)
2000
-2.0 [19]
2.1
Graphene & CNT
1750-5800 [20, 21]
-0.9 [22]
2.1
Copper metal and metal alloys are a common thermal conductor. Copper possesses
a thermal conductivity of 398 W/mK. However, its CTE is as high as 17×10-6/℃. High
138

CTE causes the material to deform significantly with a change in temperature which
is usually a problem for materials used as conductors. In addition, a density of
8.9g/cm3 prevents copper from being used in weight-sensitive applications including
aerospace. Aluminum has a lower density compared to copper and is widely used for
aircraft electronics and laptops, but its CTE is even higher. In order to lower the CTE,
various ceramic particles are developed by adding a low CTE material into metal
matrix, forming materials such as aluminum nitride and silicon carbide. The best
thermal conductor would be diamond, which has a thermal conductivity of 3000
-6

W/mK and a CTE as low as 0.9×10 /℃. However, the extremely high cost of
diamond prevents it to be widely used.
Although graphite has a thermal conductivity close to diamond and a very low CTE
and density, it is not very practical to be used as a bulk material due to brittleness
and difficulty in machining. Carbon nanotubes (CNTs), for example, can be
considered as rolled up0 sheets of graphene. Researchers [21] have found out that
the thermal conductivity of single-wall CNTs (SWCNTs) ranges from 1750-5800 W/mK.
An epoxy matrix filled with 3wt% (~1.5vol%) CNT achieves a 300% improvement in
thermal conductivity as compared to the neat epoxy resin [11]. As a comparison,
around 8.75vol% silver particles are required to have the same improvement [1].
However, similar to diamond, the high cost of CNTs makes them impractical for
composites.
Another form of graphitic materials, graphene or graphite nanoplatelets, have the
potential for composites applications. Generally, natural graphite can be exfoliated or
139

expanded into very thin layers, forming a thin (3-7 nm) platelet morphology
(diameters ranging from sub micron to 200 microns) while maintaining good in-plane
properties. Two typical methods are usually used including the graphite oxide (GO)
[14] route and the graphite intercalation compound (GIC) [23] route. The GIC route is
considered to be able to best preserve the properties of pristine graphite since the
2

basal plane of sp carbon atoms are not disrupted during the process. A novel
process consisting of acid intercalation, microwave exfoliation and ultrasonication
pulverization has been developed in the Drzal group to deliver exceptional electrical,
thermal and mechanical properties in the exfoliated graphite nanoplatelet (GNP)
2

material with a surface area up to 300 m /g, a thickness as thin as 3 nm, and a lateral

dimension up to 200 µm [24]. This scalable process has proved its potential to
provide cost-effective conductive fillers for nanocomposites applications.

Unlike the electrical conductivity of polymer composites, the percolation phenomena
and filler aspect ratio dependence have not been explicitly explained for thermal
conductivity. Mamunya et al [25] discovered a non-monotonous dependence of
thermal conductivity on the volume content of CNT filler. The minimum of
conductivity happens at the same volume percentage as the electrical percolation
threshold. However, the difference is very small with large overlapping error bars. No
“percolation” behavior is found in thermal conductivity dependence on particle
loading. Moreover, the aspect ratio of the filler always takes an important role in
electrical percolation and conductivity [26, 27], while the importance fades for
thermal conductivity. Devpura et al [6] used modeling to predict the size effect on
140

metal powder/polymer systems below electrical percolation threshold and found
that the larger filler particles exhibit higher thermal conductivity in composites. The
authors hypothesize that the effect was the result of thermal boundary resistance
between particle and matrix. However, the effect of the particle size above the
percolation threshold was not discussed.
One possible reason for the difficulty of investigating percolation phenomena and
filler aspect ratio dependence for thermal conductivity is that the two factors are
usually coupled: larger aspect ratio normally results in lower percolation threshold.
To compare composites with percolated and non-percolated network at the same
loadings, two sets of GNP/polyetherimide (PEId) composites were fabricated, one of
which was melt-blended by a twin-screw extruder and the other one made through a
pre-coating process where the GNP particles are coated onto polymer powder
surfaces and then compression molded in order to create a continuous GNP phase
[28] [Chapter 6]. The aim of this research is to investigate the importance of particle
percolation and particle size on the thermal conductivity of polymer composites.

Experimental Materials and Methods
Materials
Polyether imide (PEId) was obtained from Sabic-IP (Ultem 1010, in two forms: pellet
2

and powder). GNP-1 (particles with an average diameter of 1 micron, 100m /g
2

surface area) and GNP-15 (particles with a diameter of 15 micron, 40m /g surface
141

2

area) were prepared in the lab [24]. GNP-5 (5 micron average diameter, 40m /g
surface area) was provided by XG-Science Inc. (Lansing, MI).

Melt-extrusion
Melt-extrusion was carried on Leistritz twin-screw extruder (MIC27/6L-48D). The
screws were operated in the co-rotation mode. Barrel temperatures were set to
310℃, with a melt temperature at 340℃. The die pressure was at around 5.5MPa.
Screw speed was set at 150 rpm.

Pre-coating
GNP was first dispersed in acetone under ultrasonication. PEId powder (~300 microns
in diameter) was then added into the suspension while stirring. After evaporation of
the acetone, the mixture was further mixed in a Speed Mixer (Flextec Corporation).
The sample was dried in a vacuum oven at 120℃.

Compression Molding
After compounding, the composite material was compression molded in a CARVER
Laboratory Press (Model 2731, Fred S. Carver INC.). A picture-frame mold was used
with a diameter of 2.56 cm and a thickness of 3 mm. The composite material was
heated up to 340℃ and held for 15 min. A compression pressure of 20MPa was
applied to the material and was the mold remained closed for 45 min in order to
achieve consolidation. After that, the press was cooled by water while the pressure
142

was maintained.

Electrical Resistivity Measurement
The resistance of GNP/PEId composites is measured by 2-point AC impedance
method. The resistance value was then converted into resistivity by taking into
account of the sample dimensions.

Thermal Diffusivity and Thermal Conductivity
Thermal diffusivity of the GNP/PEI composites is measured by laser flash method
(LFA 447 NanoFlash, Netzsch) in both in-plane and through-plane directions. The
heat capacity Cp is determined by Differential Scanning Calorimetry (DSC, TA
Instruments) and the density of the samples is calculated by the rule of mixture.
Thermal conductivity was then calculated by multiplying diffusivity by Cp and density.

Results and Discussion
Electrical conductivity
The electrical conductivity of the GNP/PEI composites is shown in Figure 7.1. All the
pre-coated samples percolate at 1wt% while the melt-blended ones percolate after 5
wt% loading although some percolation behavior starts at around 3 wt%. Figure 7.2
shows the internal morphology of the well formed network of GNP fillers in the
pre-coated 3wt% GNP-1/PEI composite.

143

Figure 7.1, Electrical conductivity of GNP/PEI composites

Figure 7.2, SEM image of pre-coated 3wt% GNP-1/PEI composite showing continuous
phase created by pre-coating. The white lines indicate GNP phase while the dark part
indicates the PEI matrix. The observed surface is prepared by polishing and plasma
etching as explained elsewhere. (Scale bar: 20 micron)

144

Thermal Conductivity

Figure 7.3, Thermal conductivity of GNP/PEI composites
The thermal conductivity of the GNP/PEI composites is shown in Figure 7.3. The
melt-blended composites do not show percolation. The thermal conductivity
increases with the increase of GNP-5 loading. Second, the pre-coated GNP-5/PEI
composite shows a higher thermal conductivity than the melt-blended one. This
indicates that there is less phonon scattering at the GNP-matrix in the composite.
Although the relative improvement is around 60% as compared to melt-blended
samples, the absolute value is still orders of magnitude lower than the calculated
result from rule of mixture. If we take thermal conductivity of graphite Kg=2000
W/mK, then by the rule of mixtures, the thermal conductivity of the composite at
5wt% (around 2.5vol%) loading should be Kc = 0.025Kg + 0.975Kp = 500 W/mK,
145

where Kp is the thermal conductivity of PEI and is negligible in this equation. This
simple calculation indicates a significant role of the contact resistance between
particles. Although a continuous phase is formed by pre-coating and can generate
electrical percolation, the thermal conductivity is still highly affected by how well the
particles are touching each other. As shown in Figure 7.3, composites filled with
smaller particles like GNP-1 achieve lower thermal conductivity while the ones with
the larger GNP-15 particles show much higher thermal conductivity. The fact that the
pre-coated GNP-1 samples exhibit an even lower value than the melt-blended GNP-5
ones indicates that the size effect is more important than particle connection. Put
another way, a well-connected network with smaller particles is not as good as a
partially connected network (like in 3wt% GNP-5/PEId melt-extruded sample in
Figure 7.1 where percolation can be seen in some specimens) of larger particles
(Figure 7.4). The same conclusion can also be drawn from comparing melt-blended
10wt% GNP-15/PEId with pre-coated 10wt% GNP-5/PEId samples. At this loading,
both samples should be percolated, and the larger particle is has an advantage in the
overall conductivity of the composite.

146

(a) Smaller particles, percolated
(b) Larger particles, partially connected
network
Figure 7.4, Schematics of particle size and connectivity: (a) although the particles are
well connected, the contacting resistance is huge; (b) large particles which are barely
touching each other can still provide higher thermal conductivity than (a)
The explanation underneath this phenomenon is the fact that two connected GNP
particles are not touching each other perfectly. A gap or a very thin layer of polymer
is still spacing the particles away from each other, which will cause phonon scattering.
Thus, the thermal contacting resistance is the major factor that prevents high
thermal conductivity for percolated network.
Interestingly, the pre-coated 10wt% GNP-15/PEI composite shows a much higher
thermal conductivity (1600% improvement compared with neat PEI) than any other
samples shown in Figure 7.3. However, as shown in Figure 7.5, the thermal
conductivity of pre-coated GNP-15/PEI composites has a second order polynomial
dependence on the volume fraction of the GNP filler.

147

Figure 7.5, Thermal conductivity of pre-coated GNP-15/PEI composite

𝑦 = 726.32𝑥 2 − 2.0695𝑥 + 0.192

In Figure 7.5, the fitted line gives:

(1)

The second order polynomial dependence suggests that the contact between
particles improves with the increase of GNP loading. This can be explained by the
fact that the coated GNP layer gets thicker with higher loadings and becomes harder
to be disrupted during the following compression molding.
nd

One interesting observation of the 2

order polynomial equation (1) is that it

indicates a minimum value for thermal conductivity, at 0.14vol% GNP loading. This
result is comparable with Mamunya’s CNT-PVC system where the minimum thermal
conductivity was obtained at around 0.5vol% of CNT loading. This loading is usually
considered as the electrical percolation threshold where all the particles are
connected to form a continuous phase of conductive filler. In this case, this
phenomenon can be interpreted as that the PEId powder is fully covered by the GNP
particles at 0.2vol% loading and the coated structure can hold together even after
148

compression molding.
More interestingly, if the loading is set to 100vol%, equation (1) will give out a
thermal conductivity value at 724 W/mK. Although this value is somewhat lower
than the thermal conductivity of graphene or CNT, it is comparable with the 380
W/mK value obtained from a monolayer assembly of GNP particles on the glass slide
reported by Xiang et al [29]. The reason for the overestimation from equation (1)
may be that it overestimates the best possible contacting between the GNP particles.

Conclusions
GNP/PEId composite were made from melt-blending and pre-coating. It is found that
the pre-coated samples form a well connected GNP particle network which
percolates the composite at very low loading (1wt%). The percolated network helps
to improve the thermal conductivity to about 60% higher than that of
none-percolated composites. Larger sized GNP particles can provide their composites
with higher thermal conductivity due to reduced amount of contact points and lower
contact resistance between filler particles even when a percolated network has been
established. It is also found that the benefit of the connected network may be
negated by the disadvantage of having too many contacting points. An improvement
of 1600% was recorded for pre-coated 10wt% GNP-15/PEI composite. A second order
polynomial dependence of thermal conductivity of percolated pre-coated
GNP-15/PEId composites on GNP loading was found.

149

References

150

REFERENCES
1.

Bjorneklett, A., L. Halbo, and H. Kristiansen, THERMAL-CONDUCTIVITY OF
EPOXY ADHESIVES FILLED WITH SILVER PARTICLES. International Journal of
Adhesion and Adhesives, 1992. 12(2): p. 99-104.

2.

Kowalski, L., J. Duszczyk, and L. Katgerman, Thermal conductivity of metal
powder-polymer feedstock for powder injection moulding. Journal of
Materials Science, 1999. 34(1): p. 1-5.

3.

Mamunya, Y.P., et al., Electrical and thermal conductivity of polymers filled
with metal powders. European Polymer Journal, 2002. 38(9): p. 1887-1897.

4.

Weidenfeller, B., M. Hofer, and F.R. Schilling, Thermal conductivity, thermal
diffusivity, and specific heat capacity of particle filled polypropylene.
Composites Part a-Applied Science and Manufacturing, 2004. 35(4): p.
423-429.

5.

Mamunya, Y.P., et al., Percolation phenomena in polymers containing
dispersed iron. Polymer Engineering and Science, 2002. 42(1): p. 90-100.

6.

Devpura, A., P.E. Phelan, and R.S. Prasher, Size effects on the thermal
conductivity of polymers laden with highly conductive filler particles.
Microscale Thermophysical Engineering, 2001. 5(3): p. 177-189.

7.

Hill, R.F. and P.H. Supancic, Thermal conductivity of platelet-filled polymer
composites. Journal of the American Ceramic Society, 2002. 85(4): p. 851-857.

8.

Wong, C.P. and R.S. Bollampally, Thermal conductivity, elastic modulus, and
coefficient of thermal expansion of polymer composites filled with ceramic
particles for electronic packaging. Journal of Applied Polymer Science, 1999.
74(14): p. 3396-3403.

9.

Thostenson, E.T., Z.F. Ren, and T.W. Chou, Advances in the science and
technology of carbon nanotubes and their composites: a review. Composites
Science and Technology, 2001. 61(13): p. 1899-1912.

10.

Kuriger, R.J., et al., Processing and characterization of aligned vapor grown
carbon fiber reinforced polypropylene. Composites Part a-Applied Science and
Manufacturing, 2002. 33(1): p. 53-62.

11.

Choi, E.S., et al., Enhancement of thermal and electrical properties of carbon
nanotube polymer composites by magnetic field processing. Journal of
151

Applied Physics, 2003. 94(9): p. 6034-6039.
12.

Moniruzzaman, M. and K.I. Winey, Polymer nanocomposites containing
carbon nanotubes. Macromolecules, 2006. 39(16): p. 5194-5205.

13.

Tibbetts, G.G., et al., A review of the fabrication and properties of
vapor-grown carbon nanofiber/polymer composites. Composites Science and
Technology, 2007. 67(7-8): p. 1709-1718.

14.

Stankovich, S., et al., Graphene-based composite materials. Nature, 2006.
442(7100): p. 282-286.

15.

Fukushima, H., et al., Thermal conductivity of exfoliated graphite
nanocomposites. Journal of Thermal Analysis and Calorimetry, 2006. 85(1): p.
235-238.

16.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, Multifunctional polypropylene
composites produced by incorporation of exfoliated graphite nanoplatelets.
Carbon, 2007. 45(7): p. 1446-1452.

17.

Yang, H., et al., Improved mechanical and functional properties of
elastomer/graphite nanocomposites prepared by latex compounding. Acta
Materialia, 2007. 55(18): p. 6372-6382.

18.

Chung, D.D.L., Materials for thermal conduction. Applied Thermal Engineering,
2001. 21(16): p. 1593-1605.

19.

Bailey, A.C. and B. Yates, ANISOTROPIC THERMAL EXPANSION OF PYROLYTIC
GRAPHITE AT LOW TEMPERATURES. Journal of Applied Physics, 1970. 41(13):
p. 5088-&.

20.

Kim, P., et al., Thermal transport measurements of individual multiwalled
nanotubes. Physical Review Letters, 2001. 87(21).

21.

Hone, J., et al., Electrical and thermal transport properties of magnetically
aligned single walt carbon nanotube films. Applied Physics Letters, 2000.
77(5): p. 666-668.

22.

Schelling, P.K. and R. Keblinski, Thermal expansion of carbon structures.
Physical Review B, 2003. 68(3).

23.

Rudorff, W., GRAPHITE INTERCALATION COMPOUNDS. Advances in Inorganic
Chemistry, 1959. 1: p. 223-266.
152

24.

Drzal, L.T. and H. Fukushima, Expanded graphite and products produced
therefrom. 2004: US.

25.

Mamunya, Y., et al., Electrical and thermophysical behaviour of PVC-MWCNT
nanocomposites. Composites Science and Technology, 2008. 68(9): p.
1981-1988.

26.

Kim, H. and C.W. Macosko, Morphology and properties of polyester/exfoliated
graphite nanocomposites. Macromolecules, 2008. 41(9): p. 3317-3327.

27.

Abdel-Goad, M., et al., Preparation and rheological characterization of
polymer nanocomposites based on expanded graphite. Journal of
Macromolecular Science Part a-Pure and Applied Chemistry, 2007. 44(4-6): p.
591-598.

28.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, A new compounding method for
exfoliated graphite-polypropylene nanocomposites with enhanced flexural
properties and lower percolation threshold. Composites Science and
Technology, 2007. 67(10): p. 2045-2051.

29.

Xiang, J.L. and L.T. Drzal, Thermal Conductivity of a Monolayer of Exfoliated
Graphite Nanoplatelets Prepared by Liquid-Liquid Interfacial Self-Assembly.
Journal of Nanomaterials, 2010.

153

Chapter 8 GNP Paper: Light weight
Composite having Electrical, Thermal
Conductivity and Gas Barrier
Properties
Abstract
Paper forms (i.e. thin free-standing films) of carbon-based materials have received
increasing attention. Here we present a novel approach to fabricating a binder free,
self-standing flexible paper consisting of exfoliated graphite nanoplatelets (GNPs). It
is found that the electrical conductivity of the GNP paper can be as high as 2200
S/cm and the thermal conductivity reaches 313 W/mK. Both thermoset and
thermoplastic matrices were used to impregnate the porous GNP paper and an
extremely high tensile modulus was attained. Even with 30vol% polymer, the GNP
paper composite can still exhibit ~700 S/cm electrical conductivity thanks to the
highly continuous GNP network formed in the paper making process. The
impregnated GNP paper was also investigated as a component in carbon fiber
composite. It is found that when inserted into a layered laminate composite
construction, gas permeability can be severely reduced and electrical and thermal
conductivity can be greatly enhanced.

154

Introduction
Graphite or graphene nano platelets (GNP) have drawn a lot of interest due to their
high mechanical properties (~1TB for either single layer or multi layer graphene
-6

theoretically [1, 2] and experimentally [3-5]), excellent electrical (~1.0×10 ohm/cm
resistivity for pure graphene but more than two orders of magnitude higher for
graphite [6]), and thermal (~5000 W/mK [7-10]) conductivity. The platelet shape of
graphite also benefits the composite by providing increased tortuosity for molecular
transport which enhances barrier properties to small molecules [11, 12]. One way of
utilizing the excellent properties of GNP is to add it into a polymer matrix. The light
weight polymer nanocomposite can serve as an alternative to metal or fiber
reinforced composites with the help of GNP. However, when homogeneously
dispersed in a matrix, the GNP usually cannot fully deliver its properties to its
composite. For some specific systems like graphene-polystyrene, the electrical
percolation threshold can be as low as 0.1 vol% and around 0.01 S/cm conductivity
can be achieved at 2.5 vol% graphene loading [13-15]. However, other polymer
-3

matrices are facing much higher percolation thresholds and less improvement (~10

S/cm at 10wt% loading) [16-18]. Thermal conductivity of the GNP/polymer
composites is also much lower than graphene, usually within 5 W/mK at a filler
loading up to 20 vol% [19]. Moreover, increasing filler loading cannot always provide
significant improvement especially after electrical percolation threshold of a given
system. Meanwhile, mechanical strength would be reduced significantly due to
155

brittleness.
To fully use the conductivity properties of conductive nano fillers, a large continuous
phase of the filler is required [18][Chapter 6]. A promising approach to realize it is to
reassemble the nanometer-thick, micron-sized particles into micron-thick,
multi-centimeter sized paper or film, and then incorporate this GNP paper into the
polymer matrix. Carbon nanotube (CNT) Bucky paper [20-30] and graphite oxide (GO)
paper [31-33] have been widely studied. Paper form materials prepared by either
nanoparticle can be free-standing, self-supporting and show good flexibility [28, 31].
Electrical conductivity was reported to be 200 S/cm for Bucky paper with aligned CNT
[28] and 350 S/cm for reduced GO paper [32]. Note that the paper fabricated from
ref. [28] and [32] were claimed to be binder-free. Other types of paper with the
presence of binder would have lower values. Compared to graphene or single-walled
CNT which has a conductivity of ~2000 S/cm, these values are still one order of
magnitude lower. The reason is, for CNT, the contact resistance between each tube is
large due to the extremely small particle size for GO paper, the chemically reduced
graphite oxide is not as conductive as pristine graphite. Li et al [34] and Biswas et al
[35] both fabricated Langmuir-Blodgett thin films from exfoliated graphite. The
electrical and thermal conductivity were reported to be over 1000 S/cm [35] and 380
W/mK [36], respectively. However, the Langmuir-Blodgett approach is not suitable
for a paper form material with a thickness in the micron range and the porosity of
the film is usually very low and hard to control, making it not an adequate solution
for composites.
156

This research presents a novel route of fabricating binder-free, porosity-controllable,
highly conductive and self-standing graphite paper consisting of graphite
nanoplatelets (GNPs). The GNP particles are first dispersed in water with the addition
of polyethyleneimine (PEIn). The suspension was then filtered and could be
separated from the filter paper. Residual PEIn was removed by annealing the paper
at a high temperature above the decomposition temperature of the PEIn. The
resulting porous paper exhibits properties to other carbon based materials. Both
thermoset and thermoplastic matrices were combined into this GNP paper form and
the resulting material has potential to serve as a layer in laminate composites.
Mechanical properties, electrical and thermal conductivity, barrier properties were
measured and discussed for both GNP paper and its composites.

Experimentation
Materials
Polyetherimide (PEId) was obtained from Sabic-IP (Ultem 1010). Two thermoset
systems were used including epoxy (EPON 862:Jeffamine T403=100:49 weight) and
vinyl ester (Derakane 510A-40 with 1wt% MEKP). GNP ‘worms’ (expanded graphite
2

with a diameter over 100 microns, 40m /g surface area) and GNP-15 (particles with a
2

diameter of 15 micron, 40m /g surface area) were prepared in the lab [37]. The
exfoliated but not pulverized GNP ‘worms’ were specially fabricated in order to
achieve higher electrical conductivity. If not specified, GNP stands for GNP-15.
157

GNP Paper Fabrication
GNP

particles

were

dispersed

in

water

with

the

presence

of

PEIn

(GNP:PEIn:Water=1:0.1~10:1000 weight). Ultrasonication was applied to the
suspension to help disperse the GNP particles. After 1 min ultrasonication with a
power of 10 watts, the GNP worms were broken down to individual platelets with a
diameter ranging from 20 to 100 microns. The suspension was stirred overnight, and
then filtered with an Anodisc filter (200 nm pore size, Whatman) or glass microfiber
(1 micron opening, Whatman) filter paper under vacuum. Additional water was used
to wash the filtered GNP. GNP and the filter paper were dried overnight and then
separated. Due to its hydrophobic nature, the GNP layer was easily to be peeled off
from the hydrophilic filter paper. The paper form GNP was then annealed under
o

o

120 C for 1 hour in order to remove any residual moisture and then heated to 340 C
for 1 hour to decompose the remaining polyethyleneimine. Thermal gravimetric
analysis (TGA) indicated that the paper was binder free after the heat treatment
(Figure 8.1).

Figure 8.1, TGA result of GNP and PEIn
158

Figure 8.1 shows the comparison of TGA results between annealed GNP paper,
un-processed GNP-15 and PEIn. After about 1 hour annealing at 340 oC in air, more
than 95 wt% PEIn was decomposed. No significant weight loss was recorded before
o

500 C for annealed GNP paper under nitrogen condition. However, for un-processed
o

GNP-15, weight loss starts at around 300 C. This could be due to the residual
intercalant which might be washed away during dispersing GNP in water, so the
weight loss won’t show up in annealed GNP paper.
The filtered GNP paper possesses a porous structure. The porosity and density of the
paper depend on the conditions used for filtration, e.g. the amount of suspension
and the speed of filtration. Two grades of GNP paper were fabricated, one made
from the Anodisc filter paper with less porosity and high density (GNP-H), and the
other one made from the glass microfiber filter paper (noted as GNP-L) with lower
density (Table 8.1). The porosity was calculated by assuming the density of GNP
particle to be 2.1 g/cc. The GNP paper made from worms (GNP-W) generally showed
higher density than GNP-15 papers, due to the more ordered layer structure formed
by larger particles.
Table 8.1, Density and porosity of GNP papers, data are actual
value of the particular sample, not an average
Density (g/cc)
Porosity (vol%)
GNP-H
0.90
57
GNP-H-P
1.34
36
GNP-L
0.19
91
GNP-L-P
0.95
55
GNP-W
1.10
47
GNP-W-P
1.54
27
159

Note that the density of the GNP paper is very important to its properties. Porous
paper can be considered as a composite consisting of GNP and air. To eliminate the
effect of porosity and obtain best properties, the GNP paper was also mechanically
compressed under room temperature. After applying about 100 MPa pressure, the
pressed GNP paper (noted as GNP-x-P, x=H, L, W) presented a higher apparent
density and lower porosity (Table 8.1 (GNP-H and GNP-H-P data are from the same
paper before and after pressing), Figure 8.2). The fact that the density of pressed
paper varies according to the density before pressing suggests that higher density
paper has better alignment of GNP particles.

Figure 8.2, SEM images showing cross-section of the GNP paper (scale bars represent
10 micron)
Figure 8.2 shows the morphology of the cross-section of GNP paper with and without
pressing. In Figure 8.2(a), the paper was not pressed. The GNP particles form a long
continuous phase spaced by large pores which have a size ranging from 1 micron to
10 microns. After pressing (Figure 8.2(b)), most of the pores disappear. The GNP
particles are very closely packed and have good contact with each other. However,
voids are still visible in Figure 8.2(b).
160

The GNP paper that was fabricated in such processes exhibits very good flexibility,
even without the presence of the binder after annealing. Figure 8.3(b) shows a paper
plane folded from a GNP paper. The folding edge (Figure 8.3(d)) stays intact. This
indicates that the Van der Waals force between GNP platelets is large enough to hold
the structure that no extra binder is needed.

(a)

(b)

(c)
(d)
Figure 8.3, GNP paper and its flexibility (a) GNP paper (b) paper plane folded from
GNP paper (c) morphology of the surface of the GNP paper (d) morphology of the
folding edge of the paper plane (Scale bars: (c) 200 micron, (d) 50 micron)

In this research, the GNP-H was used to represent stand-alone GNP paper and for the
fabrication of high GNP loading paper composites (80 wt% and 70 wt%) due to its
low porosity. On the other hand, the GNP-L was mostly used for low GNP loading
161

paper composites (10 vol%) because of its higher porosity. The GNP papers used in
CF composites were all GNP-Ls.

Impregnating GNP Paper
Due to its porosity, the GNP paper can be impregnated by either a thermoset or
thermoplastic matrix. For thermoset system, the un-pressed GNP paper was soaked
in the uncured liquid epoxy mixture for 1 hour and then cured in a hot press under
o

10 MPa and 150 C (noted as GNP-E).
For the thermoplastic system, PEId/dichloromethane solution was prepared at 0.1
g/mL, 0.15 g/mL and 0.2 g/mL. The GNP paper was soaked in the solution for 1 hour
and then dried under room temperature. Depending on the concentration of the
solution, the pores can be filled with controlled amounts of PEId. The soaked paper
o

(GNP-PEId) can also be further compressed under 10 MPa and 340 C resulting in void
free GNP/PEId paper composite. The loading of GNP in the paper composite is
calculated by the weight change before and after impregnation.
The polymer impregnated GNP-E paper was used to fabricate laminate composite
with carbon fiber/polyetherketoneketone (PEKK) prepreg (CF). Structure and
notation of the composites are shown in Figure 8.4.

162

Figure 8.4, Structure and nomination of CF/GNP paper composites

Characterization
Tensile modulus was measured by Dynamic Mechanical Analyzer (DMA, TA
Instruments) with film tension clamp. The strain rate was set at 0.1%/min.
Representative stress-strain curves for both un-pressed and pressed GNP papers are
shown in Figure 8.5. Pressed GNP paper (GNP-H-P) shows a much higher modulus
and strength than un-pressed one (GNP-H). However, the elongation at break of
GNP-H is greater than that of GNP-H-P. This is because in a pressed paper, GNP
particles have better contact between each other which ensures a total force of
attraction. Also, the layered structure has fewer voids. For GNP-H, the GNP particles
have room to move and re-orient more before breaking than GNP-H-P, resulting in a
greater elongation at break.

163

Figure 8.5, Stress-strain curve of GNP paper
Electrical conductivity was measured by 4-point method. The papers were cut into
rectangular strips. Four electrode pins were pressed to the paper surface with a
spacing of 1 cm. Silver paste was placed at the contacting points between the paper
and the electrode pins in order to eliminate contact resistance.
Thermal diffusivity was measured by using the laser flash method. Thermal
conductivity was then calculated by multiplying the diffusivity by the heat capacity
and density of the paper.
O 2 permeability was measured by a MOCON OXTRAN 2/20 model (MOCON, Inc.).
The composite film was sandwiched by two pieces of aluminum masks with a 1 cm

2

opening at the center. Permeation of O 2 was measured under a pressure of 30 psi
2

and recorded with a unit of cc/m /day.

164

Results and discussions
Effect of Annealing and Pressing

Figure 8.6, Effect of Annealing and Compression on the density and electrical
conductivity of GNP-W paper. “as-made” stands for filtrated paper without annealing
or pressing. Pressure is the nominal value calculated from press/sample area
GNP-W papers were used to demonstrate the effect of annealing and compression
(Figure 8.6). After annealing, the density of the GNP-W paper decreased by 3wt%,
indicating that about 3wt% of PEIn was left in the filtrated GNP paper. Considering
the original ratio between GNP and PEIn is 1 in the suspension, it is clear that most of
the PEIn was washed away by water. Considering the 5 wt% residue of PEIn after
o

annealing under 340 C for 1 h (Figure 8.1), a total of around 0.15 wt% PEIn residue is
still in the GNP paper. However, this level is even lower than that of the un-processed
GNP particle. Thus, the annealed paper can be considered as binder free. By
removing the 3 wt% PEIn, the electrical conductivity of GNP-W paper increased
significantly, from 900 S/cm to 1200 S/cm. This huge difference indicates the
165

importance of the contact resistance. Before annealing, the 3 wt% PEIn was covering
the surface of all GNP particles, although the amount seems to be small. The
GNP-PEIn-GNP interface is clearly inhibiting the electrical conductance.
Pressing the annealed GNP-W paper resulted in further improvement of electrical
conductivity. There is a very good correlation between the density and electrical
conductivity of the GNP-W paper. After compression under 100 MPa (GNP-W-P) and
1000 MPa (GNP-W-SP), the conductivity increased to 1500 S/cm and 2200 S/cm,
respectively. The density of 1000 MPa pressed GNP-W-P reached 1.99 g/cc which is
very close to the density of natural graphite, and it is not surprising that its
conductivity is also similar to that of the natural graphite.

Properties of GNP Paper
The alignment of GNP layer and binder-free nature both benefit the properties of the
GNP paper. Tensile modulus, electrical and thermal conductivity were measured for
pressed and un-pressed GNP papers along with three kinds of commercially available
carbon-based paper form materials including TG-411, SUPER GTO (GRAFOIL®,
compressed expanded graphite from GRAFTECH) and CNT Bucky paper (NANO-LAB).

166

Figure 8.7, Comparison between carbon-based papers (a) tensile modulus (b)
electrical conductivity (c) thermal conductivity
The comparison between GNP-H, GNP-H-P, GNP-W, GNP-W-P, GNP-W-SP (pressed
under 1000 MPa nom), TG-411, SUPER GTO and CNT Bucky paper is shown in Figure
167

8.7. Generally, all the three kinds of properties are positively related to the density of
the sample, with a couple of exceptions. First, the tensile modulus of GNP-H paper is
2900 MPa, while GRAFOIL papers have a tensile modulus of about 2500 MPa.
Although the density of GRAFOIL is higher than GNP-H, unlike GNP paper where the
GNP particles are highly aligned, particles do not have preferential alignment in
GRAFOIL (Figure 8.8(a)). Since the properties of the GNP platelet is much higher in
the in-plane direction, the alignment of GNP particles can help GNP paper to achieve
maximum performance including not only modulus but also electrical and thermal
conductivity. Second, the comparison between the electrical conductivity of GNP-H-P
and GNP-W indicates the benefit of using larger particles. Despite its higher density,
GNP-H-P falls short in electrical conductivity as compared to GNP-W. As previously
stated, the size of the ultrasonicated GNP worm particles ranges from 20 to 100
microns, which is larger than GNP-15. With the same amount of material, more
contacting points are present in GNP-15 paper, resulting in higher contact resistance.
The GNP-W-SP paper has the highest electrical conductivity at ~2200 S/cm. This
result is much higher than other reported value for carbon-based paper in literature.
Under similar density, GRAFOIL® papers have similar conductivity as compared to
GNP-W. Figure 8.8(a) shows the morphology of GRAFOIL® paper. Note that the
GRAFOIL® is made from expanded graphite, so the debris morphology in Figure 8.8(a)
does not represent the actual size of the particles, and was more likely generated by
polishing during sample preparation.
CNT Bucky paper has the lowest density, modulus, electrical and thermal
168

conductivity among all the materials, 0.35g/cc, 200 MPa, 39.7 S/cm and 5.28 W/mK
respectively. From Figure 8.8(b), it can be observed that the CNT paper is very porous.
The CNTs are loosely entangled together and can be drawn apart by relatively low
force, resulting in low tensile modulus. Electrical and thermal conductivity are
hindered by the high porosity and large amount of contacting points between CNT
particles.

(a)

(b)

Figure 8.8, SEM images showing cross-section morphology of (a) GRAFOIL® and (b)
CNT Bucky paper. Samples were prepared by polishing-O 2 plasma etching process
(scale bars: (a) 200 nm, (b) 2 micron)

Impregnation of GNP Paper
To investigate the morphology of impregnated GNP paper, the epoxy fixative material
was mixed with fluorescent dye and used as support for the papers. An Olympus
FluoView FV1000 Laser Scanning Confocal Microscope with 488 Argon laser
excitation was used to detect the fluorescence signal.
169

Figure 8.9, Confocal microscopic images of GNP papers impregnated by (a) 0.1 g/mL
PEId/dichloromethane solution (b) 0.2 g/mL PEId/dichloromethane solution (c)
pressed GNP- PEId paper (d) GNP-E paper. Green indicates fluorescence signal, black
indicates GNP (scale bars represent 20 micron)

If a GNP paper is porous, the epoxy putty material can penetrate inside and give a
fluorescence signal. If the GNP paper is soaked in a dilute PEId solution, the pores
cannot be fully filled. Figure 8.9(a) shows uniform distributed fluorescent signal
throughout the GNP paper cross-section. With the increase of PEId concentration,
the area that emits fluorescent signal decreases, indicating more PEId was absorbed
in the pores preventing the fluorescent tagged epoxy for penetrating into
them(Figure 8.9(b)). Pressing the soaked GNP paper can close the pores and no
170

fluorescent signal can be seen in Figure 8.9(c). The epoxy soaked GNP paper also
shows no fluorescence signal throughout the cross-section (Figure 8.9(d)).

Figure 8.10, SEM images showing cross-section of (a) pressed GNP-PEId (b) CGC-PEId
(Scale bars: (a) 10 micron, (b) 100 micron)
One advantage of infiltrating the GNP paper with polymer is that the tensile modulus
of the composite can be greatly improved as compared with either pure GNP paper
or the neat polymer. 80 wt% and 70 wt% GNP/PEId paper show a tensile modulus of
19000 MPa and 22000 MPa respectively, which are approximately 500% to 600%
improvement to either pure GNP paper or neat PEId (Figure 8.11(a)). The tensile
strength of the GNP paper is also improved from 2.3 MPa to 22.6 MPa and 32.3 MPa
with 30 vol% and 40 vol% filled by PEId, respectively (Figure 8.11(b)).

171

Figure 8.11, Properties of GNP paper composites from soaking

172

Figure 8.12, Properties of GNP paper/CF composites
Interestingly, with 10 wt% more polymer, both modulus and strength increased for
GNP-PEId. This can be explained by looking at the volume fraction of GNP, PEId and
voids. 70 wt% GNP-PEId has similar GNP volume fraction as the 80 wt% sample but
173

more polymer to fill in the voids and bind the particles, resulting in higher modulus
and strength (Table 8.2).

Sample
Actual wt%
Density
Void vol%
GNP vol%
PEId vol%

Table 8.2, Actual composition in GNP-PEId
80 wt% GNP-PEId
70 wt% GNP-PEId
GNP-H
0.79
0.71
1
1.03
1.04
0.90
44
41
57
39
36
43
17
23
0

GNP-H-P
1
1.34
36
64
0

Moreover, adding 20 to 30 wt% of polymer to the GNP paper would not cause the
conductivity to drop significantly (Figure 8.11(c)). In fact, compared to un-pressed
GNP-H, both 80 wt% and 70 wt% GNP-PEId paper showed higher conductivity at 620
S/cm and 550 S/cm, respectively. The reason is, for pressed GNP-PEId paper, the GNP
layers are also closely packed similar to pure GNP-H-P (Figure 8.10(a), Figure 8.2(b)).
The PEId actually fills in those small voids between GNP particles. Certainly, the
presence of PEId would add extra contacting resistance, resulting in lower
conductivity for the GNP-PEId composite paper as compared to GNP-H-P. The
difference in volume fraction of GNP in the paper is within 7 vol% between 70 wt%
GNP-PEId and GNP-H (Table 8.2). Despite its 41 vol% voids, the O 2 permeability of 70
wt% GNP-PEId paper is only 1.1% of neat PEId (Figure 8.12(c)), indicating that the
voids are not connected, which can also be observed in Figure 8.9(c) where no
fluorescence signal can be observed inside the GNP-PEId paper.
The most significant effect of adding GNP paper into CF composite is the largely

174

reduced O 2 permeability (Figure 8.12(c)). The epoxy impregnated GNP-E paper
shows a permeability that is only 2% of CF prepreg. By inserting a layer of GNP-E, the
O 2 permeability of CF composite consisting of two CF layers reduced to 11% as
compared to the pristine CF prepreg.
Meanwhile, with 85 wt% filled by epoxy, the GNP-E paper retains an in-plane thermal
conductivity at 20.5 W/mK with a through-plane conductivity of 8.78 W/mK (Figure
8.12(b)). The presence of 85 wt% epoxy reduced the anisotropy of pure GNP paper
which has a maximum anisotropic factor as high as 200 (GNP-H-P in Figure 8.7(c)).
The GNP particles in this GNP-E paper were prevented from high alignment by the
spacing of epoxy. If this GNP-E paper is inserted in between of two CF prepregs, both
thermal and electrical conductivity can be improved for the composite. The most
significant improvement was attained by structuring the GNP paper to one surface of
the CF composite (Figure 8.4, Figure 8.12(a)(b)). While the CGC composite gained 200%
improvement in in-plane thermal conductivity as compared to the CC construction,
the increase for CG is over 700%. For surface resistance, the CG structure also
benefits from the exposure of GNP phase and shows lower value (0.3 ohm/sq). For
CF prepreg, the resistance in the 0 degree direction is much lower than 90 degree
due to the anisotropic nature of carbon fibers. However, when a layer of GNP paper
is inserted, the resistance becomes identical in both directions. This result indicates
good contact between CF and GNP layers which is also shown in Figure 8.10(b). Since
the conductivity of GNP is much higher than CF, the apparent surface resistance
actually mainly consists of the resistance through the thickness of CF layer (Figure
175

8.13).

Figure 8.13, Schematics of surface resistance measurement for CF/GNP paper
composite. R r and R l stand for resistance of CF through radius direction and through
length direction, respective. R g stands for resistance of GNP paper, which is far less
than the other two. N is the number of CF between electrode probes and n is that
between the probe and the GNP layer. N is usually much larger than n. When the
resistance is measured along 90 degree direction, the path for electrons with the
lowest resistance is through the thickness of CF layer and reaching GNP layer. The
total resistance can be approximated to nR r . When the resistance is measured along
0 degree direction, since the distance between two probes (10mm) is usually much
larger than that between one probe and GNP layer (~0.2mm), R l is still much larger
than nR r and the total resistance can also be approximated to nR r .

Conclusions
Binder free, self-standing flexible GNP papers have been successfully fabricated
through filtration of GNP/PEIn/water suspension followed by annealing. It is found
that the properties are highly related to the density of the paper. The pressed GNP
paper exhibited excellent electrical conductivity (~2200 S/cm) which is higher than
other reported similar type of materials. The thermal conductivity (313 W/mK) is
comparable to other commercially available graphite material. Processes have been
176

developed to impregnate the GNP paper with either thermoset or thermoplastic
matrices. The resulting paper composite showed extremely high tensile modulus
(22000 MPa) and reasonable strength (32 MPa) while still retaining high electrical
and thermal conductivity. When inserted into a multilayer laminated composite, the
additional GNP paper layer can significantly reduce the O 2 permeability and add
in-plane electrical and thermal conductivity to the composite. However,
through-plane properties were not improved significantly because of the anisotropic
behavior and alignment of GNP particles. The GNP paper composite is believed to be
a very useful addition to laminate composites, providing multifunctionality for
applications such as fuel tanks, lightning strike protection, replacement for copper
conductor for weight-sensitive area like aerospace, and flexible resistive heaters.

177

References

178

REFERENCES
1.

Chang, T.C. and H.J. Gao, Size-dependent elastic properties of a single-walled
carbon nanotube via a molecular mechanics model. Journal of the Mechanics
and Physics of Solids, 2003. 51(6): p. 1059-1074.

2.

Meo, M. and M. Rossi, Prediction of Young's modulus of single wall carbon
nanotubes by molecular-mechanics based finite element modelling.
Composites Science and Technology, 2006. 66(11-12): p. 1597-1605.

3.

Lee, C., et al., Measurement of the elastic properties and intrinsic strength of
monolayer graphene. Science, 2008. 321(5887): p. 385-388.

4.

Zhao, H., K. Min, and N.R. Aluru, Size and Chirality Dependent Elastic
Properties of Graphene Nanoribbons under Uniaxial Tension. Nano Letters,
2009. 9(8): p. 3012-3015.

5.

Wong, C.L., et al., Characterization of nanomechanical graphene drum
structures. Journal of Micromechanics and Microengineering, 2010. 20(11).

6.

Tune, L., Physicists Show Electrons Can Travel More Than 100 Times Faster in
Graphene (https://newsdesk.umd.edu/scitech/release.cfm?ArticleID=1621).
2008.

7.

Watanabe, E., et al., Ballistic thermal conductance of electrons in graphene
ribbons. Physical Review B, 2009. 80(8).

8.

Jiang, J.W., J.S. Wang, and B.W. Li, Thermal conductance of graphene and
dimerite. Physical Review B, 2009. 79(20).

9.

Saito, K., J. Nakamura, and A. Natori, Ballistic thermal conductance of a
graphene sheet. Physical Review B, 2007. 76(11).

10.

Balandin, A.A., et al., Superior thermal conductivity of single-layer graphene.
Nano Letters, 2008. 8(3): p. 902-907.

11.

LeBaron, P.C., Z. Wang, and T.J. Pinnavaia, Polymer-layered silicate
nanocomposites: an overview. Applied Clay Science, 1999. 15(1-2): p. 11-29.

12.

Yano, K., et al., SYNTHESIS AND PROPERTIES OF POLYIMIDE CLAY HYBRID.
Journal of Polymer Science Part a-Polymer Chemistry, 1993. 31(10): p.
2493-2498.
179

13.

Stankovich, S., et al., Graphene-based composite materials. Nature, 2006.
442(7100): p. 282-286.

14.

Chen, G.H., et al., Preparation of polymer/graphite conducting
nanocomposite by intercalation polymerization. Journal of Applied Polymer
Science, 2001. 82(10): p. 2506-2513.

15.

Zheng, G.H., et al., Characterizations of expanded graphite/polymer
composites prepared by in situ polymerization. Carbon, 2004. 42(14): p.
2839-2847.

16.

Zhao, Y.F., et al., Preparation and properties of electrically conductive
PPS/expanded graphite nanocomposites. Composites Science and Technology,
2007. 67(11-12): p. 2528-2534.

17.

Vadukumpully, S., et al., Flexible conductive graphene/poly(vinyl chloride)
composite thin films with high mechanical strength and thermal stability.
Carbon, 2011. 49(1): p. 198-205.

18.

Kalaitzidou, K., H. Fukushima, and L.T. Drzal, A new compounding method for
exfoliated graphite-polypropylene nanocomposites with enhanced flexural
properties and lower percolation threshold. Composites Science and
Technology, 2007. 67(10): p. 2045-2051.

19.

Fukushima, H., et al., Thermal conductivity of exfoliated graphite
nanocomposites. Journal of Thermal Analysis and Calorimetry, 2006. 85(1): p.
235-238.

20.

Liu, J., et al., Fullerene pipes. Science, 1998. 280(5367): p. 1253-1256.

21.

Cooper, S.M., et al., Gas permeability of a buckypaper membrane. Nano
Letters, 2003. 3(2): p. 189-192.

22.

Wang, Z., et al., Processing and property investigation of single-walled carbon
nanotube (SWNT) buckypaper/epoxy resin matrix nanocomposites.
Composites Part a-Applied Science and Manufacturing, 2004. 35(10): p.
1225-1232.

23.

Endo, M., et al., 'Buckypaper' from coaxial nanotubes. Nature, 2005.
433(7025): p. 476-476.

24.

Gonnet, P., et al., Thermal conductivity of magnetically aligned carbon
nanotube buckypapers and nanocomposites. Current Applied Physics, 2006.
180

6(1): p. 119-122.
25.

Gou, J.H., Single-walled nanotube bucky paper and nanocomposite. Polymer
International, 2006. 55(11): p. 1283-1288.

26.

Kim, Y.A., et al., Fabrication of high-purity, double-walled carbon nanotube
buckypaper. Chemical Vapor Deposition, 2006. 12(6): p. 327-+.

27.

Zheng, F., et al., Single-walled carbon nanotube paper as a sorbent for organic
vapor preconcentration. Analytical Chemistry, 2006. 78(7): p. 2442-2446.

28.

Wang, D., et al., Highly oriented carbon nanotube papers made of aligned
carbon nanotubes. Nanotechnology, 2008. 19(7).

29.

Xu, G.H., et al., The feasibility of producing MWCNT paper and strong MWCNT
film from VACNT array. Applied Physics a-Materials Science & Processing,
2008. 92(3): p. 531-539.

30.

Dumee, L.F., et al., Characterization and evaluation of carbon nanotube
Bucky-Paper membranes for direct contact membrane distillation. Journal of
Membrane Science, 2010. 351(1-2): p. 36-43.

31.

Dikin, D.A., et al., Preparation and characterization of graphene oxide paper.
Nature, 2007. 448(7152): p. 457-460.

32.

Chen, H., et al., Mechanically strong, electrically conductive, and
biocompatible graphene paper. Advanced Materials, 2008. 20(18): p. 3557-+.

33.

Park, S., et al., Graphene oxide papers modified by divalent ions - Enhancing
mechanical properties via chemical cross-linking. Acs Nano, 2008. 2(3): p.
572-578.

34.

Li, X.L., et al., Highly conducting graphene sheets and Langmuir-Blodgett films.
Nature Nanotechnology, 2008. 3(9): p. 538-542.

35.

Biswas, S. and L.T. Drzal, A Novel Approach to Create a Highly Ordered
Monolayer Film of Graphene Nanosheets at the Liquid-Liquid Interface. Nano
Letters, 2009. 9(1): p. 167-172.

36.

Xiang, J.L. and L.T. Drzal, Thermal Conductivity of a Monolayer of Exfoliated
Graphite Nanoplatelets Prepared by Liquid-Liquid Interfacial Self-Assembly.
Journal of Nanomaterials, 2010.
181

37.

Drzal, L.T. and H. Fukushima, Expanded graphite and products produced
therefrom. 2004: US.

182

Chapter 9 Summary and Conclusions
In summary, the possibility of using exfoliated graphite nanoplatelet (GNP) to provide
multifunctionality to a composite has been investigated. Polyetherimide (PEId) was
chosen as an exemplar matrix. The key idea of this research is to first reveal the
relationships

among

processing,

morphology

and

performance

of

the

nanocomposites, and then use the information to guide the nanostructuring of
GNP/polymer nanocomposites. The first part is based on the successful development
of sample preparation method for scanning electron microscope (SEM) quantification
of the nanoparticle reinforced composite morphology. The sectioning-polishing-O 2
plasma etching process is found to be very effective in revealing the morphology of
the GNP/polymer nanocomposites. Sharp images can be obtained by using SEM on
the sample prepared in such way. Thanks to the good contrast between nanofillers
and the matrix, the image can be processed by Image-Pro software and information
such as particle size and orientation can be gathered for further analysis. The
complementary fluorescence image can locate the epoxy fixative material used in the
preparation process and helps interpret the morphology of samples with pores or
cracks inside.
By revealing the morphology, the performance of the composites such as mechanical
modulus and strength, electrical and thermal conductivity, coefficient of thermal
expansion, O 2 permeability can be related to the distribution, dispersion and
orientation of GNP particles. It is found that high degree of alignment benefits
183

in-plane properties such as modulus and CTE but prevents particles from connecting
with each other, resulting in poor electrical and thermal conductivity. Random
orientation of the particles, on the other hand, gives the particles a better chance to
touch others and helps form a continuous phase of the filler, resulting in high
electrical and thermal conductivity. The quantified information about particle
alignment can be used to modify models such as Tandon-Weng’s equation for
Young’s modulus and Bharadwaj’s equation for permeability. The modeling of linear
thermal expansion (LTE) was also applied and explained by using the orientation
information qualitatively.
Based on the morphology-performance relationship attained from morphology study,
a few specially designed composites were fabricated by nanostructuring. First, the
orientation of the particles in the composites is modified by annealing. With 1 hour
annealing under 340oC the injection molded GNP/PEId samples show significantly
improved electrical conductivity, reducing its percolation threshold from 10 to 3 wt%.
Secondly, uniform alignment of GNP was achieved by fabricating GNP/PEId films
through extrusion casting, resulting in larger improvement in elastic modulus and
strength to the neat polymer film as compared to the improvement ratio of injection
molded composites. The alignment also benefits the barrier properties to O 2
permeation, resulting in over 65%

reduction to O 2 permeability with 10wt% GNP-5

loading and 34 stretching ratio, while only 50% reduction if the sample is prepared by
compression molding. Annealing can also be used to adjust the degree of alignment,
resulting in at least one order of magnitude lower electrical resistivity and over 100%
184

improvement in thermal conductivity.
A new efficient, solvent-less compounding method, solid state ball milling (SSBM),
has been developed. This procedure succeeds in lowering the percolation threshold
by forming a continuous GNP coating on the surface of size-reduced polymer powder
with good adhesion. Composites fabricated from SSBM approach showed similar
percolation profile as pre-coating but with better mechanical properties without
using solvent. The combination of melt-extrusion and SSBM has been found to be
capable of providing a composite with both high mechanical properties and electrical
conductivity. Thermal conductivity can also benefit from SSBM and pre-coating. A
second order polynomial relationship is found between the thermal conductivity and
the loading of GNP in pre-coated GNP/PEId composites.
Finally, in order to fully utilize the excellent properties of GNP, a novel procedure was
developed to fabricate a paper form GNP assembly and make its corresponding
composites. Binder free, self-standing flexible GNP paper has been successfully
fabricated through filtration of GNP/polyethylene imine (PEIn)/water suspension
followed by annealing. The GNP paper exhibits excellent electrical conductivity
(~1800 S/cm) which is higher than other reported similar type of materials. The
thermal conductivity (246 W/mK) is comparable to other commercially available
graphite material. Processes have been developed to impregnate the GNP paper with
either thermoset or thermoplastic matrices. The resulting paper composite (60vol%
GNP) showed extremely high tensile modulus (22,000 MPa) and reasonable strength
(32 MPa) while still retaining electrical and thermal conductivity. When applied to a
185

fiber reinforced composite laminate, the additional GNP paper layer can significantly
reduce the O 2 permeability and add in-plane electrical and thermal conductivity to
the composite. However, through-plane properties are not improved significantly
because of the alignment of the anisotropic GNP particles.
Table 9.1, Summary of properties of GNP/PEId composites and
GNP paper composites
Relative
Loading
Electrical
Thermal
Modulus Strength
& GNP
conductivity conductivity
(GPa)
(MPa)
type
(S/cm)*
(W/mK)*

Injection
molding
Compression
molding
SSBM &
Pre-coating

Extrusion
cast film

10wt%
GNP-5
Anneal
ed
10wt%
GNP-5

6.6

98

O2
Permea
bility

E-6.7
E-3.0

5.7

98

E-4.5

5wt%
GNP-15

3.9

65

1.25

10wt%
GNP-5

6.8

106

E-6.1

Anneal
ed
100%
7.0
Pressed GNP GNP-15
paper
100%
worm
GNP paper
80wt%
19
-PEId
GNP-15
GNP paper
70wt%
22
-PEId
GNP-15
GNP paper
15wt%
-epoxy
GNP-15
*: in-plane if not otherwise stated

0.40
0.85
(throughplane, 3.3
@ 10wt%)
0.29
(throughplane)

E-4.3

0.80

850

246

2200

2.8

0.35

313

23

620

32

550

0.01
20.5

186

0.02

A summary of properties of GNP/PEId composites and GNP paper composites is
shown in Table 9.1, where a good variety of GNP based composites with different
properties is presented. Based on the desired application, a composite with modulus
as high as 22 GPa, or electrical conductivity of 2200 S/cm, or thermal conductivity of
313 W/m/K, or relative O 2 permeability as low as 1.1% as compared with neat
polymer, or a combination of these properties can be produced. The possibility of
synthesizing multifunctional composites with the addition of small amounts of GNP is
feasible.

Future Investigations and Opportunities
Incorporating paper form GNP into composites seems to be a very promising
approach for developing both structural and functional materials. Due to the high
electrical conductivity, the GNP layer may be able to serve as electronic protection if
inserted properly in a laminate composite. One potential new application is whether
the GNP is conductive enough to serve as a primary lightning strike protector for
composite aircraft. Because of the increasing interest of using composite instead of
metal to construct the airplanes, the lightning strike becomes an issue. Now copper
mesh is mostly used in the airplane shell for this purpose. The advantage of using
GNP paper is its relatively lower density as compared with metal. Composite panels
with GNP paper close to surface can be fabricated through autoclave for lightning
strike testing.
The GNP/polymer composite can also be used for fuel tanks. The conductive additive
187

would help prevent static charge hazard while providing extra barrier properties to
fuel molecules.
To further improve the performance of GNP/PEId composites, surface treatment is
needed for the GNP particles. Normally, fillers with higher aspect ratio are preferred
for better properties. Although high-surface-area version GNP which has a higher
aspect ratio can be obtained, its direct use in PEId matrix hardly resulted in better
properties. The reason is, under compounding conditions such as melt-extrusion, the
thinner particles usually results in agglomerates and are hard to be dispersed well in
the polymer matrix. If an adequate surfactant for GNP can be developed to enhance
its compatibility with PEId, this problem can be solved and better performance may
be achieved by using high-surface-area GNP. Pyrene compound, for example, is a
good candidate for this kind of surfactant. The flat aromatic ring system of pyrene
has a strong interaction with the graphite basal plane. Meanwhile, molecular
branches that are compatible to the polymer matrix can be introduced to pyrene. If
the modified pyrene compound is coated to GNP particle surface, then the added
branch can help the GNP disperse well in the polymer matrix. The challenge here is
to find an adequate molecular branch that has good compatibility with PEId.

188