1:, u. 9!
gamma.

0‘
ID

 

.31...
(1:63.413 {
H ....
inshm...
V hflufimvf
1

V. . . , .. , Raunfi, Nahum
, . . . Lina}. ,-

- y .A -.....t... .. ... . 1

33.3.5... .1 ‘ a .

 

 

 

"19"“5‘18

 

AM} }, UBDARY
“1%” (p ’ Michigan State
nh ..Anr :+
I II VUI Slty'

 

 

 

This is to certify that the
thesis entitled

MODIFICATION OF THE SURFACE CHEMISTRY OF
GRAPHITE NANOPLATELETS AND THE EFFECT ON
GRAPHITE NANOCOMPOSITES

presented by

Douglas Harrison Walden

has been accepted towards fulfillment
of the requirements for the

Master’s degree in Chemical EMneering

' ,7; 1- .
l /~" /
\ ‘i ,' i ,1 ’1 ‘ / 4/:7
>4 IKLJ / , at t,

Major Professor’s Signature

 

 

\

 

DL\\\
(1»

June 23. 2003

 

Date

MSU is an Affimiative Action/Equal Opportunity Institution

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

flm'zfins

 

\\

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJClRC/Datoouepss-p. 15

MODIFICATION OF THE SURFACE CHEMISTRY OF GRAPHITE
NANOPLATELETS AND THE EFFECT ON GRAPHITE NANOCOMPOSITES

By

Douglas Harrison Walden

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE

Department of Chemical Engineering and Material Science

2003

ABSTRACT

MODIFICATION OF THE SURFACE CHEMISTRY OF GRAPHITE
NAN OPLATELETS AND THE EFFECT ON GRAPHITE NANOCOMPOSITES

By

Douglas Harrison Walden

Composite materials are heterogeneous systems where a filler is used to reinforce
a matrix. In recent years, the use of nano-scale fillers has become increasingly popular
for composite systems. Carbon nanotubes have been used as nano-fillers because they
are extremely stiff and strong, and conduct electricity and heat, but are expensive.
Exfoliated graphite has the same mechanical properties as carbon nanotubes, but has
platelet morphology. It is much less expensive than carbon nanotubes ($5/ lb. compared
to $50—500/ g). The interfacial area of exfoliated graphite is very high, and the interfacial
adhesion is critical to the success of its implementation as a nano-filler. The surface
energy of exfoliated graphite has not been extensively studied. It is the focus of this
work to characterize the surface chemistry of exfoliated graphite, modify it by various
techniques and study its effect on the physical properties of exfoliated graphite
nanocomposites. X-ray photoelectron spectroscopy and a surface energy study utilizing
capillary wicking (a procedure for which was developed specifically for this study) were
used to characterize the surface. Plasma treatments and polymer grafting were used to
modify the surface. Flexural testing was used to determine mechanical properties and
scanning electron microscopy was used to study the dispersion morphology on fracture
surfaces. Extrusion mixing improved the mechanical properties, but chemical

modification did not improve these properties.

ACKNOWLEDGMENTS

I would like to thank the following people for their invaluable assistance in the
completion of this degree. I thank my wife, Rachel, for her support and encouragement.
I thank Dr. Larry Drzal, for his tutelage and service as my advisor; Drs. Jayaraman and
Lee for their service on my committee. Thanks to Dr. Askeland and Dr. Schalek for their
assistance with XPS and ESEM. My fellow graduate students, Dr. Fukushima, Kyn'aki
Kalaitziduo, and Inwhan Do were invaluable as informal technical advisers. Thank you

NASA for funding this project.

iii

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

KEY TO SYMBOLS AND ABBREVIATIONS
CHAPTER 1: INTRODUCTION

CHAPTER 2: LITERATURE REVIEW

Section 2.1: Exfoliated Graphite Materials

Section 2.2: Surface Modification
Section 2.2.1: Plasma Modification
Section 2.2.2: Polymer Grafting

Section 2.3: Surface Energy
Section 2.3.1: Obtaining the Contact Angle
Section 2.3.2: Surface Energy Theory

CHAPTER 3: EXPERIMENTAL MATERIALS AND METHODS
Section 3.1: Graphite Exfoliation, Grinding and Surface Area
Section 3.1.1: Graphite Exfoliation
Section 3.1.2: Grinding Exfoliated Graphite
Section 3.1.3: BET Surface Area determination
Section 3.2: Surface Characterization
Section 3.2.1: X-ray Photoelectron Spectroscopy
Section 3.2.2: Column Wicking
Section 3.2.3: Environmental Scanning Microscopy
Section 3.3: Surface Modification
Section 3.3.1: Gas Plasma Modification
Section 3.3.2: Plasma Polymerization
Section 3.4: Nanocomposite Processing
Section 3.5: Thermal Characterization

CHAPTER 4: GRAPHITE SURFACE CHARACTERIZATION AND
MODIFICATION
Section 4.1: Oxygen Plasma Treatment and the Effect of Air Exposure
Section 4.2: Allylamine Plasma Polymerization Treatments
Section 4.3: Plasma Polymerization Treatment by Cycles.

CHAPTER 5: GRAPHITE NANOCOMPOSITES
Section 5.1: Epoxy Nanocomposites
Section 5.2: Vinyl Ester Nanocomposites

iv

vi

vii

@000

15
16
18

22
22
22
23
24
24
24
25
29
30
3O
31
32
34

35
35
41
43

52
52
69

CHAPTER 6: CONCLUSIONS

REFERENCES

73

76

LIST OF TABLES

Table 4.1: This table expresses XPS results indicating the exfoliated graphite
surface chemistry before plasma treatment, after treatment and the effect of
air exposure. 38

Table 4.2: Initial results from allylamine plasma polymerization experiments. 43

Table 4.3: Results from allylamine plasma polymerization experiments. 44

vi

LIST OF FIGURES
Figure 1.1: Plot of surface area of exfoliated graphite vs. particle diameter
and thickness. Surface area was based on theoretical surface area predictions.
Figure 2.1: Schematic of Exfoliated Graphite
Figure 2.2: Sessile drop contact angle schematic.

Figure 2.3: Work of adhesion at varied pH with various chemical functionality
on the surface.

Figure 3.1: Schematic illustrating the preparation of expanded graphite from
natural graphite flakes: solid lines denote graphite sheets.

Figure 3.2: Schematic of experimental set-up for the compression procedure.

Figure 3.3: Plot of compressive force vs. compression distance for packing a
capillary of exfoliated graphite powder.

Figure 3.4: Image showing the rotational mixer for use with the drill press.

Figure 4.1: Surface energy characterization for lS-um, heat exfoliated
and l-ttm microwave exfoliated graphite.

Figure 4.2: Plot of wicking results for untreated and plasma treated
exfoliated graphite indicating the increased acidic functionality that comes

from plasma treatment.

Figure 4.3: Plot showing how the work of adhesion decreases with
exposure of oxygen plasma treated graphite to air.

Figure 4.4: Chemical structure of allylamine.

Figure 4.5: Plot of surface energy of allylamine plasma polymerized graphite
compared to untreated graphite.

Figure 4.6: XPS spectra of the N l 3 peak for allylamine plasma polymerized
treated graphite when an oxygen plasma is used to prepare the surface.

Figure 4.7: XPS spectra of the N15 peak for allylamine plasma polymerized
treated graphite when a nitrogen plasma is used to prepare the surface.

Figure 4.8: TGA data showing the polymer deposition by plasma polymerization

in a cycle treatment.

vii

15

21

23

27

27

33

37

39

41

41

45

46

47

49

Figure 4.9: Plot showing the chemistry changes invoked as graphite as it is
treated by the cyclic plasma polymerization treatment. 50

Figure 5.1: Comparison of flex modulus of 3 vol. %, l-um exfoliated graphite
when it is untreated and when it is surface treated with allylamine plasma

polymerization. 54

Figure 5.2: Plot comparing thermal conductivity of untreated graphite
nanocomposites at 3 vol. % to allylamine plasma polymerization treated
graphite nanocomposites at 3 vol. %. 55

Figure 5.3: Fracture surface of 3 vol % allylamine plasma polymerization treated
graphite, processed by ultrasonic mixing. This image does not show the voids

that are known to be present in Figure 5.4. It shows the presence of agglomerates.
250X, 200 um scale bar. 56

Figure 5.4: Fracture surface of 3 vol % graphite-epoxy nanocomposite which
was intentionally fabricated with many voids. This image provides side-by-side

comparison of Figure 5.3, which does not contain voids. 250X, 200 um scale bar 57

Figure 5.5: Closer examination of an agglomeration present on the fracture

surface of allylamine plasma polymerization treated graphite. 1050 X, 45 um
scale bar. 58

Figure 5.6: Closer examination of an agglomeration present on the fracture
surface of allylamine plasma polymerization treated graphite. 500X, 100
um scale bar. 59

Figure 5.7: This plot demonstrates the effect of screw speed and residence time
on the mechanical properties of 3 vol. % exfoliated graphite-epoxy

nanocomposites made with l-ttm, untreated graphite. 61

Figure 5.8: Plot showing the effectiveness of the DSM mini-extruder to improve
the dispersion of allylamine plasma polymerization treated graphite material. 63

Figure 5.9: Plot of differential scanning calorimetry of the surface treated
graphite and epoxy extrudate. 65

Figure 5.10: Results of mixed processing with ultrasonication and twin
screw extrusion. 66

Figure 5.11: Plot of theoretical properties by Eshelby’s Method, and comparison
of current results. 68

viii

Figure 5.12: ESEM micrograph of a vinyl ester nanocomposite. This image was
taken at 3000X, with a scale bar of 15 pm. The image shows the random
particle distribution of graphite in the matrix.

Figure 5.13: Plot showing the effect of allylamine plasma polymerization and

oxygen plasma treatment on the mechanical properties of vinyl ester
nanocomposites.

ix

71

72

Abbreviation
VGCF
XPS
ESEM
um

um
VOC
ESR
TGA
GvOC
WA
XRD
GIC
PTF E
DI

DSC
MWVM
OEG
MWEG

AlAmPP

KEY TO SYMBOLS OR ABBREVIATIONS

Meaning

Vapor Grown Carbon Fiber

X-ray Photoelectron Spectroscopy
Environmental Scanning Electron Microscopy
Micrometer

Nanometer

Volatile Organic Compounds

Electronic Spin Resonance
Thermogravimetric Analysis

vanOss, Chandhury and Good

Work of Adhesion

X-ray Diffraction

Graphite Intercalated Compound

Teflon

Deionized

Differential Scanning Calorimeter
Microwave Exfoliated, Vibratory Milled
Oven Exfoliated Graphite

Microwave Exfoliated Graphite

Allylamine Plasma Polymerization

Symbol

Meaning

Mass
Time
Density
Viscosity

Surface Energy

xi

Chapter 1 : Introduction

Composite materials are heterogeneous systems where a filler is used to reinforce
a matrix. Matrix materials are most commonly polymers, but can also include metals,
ceramics, or carbons. In most cases, the filler is stiffer and stronger than the matrix and is
used to increase the modulus and strength of the matrix material. This is especially
important for polymers. Polymers are inexpensive to fabricate and lightweight compared
to metals like aluminum or steel, but cannot be used for widespread structural
applications because they are often brittle, weak or flexible. Adding stiff and strong
materials like glass or carbon fibers increases the strength and modulus of the polymer.
Because polymers have a relatively low density, a strength to weight ratio is used to
compare composite materials to conventional materials like steel. Making composite
materials that can perform better than conventional materials has long been the goal of
composites research. One difficulty that arises is anisotropy. Conventional materials are
isotropic, meaning they have the same properties in all direction. Composite materials
often times utilize aligned fiber orientation, which improves the mechanical properties in
the alignment direction, but the mechanical properties are decreased in the other
directions. This is fine for some structural applications. such as fly-fishing rods. In such
applications, the composite is exposed to very predictable forces and are designed to
withstand those forces. Most of the force sustained by a fishing rod is uni-directional
flexural force from the end of the rod. Therefore, the application does not require the
composite to perform under compression, torsion or tension. However, not all
applications are so predictable, so most composites are required to withstand many

different forces. For example, building materials must withstand the weight of the

building (compression), wind forces (flexion, especially important for tall buildings) and
debris hitting the building (impact). For these applications, anisotropic materials often
struggle to meet design parameters in off-axis stress. As the length of the reinforcement
decreases and randomness of alignment increases, the composite approaches isotropy.
Conventional composites typically refer to polymer systems reinforced with aligned glass
fibers. Advanced composites, refer to those made from carbon or aramid fibers,
ceramics, biomaterials, and nanomaterials. Advanced composites claim a meager 1% of
the world-wide composites production (incidentally, though, they claim 7% of the value).
Even with this small market-share, they occupy almost all of the research efforts in this
field[ 1 ]. With the introduction of carbon nanotubes and advances in other nanomaterials,
the field of nanotechnology has become one of the hottest areas of scientific research. In
composites, many nano-scale fillers have been used, but clays are the most common and
most researched nano-filler material. Pinnavaia, ct. a]. have published an excellent
resource that outlines the research efforts in clay nanocomposites[2]. Carbon nanotubes
are graphene sheets in cylindrical orientation. They have very high modulus, electrical
and thermal conductivity. Their introduction into the marketplace has piqued interest in
nano-filler materials in composite applications. They are of particular interest to the
conductive polymers market, but are very expensive, with costs ranging from $50 to $500
per gram. Their extremely high price makes their widespread usage very difficult.
Therefore, there is a need to develop a low-cost alternative to carbon nanotubes with
competitive properties. Carbonaceous materials such as vapor-grown carbon fibers
(VGCF), carbon black, and natural graphite provide poor physical properties even though

prices are much lower[3].

Graphite is a naturally occurring, layered material. The atomic structure of the
graphene layers is such that spZ-pi bonds hold the layers together by Van der Wall forces.
The galleries (interstitial spacing) of the graphite can be intercalated with sulfuric acid.
Rapid heating causes the acid to vaporize and provides normal forces against the layers.
The layers are expanded beyond what the pi bonds can hold them together, resulting in
exfoliated graphite. Exfoliated graphite has an advantage over other carbonaceous
nano-fillers because it provides very good mechanical properties and a more reasonable
price ($5 per pound, or $0.01 1 per gram) compared to carbon nanotubes. Since the
carbon single walled nanotube and the graphene sheet are one and the same and only
differ in their morphology, their physical properties are identical. The exfoliated graphite
has a platelet morphology, which can be just as useful as the cylindrical morphology. In
his dissertation, F ukushima provides an excellent analysis of the economic advantage of
exfoliated graphite nanocomposites[3].

Though these materials may provide a low-cost, hi gh-performance alternative to
carbon nanotubes, a critical component necessary to fabricate nanocomposites with
useful properties (and consequently industrial application) is the ‘quality’ of the interface
between the matrix and nanoreinforcement. For all composites, adequate wetting of the
reinforcement by the matrix is essential, but not sufficient, to improving mechanical
properties. Interfacial engineering can further improve the properties by increasing the
adhesion between the reinforcement and matrix. These interfaces can be modified by
introducing reactive functional groups on the filler surface, which can react with the
matrix to form covalent bonds with the polymer, or by grafting polymers on the surface

of the filler, which can form molecular entanglements with the matrix. For

nanocomposites, importance of the interface is amplified many times because the surface
area of nano-fillers is much higher than traditional reinforcements such as glass or carbon
fibers (surface area ~0.5 square meters per gram). Figure 1.1 is a plot of theoretical
surface area versus particle size and exfoliation for graphite. It is obvious that as
exfoliation is improved (which increases the aspect ratio) and as the particle size
decreases (which increases the edge area), the surface area increases dramatically.

There are two mechanisms that contribute to the success of nanocomposite
systems. The first is chemical; the second is mechanical. The surface energy of the
composite filler determines its wettability with the matrix, and the possible formation of
acid-base bonds with the matrix. Wettability is critical to composite success, but not
sufficient, and acid-base bonds improve the adhesion between the filler and matrix
thereby improving load transfer. The surface energy of exfoliated graphite has not been
extensively studied. It is the focus of this work to characterize the surface chemistry of
exfoliated graphite, modify it by various techniques and study the effect of this on the
physical properties of exfoliated graphite nanocomposites. X-ray photoelectron
spectroscopy (XPS) and a column wicking study (the procedure for which was developed
specifically for this study) were used to characterize the surface. Plasma treatments and
polymer grafting were used to modify the surface. Mechanical influences determine the
dispersion of nano-platelets into the matrix. Improved dispersion removes
agglomeration, and increases the interfacial area. With increased interfacial area, the
material transfer improves. Flexural testing was used to determine mechanical properties
and scanning electron microscopy (ESEM) was used to study the dispersion morphology

on fracture surfaces.

 

.mcocoGoE 33 Beta. 38235
no woman 33 «Ba coatsm @8533“ wow Sausage 22:3 .m> Bfiafiw 88:85 mo 83 83:8 .8 Si N: Emmi

 

3.3 .20530

«W Mg

1
4%.. _-

ooom .
Es .xefi .
om a

m

]

  

 
   

 

. y

-Om AGN<EV
-ow no; momtam

. ‘ _ ...... ...... ...... \ om?
_ if: iii/[\i. 0.3.

mmmcon... new .2955 2023a .m> ENE: moi oomtzw

 

 

 

 

 

 

 

 

 

 

Chapter 2: Literature Review

Section 2.1: Exfoliated Graphite Materials

Graphite is a naturally-occurring, layered, crystal, which can be intercalated with
sulfuric and nitric acids and heated rapidly to exfoliate the layers. This exfoliated
material forms a vermicular, worm-like structure, which can be ground by ultrasonic
mixing to platelets 15 micron in diameter and 20 nanometer thickness (see Figure 2.1).

The particle diameters can further be reduced by a variety of milling processes,
including ball milling (planetary, vibratory and rotational) and pan milling. Pan milling
is a process developed in China and is modeled after a Chinese stone mill[4]. This
process is typically used to grind polymer powders, such as nylon-6[4], but has also been
used to mill natural graphite down to 10.9 tim[5]. Another form of milling is vibratory
ball milling. For this procedure, graphite is mixed into a slurry of isopropyl alcohol and
placed in a cylinder containing zirconium oxide and shaken for 72 hours. The resulting
particles have been measured to have a 0.86 ttm average particle diameter.

Results using exfoliated graphite particles as reinforcement have been published
as mechanical and electrical reinforcements in polystyrene[6,7] in polypropylene[8] and
in epoxy[9]. Two groups have published work with in situ polymerization of exfoliated
graphite in polystyrene. Chen, et al.[6] reports an increase in tensile strength with
increasing filler loading. Their electrical conductivity properties show percolation
threshold to be at about two weight percent. These results agree with Xaio, et. al.[7],
who report the percolation threshold to be 2.5 weight percent.

Published results from the graphite nanocomposite group at Michigan State

University demonstrate similar trends in epoxy system[9] with percolation occurring at

about 1.1 volume percent (2 weight percent) for 15 um platelets. After vibratory milling,
the threshold is increased to 4.9 volume (10 weight percent) percent. This group has also
explored the use of a twin-screw DSM mini-extruder to compare nanofillers in a
polypropylene matrix[8], and improved the mechanical properties of graphite epoxy

nanocomposites by oxygen plasma treatment.

u . Emmi
8:380 @0320me .8 unmannom H m

A «Man
. “an.” _ H . m.
wwwwammo 82%:

  

Section 2.2: Surface Modification

Surface modification is critical for improving the adhesion between exfoliated
graphite and a polymer matrix. Like carbon fibers (especially highly graphitized or high-
modulus carbon fibers), untreated exfoliated graphite demonstrates poor adhesion to a
polymer matrix[lO]. Many different surface modification approaches have been
developed to improve the interfacial adhesion. Some treatments are effective by
increasing the surface energy of the filler, making it more wettable by the matrix. Other
treatments capitalize on other enhancement methods. For example, polymer grafting
utilizes surface energy and molecular interlocking mechanisms. The number and types of
surface treatments are as broad as the creativity of the scientists that develop them.
Among the many treatments that are available in the literature are: sulfonation[l 1,12,13],
fluorination[14], polymer grafting, coupling agents and plasma treatment. Of these
treatments, coupling agents (also called sizings), plasma treatment and polymer grafting
are the most common. The bulk of the research done for this document has involved
plasma treatment and polymer grafting but the technology for these techniques is still
developing.
Section 2.2.1: Plasma Modification

Plasma treatment was introduced for use as a surface modification in the 19603.
A plasma is defined as “a highly ionized gas which contains equal numbers of ions and
electrons in sufficient density so that the Debye shielding length is much smaller than the
dimensions of the gas[15].” In short, a plasma is a highly excited stream of ions,

photons, electrons and free radicals. A plasma’s properties are quite different from those

of a solid, liquid or gas, so it is often referred to as a fourth state of matter. In recent
decades, plasma technology has advanced as a useful tool for surface modification. It has
environmental advantages because VOC (volatile organic compound) emissions are low;
hazardous chemical species are often reacted to a harmless state, so hazardous waste is
often reduced or eliminated[l6]. Plasma does not tend to degrade the mechanical
properties of materials treated[17,18,19,20] hough there are some circumstances where
plasma does degrade the tensile strength of carbon fibers[21]. It has the added benefit that
many different types of molecules can be used to probe and modify surfaces, and the
same surface can have different properties depending on the plasma to which it was
exposed[21]. Virtually all gases can be used for plasma treatment, including but not
limited to, inert gases (He, Xe, air, Ar, Ne, N2), oxidative gases (H20, 02, O3), reductive
gases (H2 and NH3) and organic vapors.

There are three primary types of plasma treatment: thermal plasma, cold plasma
and hybrid plasma. Thermal plasmas are formed by and in arcs and flames. Cold
plasmas are glow discharges, and hybrid plasmas are formed from such sources as corona
discharge. All of the processes described in this document refer to cold plasma
treatments, so the term plasma will be used generically to refer specifically to cold
plasma. Cold plasma modifies surfaces in a variety of ways, including chain scission,
ablation, polymer deposition, surface cross-linking, and oxididation[22]. Ablation and
polymer deposition (also called plasma polymerization) have opposite effects, but are the
most commonly used forms of plasma modification.

Ablation is the process whereby the surface is sputtered and chemical species are

removed from the surface by the excited plasma constituents. Even in the presence of

10

“inert” plasma gases, these free radicals can react with available oxygen or water to form
an oxidative surface. Oxygen and ozone plasmas are particularly common because there
is an excess of oxidative chemicals available to react as soon as free radicals are formed
on the surface. For crystalline polymers, these fi‘ee radicals tend to be stationary with
little or no migration; amorphous polymers demonstrate the opposite effect and, in result,
tend to exhibit free radical decay more quick1y[22]. ESR studies indicate that free
radicals formed by plasma treatment can be very long lived, with reports indicating that
some radicals are “stable” for up to 24 hours[23]. This leaves ions and free radicals on
the surface, which are free to react with other species in the plasma stream or oxidize by
reacting with atmospheric oxygen or water vapor. For composite fillers, ablation has the
added benefit of increasing the interfacial area between matrix and filler, thereby
improving adhesion[21]. The use of oxygen plasma to treat carbon fiber, in particular, is
very controversial because there are some circumstances where plasma does degrade
their tensile strength[2 l ]. Because the uses of plasma are diverse, the results on carbon
fibers can be positive, neutral or negative. In most cases, the results are positive, unless
extremely plasma streams are extremely high strength or especially corrosive[24].
Plasma polymerization is the process whereby a plasma stream of an organic
vapor deposits a polymer film on the inside of the plasma chamber, and on the surface of
any test material exposed to the plasma stream. In the same way that almost any gas can
be used for plasma treatment, essentially every organic vapor (or volatile solvent) can be
used as a monomer for plasma polymerization. Aromatics, nitrogenous molecules,
alkenes, cyclic and carboxylic molecules are a few of the types of molecules that have

been used for plasma polymerization. The term “polymer” is misleading. By the strictest

ll

definition, a polymer is a chain of repeating monomer units. In plasma polymerization,
the resulting polymer is rarely, if ever, is a linear (or even cross-linked) molecule with the
elemental stoichiometric ratios of the monomer. However, the term ‘monomer’ is still
used to describe the probe gas and ‘polymer’ is used to describe the coating formed by
the plasma process. Because the polymer deposition is complex, and often times contains
elemental species that were not originally present in the monomer, the mechanism of the
polymer process is not well understood. Attempts have been made to explain the
mechanism for polymerization, but it remains, at best, esoteric, speculative, and
controversial.

Yasuda has done extensive work to describe the plasma polymerization
mechanism, by characterizing polymer films formed on glass slides by many organic
monomers[25]. The energy of the plasma stream is thought to form free radicals from the
probe gas, causing the probe gas to act as a monomer for polymer formation. The most
likely places for radical formation are double bonds, ring opening, aromatic
fragmentation and hydrogen detachment. When the free radical is formed, it is available
for reaction with any other available free radical, or can react with another component to
form more free radicals. According to Yasuda, the polymer is formed by recombination
and reinitiation[25]. Recombination is the process where a free radical of species A
reacts with the free radical of species B to form A-B. A-B is then available to form
another free radical (reinitiation) for further reaction. This proposed mechanism explains
why the polymer is highly cross-linked, allows for fragmentation to explain the
polymer’s structural difference from the monomer units, and allows other species foreign

to the monomer to be incorporated into the polymer. However, Lu disagrees, saying that

12

the incorporation of extra atomic species (like oxygen) comes from residual free radicals
which react with atmospheric oxygen and water vapor[21].

Regardless of their origin, plasma polymers usually contain chemical species that
are not present in the monomers. For example, allylamine (CH2=CHCH2NH2) does not
contain any oxygen molecules, but the empirical formula for plasma formed
polyallylamine on glass is C3H4.7N100,4[25]. The oxygen most likely comes from water
molecules that are present in the atmosphere in the chamber, or adsorbed onto the
chamber walls. Also, in some cases, a gas plasma is run prior to polymerization. This
plasma treatment effectively conditions the surface for polymerization. Since plasma
treatment can cause free radicals, peroxides and changes in surface chemistry, a pre-
polymerization plasma conditioning can modify the surface so that the polymer can easily
react with the surface. If an oxygen conditioning plasma is used, then excess oxygen can
remain in the chamber during the polymerization and react with the forming polymer to
be incorporated into the new molecule. Plasma conditioning is very important for inert
surfaces like graphite. Since surface functionality is limited on exfoliated graphite,
plasma ablation is necessary to prepare the surface. Oxygen plasma can be used if
oxidized nitrogen groups are acceptable. If oxidation is not desirable, nitrogen plasma is
also acceptable, though not as effective.

Section 2.2.2: Polymer Grafting

Polymer grafting is a technique similar in effect to plasma polymerization, but the
polymer is reacted to the surface using a “wet” chemistry approach. There are many
different processes that fall into the general category of polymer grafting. The two most

common are silane coupling agents and free radical polymerization. Silane coupling

13

agents are most commonly used with glass fibers. They are multi-functional silanes with
one to three methoxy or ethoxy functional groups. A methoxy and ethoxy group can
react with hydroxyl and carboxyl functionality to form a covalent bond with the surface.
The other functional sites on the silane are then free to react with other silanes (cross-
linking), or are free to react with the resin. Coupling techniques, like silane treatments,
can be found in any polymer surface treatment or composites textbook[26].

The other form of polymer grafting is free radical polymerization. By this
method, free radicals, which act as reaction sites, are introduced to a surface and initiate
free radical polymerization, with the polymer chain covalently bonded to the substrate.
Research efforts at Niigata University in Japan have resulted in much advancement in the
use of polymer grafting of vinyl monomers onto the surface of carbon black. In one
particular journal paper, they review their success with anionic, cationic and radical
polymerization[27], then proceed to provide evidence of polymerization caused by
surface trapped peroxide groups. These groups are introduced by decomposition trapping
on the carbon black surface, followed by polymerization. The trapped peroxide
molecules initiate a free radical polymerization of vinyl molecules to a conversion of
about 6%. Even at relatively low conversion, the dispersion stability was greatly
increased in tetrahydrofluoran, indicating that the surface free energy was significantly
changed compared to the untreated carbon black.

Alternatively, plasma radiation can be used to introduce free radicals that will
carry on polymerization at the surface. Carbon fibers have been treated this way using
argon and oxygen plasma to irradiate the surface. Yamada’s research group introduced

radicals onto the surface of carbon fibers using argon plasma, then immersed the fibers in

14

an acrylamide solution to graft polyacrylamide onto the surface[28]. The amount of
polymer grafted was proportional to the power of argon plasma exposure. Single fiber
pull-out tests of treated carbon fibers in epoxy showed that as the degree of grafting
increased, so did the pullout strength. This indicates improved adhesion from the
polymer grafting process.

As reported later, this process has been applied to exfoliated graphite
nanoparticles[3]. The particles were treated with oxygen plasma to increase chemical
oxidation, introduce peroxide functionality (albeit very low concentrations of peroxide
groups will be present on the surface) and introduce free radicals onto the graphite
surface. The graphite was immediately removed from the plasma chamber and mixed
into benzene; the dispersion was brought to the boiling point. In another flask,
acrylamide was dissolved in boiling benzene. The two flasks are mixed and allowed to
reflux for 8 hours. Upon completion, the treated graphite was filtered and dried in vaccu.
The amount of polymer on the surface, as determined by thermogravimetric analysis
(TGA) analysis, was about 30%.

Section 2.3: Surface Energy

The contact angle is typically referred to as the angle between a solid substrate

and a sessile drop on the surface of the substrate (see Fig. 2.2), and it provides a wealth of

insight into the surface character of the substrate. It provides access to the surface

«3

Figure 2.2: Sessile drop contact angle schematic.

energy, chemical functionality, and

15

wettability of the solid substrate. The information obtained depends on the model used to
analyze the contact angle data. Obtaining the contact angle for irregular geometries
requires the use of specialized techniques.

An overview of techniques to obtain the contact angle and two thermodynamic
wetting models is presented here. The Good, vanOss, and Chandhury (GvOC) model
using three probe liquids of known character is explained and compared and contrasted to
Fowkes’s model of acid-base functionality. The information presented in this section
will later be used to determine the chemical nature of exfoliated graphite surfaces.
Section 2.3.1: Obtaining the Contact Angle

The sessile drop method is simple and widely used in cases where a large flat
substrate is available. A digital camera is connected to a computer and software system
so advancing, receding and sedentary drops can easily and quickly be analyzed.
However, for solids such as powders and fibers, the luxury of a large flat surface is not
available and obtaining the contact angle is very difficult. Therefore, methods have been
developed to determine the contact angle of solids with irregular geometries.

One such method is especially powerful for powders. This method, a wicking
study, uses the Washbum equation to calculate the contact angle. For this type of
experiment, a powder is reproducibly packed into a capillary tube, which is suspended in

a probe liquid. The Washbum equation is given in (1) in its original form[29].

2 ton cost?
h =
(1) 27?

 

In this equation, the height, h, of liquid uptake is measured at time, t. The contact angle

is calculated after correcting for the effective interstitial pore radius, c, the viscosity, 7],

l6

and the surface tension of the probe liquid, )6 Because measuring the height of uptake is
difficult and often times impossible, the equation can be modified based on the density of
the liquid so that the measurements can be calculated based on the weight of liquid
uptake. Experimentally, a packed capillary is suspended from a microbalance and the tip
is immersed in a probe liquid. The microbalance measures the liquid uptake into the

column, which is used in the Washbum equation (2) to determine the contact angle.

2
(2) cos6 = fl— 77
1 Pic

Where:
mZ/t is experimental data measured as mass squared vs. time
n is viscosity of probe liquid
p is density of probe liquid
0' is surface energy of probe liquid
0 is capillary constant (also called capillarity or material constant)
Since the viscosity, density, and surface energy of the probe liquid are known, and the
experimental data is collected, the capillary constant and the contact angle are the two
unknowns. The material constant can be measured, if certain assumptions are made. If a
low energy liquid is used, such as hexane, and assuming the contact angle is zero, the left
hand side of (1) becomes the cos (0), which is one. Now there is only one unknown, the
material constant. Once the capillary constant is calculated, drying the column in a
vacuum oven and reprobing the sample with another liquid can easily determine the
contact angle between a different liquid.

This technique has been used for many different particulate materials, but has

never been applied to exfoliated graphite nanoparticles. van 055, et. al[29]. used this

procedure to determine the surface energy of polymer powders, including poly(styrene-

l7

divinyl benzene) and poly(glycerin-dimethacrylate). Extensive work has been done using
this technique at the University of Maine, Advanced Engineered Wood Composites
Center to determine the surface energy of wood particles. In one publication, they report
the surface characteristics of maple particles and evaluate the success of polystyrene-
acrylic acid block copolymer treatments[30]. In a related article, a Dutch research group
has evaluated the surface energy of two wood species (spruce and meranti) using sessile
drop, Wilhelmy plate and column wicking studies[3l]. They found that the surface
energy of powdered samples consistently gave lower yLw than measurements taken the
other methods.

Each of these research groups reports a unique packing procedure for their
respective particulates, so packing procedures vary widely. However, the procedures all
share two things in common. First, they are all very highly controlled and highly
customized for the powder being tested. Second, if known disturbances are made during
the experimental process, the change in capillary constant drastically changes the results.
2.3.2: Surface Energy Theory

Once the contact angle is obtained, the information learned about the surface
depends on the model used to analyze the information. The vanOss, Good, and
Chaudhury (GvOC) model and Fowke’s model are two common ways to analyze contact
angle information and obtain acid-base information. Both models use the work of
adhesion, which is directly related to the contact angle. The equation for work of

adhesion (WA) is derived from Young’s equation and is given in (2).
(2) WA 2 7*(1+c036)

Each model uses a slightly different approach to determine the sum effect of WA,

18

obtained by Young’s equation. GvOC considers the work of adhesion to be the sum of
dispersive, acid and base interactions. The GvOC equation is given in (3) where LW
indicates dispersive interactions, “+” indicates the acid component of the surface energy

and “-“ indicates the basic component of surface energy.

(3)149 =}'(1+COS€)=2(}éW}/§W)Q5 +2(yzy§)0.5 +2(71:}§L)0'5

By understanding (3), the contact angle between a given probe liquid and the solid is a
complicated venture. Assuming a probe liquid is chosen that has been completely
characterized (that is, the LW, “+” and “-“ character is known), there are three unknowns
yet to be determined; namely the dispersive, acid, and base components of the surface.
Hence, three probe liquids must be used and three simultaneous equations must be
solved. Equation 4 gives the experimental matrix that must be solved, based on the

choice of probe liquids.

”W57” W Jr— JET/W)
(4) W5? =2* W W W 7;
\Wfl/ _W W WA 7.; )

 

 

 

 

 

 

All of the variables in the 3x3 matrix are material properties of the probe liquids. These
properties are known for many liquids, including water, ethylene glycol,

dimethylsulfoxide, and diiodomethane.

Fowkes’s model is another common method of analyzing contact angle data.
Like the GvOC model, Fowkes considers the work of adhesion as the sum of dispersive
and acid-base interactions. However, where GvOC considers each surface to have an
esoteric “acid component” or “base component,” F owkes considers the acid-base
interaction to be a chemical reaction between acid or base species in the liquid and

functional groups on the surface (see 5).

(5) W. =1+(cos6) = W52 + W = 2(7572’)°-5 {anHAB

Where:
f is a conversion factor
n is the number of reactive sites
AHAB is the heat of reaction between the functional group and Lewis acid or Lewis base
component of the liquid.
Since this model calculates the acid-base component based on the reaction between an
acid or a base, only one probe liquid is needed. Water can be used at a variety of pH
values to provide the change in work of adhesion. For a wicking study, hydrochloric acid
(HC1) and sodium hydroxide (NaOH) are typically used to modify the pH. At moderate
ion additions (pH changes 2-14), the surface energy, viscosity, and density of the probe
liquid do not change[32], so those values can be maintained in (1).

When this approach is used, basic groups on the surface (for example —NH2) react
with HCl to increase WA; likewise, acidic groups on the surface (such as —OH or —
COOH) react with NaOH to increase the WA at high pH. Figure 2.6 demonstrates what a

plot of WA vs. pH would look like for surfaces with (a) no chemical functionality, (b)

basic only functionality, and (c) acidic only functionality.

20

 

 

 

 

 

 

 

b c
c i AB
C
E .. i ”is
J: W i
:3 St. E
'6
x i
g i a." .- \
o
[ IWet. ‘\
o
I WSL
“i .. .
WSL ' W51. 3

 

Figure 2.3: Work of adhesion at varied pH with various chemical functionality on the

surface[33].

21

Chapter 3: Experimental Materials and Methods
Section 3.1: Graphite Exfoliation, Grinding and Surface Area
Section 3.1.1: Graphite Exfoliation

The graphite material for this study was l60-50A, which was received from the
U-Carr Carbon Company, Inc. (Cleveland, OH). The as received material was
approximately 300 run in diameter and had a 5:1 ratio of sulfuric and nitric acids already
intercalated into the graphitic galleries.

Two exfoliation processes were used in this study. Both processes require rapid
heating of the graphite material to quickly volatilize the intercalated acid and force the
layers apart. Because some vapor escapes out the edges of the platelets, perfect
exfoliation is impossible to obtain. The primary mode of graphite exfoliation is by
placing the intercalated material in a furnace at 9000 C for 3 minutes. This produced a
vermicular, worm-like structure that has expanded by 50 to 100 times its original size.
Upon removal from the oven, the graphite was dispersed in acetone and ultrasonicated for
2 hours. At the end of this process the platelets are approximately 15 ttm wide and 10-20
nm thick.

Microwave radiation is the second exfoliation process, and is unique to this group
at Michigan State University matent pending). In this process, about 1.0 g of l60-50A
was placed in the bottom of a 500 mL beaker. A microwave, with 1000 W power was
used for ten 30 second cycles. After each 30-second cycle, the beaker was moved to a
new location on the turntable to protect the turntable and avoid over heating. Most of
the exfoliation occurred in the first cycle, but an X-ray diffraction (XRD) study showed

improved

22

EHNOB EM” 9/
5:3332 7%

Graphite
Flake

6)

IC Expanded

Figure 3.1: Schematic illustrating the preparation of expanded graphite from natural graphite
flakes: solid lines denote graphite sheets[6].

exfoliation with increasing time, up to five minutes. Full analysis of the control
parameters leading to the development of this procedure can be found in Fukushima’s
dissertation[3]. The same ultrasonication procedure was followed for microwave
expanded graphite as was used for furnace expanded graphite. At the conclusion of
ultrasonication, the particle dimensions were 15 um wide and less than 10 nm thick. So,
even though the exfoliation was increased, the particle diameter was the same between
these two methods of exfoliation.
Section 3.1.2: Grinding Exfoliated Graphite

To grind the particles into the submicron regime, vibratory milling was used.
This mill was a Sweco Vibro—Energy. Cylindrical zirconium oxide was used as a grinding
medium in a cylindrical, flat-bottomed jar. About 15-20 g of graphite was mixed into 4 L

of isopropyl alcohol. The vibratory mill was run for 72 hours and the milled material was

23

recovered by evaporation on a hot plate and dried for 24 hours in a vacuum oven at 70°
C. At the end of this process, the particle size was 0.86 pm with the same thickness as it
had before milling. Particle size was determined by ESEM.
Section 3.1.3: BET Surface Area determination

Surface areas were measured on a Tristar 3000. Each powder was outgassed at
250° C and 0.1 tor for 12 hours. Nitrogen adsorption occurred between P/Po of 0 to 0.2.
An average of eight data points was taken to give experimental error.
Section 3.2: Surface Characterization
Section 3.2.1: X-ray Photoelectron Spectroscopy

XPS measurements were perfomted using a Physical Electronics PHI-5400 ESCA
work station. X-Ray photons were generated from a polychromatic Mg anode (1254 eV).
The analyzer was Operated in the fixed energy mode employing a pass energy of 187.85
eV for survey scans and 29.35 eV for utility scans.

The carbon ls spectral envelope was fit using an asymmetric Gaussian-Lorenzian
mix for the main graphitic peak (fit to 284.6 eV). Additional peaks were fit using a
model of approximately 1.5 eV shift per bond to oxygen. Oxygen ls spectral envelopes
were fit with two peaks, one at approximately 533 eV assigned to hydroxyl groups and
one at approximately 531 eV assigned as oxygen bound to two carbons (ether, carbonyl).
Nitrogen ls spectral envelopes were fit with three peaks, one at approximately 400 eV
and one at 402 eV. These two peaks were assigned to nitrogen bound to carbon or

hydrogen. A third peak assigned to oxidized nitrogen, most likely nitrate, occurred at

407 eV.

24

Section 3.2.2: Column Wicking

As discussed in Section 2.2.1, this wicking method was developed for this thesis.
The procedure development was particularly important because, as mentioned in Sec.
2.2.1, reproducible packing is vital to consistent capillarity. Without constant capillarity,
it is difficult to compare repetitions of the same material. The final procedure involved
three steps. First the capillary tubes were filled. For this, 1/ 16” (ID) PTFE wire sleeving
was cut into 2 inch segments and the end capped with PTFE 0.1 pm filter paper. The
filter paper was used to prevent the loss of graphite during the packing procedure, and
was removed before immersing the column in the probe liquid. Approximately 15-20 mg
of graphite was pushed into the tubing using a stainless steel spatula.

To ensure reproducible packing, a United UTS system was used in compression
mode with a 100 lb load cell. As shown in Fig. 3.2, the capillary tube was placed in a
holding block to ensure that the coil memory of the tubing did not interfere with the
compression process. The compression mode was done in three segments. The first
segment was done at 0.5 inches/minute until the compressive force was one pound (325
psi). Then the system slowed to 0.1 inches/minutes until the compressive force was three
pounds (978 psi). Finally, the system slowed the compression rate to 0.05 inches/minute
to a final force of five pounds (1629 psi). Figure 3.3 is a graph of compressive force vs.
distance. This graph is representative of almost all of the columns that were tested. It
should be noted that there are two inflection points, one occurring at one pound, the other
at three pounds. The decision to slow the compression rate was made based on the

presence of these inflection points. It should also be noted that all samples displayed a

25

very large slope at a force of five pounds. This helps build confidence that the columns

were reproducibly packed.

26

  

L0ADCELL$
RAM ROD

HOLDING BLOCK

GRAPHITE
\\

 

 

Figure 3.2: Schematic of experimental set-up for the compression procedure.

 

Graphite Compression Curve

 

 

 

 

 

 

Force (-lbs.)

 

 

 

 

 

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Position (~in.)

 

 

 

Figure 3.3: Plot of compressive force vs. compression distance for packing a capillary of
exfoliated graphite powder.

27

After packing, each sample was measured using a set of digital calipers. Using
this method is only accurate to approximately ~0.2-0.3 mm. However, the length of the
capillary is used only as a reference to ensure the reproducibility of the packing
procedure.

The wicking procedure was conducted on a Cahn 322 microbalance with software
interface. Hexane was used to determine the capillary constant. For every 42 capillaries
(6 pH increments of one material, a complete characterization), 20 were used to
determine the capillary constant. These were suspended from the microbalance, then
brought into contact with the hexane for three minutes. Upon completion, the capillaries
were dried in a vacuum oven at 50° C for 4 hours. Capillaries used with hexane were
recycled for use with another probe liquid; comparisons were made of capillaries that had
been recycled for use with an aqueous probe after hexane, and they were
indistinguishable from capillaries that had not been used with hexane.

For tests with aqueous solutions, nominal pH values of 2, 4, 6, 10, 12 and 14 were
used. Aqueous probe liquid preparation was conducted with deionized water (DI) and
either hydrochloric acid (HCI) or sodium hydroxide (N aOH) to decrease or increase the
pH. At pH 6, DI water was used with no modification. For each pH increment, the pH
was tested before and after each test. In cases where the final pH was 0.25 more or less
than the original pH, the data was not processed and the experiment was repeated with
fresh probe liquid. For nominal pH values of 2, 4, 6, 8, 10, and 12, the actual pH was
held within 0.1 of the nominal value. For nominal value 14, the actual value 13.7-13.8.
Small ion additions of HC1 and NaOH do not affect the viscosity, surface energy or

density of water. Aqueous probes were allowed to wick for 10 minutes. After wicking

28

with an aqueous probe, the samples were discarded. In a very few cases, there was
evidence of graphitic material falling out of the capillary into the probe liquid. In these
cases, the sample was discarded and fresh probe liquid was prepared.

After the wicking was complete, the data was output to a spreadsheet file. The
mass adsorbed was squared and plotted vs. time. The slope of this line was calculated
and if the R2 value was greater than 0.97, the slope was used with no concern. In some
cases, the slope was not constant through the entire 10 minute test. As is common in
wicking studies[34], the plot was closely examined to determine the most accurate slope.
If it could be determined that there was a small portion of data that was skewing the
slope, it was ignored for purposes of calculation and the linear portion of the data was
used.

Section 3.2.3: Environmental Scanning Electron Microscopy (ESEM)

A Electroscan 2020 ESEM was used to determine the morphology of
nanocomposite fracture surfaces. Most images were taken at 30 eV, 1.85 A filament
current and a pressure of 4.0 Torr. In some cases, the voltage or water pressure was
changed to eliminate charging or increase resolution. After micrographs were taken, a
photo editor was used to increase or decrease the contrast or brightness, but no other
modifications were made to any images.

Three types of fracture surfaces were tested. The first type was the fracture
surface obtained by flexural testing. Since these composite systems exhibited
catastrophic failure, the fracture surface was examined to determine morphology of the
failure. The other type was cryogenically fractured. The samples were scored with a

razor blade and placed in liquid nitrogen until about one minute after rapid boiling

29

stopped and nucleate boiling had commenced. Upon removal, the sample was clamped
near the crack then broken by hitting with a heavy object. The final surface was highly
polished. These samples were cut by a slow-rotating diamond saw, then were polished to
a uniform thickness with 320 grit wet/dry sandpaper. The surfaces were smoothed with
2000 grit sand paper, then highly polished with an aqueous dispersion of aluminum
silicate of 0.1 pm particle size.

Plasma etching was used to ablate the “soft” resin from the “hard” graphite to
improve the clarity of particle morphology. Plasma etching was conducted on a Plasma
Science PS 0500, all etching was conducted at 0.050 Torr with oxygen plasma. Some
samples were treated at 275W for 3 minutes; others were treated at 550W for 10 minutes.
Section 3.3: Surface Modification
Section 3.3.1: Gas Plasma Modification

Plasma treatment was conducted on a Plasma Science PS 0500. The probe gases
used were nitrogen, oxygen and argon. Since oxygen plasma was the most effective at
oxidizing the surface, it was the primary gas used. The maximum power output by this
instrument is 550 W. Since a variety of tests were conducted with powers ranging from 5
to 100% of the maximum RF output and at different exposure times, each modified
sample will have a description in the results and discussion section of this work. All tests
were run at 0.050 Torr base pressure.

To prepare samples for plasma, the particles were sieved through a flour Sifter to
screen very large agglomerates. The sifted sample was scattered on an aluminum foil
sheet and covered with a stainless steel screen to help minimize fluidization during

pumping and venting.

30

Section 3.3.2: Plasma Polymerization

All plasma polymerization testing was done using allylamine (H2C=CHCH2NH2)
as the monomer unit. Testing was done using a Plasma Science PS 0500. All
polymerization treatments were preceded by either an oxygen or nitrogen gas plasma
treatment to charge the surface. The gas treatment was conducted at 275 W for 2 minutes
at a base pressure of 0.050 Torr. The polymerization step was conducted at 0.009 Torr.
Treatment time and RF power were variables used to optimize the process and thereby
changed during the experimental process. Some tests were done with the intent of
eliminating or minimizing the effect of oxygen in the plasma stream. For these tests, the
allylamine was placed in a three-neck flask. One neck was connected to the plasma
chamber. The second neck was sealed with a septum so allylamine could be injected into
the flask without opening it to the atmosphere. The third neck was connected, by way of
a three-way valve, to a belt pump and a compressed nitrogen tank. Before running the
experiment, the flask was evacuated with the belt pump and refilled with pure nitrogen
gas. The evacuation-filling process was repeated three times to reduce the amount of
oxygen in the headspace of the flask. It should be noted that the seals between the flask,
valve, pump and tank leaked at low pressures, so as soon as the mass flow controller on
the plasma chamber closed, the flask was refilled with nitrogen gas to prevent oxygen
contamination during the process.

Since plasma streams can only treat exposed surface of material, some testing was
done using a cycle method. For this, one cycle constituted a gas—plasma step (always
nitrogen for this experiment), and a polymerization step (always 5% RF for this

experiment). At the end of these two steps, the material was removed, stirred and

31

redistributed onto the foil substrate. This allowed for more reactive sites to be exposed to
the plasma and increase the overall treatment.
Section 3.4: Nanocomposite Processing

For vinyl ester testing, Dow Derkane 411-C50 was used with CoNap, ODA and
CHP-S. All composites were made at 1 volume percent of exfoliated graphite. Attempts
to process 3 volume percent composites were unsuccessful, probably because the graphite
particles blocked free radicals to prevent chain extension and cross-linkage. To avoid
excessive styrene loss by the resin, the exfoliated graphite had to be added to the 411-
C50, quickly and sufficiently mixed without causing excessive temperature rise in the
resin. Ultrasonic mixing was attempted, but the sample became very hot and did not
cure. Ten percent of the styrene in 411-C50 vaporized within 15 minutes at room
temperature and can affect the cross-link density of the final resin[35]. To prevent
styrene loss, a special, rotational mixer was devised (see Fig. 3.4). This mixer,
constructed of 304 stainless steel, was placed in the freezer overnight before
experimentation. The 411-C50 and graphite were added together in a beaker, mixed
lightly with a wooden spatula to prevent graphite fluidization, then transferred to the
reservoir in the rotational mixer. The rotational bar of the mixer was secured to a drill
press and was lowered into the resin/ graphite mixture. This was mixed for five minutes,
removed from the reservoir and reweighed so the appropriate amount of the other
components could be added. After adding the remaining components, the sample was
mixed with a wooden spatula for five minutes and poured into a silicone-rubber mold to
cure at room temperature for one hour. Caution was taken to ensure that no more than 15

minutes elapsed between when the 41 1-C50 was removed from its storage container and

32

the final resin was poured into the mold for curing. After curing for one hour, a 125° C
post cure was used for 90 minutes. At the end of the post cure, the oven was shut off and

allowed to cool to ambient temperature slowly to avoid excessive residual stress.

 

Figure 3.4: Image showing the rotational mixer for use with the drill press.

For epoxy testing, Stephan-Miller Epon 828 and Huntsman Jeffamine T—403 were
used in a manufacture-recommended ratio of 100:45. All nanocomposites were prepared
at three volume percent. The graphite was added to the Epon 828, and the two
components were mixed by ultrasonic mixing for five minutes. After outgassing, the
curing agent was added and the sample was mixed for five minutes. The sample was
outgassed again and cured at 125° C for 2 hours and post cured at 185° C for two hours.
Like vinyl ester, the oven was shut off after post cure and allowed to cool slowly to
minimize thermal residual stress.

A DSM twin-screw mini-extruder was used to mix unactivated epoxy resin and

33

exfoliated graphite. This instrument was used at 30 °C with varying screw speeds (175
and 225 rpm) and residence times (5 and 10 minutes).
Section 3.5 Thermal Characterization

Thermal gravimetric analysis (TGA) was conducted on a TA Instruments, TGA
model 2950. The sample was placed in the furnace and brought to a temperature of 25°
C. The sample was heated at 10° C per minute to a final temperature of 600° C.
Gravimetric loss was calculated based on the loss of mass between 150° C and 500° C,
and calculated as a percentage of the original sample mass.

A TA Instruments Differential Scanning Calorimeter (DSC) model 2920 was used
for calorimetry, including thermal conductivity. For thermal conductivity, samples were
highly polished and placed in the DSC. A gallium standard was placed on the sample,
which was heated at 5 C/min through the melt profile of the gallium. After cooling the
sample, the process was repeated. On the second run, the slope of heat/time was
measured through the melt profile of the gallium. The thermal conductivity was
calculated based on a linear relationship between the slope obtained using quartz and the

thermal conductivity of quartz.

34

Chapter 4: Graphite Surface Characterization and Modification

Section 4.1: Oxygen Plasma Treatment and the Effect of Air Exposure

Since natural graphite is a crystalline material consisting primarily of the inter
graphitic basal plane, the graphene sheets produced as exfoliated graphite are also
primarily chemically inert. Any chemical functionality present occurs only at the platelet
edge and arises from broken C-C bonds on the edge of the graphene sheets of chemical
groups (e. g. oxygen) that have chemisorbed at the edges. These broken bonds usually
react with atmospheric oxygen or water vapor to form C-O-C linkages or C—OH terminal
functionality. Oven exfoliated, unmilled graphite (15 um particle size) has an elemental
ratio of oxygen to carbon (hereafter referred to as the O/C ratio) of 0.022. The oxygen
that is present is completely bound into C-OH or C-O-C groups. Functional peak
analysis of XPS of shows that ~98% of the surface is carbon and ~93% of the surface is
carbon without oxygen. Only ~4% of the surface consists of carbon bound to one oxygen
with little evidence for other functional groups like carbonyl or carboxyl. Microwave
exfoliated graphite shows almost the same results; the O/C ratio is 0.06, with only C-O-C
functionality or C-OH.

Reducing the particle size increases the amount of elemental oxygen on the
surface (the edge to basal surface area increases), but does not have a large effect on the
functionality of the surface. After milling for 72 hours, the particle size is approximately
1 um, and the elemental oxygen increases by almost 100%. For microwave exfoliated,
vibratory-milled graphite, the O/C ratio is 0.055. However, even though the elemental

concentration of oxygen increases, the increased oxygen is in an unreactive state, such as

C-O-C.

35

Surface energy measurements determined by wicking experiments and analyzed
by the Washbum equation coincide with the surface composition determined by XPS
surface analysis. If the functional analysis by XPS reveals highly acidic groups like
carboxylic groups, then the work of adhesion should increase at high pH values. This
comes from the formation of acid-base complexes between the acidic functional groups
and the hydroxide ions in the aqueous solution. Functional groups like C-O-C or C=O
are relatively inert and (unless they are in high concentration) would not significantly
increase the work of adhesion.

Figure 4.1 shows the plot of work of adhesion vs. pH for 15 um and 1 pm
graphite. The l um graphite was exfoliated by the microwave process (MWVM). The
plots for 15 um graphite include material that was exfoliated by the oven process (OEG)
and by the microwave process (MWEG). There is essentially no difference in the surface
energy of these materials across particle size or exfoliation procedure. This result
combined with the XPS results, indicates two things. First, the extra elemental oxygen
on the surface of milled graphite is bound into non-reactive functional groups. Since C-
OH groups tend to react with high pH systems, and since the plot of WA vs. pH is flat
across the entire pH range, this extra oxygen is almost certainly in an unreactive, ether-
like linkages. The same conclusion holds for the 15 um materials, except that there is

less oxygen on the surface and therefore fewer ether-like linkages.

36

 

 

14o ---..- ., . - _ , ,_,_.__
120 S~-~~~ - . ”a ,. ._,,_
a _ _
. < a...
E g +15-um
. -.~
g _ ‘—I——1-um
g g . _,_--
4o 3, -
20 _~ A _a,
o _E __ a . ., ,-
0 2 4 6 8 10 12 14 16

Figure 4.1: Surface energy characterization for lS-ttm, heat exfoliated and l-ttm
microwave exfoliated graphite.

37

2:898 53 he Soto o5 new
E2585 Ste .60an5 «Emma 88mg xbmgono Beta BEafim 83:85 05 @5865 3:58 max mommoaxo 038 SE. ”:4 29C.

 

 

 

 

 

 

 

 

 

 

 

s3: s2 seem as? scam em.” s2 385
seam: s3 s3: s2 seems as? e3 3:85
saw 5

e24- see seam- s3 s5. as; e2: e2. 30?:

as _ .o- snow em. _- sea. sow- em? snow s98 amass 8.0
$320 as .280 am owcmnu as .250 EM QWEEO as .280 am omega “we .250 Ed
33 08.0 30.0 owed ~86 one we
been: a: we an em a: o assess a
a a + a a

 

 

 

 

 

 

 

 

38

Through the course of this study, many different surface modification techniques
were developed or attempted. The first of these is oxygen plasma treatment on oven
exfoliated, 15-ttm graphite. It is clear from XPS results that surface chemistry changes
are induced by oxygen plasma treatment (see Table 4.1). The untreated graphite surface
has an oxygen to carbon elemental ratio of 0.024, while after two minutes of oxygen
plasma treatment, this increases this to 0.040. The functional groups change form too.
The untreated graphite has only C-O-C or C-OH functionality. The plasma treated
material not only has a drastic increase in C-OH and C-O-C functionality, but also
contains carboxyl and carbonyl groups. Figure 4.2 shows a comparative plot of WA vs.
pH for untreated exfoliated graphite and oxygen plasma treated graphite. At high pH, the
oxygen plasma treated graphite has a higher work of adhesion, indicating that the surface

was oxidized by the plasma treatment and acid groups were introduced onto the surface.

140 , A ‘ ~~~ *
120 .-.
100 e

80 ~e -—

 

+’ Uhirééiéd ' l
60 — eeeee e +029|ast:

 

Wa (mJ/m"2)

pH

Figure 4.2: Plot of wicking results for untreated and plasma treated exfoliated graphite indicating
the increased acidic functionality that comes from plasma treatment.

39

Thus, surface energy and surface chemistry results obtained by wicking and XPS
support the conclusion that oxygen plasma treatment increases hydroxyl and carboxyl
groups on the surface. It is also clear from both characterization techniques that the
surface chemistry can change with time on exposure to air, most probably due to
interaction with atmospheric moisture. Functional group analysis is provided for oxygen
plasma treated graphite in Table 4.1. This table includes changes induced on the surface
by exposure to air, and to a humidity chamber. Though the oxygen to carbon ratio is
relatively constant with time, the functionality of the oxygen changes. The decrease in
hydroxyl functionality is clear, dropping from 43% after exposure to the humidity
chamber. On the other hand, carbonyl firnctionality increases by almost 200%. Hydroxyl
groups react more strongly in basic solutions than carbonyl groups, so the work of
adhesion at high pH should decrease as oxygen plasma treated graphite is exposed to air.
This is indeed what occurs, as evidenced by the wicking study. Figure 4.3 is a plot of WA
changes with time obtained at pH 14. It is clear from the plot that the work of adhesion
changes as the functional groups change. Since the work of adhesion is dropping, the

wicking results coincide with the previously discussed XPS results.

40

140-

120

We (mJlmAZ)
8
I—I—I

50 ;.

100 ;M,mn “ _- , ..- U ,

10

 

 

20 30 40 50

Time (hours)

60

+ untr. g;

+ ozpi‘s‘ma tr. 2

Figure 4.3: Plot showing how the work of adhesion decreases with exposure of oxygen plasma

treated graphite to air.

Section 4.2: Allylamine Plasma Polymerization Treatments

The most extensive surface treatment development completed for work done for

this study was allylamine plasma polymerization, which was performed on microwave

exfoliated, vibratory-milled (1 um) material. The objective was to increase the surface

energy by introducing basic and/or acidic functional groups onto the surface. The

hypothesis was that grafting poly-allylamine (see Fig. 4.4) to the surface of the

H2N

Figure 4.4: Chemical structure of allylamine.

graphitic material, the surface energy would increase and the interaction of the treated

41

nanoplatelets with the epoxy matrix would improve. Initial tests were conducted to
determine the proper reaction time and RF power. The intent was to find the optimal RF
power that would cleave the double bond of allylamine without splitting hydrogen off the
nitrogen, based on the assumption that higher plasma power will be more likely to split
hydroben from nitrogen. The reaction time would be determined so that the maximum
amount of polymer could be deposited on the surface in the most time-efficient manner.
Table 4.2 gives a summary of the initial experimental conditions. For initial experiments,
an oxygen plasma was run to break C-C graphene bonds and form free radicals on the
surface (see Sec 3.3.2 for detailed description of this procedure). The first two samples
demonstrate the reproducibility of the procedure, and the third run compares a higher
plasma level for a shorter reaction time. It is clear that in both cases, polymer is
deposited onto the surface. However, there is essentially no difference in the amount of
polymer deposited onto the surface (about 1.4-1 .5% after subtracting the control). The
elemental ratios are also relatively constant, with the O/C ratio being about 0.7. The
nitrogen content seems to be a little more concentrated at higher power, with the C/N
ratio increasing to 0.036 from 0.023. It may seem odd that polyallylamine contains
oxygen even though there is no oxygen in the monomer unit. Since an oxygen plasma
was run, and the reaction chamber was held at normal atmospheric conditions, there was
plenty of oxygen, as well as allylamine, in the plasma stream. Oxygen forms free
radicals and is easily incorporated into the polymer network (the grafted polymer is very

highly cross—linked and could be referred to more as a network than polymer chains).

42

 

 

 

 

 

 

 

 

 

IMWVM AlAmPP1 AlAmPP2 AlAmPP3
. Nitrogen Purged n/a No No No
Q First Plasma Step n/a 02 02 02
“>5 Plasma Poly. % RF n/a 5% 5% 25%
Plasma Poly. Time n/a 15 15 7
% C 94.02 90.67 90.241 89.19
g % O 5.7 6.89 7.41 7.21
o % N 2.03 2.09) 3.23
33 % Zr 0.28 0.37 0.26 0.37
x Zr corr O/C ratio 0.055 0.068 0.076 0.073
lN/C ratio ---- 0.0229 0.0232 0.036
TGA data % loss 0.96% 2.80% 2.49% 2.43%

 

 

 

Table 4.2: Initial results from allylamine plasma polymerization experiments.

Section 4.3: Plasma Polymerization with Nitrogen Plasma Pre-treatment

In an attempt to prevent oxygen from getting incorporated into the polymer

network and enhance the reactivity of terminal amine groups with the surface, the

procedure was modified slightly (see Section 3.2.2 for detailed discussion). The next set
of samples was purged with nitrogen and nitrogen was used as the carrier gas for the
plasma. The plasma strength and treatment time were again varied to determine the best
conditions. Comparing these results to the oxygen-rich results shows that less polymer is
incorporated into the network (implying that oxygen helps extend or cross-link the
chains). These samples have about 0.5% weight loss after correcting for the control. The
amount of oxygen on the surface remains almost constant compared to the control, and
even decreases slightly from 0.055 to 0.047 for samples 3 and 4. The concentration of
nitrogen on the surface is about the same as when oxygen is present, but the relative
concentration of nitrogen to oxygen increases dramatically, so the effect of nitrogenous
groups will be more pronounced. Another result from this test is that polymer deposition

seems to be independent of plasma strength or reaction time. This implies that the

43

plasma strength is important for splitting monomer functional groups, but that deposition

seems to occur very rapidly and does not continue significantly after the initial polymer is

deposited.

 

 

 

 

MWVM AlAmPP4 AlAmPP5 AlAmPP6
. Nitrogen Purged n/a Yes Yes Yes
E First Plasma Step n/a N2 N2 N2
.55 Plasma Poly. % RF n/a 25% 10% 10%
Plasma Poly. Time n/a 7 15 7
% C 94.02 91.82 92.83 91.62
,3 t7:: 0 5.7 5.12 5.21 6.00
o % N 2.64 1.55 2.00
(0’3 % Zr 0.28 0.42 0.41 0.38
x zr corr 0/c ratio 0.055 0.04? 0.047 0.057
IN/c ratio 0.0288 0.0167 0.0218
TGA data % loss 0.96% 1.48% 1.64°/ 1.64%

 

 

 

 

 

 

 

 

Table 4.3: Results from allylamine plasma polymerization experiments.

Surface energy results for this material confirms that the plasma polymerization
treatment successfully increases the surface energy of the graphite particles. Fig 4.5
below shows the wicking results from sample AlAmPP3 in Table 4.2, which is a sample
prepared in the presence of oxygen. It is very obvious from the graph that this treatment
successfully implants basic and acidic groups on the surface. The work of adhesion
increase at pH 2, and drops at pH 4 and 6, indicates that there are strong basic groups on
the surface, such as —NH2. The surprising result is the incredibly high work of adhesion
at pH 12 and 14. This indicates that there are many, very strong acidic groups on the
surface, such as COOH. The presence of hydroxyl terminality is a given. Since the
terminal functionality influences the work of adhesion, it is safe to conclude that the

oxygen introduced into the polymer network is not only bound as cross-linked sites, but

also forms strong acidic groups like COOH on the surface. However, the nitrogen that is
bound to the polymer network is most certainly terminal amines. Figure 4.6 gives an
image of the N13 peak for this treatment. Oxygen bonds shift the binding energy to the

left, so the lack of a double-hump peak indicates that all of the nitrogen is bound as —NH-

01' —NH2.

160 -

nab
N
O

4&— untr. ,
+ AlAmPP

Wa (mJImAZ)

(D
O

 

0 2 4 6 8 10 12 14 16
pH

Figure 4.5: Plot of surface energy of allylamine plasma polymerized graphite compared
to untreated graphite.

45

 

9000 _

 

 

8500 ~
{3, 8000 —
7500 —
li'l til
I] 1A -4 AL;
7000 __ v VII

 

 

 

1 J 1 l 1 l I I _ . 1 1 1
414 412 410 408 406 404 402 400 398 396 394
Binding Energy (eV)

Figure 4.6: XPS spectra of the NIS peak for allylamine plasma polymerized treated
graphite when an oxygen plasma is used to prepare the surface.

46

I

9000

8800 _

8600—

015

8400 -

8200

I

 

8000 e

 

I

7800

 

J

 

 

l l 1 l l I I 1 l 1
41 4 41 2 410 408 406 404 402 400 398 396 394
Binding Energy (eV)

Figure 4.7: XPS spectra of the N18 peak for allylamine plasma polymerized treated
graphite when a nitrogen plasma is used to prepare the surface.

47

Interesting results are obtained when functional analysis is carried out on these
samples. The N15 peak of AlAmPP3 is compared to AlAmPP6 (see Figs. 4.6 and 4.7).
Oxidation of nitrogen shifts the peak to higher energy levels (to 407 from 400 eV).
AlAmPP3 pre—treated with oxygen plasma, had a significant increase in surface energy at
high pH values. AlAmPP6 was expected to be different as a result of eliminating oxygen
from the polymer and prevent hydrogen from splitting from the nitrogen. Comparison of
each specimen’s le peak shows that the AlAmPP3 has no oxidized amine, whereas
AlAmPP6 has about 16% of its nitrogen in an oxidized state. Since oxygen and water
vapor cannot be completely eliminated from the plasma chamber, they will always be
competing in the polymerization reaction. If oxygen is more highly selective than water
vapor, drastically reducing the oxygen in the chamber will allow water to more easily
react with the forming polymer. Water is less likely to form additional free radicals, so
once it reacts with nitrogen, it will become an end group for the polymer.

Section 4.4: Plasma Polymerization Treatment by Cycles.

Since plasmas only react with the exposed surface of 8 treated material, an
alternative procedure was tested. For this procedure, graphite was polymerized (in the
absence of oxygen) for short times and low powers. After the polymerization, the
graphite was removed, stirred to expose new reactive sites then retreated. Figure 4.8
shows the TGA results when 5% RF power was used. The plot shows that as the number
of treatments increases, the amount of polymer on the surface increases also. The
polymer deposition increases rapidly to 1.5%, then slowly increases with further

treatment. This seems reasonable because, 1.5% was added to samples 4, 5 and 6 above,

48

which were treated in one step at varying times. It also indicates that very little polymer
growth occurs after 1.5%, with the sample taking 36 minutes (or 12 intervals) to reach the

2.5% weight loss.

3.00% - --., - ,-

 

2.50% -4
2.00% 'i e-

1 50%

 

Wt. Loss (%)

1.00%

0.50% -- WW -- - I

0 1 0 20 30 40

Time (min)

Figure 4.8: TGA data showing the polymer deposition by plasma polymerization in a
cycle treatment.

49

Figure 4.9 provides XPS elemental analysis of these tests. The graph plots the
WC ratio and the N/C ratio for this the test conducted at 10% RF and 5% RF. The initial
values of O/C for each RF power are very similar to the O/C ratios for samples 4 through
6 in Table 4.1. At 10% RF, the O/C ratio oscillates around the initial value of 0.06, with
an overall upward trend. At the lower power, the O/C and N/C ratios both track the

increase in deposited polymer as seen in the TGA plot.

:57. RF O/CT:
E—l—-5% RF N/C ;_
A 110% RF O/C

197951: ”/9 1

 

15

Exposure Interval Number

Figure 4.9: Plot showing the chemistry changes invoked as graphite as it is treated by the
cyclic plasma polymerization treatment.

These results indicate that more polymer can be grafted onto the surface by
repeating the plasma treatment. Mixing the graphite exposes new reaction sites and
polymer reacts onto the surface. However, the change is not drastic, only increasing the
polymer on the surface by 1 wt. %. It should also be noted that the increase in polymer
deposition comes at a material cost. The pumping and venting cycles of the plasma
chamber fluidize the very small particles. The stainless steel mesh that is used to cover
the graphite does not prevent the smallest particles from leaving the chamber. The

particles that remain have more polymer on them than in the one-cycle treatment and the

50

elemental increase in oxygen and nitrogen tracks the polymer deposition. At higher
plasma strength, nitrogen appears to get incorporated into the network more quickly and

the incorporation of oxygen is independent of plasma strength.

51

Chapter 5: Graphite Nanocomposites

The surface modifications developed in Chapter 4 were evaluated by flexural
testing of nanocomposites made of epoxy, and vinyl ester. It was planned to mix the
nanoplatelets into the uncured, unactivated epoxy resin by ultrasonic mixing and to
further mix using DSM twin-screw mini-extrusion. Because vinyl ester is prone to
styrene loss, neither of these processes would work for this resin. Ultrasonication
increases the resin temperature very quickly, causes excessive styrene loss, and the
system does not cure. Similarly, the DSM mini-extruder would require too much time
and too much styrene would evaporate before the peroxide activator could be added. For
this vinyl ester matrix, a rotational mixer was developed and used to process the

nanocomposites (see Section 3.4).

Section 5.1: Epoxy Nanocomposites

Flexural results from the allylamine plasma polymerized graphite in epoxy (that
is, the material that was treated with the multi-cycle treatment) were very poor when
mixed with ultrasonic mixing. Figure 5.1 compares the flexural modulus of untreated
and allylamine plasma treated graphite when each is processed by ultrasonic mixing.
Since modulus is a material property, a physical or morphological artifact must be
responsible for preventing the stiffness of the graphite from contributing to the modulus
of the composite. As corroborating evidence, the thermal conductivity was taken for
these same materials. Figure 5.2 is a comparison of the thermal conductivity. The
thermal conductivity and modulus both show the same trend. Electron Scanning Electron
Microscopy was used to examine the fracture surface and determine the morphology of

the nanocomposites.

52

The first hypothesis is that the samples made with surface treated graphite were
improperly made and had voids. However, this is not the case. Figures 5.3 and 5.4 show
the fracture surface of the allylamine plasma polymerization treated surface and the
fracture surface of a composite known to contain voids. The voids in Figure 5.4 are
readily apparent, and are nonexistent in the fracture surface of the treated composites.

Another possibility is that the polymer grafted onto the surface of the graphite is
softer than the matrix and lowers the modulus signifcantly. However, this is not a
reasonable explanation because there is only 1.5 weight percent of polymer on the surface
of the filler, so the polymer contribution to the modulus is negligible compared to the
stiffness of the graphite.

One obvious difference between Fig 5.3 and Fig 5.4 is the presence of large
agglomerates. The agglomerates are approximately 15-30 pm, and are comprised of
many small particles that are < 5 pm (see Figs 5.5 and 5.6). Also, the presence of very
small, randomly distributed particles can be observed in the background of Fig 5.5. This
indicates that not all of the graphite is bound into agglomerates, but only a small amount
was able to disperse into the matrix. These images, along with the mechanical and
thermal characteristics of the composites support the conclusion that plasma
polymerization forms a “net” of polymer around some of the graphite particles and form
agglomerates of about 15-30 pm. These agglomerates are randomly dispersed through
the matrix, but ultrasonic mixing is not sufficient to break these agglomerates. Therefore,

the graphite is prevented from contributing to the modulus or thermal conductivity.

53

 

      

Neat Epoxy Untreated AlAmPP

 

 

 

Figure 5.1: Comparison of flex modulus of 3 vol. %, l-ttm exfoliated graphite when it is
untreated and when it is surface treated with allylamine plasma polymerization.

54

 

 

 

 

 

 

 

 

 

 

 

 

W/(QK)

 

a":WWWWPat-swmit-fi‘i'fim'ma'r "first! rev" ; H .
-_ - '. . -_..-v_ '.- ... -"

 

r:
Fit... -

”a“
V. .
r ‘.

 

Untreated AlAmPP

 

 

 

Figure 5.2: Plot comparing thermal conductivity of untreated graphite nanocomposites at 3 vol.
% to allylamine plasma polymerization treated graphite nanocomposites at 3 vol. %.

55

 

Figure 5.3: Fracture surface of 3 vol % allylamine plasma polymerization treated
graphite, processed by ultrasonic mixing. This image does not show the voids that are
known to be present in Figure 5.4. It shows the presence of agglomerates. 250x, 200 um
scale bar

56

 

Figure 5.4: Fracture surface of 3 vol % graphite-epoxy nanocomposite which was
intentionally fabricated with many voids. This image provides side-by-side comparison
of Figure 5.3, which does not contain voids. 250x, 200 um scale bar

57

 

Figure 5.5: Closer examination of an agglomeration present on the fracture surface of
allylamine plasma polymerization treated graphite. 1050X, 45 um scale bar.

 

Figure 5.6: Closer examination of an agglomeration present on the fracture surface of
allylamine plasma polymerization treated graphite. 500x, 100 um scale bar.

To overcome the agglomeration issue, other processing techniques were
explored. The first technique was the use of the rotational mixer used to process vinyl
ester composites. This mixer did not improve the dispersion, and did not provide
sufficient shear to completely disperse the particles.

The DSM mini-extruder was also tested for its ability to improve particle
dispersion. Before testing it with composites, the neat resin was tested at a screw speed
of 225 rpm, and 30 °C for 10 minutes (the most extreme experimental condition tested in
this study). The strength and modulus of neat epoxy extruded in the DSM was the same
as the resin mixed by ultrasonication, so this study was conducted with confidence that
the processing did not adversely effect the matrix. Untreated, l-ttm graphite was mixed
at 3 vol. % in the DSM at 175 and 225 rpm for 5 and 10 minutes. The results are plotted
in Figure 5.7. This plot shows no difference between each condition in the experimental
matrix, indicating that there is no improvement that comes from increased screw speed or
residence time. However, comparing these mechanical properties to the properties of
nanocomposites fabricated by ultrasonic mixing indicates a much lower modulus. This is
due to procedural design. For the first set of experiments, the graphite and epoxy were
placed in the hopper and mixed with a wooden spatula before injecting into the barrel.
This resulted in many large agglomerates getting stuck to the top of the screws, the large
agglomerates therefore did not get mixed into the bulk dispersion and did not get
included into the final composites. For a second experiment, the graphite and epoxy were

premixed by ultrasonication, then placed in the hopper and injected into the barrel.

60

 

 

,._

4000 Modulus +— 200

 

 

 

 

 

' I+Stren th1 :
i I * g-- a 7+ 180
3500 +-—- ~~— v __-___,-__ _ ~*~ —-+~—~--—-,
l I I l
I + 160
3000 ...- - __-_- ——— ~~— -—5
1 -l- i 140
8 2500 _._- r - - --: 120 r
E. i E
3 2000 W 100 g
s , 1 s
5 1500 L» * --- ,-4- I 80 3?,
. ~- 60
1000 5 -- - ,-
' 40
500 4—— .-_-_ vm h- —-—;_ 20
0 _ _ _.-_-___ -___._ , . . __ :_ 0
Untreated 5 min 10 min 5 min 10 min
sonic

 

 

 

Figure 5.7: This plot demonstrates the effect of screw speed and residence time on the
mechanical properties of 3 vol. % exfoliated graphite-epoxy nanocomposites made with

l-ttm, untreated graphite.

61

Allylamine plasma polymerized graphite was also mixed in the DSM to test
whether the high shear from the twin-screw extrusion would be sufficient to break the
agglomerates and improve the mechanical properties. Using graphite treated with 12
cycles of nitrogen plasma and allylamine plasma polymerization, the treated material was
mixed at 3 vol. % into the epoxy. The slurry was mixed at 175 and 225 rpm for a
residence time of 10 minutes. Results from this experiment reveal that agglomerates are
broken and dispersed by this method. Figure 5.8 indicates the effect of twin-screw
extrusion on the mechanical properties of allylamine plasma polymerization treated
graphite composites. It is obvious that the mini-extruder is superior to ultrasonic mixing,
for this application.

Though the agglomerates are broken and the particles dispersed, the surface
treatment does not improve the adhesion. As seen in Figure 5.8, the strength of the
composites is not differentiable from the untreated graphite composites. Since the
surface energy of the surface treated material was very high and there were many
nitrogen groups on the surface, the hypothesis was that the surface treatment would
improve interfacial adhesion. However, this experiment indicates that the surface
treatment has no effect on the adhesion. This could result from three possible causes.
First, the polymer deposition by the plasma stream is limited to exposed surfaces, so
agglomerated particles would only receive surface treatment on the outermost particles.
Breaking these agglomerates would expose and disperse many untreated particles, which
would mask the effect of the surface treatment. Second, the reaction between the amine
groups could be weak, and not occur at room temperature. Since the polymer is mixed in

the DSM at room temperature, and the

62

8:858 BEQEw 8588:
8888838 «E83 85:83:“ .8 87.886 05 9895 8 8258.88 Ema 85 .20 8858888 05 wfiaonm 8E ”Wm 8:me

 

 

 

 

 

 

28.: SB: 2:8 2:8
8:2,: 8:51: 8:22 888:: 588282
................................. r o
.l 08
.. 11-88
5 00 t M W
m 8 -- 89 m
.u. 09 - 88 m
m 8; -- 88 m
w oi -- w
. . 0008
8_ i .............
of 4 88
8m 88

 

 

 

 

 

63

polymer is not heated until after the curing agent had been added, it is plausible that the
reaction between the curing agent is more highly-selective than the reaction with the
treated graphite. Third, the mechanical properties from untreated graphite could be very
good, so changes to the surface chemistry will not greatly increase the adhesion. If this is
the case, the only successful surface treatments would be wet-chemistry approaches that
capitalize on molecular entanglements.

To test the first hypothesis, a change to the surface treatment was made. It was
found that the lZ-cycle treatment loses approximately 60% of the graphite during the
pumping and venting process in the plasma chamber. Since the particles that are lost will
tend to be the smaller particles, the only particles remaining would be the large
agglomerates. So, the graphite was treated in a one-cycle treatment at 25% RF for 10
minutes. The graphite loss was much smaller (about 20%), so more, smaller particles
would remain. This graphite was mixed into the epoxy resin with ultrasonic mixing,
then injected into the DSM mini-extruder. Flexural results indicate not improvement in
the mechanical properties.

Two test the second hypothesis, a DSC was run on the DSM extrudate of epoxy
and surface treated graphite. The mixture was heated to 200 0C to test for any exothermic
reaction between the epoxy and amine-functionality of the surface-treated graphite. The
results of the DSC are provided in Figure 5.9. Since there is apparently no reaction
between the materials, it can be concluded that the reaction between the curing agent and
the epoxy is the only reaction occurring during the curing process and that any and all

adhesion comes from wetting and molecular incorporation.

 

 

Heat Flow (mW)

 

0 50 100 1 50 200
Temperature (C)

Figure 5.9: Plot of differential scanning calorimetry of the surface treated graphite and
epoxy extrudate.

65

During the initial experiment with the DSM mini-extruder, it was observed that
not all of the graphite was loaded into the barrel. The material is loaded into the barrel
above the point where the recycle is reintroduced to the barrel. When DSM processing
was used alone, there were large clumps of graphite that remained on the screws at the
point of loading. So, the results in Figure 5.8, which show lower modulus for DSM-
processed material than ultrasonic processed material occur because the actual
composites are not three volume percent. Though 3 vol % was loaded into the barrel, the
extrudate had a lower volume percent. To overcome this, ultrasonication and DSM twin-
screw extrusion processes were combined. For this experiment, ultrasonication was used
to “homogenize” the epoxy and graphite, then the DSM was used to further disperse the

particles. The results are astounding, and shown in Figure 5.10.

4000

Modulus :

3000 70:89am?

2000

Modulus (MP3)

1000

 

O

 

Figure 5.10: Results of mixed processing with ultrasonication and twin screw extrusion.
As seen in Figure 5.10, the modulus of the composite greatly increases (this is the

average of 3 reproductions, each with 4 flex bars for repetition and is the average of 12

66

values). These results show that mechanical shear of a pre-mixed filler-resin system can
improve the mechanical properties well above what other processing methods are able to
achieve. The improved mechanical properties are well above the theoretical results fi'om
Eshelby’s method. Figure 5.11 is a plot of the theoretical values obtained from Eshelby’s
method, and shows the improvement obtained by the current results.

The stark improvement probably arises from improved dispersion and improved
exfoliation. Since the results are higher than the theoretical results for an aspect ratio of
78, the particles probably have a higher aspect ratio than when they were originally
loaded into the extrudcr. This increase in exfoliation will increase the surface area, and
allow the graphene sheets to have more interaction with the matrix and further improve

the mechanical properties.

67

 

5000 ; :-__._______ <-;_- ____ W -. W. -W- :_.

ii +Experimental Data

l I

 

 

 

 

 

l E 4500 _;__. ; X Current Results‘
,2 i ; +Aspect ratio=78 l
§§4000~ WW W _
; 3
2'o
f O
1 E 3500 , :---——~
* 2
3'6 -
: C .
1.2 3000 W- ,. - - -_.W__.-_..-_ W

2500 : f - _W .. _ -___.

0 1 2 3 4
Reinforcement Content (Vol%)

 

 

Figure 5.11: Plot of theoretical properties by Eshelby’s Method, and comparison of
current results.

68

 

Section 5.2: Vinyl Ester Nanocomposites

Vinyl Ester is formed by free-radical polymerization. The use of carbon black,
carbon fibers and other carbonaceous fillers is limited because these materials tend to act
as free radical blockers and prevent the vinyl ester from fully curing. This was observed
with exfoliated graphite. When the graphite loading was 3 vol. %, the post-cured resin
was soft, tacky and still fluid in some areas. When the loading was reduced to 1 vol %,
the samples cured, but the modulus remained the same as the neat resin. Since the
modulus remained the same afier adding 1 vol. % of graphite, and the particle dispersion
appears random and well-dispersed (see Figure 5.12), the graphite probably inhibits the
curing of the resin.

Two surface treatments were tested to improve the mechanical properties of the
nanocomposites. Allylamine plasma polymerized graphite was hypothesized as a
potential improvement because the polymer could coat the particles and prevent free
radical blocking. Oxygen plasma treatment was also tested because, as mentioned in
Chapter 2, free radicals are induced on the surface of the crystal and may not decay for up
to 24 hours. The graphite was plasma treated, and immediately mixed into the
unactivated resin. Since the surface probably contains residual free radicals, these should
act as chain initiators to improve the cross-link density of the polymer and also improve
interfacial adhesion.

Figure 5.13 shows the effect of each of these surface treatments. Both improve
the modulus of the material, but not the strength. This indicates that that the surface
treatments are effectively eliminating free-radical blocking, but adhesion between the

particles and matrix is poor. This result is especially surprising for oxygen plasma

69

treatment. Residual free radicals should act as initiator sites and improve adhesion.
However, this view may be an over simplified View of the polymer chemistry. Styrene
content is critical to chain extension and cross-linking of the polymer. When the graphite
is added to the unactivated resin, it is possible that styrene reacts almost instantly with the
free radicals on the surface of the graphite. This could reduce the overall styrene content,
or attract more styrene and create a layer of polystyrene around the particles. Also, the
chains that are initiated could terminate themselves and, in effect, isolate them from the

curing matrix.

70

 

Figure 5.12: ESEM micrograph of a vinyl ester nanocomposite. This image was taken at
3000X, with a scale bar of 15 pm. The image shows the random particle distribution of
graphite in the matrix.

71

 

 

 

Strength (MPa)

    

Neat Vinyl Ester Untreated AlAmPP Oxygen Plasma

Modulus (MPa)

Figure 5.13: Plot showing the effect of allylamine plasma polymerization and oxygen
plasma treatment on the mechanical properties of vinyl ester nanocomposites.

72

Chapter 6: Conclusions

This study was based on the ability to characterize the surface energy of graphite
nanoplatelets, and disperse the graphite into polymer matricies to form nanocomposites.
The testing was designed to determine whether chemical modifications or mechanical
shear will improve the mechanical properties of graphite nanocomposites. The nano-
platelets were successfully characterized by wicking study and XPS. Chemical
modification was successfully completed by allylamine plasma polymerization. The
effect of mechanical shear was completed by ultrasonic mixing, and twin-screw
extrusion.

UTS and a consistent compressive force reproducibly compressed the platelets
into the capillary tube. Wicking analysis by the Washbum equation accurately
characterizes the surface energy of graphite nanoplatelets. Oxygen plasma adds acidic
groups on the surface, but the functionality changes with time exposure to air.
Allylamine plasma polymerization successfully grafis polymer onto the surface of
exfoliated graphite. Pretreating the graphite with oxygen plasma causes primary and
secondary amines, while pretreating with nitrogen plasma causes some amines to be
oxidized. Treating the graphite in cycles, with mixing between cycles increases the ,
amount of polymer grafted on the surface, but causes large agglomerates and graphite is
lost through the pumping and venting process. The surface energy of allylamine plasma
polymerized graphite increases, with the presence of both basic and acidic functional
groups.

Graphite vinyl ester nanocomposites show decreased mechanical properties, due

to free radical blocking. Surface treatment with allylamine plasma polymerization or

73

oxygen plasma improves the modulus, indicating that the polymer cures properly, but
decreases the strength of the composites.

Allylamine plasma polymerization forms agglomerates of exfoliated graphite,
which cannot be dispersed by ultrasonic mixing. Extrusion mixing overcomes
agglomeration, but untreated particles inside the polymer network mask any
improvement from the surface treatment.

Combining extrusion mixing and ultrasonic mixing greatly improves the
mechanical properties of epoxy nanocomposites. The modulus increases drastically,
meaning that the combination of the processing methods improves dispersion, possibly
by improving the degree of graphite exfoliation. The values for modulus are very near
the theoretical values for modulus.

Thus, the chemical modifications that were tested do not improve the mechanical
properties. However, increased shear through extrusion mixing does improve the
properties.

Through the course of this study, many different things were tested, more than
could fit in this volume. Of the chemical modifications that were tested, the use of the
microwave was the most innovative. Unfortunately, the method was not successful at
modifying the surface functionality. Plasma polymerization is, for all intensive purposes,
a successful treatment method. Unfortunately, the loss of graphite by fluidization, and
short grafted chains hinder the effectiveness as a sizing for the graphite in epoxy systems.
I think that successful chemical modifications will have common characteristics. They

will probably be easily controllable and predictable (like wet chemistry approaches and

74

unlike plasma polymerization). They will probably utilize molecular entanglements by
grafting long polymer chains on the surface.

The extrusion mixing process opens a new world of research. There needs to be
further work done to characterize the results I obtained with untreated graphite.
Obtaining the aspect ratio of the extruded graphite will be essential to understanding
these results. Also of interest will be the effect of polymer viscosity and barrel charge
size. I think that successful polymers will be those that are not subject to

homopolymerization and have relatively low viscosities.

75

 

 

 

REFERENCES

1 Comprehensive Composite Materials, Vol. 6, Sec. 6.1, A. Kelly and C. Zweben editors-
in-chief, Amsterdam; New York: Elsevier, 2000.

2 Polymer-Clay Nanocomposites, T. Pinnavaia and G. Beall, editors, New York: John
Wiley and Sons, Ltd., 2000.

3 Fukushima, H., “Graphite Nanoreinforcements in Polymer Nanocomposites” Ph. D.
dissertation, Michigan State University, 2003.

4 Chen, 2., C. Lui, Q. Wang, Polymer Engineering and Science, 41(7):1187-1 195 (2001).
5 Kanshie, L., Q. Wang, Acta Polymerica Sinica, 6:707-711 (2002).

6 Chen, Guo-Hua et. Al. Polymer International 50:980-985(2001).

7 Xiao, Peng, et. A1. Polymer 42:4813-4816(2001).

8 Kalaitzidou, K. and L. Drzal, Proceedings for the 14:}: International Conference of
Composites Manufacturing, San Diego, 2003. in press.

9 Fukushima, H., L. T. Drzal, Proceedings of the American Societyfor Composites, l7th
Technical Conference, 160 (2002).

10 Dilsiz, N., N. Erinc, E. Bayramli, G. Akovali, Carbon, 33(6):853-858 (1995).

11 B. L. Erickson, H. Asthana, L. T. Drzal, Journal of Adhesion Science and Technology,
11(10):1249-1267 (1997)

12 H. Asthana, B. L. Erickson, L. T. Drzal, Journal of Adhesion Science and Technology,
11(10):1268-1288 (1997).

13 Cheikh, R., P. Askeland, R. Schalek, L. T. Drzal, Journal ofAdhesion Science and
Technology, in press.

14 Y. Chong, N. Watanabe, H. Idogawa, A. Wakata, Journal of the Chemical Society of
Japan, 6:746-751 (1993).

15 Dictionary of Scientific and Technical Terms, 5”n ed., S. Parker, editor-in-chief,
McGraw-Hill, pg 1521 (1994).

16 Press release from Fraunhofer 1GB, www.igb.thg.de 2001.

17 Dilsiz, N., N. K. Erinc, E. Bayramli, and G. Akovali, Carbon, 33(6):853-858 (1995).

76

18 Farro, G. J ., and C. Jones, Journal of Adhesion, 45:29-42 (1994).
19 Nirrnala, P., Vaidyanathan, V. N. Kabadi, Journal of Adhesion, 48: 1-24 (1995).

20 Montes-Moran, M. A., A. Martinez-Alonso, J. M. D. Tascon, and R. J. Young,
Composites Part A, 32:361-371 (2002).

21 Polymer Interface and Adhesion, S. Wu, Marcel Dekker Inc., pg. 298-321 (1982).
22 Bamford, C., and Ward, J ., Polymer, 2:277 (1961).

23 Sowell, R., N. DeLollis, H. Gregory, and O. Montoya, Journal of Adhesion, 4: 15
(1972).

24 Li, R., L. Ye, Y. Mai, Composites Part A, 28A: 73-86 (1997).

25 Yasuda, H., M. O. Bumgarner, H. C. March and N. Morosoff, Journal of Polymer
Science: Polymer Chemistry Ed., 14: 195-224 (1976).

26 An Introduction to Composite Materials, 2““I ed. D. Hull and T. Clyne, Cambridge
Solid State Science Series, 1996.

27 Hayashi, S., S. Handa, and N. Tsubokawa, Journal of Polymer Science Part A:
Polymer Chemistry, 34:1589-1595 (1996).

28 Yamada, K., T. Haraguchi, T. Kajiyama, Journal of Applied Polymer Science, 75:284-
290 (2000).

29 vanOss, D., R. Giese, Z. Li, K Murphy, .I. Norris, M. Chaudhury and R. Good,
Journal of Adhesion Science and Technologi, 6(4):4l3-428 (1992).

30 Walinder, M., D. Gardner, M. Kosonen, and G. Caneba, submitted to Holforschung.
31 Meijer, M., S. Haemers, W. Cobben, and H. Militz, Langmuir, 16:9352-9359 (2000).
32 Chemical Engineers’ Handbook 7th ed., Perry, editor.

33 Huttinger, K., el. Al., Journal of Adhesion Science and Technology, 6(3):3l7-33l
(1992)

34 Rulison, C. “Wettability Studies for Porous Solids Including Powders and Fibrous
Materials” Kruss Technical Note #302.

35 Xu, L. “Engineering the Interface of Carbon Fiber and Vinyl Ester Composites”, Ph.
D. dissertation, Michigan State University, 2003.

77

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIII

ll‘IIllllllzllllllslllllzllflllll1111111111