 

 

 

 

, :1

.1.

11:...

a.»

$.___
ﬁiﬁab

«LE...»

1‘
ran .31

a...
An

.k

.40.. 3mm;

1

 

 

 

THESIS

IlllllilllllllllilllllHHlHlllUlil‘llllHUlllHlﬂHHIHI

3 1293 01688

This is to certify that the

thesis entitled

Adhesion of Vapor Grown Carbon Fibers to Epoxy Matrices
And its relation to Fiber and Composite Properties

presented by

Arvind Krishnaswamy

has been accepted towards fulﬁllment
of the requirements for

 

M. S. degree in Chemical Engineering

9%UWW

Major professor

 

Date 33 / W

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

LIBRARY

M'cmgan State .
Unlverslty

 

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINE return on or before date due.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/98 mm“

ADHESION OF VAPOR GROWN CARBON FIBERS TO EPOXY MATRICES
AND ITS RELATION TO FIBER AND COMPOSITE PROPERTIES
By

ARVIND KRISHNASWAMY

A THESIS

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Chemical Engineering

1998

ABSTRACT

ADHESION OF VAPOR GROWN CARBON FIBERS TO EPOXY MATRICES
AND ITS RELATION TO FIBER AND COMPOSITE PROPERTIES

BY

ARVIND KRISHNASWAMY

Vapor grown carbon ﬁbers of sub-micron dimensions were investigated for their
potential as reinforcements for polymeric matrices. The effect of surface treatment of
these ﬁbers, on their adhesion with an epoxy matrix, and the effect of adhesion on the
mechanical properties of vapor grown carbon ﬁber / epoxy composites was studied. It
was found that the ﬁbers possess a low elastic modulus and hence do not have a high
reinforcing ability. It was also seen that the microstructure of the concentric graphene
layers of these vapor grown carbon ﬁbers did not possess the three-dimensional order
typical of graphite and the low ﬁber modulus was attributed to it’s high graphitic
interlayer crystal spacing. Furthermore, it was found that usually effective surface
treatments applied to the vapor grown carbon ﬁbers did not improve adhesion nor affect

the composite mechanical properties.

ACKNOWLEDGEMENTS

I wish to thank Dr. Lawrence T. Drzal for his guidance, support and ideas
throughout my work here. I have learnt a lot in research as well as in professionalism and
in presentation skills ﬁ'om him.

I wish to thank Mike Rich, Phil Culcasi and other members and staff of the
Composite Materials and Structures Center, for their help in carrying out research and
Operation of equipment. I especially thank Venkatkrishna Raghavendran, Shawn Corbin,
Juergen Ludwig, and Dr. Richard Schalek of the CMSC for those long after-hours
discussions about my research and theirs.

I am grateful to Dr. Thomas Bieler for all the help rendered with X-ray
diffraction, Dr. Krishnarnurthy Jayaraman for giving me full freedom in the usage of his
laboratory, and Dr. Dennis Miller for his advice on oxidation of carbons. I am grateful to
my friends Padmesh Venkitasubramanian and Vinita Singh for help with organic
chemistry. I also thank the staff members of the Chemical Engineering Department for
helping me time and again, and especially Faith Peterson, for scanning in all the images I
needed for my thesis.

Finally, I thank all my friends who have been a source of inspiration for hard

work, especially, P. T. Ramakrishnan, Aditya Vailaya, Sunil Chandran.

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.8.1

1.8.2

1.9

2.1

2.2

3.1

3.1.1

3.1.2

3.1.2.1

TABLE OF CONTENTS

Introduction

Motivation

Background

Applications

Micromechanics

Halpin-Tsai Equations and Random In-Plane Orientation
Adhesion

Methods to Improve the Interphase
Carbon Fibers

Ex-Polymer Carbon Fibers

Vapor Grown Carbon Fibers

Objective

Materials

Epoxy Resin

Vapor Grown Carbon Fibers
Experimental Strategies and Methods
Strategies for Surface Treatment
Extraction of Weak Boundary Layers
Fiber Surface Treatments

Oxidation

iv

12

12

12

17

19

19

20

23

23

23

26

26

3.1.2.2
3.1.2.3
3.2
3.2.1
3.2.2
3.2.3
3.2.4
3.2.5
3.2.6
3.3

3.3.1
3.3.2
3.3.3
3.3.3.1
3.3.3.1.]
3.3.3.1.].1
3.3.3.1.l.2
3.3.3.1.2
3.3.3.2
3.3.3.2.]
3.3.3.2.2
3.3.3.2.3

3.4

Amination

Plasma Treatment

Strategies for Analyses

X-ray Photoelectron Spectroscopy
Environmental Scanning Electron Microscopy
Energy Dispersive Spectrometry

Dynamic Mechanical Analysis
Thermogravimetric Analysis

X-ray Diffraction

Experimental Methods

Fiber Variants

Extraction of Polynuclear Aromatic Hydrocarbons
Surface Treatment Methods

Wet Chemical Treatments

Amination

Friedel Craﬁs Alkylation

Ethanolamine Treatment

Oxidation

Plasma Surface Treatments

Methane Plasma Treatment

Air Plasma Treatment

Argon Plasma Induced Allyl Glycidyl Ether Treatment

Composite Preparation

27

28

30

30

31

33

34

36

36

38

40

40

41

41

41

41

41

41

42

42

43

43

44

4.1

4.2

4.3

4.4

4.5

4.6

4.7

Results and Discussion

Energy Dispersive Spectrometry : Efﬁcacy of Wet Chemical
Treatments

X-ray Photoelectron Spectroscopy : Plasma treatment
Environmental Scanning Electron Microscopy

Dynamic Mechanical Analysis

Thermogravimetric Analysis

X-ray Diffraction

Summary

Conclusions

Bibliography

vi

46

46

52

54

58

62

68

75

78

81

INTRODUCTION

1.1 Motivation

Vapor grown carbon ﬁbers are short carbon ﬁbers that have potential use as
reinforcements for polymers, since their composites may be processed using techniques
such as extrusion and injection molding, even into complex shaped parts, due to their
sub-micron dimensions. Although the directional properties of short ﬁber composites are
lower than their continuous ﬁber counterparts, the above-mentioned advantages, along
with the availability of low cost vapor grown carbon ﬁbers, make these ﬁbers into
particularly interesting candidates for reinforcement. When compared with regular carbon
ﬁbers, vapor grown ﬁbers are different in that they can be made in sub-micron diameters
(regular carbon ﬁbers are about 7 pm in diameter on an average ) by a vapor-phase
technique. They also have a unique structural morphology that exhibits a greater degree
of graphitic orientation than other carbon ﬁbers produced at comparable temperatures.
However, as shown by reinforcement theory, their potential cannot be fully exploited
without adequate interfacial adhesion between ﬁbers and matrix, optimized ﬁber aspect
ratio (L; /d) and acceptable levels of dispersion. Composite mechanical properties depend
upon the reinforcing ability of the ﬁber, which is a strong function of ﬁber aspect ratio.
With increased adhesion, more and more of the ﬁber is involved in load transfer, thus
increasing the effective aspect ratio and composite mechanical properties. Unfortunately,
often enough, incompatibility between the polymer phase and the ﬁbrous reinforcement

results in composites with poor performance.

1.2 Background

Composites consist of two or more physically distinct and mechanically separable
materials. They are made by mixing the separate materials in such a way that the
dispersion of one material in another can be done in a controlled way to achieve optimum
properties. The properties are superior or possibly unique in some speciﬁc respects to the
properties of the individual components.

In many composites, a strong and stiff component is present as a dispersed phase,
often in elongated form, embedded in a continuous phase of softer constitution, known as
the matrix phase 1. The matrix keeps the stiff component in the desired location and
orientation, acts as a load transfer medium and protects the dispersed stiffening
component from environmental damages 2. Such composites commonly exhibit
anisotropy — that is to say, their properties vary signiﬁcantly when measured in different
directions. This usually arises because the stiffer constituent is in ﬁbrous form, with the
ﬁber axis preferentially aligned in particular directions. In addition to that, one or more of
the constituents may exhibit inherent anisotropy as a result of the constituent
microstructure. These reinforced composites are classiﬁed based on the nature of the
ﬁbrous phase, or the nature of the matrix phase, or on the alignment of the dispersion. For
example, the ﬁbers may be aligned (in one or more directions) or random. If the ﬁber
axial length spans the entire length of the part or if the aspect ratio of the ﬁber is high
enough, the composite behaves like a continuous ﬁber composite; otherwise, it’s a short

ﬁber one.

1.3 ApplicaLions

Composites typically ﬁnd application where high strength to weight ratios, high
modulus to weight ratios, high dimensional stabilities, high internal damping capacities
or low thermal coefficients of expansion are required. The aerospace market is the largest
sector of composites users and accounts for 60 to 70% of current world-wide output of
carbon ﬁbers. Even there, the highest portion of use is for carbon ﬁber with epoxy matrix
systems. Also used are Boron / epoxy and Boron / Aluminum systems. Apart from
aerospace, composites are also found in marine, land transport and sporting goods
applications 3 .
1.4 Micromechanics

For a continuous ﬁber composite subjected to longitudinal tensile loading, we
can develop a model assuming the composite is linear elastic in the strain range

considered. If E and V stand for Young’s modulus and volume fraction, and if subscripts

c, f, and m represent properties of the composite, ﬁber and matrix respectively, then :

EC = Efo + Eme (1)

This equation assumes compatibility between ﬁber and matrix, i. e. , ﬁber and matrix are
so well bonded together that there is no relative slippage 2.

For an oriented short fiber composite subjected to a tensile load along it’s
longitudinal axis, since the matrix has a lower modulus, the longitudinal strain is higher
in the matrix than in the adjacent ﬁbers. If a perfect bond is assumed between the two

constituents, the difference in the longitudinal strains creates a shear stress distribution

across the ﬁber-matrix interface. Ignoring the stress transfer at the ﬁber end cross-
sections and the interactions between neighboring ﬁbers, we can derive a simple

equilibrium equation for the development of the ﬁber stress along it’s length.

86f 417i

 

ax df ‘2’

Here 0 f , t ., x and d I represent tensile stress in the ﬁber, shear stress across the

interface, variable distance from the ﬁber end up to it’s mid-section, and the diameter of

the ﬁber. This equation Shows the development of the ﬁber stress from a zero value at the
ﬁber ends, to a maximum value at the central region. This stress development occurs over
half a ﬁber load transfer length “ L t” from each end. Thus the ﬁber load transfer length is

the minimum ﬁber length in which the maximum ﬁber stress is achieved 2. Also, “L c” or
the critical ﬁber length, is the minimum, required for the stress in the ﬁber to attain the

ultimate ﬁber strength value ( o m ), at ﬁber mid-length. This can be represented as the

integrated form of equation (2).

(If ]max = Z‘ci — (3)

Gfu = 2T' — (4)

For effective ﬁber reinforcement, that is, for using the ﬁber to it’s potential

strength, one must select ﬁber length, L; >> L. . For a given diameter and ﬁber
strength, L. can be controlled by increasing or decreasing t. . For example, in glass-
ﬁber composites, a matrix-compatible coupling agent may increase the interfacial shear
strength, in turn decreasing L. . If L. can be reduced relative to L; through proper ﬁber
surface treatments, effective reinforcement can be achieved without changing the ﬁber
length. This treatment assumes that the interfacial shear stress is constant along the ﬁber
length.

When the matrix is in the elastic state and the ﬁber-matrix bond is still unbroken,
the interfacial shear stress is not a constant but varies with ﬁber length. Using equation
(2) and a shear lag analysis, Cox 4’ 5 Showed that a higher ﬁber aspect ratio
(length/diameter) as well as a high ratio of Gm/Ef is needed, where G.n , is matrix shear

modulus.

of = Efsc[1— cosh(2nx / df )Sec h(2an /df )] (5)

where, n = [4Gm /Ef ln(1/Vf)]1/2

The factor 8 . , is the overall composite strain.
1.5 Halpin -—Tsai EMons and Random In—plane Orientation
Often composites with short ﬁbers randomly distributed in a three-dimensional

array, are modeled as random in a plane 82. The following equations then hold true.

EC : (3/8)Ec11 +(5/8)Ec22 (6)
PC :Em[(l+gpnpvf)/(1_Tlpvf)] (7)

HP =[(Pf/Em)-1l/[(Pf/Em)+€p] (8)

Here, the composite modulus is an approximate function of the oriented
composite properties in the longitudinal ( ll ) and transverse ( 22 ) directions. P . is the
oriented short ﬁber composite modulus ( either E .11 or E . 22 ) whereas P f is the
corresponding individual ﬁber property in that required direction. The measure of

reinforcing ability ( C; p ) for ﬁbers is given by :
10
SE11 = 2(Lf /df)+4OVf (9)

C31322 = 2 + 40Vf10 (10)

All these treatments assume a good interfacial bonding between ﬁber and
matrix. The aspect ratio (14” df) considered throughout is that which is well bonded
and takes part in load transfer. It is indeed known that the level of adhesion between
ﬁber and matrix affects both the longitudinal and the off-axis mechanical properties of

the composite 10. Thus good adhesion builds in a strong interface which is beneﬁcial to

the system though there are exceptions as in brittle systems, where poor bonding can
have a toughening effect 1.
1.6 Adhesion
Interfacial adhesion can be classiﬁed as:
1.6.1 Physical Adhesion due to wetting of the ﬁber of surface tension 7 f by the

matrix resin of surface tension 7 m , to form an interface of tension 7 f... , thereby

lowering the free energy of the system 7’ 8. Separation of ﬁber and matrix results in
creation of ﬁber and matrix surfaces, at the cost of the ﬁber-matix interface, thus

deﬁning a reversible work of adhesion W. as:

Wa=Yf+Ym"Yfm (11)

Thus, a low interfacial tension indicates a strong interface that requires greater
work to separate into the starting phases. Surface tension is a driving force for wetting

phenomena. A Spreading Coefficient 3 for the above system is given by:

Sf/mz'Yf—Ym“7fm (12)

If S r/ m is positive and deﬁnite, the matrix spreads and wets the ﬁber and does
not merely “sit” on it. Thus wetting is required for good adhesion though kinetic

constraints also demand a low viscosity for the matrix resin.

1.6.2 Chemical Adhesion as deﬁned by the adsorption theory, subscribes to the view
that adhesion to surfaces is due to surface forces amongst atoms and molecules when
there is molecular contact at an interface. These forces include Van der Waal’s and
London interactions, donor-acceptor bonds (acid-base and hydrogen bonding), and
primary chemical bonding 9. These same surface forces are also responsible for the
surface tension driving forces that constitute physical adhesion.

1.6.3 Mechanical Keying or interlocking due to surface irregularities also give rise to
frictional forces that strengthen adhesion. This can however be detrimental too, as for
stiff ﬁbers, irregularities can be areas of high stress concentration, arising to fracture
zones 9.

1.6.4 Diffusion of long chain molecules and consequent molecular entanglement play
an important role where both components are polymeric. For this to occur at a
meaningful rate, the polymers must be above their glass transition temperatures so that
chain segmental motion is not frozen.

However, there is little quantitative connection between the net effect of these
forces and the resulting strength of the bond 9, or extent of adhesion 10. A body of
experimental evidence has expanded the concept of the ﬁber-matrix interface as a two-
dimensional boundary into that of an interphase that exists in three-dimensions. This
interphase exists from some point in the ﬁber where the local properties begin to change
from their bulk values, through the actual interface and into the bulk matrix region 6. The
interphase itself can extend from a few to a few thousand nanometers in depth 6. It
includes chemical and morphological features of ﬁber and matrix that are different from

their bulk regions, additives, impurities, voids, adsorbed gases, etc. Apart from these, the

thermal, chemical and mechanical environments during processing of the composite, also
make their presence felt ‘0. Although it is not yet possible to quantify the interphase from
ﬁrst principles, thermodynamics and experimental data can be used to qualitatively
design the interphase to beneﬁt the composite properties. For this, wetting of the ﬁbers by
the matrix is a prerequisite for which the ﬁber surface free energy will have to be larger.
Solid surfaces exposed to their ambient environment want to minimize their surface
energy and do so by adsorbing material or growing oxides, ending up with a “native”
surface energy lower than the polymer matrix '0. In order to increase this, surface
treatments, sizings and coupling agents have been developed to enhance solid wettability
and may even contribute to chemical adhesion by bonding.
1.7 Method; to Improve the Interphge

The aim of surface treatment is to eliminate weak boundary layers, improve
wetting and achieve molecular contact, alter surface chemistry, and increase surface
roughness (which favors mechanical interlocking and increases surface area for contact).
Generally more than one parameter gets altered at a time. Some of the surface layers may
get etched away. Care is exercised so as not to weaken the material itself.
1.7.1 Solvent cleaning is a common method for the removal of contaminants and
loosely held particles from the surface. It is a good pre-treatment step to expose the
surface to further treatments.
1.7.2 Mechanical abrasion as by grit blasting or by the use of abrasive pads has been
used for ﬁber-reinforced thennosets. Removal of contaminants and increased roughness

account for increased performance but is likely to be operator dependent.

1.7.3 Chemical modiﬁcations are the most widely used of surface treatments and
include wet chemical methods, plasma attack, grafting, UV treatment, etc.

Acid etching causes oxidation of the surface and may introduce speciﬁc
functional groups, as do other wet chemical methods.

Plasma treatments are increasingly ﬁnding use as surface treatments. Plasma is an
ionized gas with an essentially equal density of positive and negative charges 1‘. It can
exist over a wide range of temperatures and pressures. Usually low pressure plasma or
glow discharge is used ( 0.1 to 1 torr ). Most of the positive charges are ions and most of
the negative charges are electrons. Ions nearest the electrodes or material surfaces, can be
accelerated by sheath voltages to 10’s of eV’s “. These cause bond breaking and generate
free radicals on the material surface. Thus the plasma gas contains ions in ppm, 2% - 20%
free radicals, and a large amount of extremely energetic vacuum-ultraviolet ( VUV ) light
11. All signiﬁcant reactions are based on free-radical chemistry. Glow discharge creates a
high density of free radicals, both in the gas phase as well as on the surface of organic
materials, including the most stable polymers. Radio frequencies ( at 13.56 MHz ) are
usually used as the quench time of the plasma species is much longer than the time
between half-cycles of the excitation ”. The sought-after effects are ablation and surface
cross-linking (noble gases like Ar), etching and introduction of functional groups (
reactive gases like 0 2 , SO 2 , NH 3 , N 2, etc. ), and deposition of cross-linked ﬁlms (
using polymerizable gases like methane, methanol, hexaﬂuoroethane, acetylene, ethylene
oxide, vinyl and acrylate gases, etc. ).

Grafting of polymers or oligomers onto a surface is done to increase functional

group interaction form of chemical adhesion, and to facilitate diffusion adhesion by

10

interrningling of chains, which is especially true for thermoplastics. Approaches to
graﬁing are by coupling via side chains, polymer backbone itself, via reactive end-
groups, or by growing the polymer from the surface. The ﬁrst three examples are those of
grafting by termination. By termination mode, good control over the grafted chain
molecular weight is obtained though it is tedious to achieve for species formed by radical
mechanisms 52. By the initiation technique, control over molecular weight is tough as
transfer reactions easily occur. Also polymerization may occur in the bulk rather than on
the surface though crowding though steric effects are less, initially.

UV treatment is also used for initiating photochemical processes to achieve all the
results undertaken by other wet chemical methods. The key is to use UV radiation of
lower wavelengths, that is to say, of higher energy, so as to effect bond cleavage.
Advantages include spatial and temporal control, selective bond cleavages by appropriate
choice of sensitizers and radiation wavelengths, non-contact and clean processing, and
very fast time-scales as UV excitations occur in femtoseconds!

1.7.4 Whiskerizing or, the use of whiskers which are defect-free sub-micron scale ﬁbers
as supplementary reinforcement, is a surface treatment, as the whiskers are directly
grown on the surfaces. This is usually done for carbon ﬁbers and greatly improves
composite interlaminar shear strength '2’ '3.

Other treatments include ﬁber ﬁnishes, ﬁber sizings, coupling agents, etc. Sizings
contain ﬁlm-formers, anti-static agents, lubricants and coupling agents. Their aim is also

to protect the ﬁber from damage during manufacture and processing '0’ '4.

11

1.8 Cagon Fibers
1.8.1 Ex-Polvmer Carbon Fibers 83

Commonly used carbon ﬁbers are made from polyacrylonitrile or from pitch. Ex-
polymer ﬁbers are manufactured by extruding the polymer through a nozzle into a
continuous ﬁlament, stabilization treatment at about 200 to 350 0C in air followed by a
heat treatment at temperatures on the order of 1000 0C to carbonize the ﬁlaments and to
remove elements H, O, N, etc. Further heat treatments in inert atmospheres are done
between 1300 to 3000 0C to modify the mechanical properties. Heat treatments done
above 2200 0C are referred to as “graphitization” steps. These carbon ﬁbers are produced
as continuous ﬁlaments of bundles of ﬁbers called “tows”, with each tow containing
several thousand individual 7 urn diameter ﬁbers. Each ﬁber has a roughly circular or a
dog-bone Shaped cross-section consisting of a disordered arrangement of ribbons, each
ribbon comprising of a sheet or sheets of partially graphitized layers of hexagonal carbon.
Theses planar structures lie roughly aligned in the ﬁber longitudinal direction due to the
drawing and the heat treatment while manufacture, and are responsible for the ﬁber
mechanical properties. These ﬁbers, depending upon their precursor material and
manufacture, possess moduli of 240 to 500 GPa, tensile strengths of 2 to 5 GPa, and
strains-to-failure of 0.3 to 1.5 %.
1.8.2 Vapor Grown Carbon Fibers

Vapor grown carbon ﬁber ( VGCF ), is a class of carbon ﬁber that is distinctly
different from other types of carbon ﬁber in it’s method of production, it’s unique
physical characteristics, and the prospect of low-cost fabrication 21’ 22. The manufacture

is by a combination of catalytic and chemical vapor deposition processes 23. The

12

lengthening of the ﬁber can be independent of the thickening, in the reactor 2'. VGCF’S
may be obtained with diameters from 0.1 to over 100 microns, and with lengths varying
from 50 microns to several centimeters. Some of the properties of the 6 micron
diameter ﬁber are given in Table l . Thus thermal management in high performance
applications ﬁgures as a potential use for these ﬁbers for purposes such as electronic
heat sinks, radiator ﬁns, etc. The small diameter ﬁbers may also ﬁnd use as random
reinforcements, forming isotropic structures, for use in carbon / carbon composites for
pistons, brake pads and heat sink applications. Toughness in composites would also be
beneﬁted due to the non-catastrophic failure mode of these ﬁbers. These ﬁbers, if
intercalated, possess electrical conductivities approaching that of copper and hence, are
candidates for EMI shielding applications. Also, due to the small size of VGCF’S, net
or near-net shape fabrication of parts can be carried out by a variety of molding
techniques.

Carbon ﬁbers can be made from the vapor phase, from various compounds
including benzene, hexane, allenes, and carbon monoxide 22. Macroscopic and
microscopic characteristics of these ﬁbers also depend upon the carbon source 22. Vapor
grown carbon ﬁbers are generally shorter than a few decimeters. Unlike ex-PAN and ex-
pitch ﬁbers, they are discontinuous.

VGCFS have been known since 1890, while studying the thermal decomposition
of hydrocarbons and benzene, and the disproportionation of carbon monoxide. VGCFS
are grown in two ways — by a substrate seeding method, and by a ﬂuidized seeding one.
By the ﬁrst method, ﬁne catalyst particles of iron or other relevant metals are dispersed

onto a substrate. A mixture of hydrogen and hydrocarbon is also admitted into the

13

reaction tube in an electric furnace 22 at around 1100 OC. Hydrogen is essential as it
reduces the catalyst to iron, preventing poisoning by oxygen, etc. Initially a cylindrical
carbon ﬁlament forms over the catalyst. This involves dissolution and diffusion of carbon
at the exposed surface of the catalyst and precipitation at the metal-support interface. The
most effective catalysts are particles about 15 nm in size, or less 23.This ﬁrst ﬁlament has
a hollow core about the same diameter as the catalyst size, and grows to an outer
diameter of up to 85 nm. Lengthening of the ﬁlament stops when the catalyst surface
gets covered with a carbon layer. A suitably large hydrogen content also prevents
excessive thickening and promotes lengthening 23 . Raising the temperature and increasing
the hydrocarbon concentration promotes thickening. By this model (proposed by Gadelle
et. al.), the catalyst particle ends up at the ﬁber tip, and hence must be a non-solid, molten
drop consisting of an iron-carbon eutectic 23 . This was conﬁrmed by Tibbets et. al. 34,
who also showed that presence of sulfur increases the yield by forming a eutectic with the
catalyst (at temperatures depending on the sulfur /iron amounts) and causing it’s melting,
thereby increasing ﬁlament nucleation considerably. Surplus sulfur curtails ﬁber
lengthening from 25 pm to 2 pm 34. Filaments grown with sulfur dissolved in the iron
particle have a larger diameter than usual, and a different orientation of the graphene
planes 34. Excess sulfur produces low-quality ﬁbers 34. A second model proposed by
Madronero for the production of long VGCFS states that the liquid drop is a carbon
material coronene ( C 24 H 12 ), implying that the metal remains on the substrate 23. It is
seen that the net VGCF structure is very complex, consisting of a central nanometric

ﬁlament, a micrometer inner core, and an outer pyrolytic carbon coating 35 .

l4

In the ﬂuidized seeding process, the catalytic particles are either incorporated in
the feedstock or produced in the reactor by decomposition of an organometallic. The
carbon potential may be adjusted to a compromise between lengthening and thickening,
allowing ﬁbers a few microns long and a fraction of a micron in diameter to be produced
36. The VGCFS produced in this manner differ from the substrate seeded ones in that they
have smaller hollow tubes ( 2 to 3 nm )22. This method can produce ﬁbers ranging from
carbon black-like with an aspect ratio of unity, to those with high aspect ratios, with a
growth rate as rapid as 1 mm / min 37. Here again, the ﬁber ceases to grow once the
catalyst is completely covered.

Fibers with diameters of atleast 5 pm have been studied for their characteristics.
The ﬁber’s cross-section resembles that of a tree with concentric annular rings (ﬁgure 2).
The initial ﬁlament lies at the center of symmetry of this structure, with it’s hollow core
lying along the axis of the macroscopic ﬁber. X-ray scattering has shown that these are
partially aligned turbostratic carbon layers 36, with the average basal plane misaligned
from the ﬁber axis by about 10 0 to 15 0 at the central and outer or cortical regions 37. In
many cases, upon cooling from the thickening temperature where the ﬁbers are still
smooth, since the pyrolytic carbon contracts more rapidly in the radial than in the
circumferential, high hoop stresses cause the surface to become crenulated 36. VGCF is
composed of two different micro-structures formed by different mechanisms. The tube-
wall at the core region is composed of linearly extended carbon layers with 2-
dirnensional order, whereas at the ﬁber periphery, two or three carbon basal networks are
layered, indicating micro-crystallinity 22. The ﬁbers show strain stiffening. Whereas other

carbon ﬁbers show a brittle failure, VGCFS with their nested planes, fail

15

non-catastrophically in a sword-in-sheath manner 36, especially aﬁer heat treatment. It is
seen that the tensile strength and modulus of the ﬁber increase with decreasing ﬁber
diameter, from 30 um to 7.5 um 37. Tibbetts et. al. 37 believe that this phenomenon has its
origins in the rates at which chemical vapor deposition (thickening) is carried out. Fibers
thickened by rapid deposition have a lower degree of preferred orientation and hence are
less perfect; so thicker ﬁbers have poorer ordering of basal planes 37, slightly in the radial,
and more so in the axial direction. A more recent investigation has shown a triplex
structure for ﬁbers of diameter 3 pm or greater, comprising of a central carbon ﬁlament,
an inner core and an outer shell and that graphitic radial ordering as shown by the d-
spacing is essentially the same between ﬁbers differing in diameter 35. Madronero and
others 28, have Shown that thick ﬁbers have more hollow spaces or cavities (as in the
Ruland model for carbon ﬁbers) at the cortical zone, than thin ones. Thus when the
pyrolytic coating is very thick, the number of faults increases and weak walls could exist,
which may account for the modulus and strength dependence on diameters. They also
infer that the tree-trunk structure is not a pre-existing condition and is formed during
tensile failure of the ﬁber 28. Their work may also explain the results of Tibbets and co-
workers 38 who found ﬁber density to increase with decreasing pyrocarbon layer
thickness. By comparing surface area measurements from BET measurements of
adsorbing N 2 gas and from geometric considerations of photographs, it was concluded
that the surface has low porosity and is relatively inactive, as far as gaseous adsorption
and possibly adhesion to a polymer matrix is concerned 38. Madronero and Verdu have
reported that the hydrogen content of VGCFS is higher in their core, than in their cortical

layers 39. Serp et. al. have also shown that tensile strength decreases with increasing

16

diameter, and that crenulated ﬁbers have only 45% of the tensile strength of straight
cylinder ﬁbers 40.

The investigations carried out so far on sub-micron diameter vapor grown carbon
ﬁbers, have been on grafting of polymers by termination, and by polymerization from the
ﬁber surface, as measured by improvement in dispersability in a good solvent for the
grafted polymer 47’ 59’ 60. Radical graft polymerization of vinyl monomers has been
achieved initiated by forming azo- and peroxyester groups, the latter indirectly attached,
on the ﬁber surface 59’ 60. Acylum perchlorate groups have been reacted to the ﬁber
surface to effect cationic graft polymerization of vinyl monomers and ring-opening
cationic polymerization of cyclic monomers. Anionic graft polymerization and anionic
ring-opening alternating copolymerization of epoxides with cyclic anhydrides has been
carried out using butyl lithium with crown ethers, and using potassium carboxylate
groups has been accomplished 47. Grafting by termination has been carried out by
condensing amino- and hydroxy- terminal polymers with surface carboxyl groups of
VGCFS, using condensing agents like thionyl chloride with N-methyl pyrrolidone.

Whereas adhesion was not measured, dispersability was seen to improve markedly.

1.9 Objective

The aim of the project is to investigate the role of ﬁber-matrix adhesion for vapor
grown carbon ﬁbers of sub-micron diameters and to ascertain the effect of alteration of
ﬁber surface variables, on the ﬁnal composite mechanical properties. A secondary
objective is to conﬁrm the mechanical properties of the 0.1 to 0.2 pm vapor grown

carbon ﬁbers. Vapor grown carbon ﬁbers are too small to use conventional methods for

measuring adhesion. Reinforcement theory says that for a randomly distributed short
ﬁber composite, the composite modulus is a function of the ﬁber modulus, ﬁber content
and, the aspect ratio (length/diameter) of the ﬁbers. The ﬁber aspect ratio is related to
adhesion. Previous work on large diameter carbon ﬁbers has shown that adhesion can be
improved by suitable surface treatments and coupling agents, that eliminate weak
boundary layers, improve wetting and help achieve molecular contact. Therefore,
applying these methods to the vapor grown carbon ﬁbers, fabricating epoxyNGCF
composites and measuring the composite modulus will allow the project goals to be

achieved.

18

2 Materials
2.1 Epoxy Resin

The term “epoxy” refers to a reactive chemical three-membered cycle, one of
which is oxygen 29. This cyclic is also called the oxirane ring. Standard epoxies are
formed by the reaction of bisphenol A and epichlorohydrin to produce a prepolymer with
two epoxide end groups (ﬁgure 1). They are difunctional bisphenol-A-diglycidyl ether
epoxides (DGEBA). The ether linkages account for their chemical resistance, while the
aromatic groups give strength, rigidity and thermal stability (up to 300 F) 31.

The matrix material used for this research is a DGEBA resin made by Shell
Chemical Co., called EPON 828. The value 33 of the repeat unit “11” is around 0.15. It’s
molecular weight is 378 and it’s epoxide equivalent weight is 189. The curing agent used
is meta-phenylene diamine (ﬁgure 1). It’s molecular weight is 108.14, and it’s equivalent
weight is 27. Hence the stoichiometric quantity of curing agent required is the equivalent
of curing agent divided by that of the resin as this is a one equivalent per one equivalent
system. This is expressed as parts curing agent per hundred parts resin (phr). For mPDA
(meta-phenylene diamine), the stoichiometric quantity is about 14.5 phr 3 1’ 6. The glass-
transition temperature “T g” for the stoichiometrically cured epoxy is about 160 0C, if
cured using a standard cycle of 75 0C - 2 hours and 125 OC — 2 hours 6’ 33. Even when
prepared from stoichiometric reacting mixtures, these epoxies are not highly cross-linked
glasses 32. For this system of DGEBA with mPDA, Rao 30, found that conversion-at-gel or

“or 3..” ranges from 0.67 — 0.73, before post-cure.

l9

2.2 Vapor Grown Carbon Fibers

The present work uses Pyrograf III vapor grown carbon ﬁbers supplied by
Applied Sciences, Inc. The ﬁbers were grown by the ﬂuidized-seeding process, with in-
situ addition of equimolar levels of hydrogen sulﬁde and iron pentacarbonyl catalyst 84,
and were produced in 1997. These ﬁbers 42 have a size range of 0.1 to 0.3 pm in
diameter, with aspect ratios ranging from 40 to 200 84. However, the average aspect ratio
and it’s distribution for this batch of ﬁbers, is not known. The BET speciﬁc surface area
as found by nitrogen adsorption 42 is 11 mz/g. If pyrolysed in nitrogen at 1000 0C, the
surface area becomes 9.8 mz/g suggesting the presence of very few pores. Based on
density and geometry, the calculated values of the ﬁber surface area agree with the
measured values, indicating that the ﬁber surface is smooth, and with very few pores, if

any 42.

20

EPON 828

H20\-—/CHCH 2. C( CH3 )2.0 -—— R CHch— CH2
0 \ /

n

o
R = CH2CH(OH)CH 2.. C( CH3 )2

mPDA

 

 

 

NH2

.NHZ

Figure l : Matrix resin and curing agent

21

    

' A
Figure 2 : End-on view of a 10 micron diameter VGCF

Table 1 : Physical properties of a single VGCF of diameter 7 pm 27’ 4'

 

 

 

 

 

 

 

 

 

 

 

Property Value Units
Length 10'3 to 30 cm
Diameter 10'3 to 0.3 cm
Density 1.8 gin/cc
Tensile Modulus 240 GPa
Tensile Strength 2.9 GPa
Ultimate Strain 1.5 %
Electrical 10'3 9 cm
resrstrvrty
Thermal 20 W/m/K
conductivity
Cost per kg 2 - 10 $

 

 

 

 

22

3 Experimental Strategies and Methods

To maximize composite properties, adhesion between ﬁber and matrix
components must be at an optimum level. Since the matrix is in a ﬂuid state and the ﬁber
is in solid state during processing of the composite, the surface of the ﬁbers is often
subjected to treatments prior to processing to increase the surface energy, so that the
ﬁbers be wetted, and adhere well, with the matrix. In this project, two surface treated
variants of the baseline ﬁbers were obtained from the ﬁber manufacturer itself and used,
and as-received vapor grown carbon ﬁbers were subjected to various surface treatments,

prior to preparation into composites with epoxy resin as matrix.

3.1 Strategies for Surface Treatment
3.1.1 Exgaﬂon of Wea_k Boundary Lavers

The VGCFS are always coated with thin layers of polynuclear aromatic
hydrocarbons (naphthalene and other homologues) probably due to imperfect graphene
deposition, or due to re-condensation of furnace vapors 44. This causes weakly bonded
carbonaceous debris to form on the surface, which is an impediment to adhesion of
VGCF and polymers. An effective surface treatment must remove this coating.
Methylene chloride is a very good solvent for polynuclear hydrocarbons. Applied
Sciences Inc. has reputed the extraction of polynuclear aromatics from VGCF surfaces. 1
gm of ﬁbers extracted with 200 mL acetone for 24 hours obtained 4.89 mg of

polynuclear aromatics from Variant 1 VGCFS 44’ 4°.

23

 

 

A ID AND BA E ATTACK OF EPOXY RING

 

”AER

\1

 

 

0H- Counterion CH2— CH
OH
1
C|)H
CH2 —CH
counterion

 

Figure 3 : Acidic and basic cleavage of oxirane ring

24

 

 

Mechanism of Alkylation and Amination

 

CH2C12 + AlCl3

VGCF
/

/ + ClCl-lz

/ H

’\

d+ d '
Cl-CH2-Cl ....... AIC13

ll

 

 

+ -
ClCHz + AlCl4

/ + AlCl3

==—- / CH2c1

+ HCI

/

' HCI

/ $520 + :NCH2CH2OH—' / CH2NHCH2CH2O

./

Figure 4 : Mechanism of Alkylation and Amination of VGCF

25

3.1.2 Fiber Surface Treatments

The role of surface treatment is to create a surface on the ﬁber that can be easily
wet by the matrix, and maybe, that chemically bonds with the matrix, forming a
beneﬁcial interphase. VGCFS have a low-energy surface. It has been found 58 that the
surface free energy of Variant 1 VGCFS and of methylene chloride extracted Variant l
ﬁbers are quite low, at 24.7 mJ/m2 and 25.7 mJ/m2 respectively. For thermodynamic
wetting to occur, the ﬁber’s surface energy needs to be increased to a value greater than
that of the epoxy matrix ( ~ 35 mJ/m2 ). This may be achieved by introduction of surface
oxygen groups. Chemical bonding can be induced by introducing functional groups that
react with the epoxy moiety, or with the curing agent. The oxirane ring of the DGEBA
matrix resin is highly strained 53 and is easily cleaved by many acidic and basic groups
(ﬁgure 3). So the aim of the surface treatments attempted is to incorporate chemical
groups on the VGCF surface, that have the potential to bond with epoxy moieties, and to
increase the wetting characteristics of the matrix.
3.1.2.1 Oxidation

This is a straightforward procedure for generating surface oxygen groups.
Oxidative treatments can increase the number of edge planes and create more edge pits
by breaking aromatic bonds. They also increase the surface oxygen content as measured
by the number of hydroxyl, ketonic and carboxyl groups 47. Surface oxygen content
increases the surface free energy of the ﬁber. Carboxyl groups can attack the oxirane
ring, for chemical bonding, as shown in ﬁgure 3. They can also react with the amine
curing agent. Tsubokawa et. al. 47 ﬁnd that treatment with 35% nitric acid at 110 C

increases the number of carboxyl groups on the surface of vapor grown carbon ﬁbers of

26

similar dimensions made by Asahi Chemical Industries Ltd., Japan. They used the
surface carboxyl groups to graft and co-polymerize epoxides with cyclic acid anhydrides
on VGCFS 47.
3.1.2.2 Amination

The aim is to introduce amine groups onto the ﬁber surface because amine groups
being highly basic, cleave the epoxy ring easily and effect chemical adhesion. The
arnination mechanism is depicted in ﬁgure 4. VGCF in methylene chloride is mixed with
anhydrous aluminum chloride. The electron pair deﬁcient Lewis acid (aluminum
chloride) abstracts a chlorine anion from the solvent forming an ion-pair, via a transition
state. The cation of this pair is an electrophile while the conjugated graphene layers of the
VGCF surface are electron rich. Thus in the event of a collision between the two, the
VGCF would undergo electrophilic aromatic substitution at an edge plane, with a proton
getting abstracted in the process. This is a typical Friedel-Crafts alkylation mechanism.
The catalytic power of the Lewis acid varies as 54 : AlBr3 > AlCl3 > GaCl3 > FeCl3 >
SbCls > TiCl4 > ZnCl2 > SnCl4 > BF3, SbCl3, BCl3. Aluminum chloride itself has been
used with methylene chloride for Friedel-Crafts alkylation 55. As the edge planes of a I
graphene layer can be modeled as polyaromatic, like naphthalene, pyrene, chrysene, etc.,
it is useful to look at their reactions to gage the reactivity of the carbon ﬁber. Several
researchers have studied the reactions of polyaromatics 56. Price et. al. 57, have used
equimolar amounts of AlCl3 and polyaromatic hydrocarbon and an equal weight of the
alkyl halide with a reaction period of 4 to 5 hours at room temperature and below. The
catalyst is usually added last 6'. In this research, the weight of VGCF used per run is only

25% more than the A1C13 used. Also, the amount of solvent used is such that the

27

transitional collision is favored by better dispersion of the VGCF in the solvent. Also for
stirring to occur, a certain minimum level of solvent was found necessary. The second
step in this two-step reaction is the treatment of alkylated ﬁbers with ethanolamine.
Ethanolamine was chosen for it’s easy availability in liquid form. The mechanism here is
one of nucleophilic substitution at the alkyl carbon of the VGCF. Amines are strong
bases and hence good nucleophiles 53. It must be noted that aromatics can be chlorinated
very easily. Direct chlorination instead of an alkyl chloride group incorporation will be
easier, but the resulting aromatic chloride would be resonance stabilized and unreactive
to nucleophilc substitutions 53 unless used with extremely strong bases such as a mixture
of sodamide and ammonia. Hence the alkylation step is essential. A graphic for the attack
of an oxirane ring by a basic group is depicted in ﬁgure 3.
3.1.2.3 Plasma Treatment

Cleaning processes designed to remove contaminants from the surfaces of
materials play a central role in many industries. Due to solvent removal requirements and
due to environmental and health reasons, common organic solvents are not preferred.
Cold gas plasma of argon has been used in the past for surface cleaning and has been
shown 48 to remove surface contaminants and weakly-bound layers, and produce free-
radicals on the surface, and to create cross-linking of organic surfaces. Plasma gases are
categorized 50 as ( i ) chemically non-reactive plasma, ( ii ) chemically reactive, and ( iii )
polymer-forming plasma. Non-reactive gas plasmas like argon mainly sputter surfaces.
Reactive ones like oxygen, nitrogen and carbon tetraﬂuoride do not polymerize but
chemically react with the surface. Both types ablate the surface. Ablation for type ( i ) is

only by momentum exchange. For type ( ii ), example CF4 , metal ﬂuorides are quite

28

volatile and hence both momentum exchange processes and chemical etching become
important. In plasma forming polymers, ablation and deposition are competing effects.
For a given product of monomer ﬂow rate and molecular weight, the deposition rate rises
to a maximum, saturates, and then drops off, with respect to discharge wattage. This peak
value and saturation period before drop-off, may be increased, by raising the ﬂow rate so.
The polymerization by plasma does not proceed by a chain mechanism because, in high
vacuum conditions under which glow discharge occurs, the entropic loss due to
polymerization is higher than in usual liquid phase reactions, thus having associated with
it, a very low ceiling temperature 50 (for depolymerization). Also, for plasmas, it is a non-
equilibrium temperature distribution as the temperature of a free electron is usually a few
orders of magnitude greater than the ions and radicals. Although the reactive species are
free-radicals, the mechanism is described as a rapid step-growth polymerization. As the
ceiling temperature is a limiting factor of the thennodynamics of vacuum polymerization,
the polymer deposition is enhanced at lower temperatures by two factors : the magnitude
of the difference between ceiling and substrate temperatures, and the enhanced adsorption
or condensation rates due to lower substrate temperatures 50. This alone is not a true
picture of plasma polymerization as ionization is necessary to create the plasma. But
since bond energies are much lower than ionization energies, ionization effects are low
and ions are fewer than radicals for molecules in the plasma state by about 5 orders of
magnitude 50. The energy of the electrons is 10 — 20 eV whereas bond energies are
typically lower (1 — 8 eV). Oxygen containing molecules ( -COOH, -CO-, -OCO-, OH,
etc.) show a reduced tendency to deposit polymers due to poisoning by oxygen. Non-

polymerizing gases may be copolymerized with inherently polymerizing ones. Thus, N2 ,

29

CO and H20 have been incorporated in the polymer chain of styrene 50. In general, the
polymer deposits formed are pin-hole free, and highly networked and cross-linked 50’ 51.
Treatments with air plasma conducted on ex-PAN and ex-pitch carbon ﬁbers
show that this creates edge pits of a size that are not seen by scanning electron
microscopy and are detected instead by scanning tunneling microscopy. Also this
increases the surface oxygen content, and evens burns the outermost ﬁber surface 63 .
Methane Plasma is known to produce a highly cross-linked -CH2- coating on
substrates. The structure may have occasional double bonds and unsaturation, but

aromatization does not occur 50.

3.2 Strategies for Analyses

3.2.1 X-ra Photoelectron S ectrosco ”'65

XPS is a surface sensitive method as it only interrogates the outer 100 A of the
ﬁber surface. A photon of energy hv penetrates the surface, and is absorbed by an
electron with a binding energy E. below the vacuum level, which then emerges from the
solid with a kinetic energy hv - Eb . There are other complications as the probabilities of
photons being absorbed by electrons in different states are not the same, and due to
chemical shifts and relaxational shifts '5. Any photon of energy greater than the work
function of the solid surface being studied may be used for XPS. The common X-ray
sources used are the Al and the Mg Ka emissions at 1486.6 eV and 1253.6 eV
respectively. The XPS spectrum is a plot of intensity in arbitrary units versus binding
energy. It consists of the core region and the valence band region. Core region (with

binding energies greater than 30 eV) spectra are the easiest to interpret since each type of

30

atom has core electrons in a characteristic region whose shifts are representative of
chemical interactions and hence contain information about chemical bonding. Sometimes
core shifts cannot distinguish between subtle chemical differences and thus the valence
band comes into play. To interpret the valence band spectrum, one needs reliable spectra
from model compounds together with a model ( such as the relaxational potential model)
to reliably predict the spectra. Careful data analysis is needed whenever the core spectra
have overlapping features, and peak ﬁtting techniques are used 65.

XPS was used to gage the effect of plasma treatment on aluminum foil and on
regular carbon ﬁbers ( AS4 — variety ). It was hoped to gain an idea on what the
treatments would yield, on vapor grown ﬁbers. VGCFS as such could not be analyzed as
they readily become airborne, and could damage the spectrometer or its components.
3.2.2 Environmerial Scinnr_'_r_1g Electron Microscopy 66

For high magniﬁcations and resolution, scanning electron microscopy (SEM) is
used. The resolution is possible upto 20 — 50 A when bulk objects are examined. Another
important feature of the SEM is the three-dimensional effect due to the large depth of
ﬁeld, as well as to the shadow-relief effect of the secondary and back-scattered electron
contrast. The ﬁnal beam diameter, called the spot size or probe size, limits the best
possible image resolution. The amount of current in the ﬁnal probe determines the
intensity of emitted signals such as secondary electrons, back-scattered electrons, or x-
rays. Unfortunately, the smaller the spot size, the smaller is the available spot current. So
an efficient optimization must be carried out for best results. The basic principle of
Operation is that an electron gun gives off electrons by photoelectric effect. This gun also

accelerates the electrons to energies in the range 1 — 40 eV. The gun’s beam diameter is

31

too large to create sharp images at high magniﬁcations. So electron lenses or condensers
are used to place a small focussed beam on the specimen. Once the electrons interrogate
the surface with a spot size of between 2 nm to 1 pm, elastic collisions produce back-
scattered electrons whereas inelastic collisions give rise to secondary electrons, UV, IR,
lattice vibrations, Auger electrons, x-rays, etc. The detector consists of a scintillator (that
produces light upon impingement by electrons), and a photo—multiplier tube with a 300 V
positively charged screen before it, to attract the electrons. Inherent limitations of the
SEM, of imaging only vacuum tolerant and electrically conductive materials are partly
overcome by applying conductive coatings (gold), though these could be time consuming
and destructive (to the sample’s surface structure) by creating preparation artifacts. In the
environmental SEM or ESEM, imaging of vacuum-intolerant materials such as hydrated,
dirty, oil-bearing ones, and of insulating materials is possible 67. This is because, electron
detectors here are compatible with a non-vacuum environment and the pump system
separates the low-pressure electron gun chamber ( 10'7 torr) from the higher pressure
specimen chamber (2 to 5 ton) by two intermediate manifolds at 10 '4 torr and 0.1 torr.
Electron microscopy, has been used to qualitatively classify the degree of
adhesion in ﬁber / matrix composites on the basis of fracture surface Observations using
the degree of ﬁber pull-out and the ﬁber surface appearance as relative measures of
interfacial bond-strengths 19. Also it has been shown that short ﬁbers imbedded in
polymeric matrices arrest crack growths 68, and that in well-bonded systems, fatigue
crack propagation becomes an important failure mode whereas in poorly bonded cases,
fast fracture occurs 19 , with higher ﬁber pull-out lengths. Horst and Spoorrnaker 20, have

demonstrated that fatigue failure mechanisms reveal the initiation of damage at ﬁber

32

ends, followed by growth of damage into voids and subsequent debonding, growth of
voids into micro-cracks with bridging by drawn matrix and unbroken ﬁbers, further
debonding to ease the bridging, and ﬁnally growth of bridged crack till a critical size
causes specimen failure. The propagation of cracks at the interface, or in the matrix
phase, is decided by interfacial shear strength 20. Extensive matrix shearing with matrix
material still sticking on to ﬁber surfaces is an indicator of good adhesion 20. The pull-out
aspect ratio is also an adhesion indicator '9. As the environmental SEM or ESEM has a
tensile stage capability, an experiment was conducted to fracture composite samples and
image the fractured VGCF-matrix interface, as well as to see the effect of interfacial
strength on ﬁber pull-out.
3.2.3 Energy Dispersive Spectrometﬁ 66

It ( EDS ) is essentially an additional feature of SEMS since interrogation of
surfaces by electrons also produces x-rays. So an x-ray detector interfaced with an
electron beam instrument is often used in mid-energy ( 1 — 12 eV ) spectroscopy. The
solid-state detector consists of a p-n type lithium-layered silicon semiconductor crystal
that absorbs x-ray photons, ejecting photoelectrons. This is then converted to a voltage
pulse, proportional to the energy of the incoming photon. The pulse is ampliﬁed before
data analysis. To prevent thermal effects, the detector crystal is kept at close to liquid
nitrogen temperature. To correlate peak intensities with concentrations, standards with
the same constituent being analyzed and of known composition are used. The
concentrations are then directly proportional to the intensities. Still matrix correction
factors have to be applied to account for the differences between “standar ” and the

specimen. These arise due to x-ray ﬂuorescence, x-ray absorption and atomic number.

33

The specimen must be homogeneous within the volume treated. The specimen must have
a ﬂat, highly polished surface, placed at known angles to the incident beam and detector.
Then the only difference between standard and sample is one of composition. As none of
these issues, and especially the mass-absorption coefﬁcient, can be met with consistency
from sample to sample, in the case of VGCFS, only qualitative analysis is sought. EDS
was used to see if certain chemical reactions on the VGCFS were successﬁrl.
3.2.4 Dynamic Mechanical Analysis

It is often used to study the structure of polymers since dynamic mechanical
properties are sensitive to glass transition, crystallinity, cross-linking, ﬁlling and
reinforcing systems, molecular aggregation, phase separation, etc 69. In dynamic
mechanical tests, either a vibrational force or a deformation is applied to a specimen and
the sinusoidal response of either the force or the displacement is measured. These
properties can be measured as a function of frequency or temperature. Viscoelasticity is
usually represented by a spring and dashpot connected in series (Maxwell model) or in
parallel (V Oigt model). For a material in a linear viscoelastic region, if a sinusoidal strain
is applied to the specimen, i. e., s = so sin(0)t) , using the Maxwell model, we get the

following equations:

d8 0 1 do

_ — + — equation (13)
dt

11 E E

[(E' )2 '1' (En )2 ]1/2 Sln((1)t+ O) equation (14)

34

E0021:2

 

where, the Storage Modulus, E =

 

2 2 , the Loss Modulus,
1+ 0) t
.. Ewr
E = 2 2 and the Damping Coefficient, tan O = E”/ E'
1 + (1) I

Here, r is n / E , n is the viscosity, E, the Young’s Modulus, and o, the applied
stress. Dynamic properties may be tested in tensile, ﬂexural and shear modes. In this
research, the single cantilever bending ﬂexural mode was used. In the single cantilever
conﬁguration, the specimen is in the form of a rectangular bar, placed horizontally on it’s
wide face. The specimen is clamped to a rigid support on one end and is clamped to a
movable arm on the other. The movable arm applies a variable downward load
(sinusoidal), bending the specimen to a pre-decided value, for which the force response is

measured. The following equations are valid in the single cantilever mode :

0' = K F and S = K D equation (15)

where, KO. = 31.4/W't2 and K8 = 3I/L2

The symbols F, D, L, w, and t stand for measured force, displacement, span
(distance between supports), width and thickness. An important condition is that the span

to thickness ratio be atleast 15 : l, to avoid shear effects appearing in the data.

35

Dynamic moduli have been used to compare reinforcing abilities 17’ ‘8‘ 69. Others
have shown that ﬂexural modulus is a good indicator of adhesion improvements by
surface treatments 70. Flex modulus has been used to compare adhesion levels in interface
modiﬁed and unmodiﬁed composites 7'. In this research, ﬂexural modulus in a dynamic
mode was used to gage the effect of surface treatments on the reinforcing ability of
VGCFS in epoxy composites.

3.2.5 Thermogravimetric Analysis 72

This is a continuous measurement of a sample’s mass as the temperature and/or
time is changed. The sample may be heated or cooled at some selected rate, or it may be
maintained at a ﬁxed temperature, i. e. , isothennally. Commonly, the sample is heated at a
constant rate in the range from 5° to 200 C per minute. In many cases, the sample in a pan
is suspended from the balance beam to hang down into a furnace. To prevent heating
from affecting the balance, or thermal drift, a downward ﬂow of purge gas is helpful. The
instrument can also be run in a high-resolution mode (Hi-Res TG) where the heating rate
slows down considerably or is even zero (isothermal) during active periods when weight
changes occur. This is done because slow rates and isothermal runs can resolve
neighboring peaks easily, whereas fast rates are required for economy of time. TGA has
been used for characterization of carbon ﬁbers (ex-PAN and ex-pitch) 72.

3_.2.6 X-ray Diffraction 73

The starting point for x-ray diffraction is always Bragg’s law:

n). = ZdhkL Sin 9 (16)

36

where d u I. L is the distance between parallel hkl planes of atoms making up a crystal, 0
the angle of incidence, and A the x-ray wavelength. For ﬁrst order reﬂection, n equals

unity. It is useful to note that atoms arranged randomly in space (as in a gas) scatter in all
directions and the effect is weak — intensities add. Scattering by atoms arranged
periodically in space as in a perfect crystal, is strong in very few directions (where
Bragg’s law is satisﬁed) and is called difﬁ'action. Amplitudes add. Hence the resulting
intensity is even larger than if intensities were to add. In most other directions that do not
satisfy this law, scattered rays cancel one another out. A diffracted beam ﬁ'om a crystal is
built up of rays scattered by all atoms that lie in the path of the ray, and not from the
surface layer alone. In this, diffraction differs from reﬂection. Also diffraction takes place
only at speciﬁc angles, called Bragg angles. Diffraction directions are determined solely
by the shape and size of the unit cell. Intensities of diffracted beams are determined by
the positions of atoms within the unit cell. Destructive interference is a consequence of
the periodicity of atom arrangement, as is the constructive case. Thus if the periodicity is
not of long enough range, line broadening occurs, i. e., diffraction occurs at angles near to
but not equal to, the exact Bragg angle. Thus only an inﬁnite crystal would give a
diffracted beam like an impulse function. Of the several techniques for recording x-ray
diffraction, Powder Diﬁ’raction is commonly used for its’ versatility of technique. In this
method, the crystal to be examined is reduced to a ﬁne powder and placed in a beam of
monochromatic x-rays. Each powder particle is a crystal, or an assemblage of crystals,
oriented at random with respect to the incident beam. For each Bragg angle and its
characteristic hkl plane, there will be some crystals in the random mass, that are perfectly

oriented for diffraction to occur. The mass of powder is equivalent to a single crystal

37

rotated about all possible axes. Thus a single monochromatic unidirectional ray scatters
in every direction, for all possible Bragg angles that constitute the crystal structure of the
sample. So in order to be able to detect, the x-ray gun or the detector moves around the
sample in a circular path. The rate of this motion is important for resolution purposes.
The full-width at half the maximum intensity of a peak, is used to calculate the average
size of a crystallite, by the Scherrer formula 73. Amorphous liquids and solids have
structures characterized by an almost complete lack of periodicity and give a broad curve
with one or two wide humps. Polycrystalline samples, and samples with mosaic structure
(with disorientations) or with other defects give broader peaks (due to the disorientation).
Graphitic crystals have a hexagonal structure. Therefore the Bragg formula
combined with that relating the value of d I. k L in the set (hkL) with the other crystal

parameters is used. The relation for a hexagonal system is:

1 4 h2+hk+k2 £3

+
(11211.1 3 a2 c2

 

equation (17)

Here, a is the length of the smaller edge of the crystal, and c is the length of the
larger.
3.3 Experimental Methods

The following table gives a summary of surface treatments conducted, and of

surface treated ﬁbers, obtained, which were used for composite preparation.

38

Table 2 : List of Surface Treatments

 

VGCF Treatment

Description

 

Variant 1

Baseline ﬁber (as-received) for surface treatment

 

Variant 2 —- from supplier

Variant 1 treated in air at 415-500 C for 20 min.

 

Variant 3 — from supplier

Variant 1 treated in CO2 at 900 C for 20 min.

 

 

 

Soxhlet extraction Extraction of poly-nuclear aromatic hydrocarbons by
reﬂuxing with methylene chloride
Amination Incorporation of amine groups on extracted VGCF by an
alkylation step followed by treatment with ethanolamine
Oxidation Increasing the surface oxygen groups by reﬂuxing with

35% nitric acid

 

Methane plasma

Creating a low energy cross-linked hard hydrocarbon

surface layer by deposition from methane plasma

 

 

 

Air plasma Surface oxidation and creation of surface oxygen
functional groups using air plasma
Argon plasma with Incorporation of oxide and epoxy groups on the VGCF
Allyl glycidyl ether surface using argon plasma and allyl glycidyl ether

 

 

39

 

3.3.1 Fiber Varian_ts43

Different variants of Pyrograf III VGCFS were obtained from Applied Sciences,
Inc., USA, namely, variants 1, 2 and 3.
Variant 1

The VGCFS may be produced at slow, modest or high rates of growth. This
variety is grown at a Slow rate, and was supplied as the baseline vapor grown carbon
ﬁber.
Variant 2

As-grown variant 1 ﬁber post-treated in ﬂowing air at 425 — 500 C for 20 min.
This treatment was carried out to increase ﬁber surface energy by creating oxide groups
on the VGCF surface. These ﬁbers were obtained from the manufacturer.
Variant 3

As-grown variant 1 ﬁbers were post-treated in ﬂowing carbon dioxide at 900 C
for 20 minutes. This was done in order to strip the ﬁber surface of weakly attached
polyaromatic hydrocarbons. These ﬁbers were also obtained from the manufacturer.
3.3.2 Extraction of Polynuclear Aromatic Hvdrocarbona

Prior to surface treatments, the Variant 1 ﬁbers were extracted with methylene
chloride in a soxhlet apparatus. In each case, approximately 1 gm of VGCF was wrapped
in a 0.2m pore size, 142 mm diameter TeﬂonTM membrane and reﬂuxed for 36 hours.
After extraction, the ﬁbers were dried at 60 C for 12 hours before use for surface

treatments.

40

3.3.3 Surface Treaﬂrent Methods

All surface treatments were performed on methylene chloride extracted Variant 1
ﬁbers.
3.3.3.1 Wet Chemical Treatments
3.3.3.1.1 Amination
3.3.3.1.1.1 Friedel Crafts Allaylatﬂa

About 1g of VGCF was taken in dry round bottom ﬂask with 150 cc of methylene
chloride and 0.75 g of anhydrous aluminum chloride. This was stirred vigorously for 10
hours and the temperature maintained below that of the room. The reaction nrixture was
then ﬁltred and the ﬁltrate in the ﬁlter media itself, washed with 1500 cc of de-ionized
water in 20 cc batches
3.3.3.1 .1 .2 Ethanolamine treatment

The alkylated ﬁber mass (about 1 g) was stirred with 100 cc monoethanolamine
for 10 hours, at below room temperature, in a round bottomed ﬂask. It was then ﬁltered
and the ﬁltrate washed with 3000 cc de-ionized water in 20 cc batches. Then the washed
VGCF was dried at 105 C for 24 hours. To see the efﬁcacy of the friedel-crafts alkylation
in helping incorporate the amine group, direct ethanolamine treatment of soxhlet
extracted was tried at the same conditions as above.
3.3.3.1.2 Oxidation 47

VGCFS were subjected to oxidative treatment as follows 47. About 1 g of ﬁbers
were mixed with 150 cc of 35% HNO 3 and maintained at 110 C for 5 hours with reﬂux

condensation and thorough stirring, in a round bottomed ﬂask. The reaction mixture was

41

ﬁltered and the ﬁltrate washed with about 1500 cc de-ionized water, and dried at 120 C
for 24 hours.
3.3.3.2 Plasma Surface Treatments

This work was carried out using a Plasma Sciences, Inc. PS 0500 batch reaction
system for modiﬁcation of materials by cold gas plasma. The plasma is excited by high-
energy RF power at low pressures (in a vacuum). Plasmas at low pressures produce very
little heat and thus minimize thermal effects. The reaction chamber is a rectangular
aluminum box of internal dimensions 30x16x20 inches, with an RF and UV shielded
view port in the front. It has a 13.56 MHz RF generator of 550 watts maximum power.
Provisions exist for gas inlets up to three at a time.
3.3.3.2.1 Methane Plasma Treatment

The ﬁbers were treated in batches of 0.75 g each. About 0.75 g VGCF were
crushed by hand in a zip-lock bag, and spread out in a 15cm diameter glass petri-dish, 7.5
cm in height. This was to ensure that the ﬁbers have a uniform exposure and don’t ﬂy out
of the dish when the chamber is evacuated. The plasma chamber was then evacuated to
0.05 torr base pressure, followed by an argon purge for 1 min.(till the pressure rises to
2.03 torr). Again it is evacuated to base pressure. Then the plasma treatment occurs with
argon gas at 1450 std. cc / min. for 2 minutes at 300 watts plasma level, the pressure
being maintained at 0.675 torr. Then the chamber is evacuated (base pressure), purged
with argon, evacuated and then treated with methane gas at 72 std. cc / min. for 10
minutes at 300 watts plasma level with the pressure at 0.054 torr. This is followed by the
cycle of base pressure, purge (argon), base pressure and venting. The temperature was

approximately 32 to 34 0C.

42

To check the efficacy of plasma treatment, the combined steps of argon and
methane plasmas, the treatment was also conducted on an aluminum foil substrate 49.
Also, this plasma treatment was also tried on AS4 12000 tow carbon ﬁbers 49.
3.3.3.2.2 Air Plasma Treatment

The ﬁbers were treated in batches of 0.75 g each. About 0.75 g VGCF were
crushed by hand in a zip-lock bag, and spread out in a 15cm diameter glass petri—dish, 7.5
cm in height. The chamber was evacuated to 0.05 torr base pressure and purged with
argon at 2.03 torr for 1 minute. The chamber was again evacuated to base pressure and
then treated with argon plasma at 300 watts for 2 minutes; the ﬂow rate of argon was
1450 std. cc / min. and the pressure in the chamber, 0.675 torr. The chamber was then
evacuated to base pressure, purged with argon as in the beginning, evacuated, and then
treated with air plasma at 300 watts for 1 minute. Here the ﬂow rate was 100 std. cc/min.
and the pressure was 0.07 torr. The system was then evacuated to base pressure (0.05
torr), purged with argon for 1 minute (2.03 torr) and evacuated before venting.
3.3.3.2.3 Argon Plasma Induced Allyl Glycidyl Ether Trgrtment

Allyl glycidyl ether (98%) was mixed with acetone and made into a 30% solution
to ensure equal contact with all the ﬁbers. Allyl glycidyl ether was removed by syringe
with Simultaneous injection of nitrogen gas at l psig into the reagent container for storage
of the remainder, as required for handling of air-sensitive reagents. 0.75 g VGCF crushed
by hand in a zip-lock bag, was taken in a 15cm diameter by 7.5 cm high petri-dish with
100 cc of 30% acetone solution of allyl glycidyl ether and placed in the plasma chamber
per run. The chamber was evacuated to a base pressure of 3 torr. This time a higher base

pressure was chosen as the reagent boils at 1540 C. In a separate but similar experiment it

43

was seen that allyl glycidyl ether does not boil off completely before the end of the run.
Purging the chamber with argon, after the ﬁrst evacuation, could not be done as there was
no provision to let in the moderately low-boiling solution after the purge and evacuation
cycles had been completed and before the start of the plasma. Therefore, immediately
after the ﬁrst evacuation step, argon was let in at 1450 std. cc / min. for 10 minutes with
a plasma RF level at 300 watts, the pressure during plasmation being about 1 to 1.3 torr.
The chamber was then evacuated, and vented. The resulting ﬁbers were then dried at 160
0C for 24 hours to remove physisorbed allyl glycidyl ether. It was seen that the ﬁbers
were in part colored yellow-brown.
3.4 Composite Preparation

Composites were prepared with various surface treated and as-received VGCFS.
Vapor grown carbon ﬁbers inherently exist as agglomerates with an apparent density 74 of
about 5 to 50 ft 3 / gm. This renders mixing with resins very difficult. For good
dispersion, the agglomerates also need to be broken down by shear or other techniques,
which will also increase matrix wetting during composite formation. Hence the VGCF
were compressed at 120 to 140 MPa, to increase their bulk density, making them more
amenable to processing. This was done with a stainless steel cup and ram arrangement,
using a compression molder (Wabash Metal Works) for applying the pressure. The epoxy
resin (Shell EPON 828 DGEBA epoxy resin) was mixed with 14.5 phr (14.5% of resin
weight) of curing agent (meta-phenylene diamine), melted and mixed at 680 C. The ﬂuid
mixture was then added to the compressed VGCF in the ratio required for the ﬁber
loading sought, and mixed by hand. The mixture was warmed at 680 C to keep the resin-

VGCF system ﬂuid enough for vacuuming purposes. The mixture was poured into a

44

silicone (RTV 664, GE Silicones) mold of dimensions 3.0” by 0.5” by 0.125” and
vacuumed at 30” Hg (vacuum) for 1 hour. The contents were then taken out of the
vacuum and a polyimide sheet was placed on it, followed by 6 kg mass. This exerts about
5 kPa pressure on the resinous mixture, prevents it from overﬂowing during subsequent
out-gassing cycles and gives a smooth surface ﬁnish. The system was again vacuumed
for 2.5 hours. Following this, a total weight of 18 kg was placed on the polyimide sheet
covered mixture, amounting to a pressure of 15 kPa, and the system was cured in an
oven. The cure cycle used was 75 0C for 2 hours followed by 125 0C for 2 hours with a 15
minute constant ramp in-between. After cure, the sample was left as it was, and reached
room temperature in 5 to 6 hours, thus minimizing shrinkage stresses due to fast cooling.
As gelation occurs during the lower temperature cycle, this again minimizes thermal
stresses 6. Samples were made with variants 2, 3, 4, and surface treated ﬁbers with

loadings of 15 wt. % and 20 wt. % in the epoxy matrix.

45

 

4 Results and Discussion

The results of the experimental program are taken together to explain and analyze
the VGCF project goals. VGCFS that have been treated by wet chemical methods have
been analyzed using energy dispersive spectrometry ( EDS ) to ascertain the atoms
incorporated on the surface. A more surface sensitive and quantitative technique namely,
x-ray photoelectron spectroscopy ( XPS ) has not been used due to the hazards posed by
VGCFS to the electrical equipment. However, XPS has been carried out for pilot plasma
treatments undertaken on aluminum foils and on AS-4 carbon ﬁbers, to determine the
effect of plasma treatment on these substrates and as a model for VGCF analysis.
Dynamic bending modulus tests conducted in a single cantilever mode on composite
samples made at 15 wt.% and 20 wt.% loadings of each of the surface treated VGCFS
have been used to indicate the result of ﬁber surface treatments on adhesion to the epoxy
matrix. Thermogravimetric analysis (TGA) and x-ray diffraction ( XRD ) of the soxhlet
extracted VGCFS compared with the results of those for a commercial graphite, were
useful in explaining the true nature of the VGCF microstructure, and to determine if the
assumptions about the VGCF modulus can be substantiated.
4.1 Energy Dispersive Spectrometry Efﬁcacy of Wet Chemical Treatmenta

Vapor grown carbon ﬁbers cannot be analyzed by X-Ray Photoelectron
Spectroscopy as they are readily air-home and present hazards to the electrical
equipment. Therefore they have been characterized using EDS. Although factors like
path length, ionization cross-section, etc. cannot be maintained constant from sample to
sample in this case, preventing any quantitative analysis, qualitative elemental mapping

can be done. This, with the reaction steps, help gage the success of the reaction.

46

 

The analyses were performed on Electro Scan Model 2020 Environmental
Scanning Electron Microscope (ESEM) with X-ray microanalyzer and Link ISIS
Detector. The lithium-drifted silicon detector resolution is 133 eV. The acquire time was
200 seconds and the Spot size was less than a micron. In each case, the compressed ﬁbers
were imaged at about 15 kV beam voltage, by sticking them onto an aluminum stub with
an adhesive, masking all signs of the stub and adhesive. Also, since the voltage was not
high, the path length of the accelerated electron was also less and only the ﬁbers were
imaged. Initially, the bare stub and the stub with just adhesive on it, were imaged. It is
pointed out that “matrix eﬂects” 66 impede the quantitative analysis of VGCF samples.
This is because, factors pertaining to the interrogated surface such as ionization cross-
section, density, path length, etc. cannot be maintained the same from run-to-run. The
samples analyzed were soxhlet extracted VGCF, friedel-crafts alkylated VGCF, ﬁbers
with post-alkylation ethanolamine treatment, and ﬁbers without alkylation but with direct
ethanolamine treatment. Since the ﬁbers already contained carbon and oxygen, and since

quantitative analyses could not be performed on VGCFS, oxidative and plasma surface

treatments were not analyzed by EDS.

47

Wled'llrde'nlok

 

 

 

 

 

 

 

 

ClientzArvind
Jobzcmma'owtusken
sordid m (wmsday. January 14. 1998 09:52)
cps
: i
1
"H
a:
3 s
6—2
4—2
: ' '
21‘ o
o I I I I I I I I I I I I I I I I I T I I I l l I I r T I I r U f
2 4 6
Enemytkovl

Figure 5 : EDS of soxhlet extracted VGCF

48

Operator: Ridrarderalek

Cliont:Arvlnd

Job:Carbon Micmmlam

tct rye (Thursday. January 15. 199810222)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

q?!
" AI
i
10—
5..
.l
i 0
l
c I I I I l I I I I I I I I I I I I I I T ﬁj I r T I I r I I I I I
2 4 6
50.79! WOW

Figure 6 : EDS of VGCF after Friedel-Crafts reaction

49

Operator: Richard Schaiok
' Client: Arvind
Job: Carbon Miamntriskera
otht try4 (Thursday. January 15. 1998 10:55)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GP!
- C
: Fe
3—-1
2...
_‘ CI
3 O
.1
o [IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
O 2 4 6 8

Energy (koV)

Figure 7 : EDS of VGCF after amination

50

Operator : Richard Schalek

Cient : Arvind

Job: Carbon Microwhiakere

eh21ry1 (Sunday. February 15. 1998 13:44)

i,

 

lleleljl

 

llll

Ilillll

 

 

 

llJllll

I L

 

 

 

 

 

 

 

lIIIIIIﬁIIIIrjIWﬁII

o 2
Energy (keV)

Figure 8 : EDS of VGCF directly treated with ethanolamine

51

Figure 5 shows the bare, soxhlet extracted VGCFS. The Fe and the S peaks arise
from their manufacturing process. The effect of the F reidel-Craft alkylation is shown in
ﬁgure 6. The big Al peak is due to the associated aluminum counter-ion probably in a
hydrated form, from the washing step. Figure 7 shows the alkylated VGCF after ethanol
amine treatment. The large Al peak at about 1.5 eV binding energy is missing. Secondly,
and most important, there is a smaller but deﬁnite Nitrogen peak, indicating nitrogen
incorporation, possibly in an amine form, from the reaction step. Nitrogen is hard to
detect as the peak ( 0.282 eV) almost overlaps with that of carbon (0.392 eV). Yet, it is
distinct, due to the right shoulder on the carbon peak. To ascertain if the alkylation step is
essential for amine incorporation, the VGCFS were directly treated with ethanolamine,
washed, dried, and analyzed. The results are shown in magniﬁed form, in ﬁgure 8. There
is a marked absence of nitrogenous groups. These results Show that the two-step
amination reaction incorporates nitrogenous groups on the VGCFS, presumably as amine-
functionalities.

4.2 X-ﬁrayPhotoelectron Spectroscopy : Plasma Treatment

As the plasma treatments used mainly incorporate carbon and oxygen atoms, EDS
is not of much value since quantitative analysis is not possible. Hence, an AS4 12000 tow
carbon ﬁber bundle was used as a model to judge the efﬁcacy of the methane plasma
treatment, and evaluated using x-ray photoelectron spectroscopy. All XPS work 64 was
carried out at a 45 0 angle, with a Mg K01 x-ray source at 300 W, acquire time of 3.67
minutes, pass energy of 89.450 eV . Presented below are the results for aluminum foil
treated with CH4 plasma and the XPS results for 12000 tow AS4 as-received, argon

sputtered and CH4 plasma treated carbon ﬁbers.

52

Firstly, aluminum foil was treated with methane plasma. Methane plasma is

known 50 to form a highly cross-linked —CH2- ﬁlm on substrates.

Table 3 : XPS of methane plasma treated aluminum foil

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Untreated Al foil CH4 plasma-Al foil
Element Sensitivity factor Concentration % Concentration %
C Is 0.296 33.29 95.72
0 Is 0.711 48.75 4.28
A1 2p 0.234 17.96 0.00
Table 4 : XPS of plasma treated AS4 carbon ﬁbers
Element on Sensitivity As received Ar-sputtered CH4 plasma
AS4 factor Concentration Concentration Concentration
% % %
C Is 0.296 88.62 91.54 95.38
0 Is 0.711 7.41 7.47 2.94
N Is 0.477 3.97 0.99 1.68

 

 

 

 

 

As shown in table 3, untreated aluminum foil forms an oxide layer, and also has a

lot of physisorbed carbon dioxide on it, hence the carbon peak seen on the untreated foil.

But after methane plasma treatment, no aluminum can be detected, indicating a uniform

coating. The methane-plasma sputtered coating, when obtained on glass slides was found

to be red-yellow in color and extremely tenacious as it could not be removed with a razor.

53

 

 

XPS results for AS-4 carbon ﬁbers subjected to several plasma treatments are shown in
table 4. Once again, it is seen that CH4 plasma coats the surface. Ar sputtering seems
merely to reduce the nitrogen content. The effects of oxidation by air plasma 50, and of

attaching chemical moieties using plasma attack at an unsaturated site 50’ 5 I

, are well
documented.
4.3 Environmental Segm'aLElectron Microscopy

The propagation of cracks at the interface, or in the matrix phase, is decided by
interfacial shear strength 20. Extensive matrix shearing with matrix material still sticking
on to ﬁber surfaces is an indicator of good adhesion 2° The pull-out aspect ratio ( aspect
ratio of the ﬁber sticking out from the surface ) is also an adhesion indicator 19 with
higher ﬁber pull-out lengths in interphases poor bonding. The ESEM to fracture
composite samples and image the fractured VGCF-matrix interface, as well as to see the
effect of interfacial strength on ﬁber pull-out.

The VGCF-composites were fractured in a tensile fashion and the fracture
surfaces were imaged in an Electroscan Model 2020 ESEM with a lanthanum hexaboride
ﬁlament. The imaging was done with beam voltages of 20 - 25 kV, at 3 — 5 torr pressure.
Typical fractographs of notched composite samples broken in a tensile mode are shown
for variant 3 and for the soxhlet extracted ﬁbers (ﬁgures 9 and 10). The resolution of the
ESEM, which is a function of the spot size, is variable, but less than a micron. The ﬁber
diameter being only 0.2 microns (on an average), the resolution is not ﬁne enough to
evaluate subtle differences in the interfacial region of the ﬁber. It is also not possible to

even roughly establish the number of pull-out holes in a given fractured surface. Also

since the ﬁbers are curved, sometimes crenulated , having a whole range of lengths and

54

diameters, and random in a three-dimensional plane, no pull-out length values are
meaningful. Hence the ESEM cannot be used to get qualitative interphase information in
this case. However, qualitatively, there are several pull-out lengths of bare ﬁber. But
these are indicative of poor adhesion in a relative sense only if meaningful comparisons
can be made between composite samples with ﬁbers of different surface treatments,

showing marked differences in pull-out lengths.

55

FDURDG

 

Figure 9 : ESEM of variant 3 VGCF in epoxy matrix

31:1 1 '..' .
Philips. _. . 313313

 

Figure 10 : ESEM of soxhlet extracted VGCF in epoxy matrix

57

4.4 Dyaamic Mechanical Analysis

Dynamic mechanical analysis was used to determine the storage modulus of the
composite samples made with the surface treated VGCFS and an epoxy matrix. The
storage modulus, like the elastic modulus from a static test, depends on the level of
adhesion between ﬁber and matrix ( equations 6 - 10 ). The ﬂexural properties of the
composite were tested in a Rheometics Scientiﬁc Mark IV Dynamic Mechanical Thermal
Analyzer (DMTA). Composite samples with 15 wt. % and with 20 wt. % VGCF loading
were tested. The samples were 22 to 24 mm in length, 0.52 mm thick and on an average,
3.8 mm wide. Samples were placed with their wide face down, and clamped to a Single
cantilever ﬁxture. The clamping was done at with an applied torque of 28 centi-
Newton(cm). The test was conducted at room temperature, at 1 Hz frequency with a
strain-amplitude of 4 um. The span of the sample for test purposes was 8 mm.
Composites with ﬁber variants 1, 2, 3, soxhlet-extracted VGCF, aminated VGCF, nitric
acid etched variant, and all the plasma treated varieties.

Figure 11 shows the ﬂexural storage modulus of the composite with ﬁber
loading, for various surface treatments. The value, for the neat matrix material is 3.18
GPa. There is a slight improvement in stiffness from 0 wt. % ﬁber ( pure matrix at 3.18
GPa ) to those at different loadings. The change in modulus from 15 wt. % to 20 wt. %
ﬁber weight fraction is only minor. Also, within the experimental error, the storage
modulus of the composite is not affected by the different ﬁber surface treatments,
remaining at 4.5 GPa at 15 wt.% and 5 GPa at 20 wt.%.

Various surface treatments produce different chemical effects. Soxhlet extraction

removes weakly bound layers. Amination incorporates amine groups which can react

58

with epoxy resins. Oxidation treatment not only increases wettability and creates
carboxyl groups for bonding, but also creates etch pits on the ﬁber surface that may
weaken the ﬁber itself. Air plasma also oxidizes the surface. Methane plasma coats the
surface with a hard and extremely low-energy layer that must not be very wettable. Allyl
glycidyl ether treatment via argon plasma may introduce oxirane rings and hydroxyl

groups on the surface, thus effecting chemical adhesion.

 

Dynamic Bending Modulus vs Fiber Wt%

 

 

l[Ivariant 1

uvariant 2

‘ljwrriant 3

‘Isoxhlet

‘Iamination
‘onidation

(.methane plasma

1 Clair plasma

‘la.g.e + argon plasma

Modulus in GPa
—I N (A) & 0| 0)

 

 

 

O

0 15 20
Fiber Weight % lip-heat matrix

 

 

 

Figure 11 : Dynamic mechanical analysis of composite samples

In a recent paper 80, Darmstadt et. al. , when working with the same PYROGRAF
III VGCFS, showed that surface treatments not only treat the surface of these 0.2 pm
carbon ﬁbers, but also affect the bulk chemistry and structure. Yet, none of these surface
treatments show any change in the apparent reinforcing effect. The reasons for this may

be that since the interphase is a three-dimensional structure from the pre-outer layers of

59

the ﬁber, in to the bulk region of the matrix surrounding it, a region diﬂerent from the
interface between ﬁber and matrix, is controlling.

It is to be noted that Patton et. al. 42, have prepared composites of Pyrograf HI
VGCFS with a different epoxy matrix material and have comparable values for ﬂexural
modulus as determined by a static test, with a lot of scatter. They have shown 8 to 10 %
improvement in the ﬂexural modulus obtained, with better mixing and processing
strategies. The authors suggest a ﬁber modulus between 88 to 150 GPa obtained from the
composite ﬂex modulus values, using the rule of mixtures with a multiplicative factor 75’
82 for the ﬁber stress term, called the reinforcing efficiency K. It’s value for a 3-D

random ﬁber composite is 0.2.

E0 = KEfo +13me (18)

By the same treatment, for Pyrograf III, using the dynamic bending modulus
value of 4.5 GPa for the composite strength at 9.4 volume % loading, a ﬁber modulus of
88 GPa is obtained. For a ﬁber loading of 12.8 vol. %, using a modulus of the composite
as 5 GPa, the ﬁber modulus obtained is still 88 GPa. The above equation is very
approximate one and does not include an L; Id term, or any other intrinsic term for
adhesion.

Alternatively, using equations 6 — 10, for short ﬁber random in-plane composites,
the results obtained are shown in ﬁgure 12. Dynamic bending modulus values may be
used as elastic modulus data. This model requires values for the ﬁber transverse modulus,

which was assumed to be 14 GPa, an average number for many regular carbon ﬁbers 83 .

60

Plots are Shown for the variation of composite modulus with ﬁber aspect ratio, for
different values of ﬁber longitudinal modulus, for a 12.8 % volume fraction composite.
An aspect ratio range of 40 to 200 was used 82. The only value of the ﬁber longitudinal
modulus that ﬁts the composite dynamic bending modulus data at 4.5 GPa and 5 GPa ( at
9.4 % and 12.8 % ﬁber content by volume ), for the entire range of aspect ratios
considered, is 34. 5 GPa. At this value of E f , ﬁgure 12 shows that the composite elastic

modulus is independent of aspect ratio from 40 to 200.

 

composite modulus vs aspect ratio for 12.8% fiber
volume

 

 

 

..................................
........
o'-
0
ﬂ
.

EcinGPa

 

 

 

 

______-._ __v———_.______.

0 50 100 150 200 250 300

 

 

Ild
—-EfL=1SOGPa ------- EfL=88GPa EfL=34.SGPa

 

 

 

 

 

 

 

 

 

Figure 12 : Composite modulus vs L/d for various ﬁber moduli

 

 

61

4.5 Therrnoggrvimetric Analysis

Thermogravimetric analysis ( TGA ) was conducted in order to gain some insight
into the microstructure of the VGCFS used in this research. It is well known 72 that
graphites and crystalline forms of carbon have a much higher oxidative resistance at
elevated temperatures. So TGA runs in oxidative atmospheres were conducted to
ascertain the structural entity of the VGCF used. It is also known that pure carbon forms
have a good thermal stability in inert atmospheres, whereas impurities such as other
chemical groups, that may be associated with the carbon in the VGCF would not remain
on the surface and would desorb at elevated temperatures. Hence inert atmosphere runs
were conducted.

Thermogravimetric analysis was conducted using a TA instruments TGA 2950
instrument. Air oxidation runs were performed on as-received GRAF OIL ®, a ﬂexible
natural graphite from UCAR Carbon Company, Inc. and on soxhlet extracted compressed
VGCF. The runs were carried out at 50 C / minute from ambient to 10000 C with a Hi-
Rez 4 m option. This is a high resolution option that changes the mode to isothermal
when weight changes are occuring, but ramps the temperature otherwise. TGA runs were
also done for the two samples in an atmosphere of nitrogen (inert) from room temperature

to 10000 C at 200 C per minute, again with Hi-Rez 4 m option.

62

coo“

 

 

 

 

 

a
comm scoaao «$.m> «we .oov mcaumcm ewe
com com cow com o
. _ p s _ r p . OOal
rom.
o
a m.
m
U.
MW
tom
too“
#
om“
oneaono aHau moooo"-Vqu qusz\uom “acmesoo
mm om mmlcmzlvm “mumo can «NucHx cacao“ oe sz\uom ”nocbmz
ozH>a< ”Losmtmuo /\ mw.;r as onm«.v$ “mNHm
«oo.aH<nHou<¢w u “mﬂnu weHzaamolow>Hmoma malnHouaam ”mmoemm

f Grafoil

ran 0

: TGA of Air Oxidat

Figure 13

63

comm accuse <«.m> <0»

000“

h

oom
L

“may
owe

mcaumcmasmh

oov oom
p b

 

 

 

 

 

 

 

 

mv ma mmicmzlvm

“mama cam

ozH>a< “Leumcmao

Hoo.¢H<xomUU

“mnau

onp<ono EH41 Uoooo«1VNwa HIIZH2\Uom

4m:

VNmaH: 0.000“ on sz\Uom
as o_mm.mm
aaa10mmmmaazoo-u<mooa>a-1xom

o
0
Tom
low
. M
3
m
row w
m
Tom
Ioo«
omﬁ
“ucmeeoo
“nonumz
Hm~ﬁm
“mﬁaemm

TGA of Air Oxidation of VGCF

Figure 14

comm

ucoano <H.m> 40h

oooﬁ

owm

”may
owe

mcaumcmasmp

OOV
_

mm

 

 

 

1mm

1mm

luﬁtaM

1%)

Imm

loo"

 

«ow

 

Mahm

« mmico<1mm Hmumo cam
ocﬁ>c< ”coumcmao

aoo.NZJHOu<meU

“maﬁa

cmoocuac ca uoooﬁrv~ma HI: CHE\Q om

<m:

VNmaHI 0000“ cu CHE\om
ms ommm.mﬁ
ucmecocn>cm mzrﬁwoamco

“ucmeeou

“Documz
”mNAm
“mﬁasmm

ironment

rt env

l heated in ine

: TGA of Grafoi

Figure 15

65

comm

ucoaao U«.v> Hmcmcmo
ooou owm

Ace.
owm

mcaumcmaem»
owv oom

 

 

 

mm

3U5I8M

(%l

 

 

mo mm.mmrca<-mm ”mama cam
ucﬁ>c< ”caumcmao

«oo.NZIxom ”maﬁa

cmuocuwc c“ oooouivumm H11 C«E\o om
VNNEHI coco“ 0» CHE\ON

< D . a... 834...;
.Hr . ucmscoeﬂ>cm mzrcxom

«Ca

”ucmseoo
”Documz
”mNAm
“mnaemm

ironment

rt env

mine

TGA of VGCF heated

Figure 16

66

Oxidation resistance of soxhlet extracted VGCF is shown (ﬁgure 14) as against
that of GRAFOIL, a commercially available graphite (ﬁgure 13). The VGCF show
marked oxidation at 550 0C, whereas GRAFOIL markedly oxidizes at about 700 0C. It is

well known that the rate of oxidation is dependent on the particle size 72’ 35

and smaller
particles have a much higher surface to volume ratio and hence oxidize at much higher
rates than larger particles of the same nature. Also the presence of iron might have a
catalytic effect on the oxidation of the VGCF. Smith et. al. 76 at the General Motors
research Laboratories, USA, found that PYROGRAF ﬁbers in an as-produced state Show
more oxidation resistance (indicated a by a higher onset temperature) than ex-PAN and
rayon-based carbon ﬁbers. They attribute this to a higher graphitic ordering in the ﬁbers.
It must be noted that they worked with 30 um diameter ﬁbers, which is 3 to 5 times larger
than the ex-PAN carbon ﬁbers and hence is likely to Show a much lower oxidation rate
and higher oxidative resistance. The study here also raises doubts as to the true nature of
this graphitic ordering on these pyrolitically grown carbon ﬁbers and on the presence of
impurities. The latter was checked, by running TGA experiments in an inert atmosphere
up to 1000 0C as purely carbonaceous forms have good stability in inert atmospheres 72.
The results of a 20 0C / minute ramp nitrogen atmosphere tests are shown in ﬁgure 15
and ﬁgure 16 for GRAFOIL and for VGCF respectively. Huttinger et. al. 76, have shown
by temperature programmed desorption runs, the composition of gases that evolve from
carbon ﬁbers at what temperature ranges, and what kind of surface functional group and
in what state of bonding that represents. From the TGA runs performed here, it is seen

that there is not much loss of mass from the VGCF (3.5 % at the ﬁnal temperature)

compared with GRAFOIL (3 %). Hence the composition of VGCF is mostly carbon. For

67

the VGCFS, the small weight loss ( 1 to 1.5 % ) between 200 °c and 250 0c is thus
attributed to CO2 desorption rates by strongly acidic groups while the loss of another 1.5
% between 900 0C and 950 0C is possibly due to CO desorption from carbonyl and
quinonic groups 76.

Whereas the inert atmosphere tests Show that VGCFS are indeed carbonaceous,
the lower oxidation stability of VGCF as against graphite, may be due to particle Size, or

due to lack of graphitic ordering, or both.

4.6 X-Rav Diffraction

X-ray diffraction ( XRD ) was canied out to ascertain the microstructure and
especially, the three-dimensional ordering and interlayer spacing for the VGCF
crystallites as that value for graphite is documented.

X-ray diffraction has been used to study vapor grown carbon ﬁbers in the past.
Pure grahite has a c spacing (height of a hexagonal crystal) of 6.7080 A which gives it a
d spacing of 3.3540 A as the multiplicity factor for a 00L reﬂection in a hexagonal
crystal is 2. Koyama and others 77 obtained a d spacing of 3.49 A for their VGCF of
diameter greater than 5 pm. Endo et. al. 25, also studied pyrolitic ﬁbers with diameters of
about 0.8 microns, by small angle diffraction and concluded that the ﬁbers have an axial
ordering but show no h k . L reﬂections, indicating an absence of three-dimensional
order. Hence the structure is only turbostratic.

Endo 22 has also reported for higher diameter ﬁbers, that the core regions are more
aligned than the cortical or outer regions though neither has 3-d order in an as-produced

state. In a more recent paper, Endo and others 78, working with pyrolytic carbon

68

nanotubes (VGCF nanotubes) of diameters of about 5 nm or less, report a high interlayer
d spacing of 3.4 A. Thicker ﬁbers seem to have lower spacing as their curvature is less.
They also reported 78, that the nanotube tips have even greater curvature and hence have
greater interlayer spacing. Tibbetts 36 has reported a d spacing at 3.416 A for 6 pm ﬁbers.
XRD was obtained by a powder technique using a SCINTAG XDS 2000 system.
Copper Ka at 1.54056 A was the radiation used. The runs were conducted at 0.3 degrees
per minute and 0.03 degrees per data point. XRD patterns were obtained for as-received
GRAFOIL m commercial graphite, as well as for soxhlet extracted and compressed
VGCF. In each case the sample was stuck onto the stage with petroleum jelly that is
amorphous and that has no peaks in the range considered. From the Powder Diffraction
Files (J CPDS ﬁles), it was determined that graphite has a 002 peak and a 101 peak at 20
values of 26.60 and 44.670. SO the range of 10 0 to 49 0 was used, to include broadening

effects, etc.

69

 

1m ’ j 1“)

 

 

 

 

 

 

900000 : ~—«

T 14000
81100130 - _.

~~ 12000
700000

.. 10000
600000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

400000 f
~~ 6000
300000
~~ 4000
I
«~ 2000
100000 —-—-
o r J Y L I 0
0 10 Z) 30 40 50 60

2M (@9383)

 

—mrero=w=01 —xmrervocr=l

 

Figure 17 : XRD of Grafoil (primary axis) and VGCF (secondary axis)

 

 

 

70

 

 

nuansny «wbnnuy)ﬂu13RAfIMl

1 000000
900000
800000
700000
600000
500000
400000
300000
200000
1 00000

0

XRD for GRAFOIL

 

20 30 40 50
2theul (degnuun

 

l— XRD for GRAFOIL

l
l
L

 

 

 

 

Figure 18

 

7l

 

 

 

)RDIO‘WIZF

 

 

 

 

 

 

 

 

intensity (arbitrary)
.. a 8 E 3
:==--

 

 

O 10 2) 3D 4) 50 00
21mm (thg'eea)

._x=orer\m=

 

 

 

 

 

 

 

Figure 19

 

 

 

In the present study, XRD patterns for VGCFS that were soxhlet extracted are
shown with GRAFOIL, a commercial graphite. From the position of the ﬁrst 20 peak
which is the 002 peak, the interlayer spacing can be obtained using equations 16 and 17,

and one can arrive at the spacing value for a h k L peak corresponding to 002. The

72

wavelength here is that for Cu K011 and is 1.54056 A. Using this, the following results

are obtained (ﬁgures 17, 18, and 19) and are shown in the table below.

 

 

 

 

Table 5
material 2 0 h k L intensity d-spacing
GRAFOILTM 26.49 0 o 0 2 887800 3.362 A
VGCF 26.16 0 0 0 2 14360 3.4036 A

 

 

 

 

 

 

The vapor grown carbon ﬁber used is hence proven to be turbostratic in nature
and does not have the three-dimensional order exhibited by graphite. Several authors 79

describe a linear dependence of degree of graphitization P and interlayer spacing d o 0 2 .

P = [3.44—d002]/[3.44—3.354] (19)

By this relation, the degree of graphitization for GRAFOIL works out to be
90.733 % whereas that for the Pyrograf III VGCF is 42.3 %. Iwashita and Ingaki 79 have
shown that not only is this linear relationship in error, but also changes with the carbon
type. Obtaining the true degree of graphitization (i. e., the volume fraction of 3-
dimensional graphitic stacking) by fourier analysis of XRD proﬁles, and comparing with
the d spacing, for different extents of heat treatment, the authors conclude that the trend
has a downward swing (below the linear proﬁle) for d-spacing in the ordinate versus P on
the abscissa. They show that between 3.40 A and 3.37 A d—spacing, there is very little
changeof the degree of graphitization from about 10 % to 40 %. This swing is least in

the above mentioned region for commercial cokes, and maximum, for vapor grown

73

carbon ﬁbers of diameters 8 to 13 um and in the range 0.1 to 1 pm, as well as for some
pitch-based carbon ﬁbers. Using that trend 79, for a d spacing of 3.4036 A for the VGCFS,
a degree of graphitization of about 22 to 25 % is obtained. A qualitative comparison of
the XRD patterns of VGCF and Grafoil also provides some insight. The full-width at half
maximum for Grafoil is much less and the peak is sharp and intense. In comparison, the
VGCF peak at 26 0 is broad and diffuse. Line broadening is due to crystallite size (only an
inﬁnite crystal would give a impulse peak) and due strain caused by ﬂuctuations in the
interlayer spacing 79. As is clearly seen, the peak at 26 0 (002) is unsymmetric and shifted
to a lower 20 value, indicating ﬂuctuations in d spacing towards a higher Spacing (from
Bragg formula). Although the most intense peak for the VGCF is at 26.160, there may be
other peaks at lower angles (higher spacing). Also, the broad hump at 20 of about 15 0 is
typically due to the partial amorphous nature of the VGCF 73.

The graphene layers of as-produced VGCFS are concentric and circular. It is
known that for these layers to form graphitic planes, they have to be polygonal in shape
81,79,25. The high interlayer spacing in as produced ﬁbers may also lead to poor properties

for the ﬁber due to internal shearing of concentric graphene layers.

74

4.7 Summag

In order to estimate the effect of ﬁber surface treatment on the adhesion of
VGCFS to epoxy matrices, several chemical and plasma treatments were carried out.
These treatments were conducted on VGCFS that were extracted with dichloromethane
(to remove physisorbed poly-nuclear aromatics on the ﬁber surface). The two chemical
treatments used were, oxidation by nitric acid and, amination. Due to the small size of
these ﬁbers and their ability to be easily air-borne, traditional surface analytical methods,
like XPS, cannot be used. EDS was used to evaluate the results of some of these surface
treatments, to provide qualitative analytical information. As the VGCF S already possess
carbon and oxygen, EDS was not used to analyze the ﬁbers oxidized by nitric acid as
quantitative differences in these chemical groups cannot be obtained. Likewise, none of
the plasma treated VGCFS were analyzed by EDS. Since the amination reactions used
aluminum, chlorine, and nitrogen, EDS was used to gage the efﬁcacy of the amination
procedure, and it confrrrned the incorporation of nitrogenous groups, presumably as
amine-functionalities, on the VGCF surface. XPS was used on model aluminum and AS-
4 carbon ﬁber substrates, to gage the effectiveness of methane plasma treatment on
VGCFS. Overall, a series of treatments that vary the surface chemical composition and
free energy, were conducted to ascertain their effect on adhesion of the treated VGCF to
the epoxy matrix.

Conventional methods of evaluating ﬁber-matrix adhesion cannot be used with
these sub-micron sized randomly distributed ﬁbers. It is known that good interfacial
adhesion improves composite mechanical properties, so dynamic mechanical analysis

was used to evaluate the storage modulus of composite specimens with 9.4 % and 12.8 %

75

ﬁber fraction (by volume), as a function of ﬁber surface treatment. Dynamic mechanical
analysis showed that within experimental error, storage modulus of the composite was
independent of surface treatment carried out on the ﬁber, and hence that no
improvements in VGCF adhesion to epoxy matrices occurred, with ﬁber surface
treatment. Using the random-in plane model for random short ﬁber composites, it was
determined that the ﬁber had a modulus of 34.5 GPa, which is well below the reported
value.

Estimation of ﬁber pull-out lengths of a fractured composite surface, and
examination of ﬁber/matrix interface, were also used to qualitatively evaluate the level of
adhesion in the VGCF/epoxy composites by environmental scanning electron
microscopy. Due to limitations of equipment resolution, due to the random three-
dirnensional orientation of the VGCFS , and, since the ﬁbers possess a range of aspect
ratios and also are curved and crenulated, it was found that little quantitative information
on differences in pull-out lengths between composite samples of ﬁbers differing in
surface treatment, could be obtained.

To explain the low value of ﬁber elastic modulus obtained, the VGCF micro-
structure was analyzed. Thermogravimetric analysis and x-ray diffraction were conducted
on bare VGCF and compared with the results for a commercial graphite, to ascertain the
nature of the VGCF micro-structure. TGA experiments in an inert atmosphere conﬁrmed
that VGCF was carbonaceous as it had the high temperature stability that graphites do.
However, elevated temperature TGA runs in air, showed that VGCF had lower oxidation
resistance than graphite, and this may be due to the partial amorphous nature of the

VGCF carbon, VGCF size effects, or both. X-ray diffraction revealed that the

76

microstructure of VGCF lacked the three-dimensional order of graphite, and had a higher
interlayer crystal spacing. It was seen from the non-gaussian shape of the XRD pattern
that the VGCF possessed more number of crystallites of interlayer spacing higher than
the peak value, than crystallites with lower values. The XRD pattern also showed a broad
hump, typical of partially amorphous carbons. This proved that VGCFS had a
considerable amount of amorphous carbon, and lacked three-dimensional order. This
micro-structural conﬁrmation justiﬁes the low modulus of the sub-micron diameter vapor

grown carbon ﬁbers.

77

5 Conclusions

Vapor grown carbon ﬁbers of average diameter 0.2 pm have been investigated to
determine their ability to reinforce polymers, to determine the effect of surface treatment
on their composite properties, and to estimate the ﬁber mechanical properties. It was
assumed that the sub-micron vapor grown carbon ﬁbers would have similar mechanical
properties and reinforcing ability as the larger diameter VGCFS.

Since ﬁber surface treatments improve the ability of the ﬁber to adhere to the
matrix (epoxy), several potential chemical and plasma surface treatments have been
identiﬁed and used to treat the VGCF. The treatments include, (i) solvent extraction of
weakly bound surface poly-aromatic layers, (ii) chemical oxidation by nitric acid, (iii)
incorporation of surface amine groups, (iv) oxidation by air-plasma, (v) deposition by
methane-plasma, and (vi) treatment with allyl glycidyl ether and argon plasma. While
methane plasma is known to deposit a hard low-energy surface layer, treatments (ii), (iv)
and (vi), increase the surface oxygen content, and treatments (iii) and (vi) incorporate
surface chemical groups that react with the matrix resin or curing agent. Several
treatments ( e. g. (ii), (iii), (iv) and (vi) ) conducted on larger diameter carbon ﬁbers have
been shown to improve the interfacial adhesion of those ﬁbers to polymers.

Due to the small size, the crenulated and curved shape of VGCFS, and as a result
of the random nature of the reinforcement orientation, conventional methods of
evaluating ﬁber/matrix adhesion cannot be used. However, since improvements in
adhesion have been Shown to increase the modulus of the resultant composite, the
ﬂexural storage modulus was determined for composites of each of the surface treated

VGCF samples, at 15 % and 20 % ﬁber weight fractions.

78

The VGCF-epoxy composite bending modulus increased with ﬁber content, but
was independent of the ﬁber surface treatment applied to the VGCFS. Using the
random-in-plane model for short ﬁber composites, and assuming a transverse ﬁber
modulus of 14 GPa, the longitudinal modulus of these vapor grown carbon ﬁbers was
calculated ﬁ'om the model, to be only 34.5 GPa.

The efﬁcacy and level of adhesion in a composite may also be qualitatively
Obtained by examining a fracture surface for ﬁber pull-out lengths. However, due to the
random nature of reinforcement orientation, it was not possible to detect differences in
levels of adhesion using the environmental scanning electron microscopy.

To determine if the low ﬁber modulus value was reasonable, the ﬁber was studied
using thermogravimetric analysis and x-ray diffraction. It was seen that the ﬁbers had a
lower oxidative resistance than that of graphite, and their microstructure, as studied by X-
ray diffraction, lacked the three-dimensional order present in graphite. The interlayer
spacing between crystal planes was determined to be higher than that of graphitic crystals
and the VGCFS had a partial amorphous content.

The high interlayer spacing in these 0.2 pm VGCFS, coupled with their concentric
ring micro-structure, may allow internal shearing of the concentric graphene layers and
cohesive failure mode within the ﬁbers, resulting in a low ﬁber longitudinal elastic
modulus, and poor adhesion. Hence surface treating the outer layer of these ﬁbers may
not affect adhesion. Another possible explanation may be that, as VGCFS have a hollow
core of approximate dimensions of the catalyst particles used in its production, the micro-
ﬁbers may have a stiff shell structure of carbon with a central hollow tube, that may

result in a net low material modulus. It is known that for larger diameter VGCFS, the

79

micro-structure changes from turbostratic and non-crystalline at the hollow tube wall, to
micro-crystalline at the ﬁber periphery 22. The stiffening ability of the ﬁber depends upon
it’s modulus, which will increase if the ﬁber is more graphitic 24’ 25’ 26. For the sub-micron
VGCFS, at their low value of ﬁber elastic modulus, there is very little dependence of
composite modulus of a randomly oriented short ﬁber composite on ﬁber aspect ratio and
hence, on adhesion.

Hence it is concluded that the sub-micron vapor grown carbon ﬁbers may have
very good thermal and electrical properties, but they do not have the modulus required

for structural reinforcing ability.

80

BIBLIOGRAPHY

10.

11.

12.

13.

14.

BIBLIOGRAPHY

Hull, D. and Clyne, T. W., An Introduction to Composite Materials, 2nd edition,
1996.

Mallick, P. K., Fiber-Reinforced Composites, Marcel Dekker, Inc., 2“‘1 edition,
1993.

The Plastics and Rubber Institute, London, England, Carbon Fibers —
Technology, Uses and Prospects, Noyes Publications, 1986.

Cox, H. L., “The elasticity and strength of paper and other ﬁbrous materials”,
British Journal of Applied Physics, 3:72, 1952.

Clyne, T. W., “A Simple Development of the Shear Lag Theory Appropriate for
Composites with a Relatively Small Modulus Mismatch”, Materials Science &
Engineering, A122, p 183-192, 1989.

Drzal, L. T., Rich, M. J. and Lloyd, P. F., “Adhesion of Graphite Fibers to Epoxy
Matrices: I. The Role of Fiber Surface Treatment”, Journal of Adhesion, v 16, p
1-30, 1982.

Osipow, L. 1., Surface Chemistry: Theory and Industrial Applications,
Reinhold, 1962.

Adamsom, A. W. and Gast, A. P., Physical Chemistry of Surfaces,
Wiley-Interscience, 6th edition, 1997.

Sharpe, L. H., “Short Course on Adhesion”, Annual Meeting of The Adhesion
Society, South Carolina, USA, February 17, 1996.

Drzal, L. T., “Fiber-Matrix Adhesion in Composites”, presented at the Tenth
Annual Intensive Short Course “Chemistry and Physics of Adhesion”, Akron,
USA, April 1992.

Liston, E. M., “Plasma Treatment for Improved Bonding: A Review”, Journal of
Adhesion, v 30, p 199-218, 1989.

Milewski, J. V. and Katz, H. S., “Whiskers”, Handbook of Fillers and
Reinforcements, 1987.

Prewo, K. M., Journal of Material Science, 15, p 2458, 1980.

Kardos, J. L. and Tang, L., “A Review of Methods for Improving the Interfacial

81

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

Adhesion between Carbon Fiber and Polymer Matrix”, Polymer Composites,
v18, n1, p100,1997.

Woodruff, D. P. and Delchar, T. A., Modern Techniques of Surface Science,
Cambridge University Press, 2“d edition, 1994.

Drzal, L. T. and Herrera-Franco, P. J ., “Comparison of methods for the
measurement of ﬁber / matrix adhesion in composites”, Composites, v 23, n 1,
p2-27, January 1992.

Favier, V., Canova, G. R., Shrivastava, S. C. and Cavaille, J. Y., “Mechanical
Percolation in Cellulose Whisker Nanocomposites”, Polymer Engineering and
Science, v 37, n 10, p 1732-1739, October 1997.

Helbert, W., Cavaille, J. Y. and Duﬁesne, A., “Thermoplastic Nanocomposites
Filled with Wheat Straw Cellulose Whiskers. Part 1: Processing and Mechanical
Behavior”, Polymer Composites, v 17, n 4, p 604, August 1996.

Lang, R. W., Manson, J. A. and Hertzberg, R. W., “Mechanisms of
fatigue fracture in short glass ﬁbre-reinforced polymers”, Journal of Materials
Science, v 22, p 4015-4030, 1987.

Horst, J. J. and Spoorrnaker, J. L., “Mechanisms of Fatigue in Short Glass Fiber
Reinforced Polyamide 6”, Polymer Engineering and Science, v 36, n22, p 2718,
November 1996.

Ting, J -M. and Lake, M. L., “Vapor-Grown Carbon-Fiber Reinforced Carbon
Composites”, Carbon, v 33, n 5, p 663-667, 1995.

Endo, M., “Grow carbon ﬁbers in the vapor phase”, Chemtech, p 568, September
1998.

Serp, Ph. and F igueiredo, J. L., “A microstructural investigation of vapor-grown
carbon ﬁbers”, Carbon, v 34, p 1452, 1996.

Endo, M., Koyama, T. and Hishiyarna, Y., “Structural Improvement of Carbon
Fibers Prepared ﬁ'om Benzene”, Japanese Journal of Applied Physics, v 15, n 11,
p 2073-2076, November 1976.

Endo, M., Oberlin, A. and Koyama, T., “ High Resolution Electron
Microscopy of Graphitizable Carbon Fiber Prepared by Benzene
Decomposition”, Japanese Journal of Applied Physics, v 16, n 9, p1519-1523,
September 1977.

Saadaoui, H., Roux, J. C. and Flandrois, S., “Graphitization Of Pyrocarbons — An
STM Study”, Carbon, v 31, n 3, p 481, 1993.

82

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

Ting, J-M., “Processing-Microstructure-Tensile Property Of Vapor Grown
Carbon Fiber Reinforced Carbon Composite”, Materials Research Society
Symposium Proceedings, v 383, 1995.

Madronero, A., Ariza, E., Verdu, M., Brandl, W. and Barba, C., “Some
microstructural aspects of vapour-grown carbon ﬁbers to disclose their failure
mechanisms”, Journal of Materials Science, v 31, p 6189, 1996.

Morgan, R. J. and Walkup, C. M., “Epoxy Matrices for Filament-Wound Carbon
Fiber Composites”, Journal of Applied Polymer Science, v 34, p 37, 1987.

Rao, V., “Interfacial Changes during the Processing of a typical Carbon Fiber /
Epoxy Composite Material”, Ph.D. Dissertation, Department of Chemical
Engineering, Michigan State University, 1991.

Shell EPON 828 epoxy resin catalogue, Miller Stephenson Chemical Company,
Incorporated, Connecticut, USA.

Morgan, R. J. and O’Neal, J. E., “A Review Of The Relation Between The
Physical Structure And Mechanical Response of Epoxies”, Chemistry and
Properties of Cross-linked Polymers, editior : Santokh S. Labana.

Drzal, L. T., Gupta, V. B., Lee, C. Y-C. and Rich, M. J ., “The Temperature-
Dependence of Some Mechanical Properties of a Cured Epoxy Resin System”,
Polymer Engineering and Science, v 25, n 13, September 1985.

Tibbets, G. G., Bernardo, C. A., Gorkiewicz, D. W. and Alig, R. L., “Role of
Sulfur in the Production of Carbon Fibers in the Vapor Phase”, Carbon, v 32, n 4,
p 569, 1994.

Serp, Ph. and Figueiredo, J. L., “An Investigation of Vapor-Grown Carbon Fiber
Behavior towards Air Oxidation”, Carbon, v 35, n 5, p 675, 1997.

Tibbetts, G. G., “Vapor-Grown Carbon Fibers”, Carbon Fibers, Filaments and
Composites, edited by: F igueiredo, J. L., Bernardo, C. A., Baker, R. T. K. and
Huttinger , K. J ., Applied Sciences : E, NATO ASI series, 1989.

Tibbetts, G. G. and Beetz, C. P., Jr., “Mechanical properties of vapour-grown
carbon ﬁbres”, Journal of Physics : D, v 20, p 292, 1987.

Tibbetts, G. G., Doll, G. L., Gorkiewicz, D. W., Moleski, J. J ., Perry, T. A.,

Dasch, C. J. and Balogh, M. J ., “Physical Properties Of Vapor-Grown Carbon
Fibers”, Carbon, v 31, n 7, p 1039-1047, 1993.

83

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

Madronero, A. and Verdu, M., “Hydrogen Content Evaluation in Vapor-Grown
Carbon Fibers by SIMS”, Carbon, v 33, n 3, p 247-251, 1995.

Van Hattum, F. W. J ., Serp, Ph., F igueiredo, J. L. and Bernardo, C. A., “The

effect of morphology on the properties of vapour-grown carbon ﬁbres”, Carbon,
V 30, n 6, p 860, 1997.

Tibbetts, G. G., “Vapor-Grown Carbon Fibers : Status and Prospects”, Carbon, v
27, n 5, p 745-747, 1989.

Patton, R. D., Pittman, C. U. Jr. and Wang, L., “Vapor Grown Carbon Fiber
Composites With Epoxy And Poly(phenylene sulﬁde) Matrices”, 29th
International SAMPE Technical Conference Proceedings, p 77, 1997.

Proprietary information of Applied Sciences, Inc. Data accurate and current as of
September 24‘“, 1996.

Hager, J. W., Ting, J-M. and Lake, M. L., “Interphase Challenges In The
Fabrication Of Polymer-Matrix Composites Reinforced With Vapor Grown
Carbon Fiber”, Advanced Materials: Development, Characterization, Processing,
and Mechanical Behavior (Book of Abstracts), MD-Vol. 74, ASME 1996.

Private correspondence with Dr. Paul Loconto, Department of Environmental
Engineering, Michigan State University, April 1996.

Hager, J. W., “VGCF Composites For Automotive Applications”, ATP Quarterly
Technical Progress Report, Applied Sciences, Inc., January 31“, 1997.

Tsubokawa, N. and Yoshihara, T., “Grafting of Polyesters from Carbon Whisker
Surface: Copolymerization of Epoxides with Cyclic Acid Anhydrides Initiated by
COOK Groups Introduced on to Surface”, Journal of Polymer Science: Part A:
Polymer Chemistry, v 31, p 2459, 1993.

Kaplan, S. L. and Rose, P. W., “Plasma Treatment Upgrades Adhesion in Plastic
Parts”, Plastics Engineering, v 44, n 5, p 77, 1988.

Work conducted with V. Raghavendran, Composite Materials and Structures
Center, Michigan State University, East Lansing, Michigan, USA, 1998.

Yasuda, H., Plasma Polymerization, Academic Press, 1985.
Yasuda, H., “Plasma Polymerization and Plasma Interactions with Polymeric

Materials”, Journal of Applied Polymer Science: Applied Polymer Symposia 46,
April 1990.

84

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

Rempp, P. F. and Franta, E., “Macromonomers : Synthesis, Characterization and
Applications”, Advances in Polymer Science, v 58, p 1-53, 1984.

Kemp, D. S. and Vellaccio, F ., Organic Chemistry, Worth Publishers, Inc.,1980.
Taylor, R., Electrophilic Aromatic Substitution, John Wiley and Sons, 1980.
Adams, R., Organic Reactions, v 3, 1946.

Clar, E., Polycyclic Hydrocarbons, v 2, 1964.

Price and Tomisek, Journal of the American Chemical Society, 65, 439,1943.
Work carried out by Ming Xie, Research Associate, Composite Materials and

Structures Center, Michigan State University, East Lansing, Michigan, USA,
1997.

Tsubokawa, N. and Ueno, H, “Radical graft polymerization from carbon whisker

initiated by peroxyester groups introduced onto the surface”, Composite
Interfaces, v 3, n 3, p 209-20, 1995.

Tsubokawa, N. and Ueno, H., “Radical graft polymerization initiated by azo
groups introduced onto the surface of carbon whisker”, Composite Interfaces, v 4,
n1, p11-20,1996.

Trost and Fleming, Comprehensive Organic Synthesis, v 2, 1991.

Barton, D. and Ollis, D. W., Comprehensive Organic Chemistry, v 2, 1979.

The Polymeric Materials Encyclopedia, CRC Press, Inc., 1996.

Work carried out by Askeland, P., Research Associate, Composite Materials and
Structures Center, Michigan state University, East Lansing, Michigan, USA,
1998.

Sherwood, P. M. A., “Surface analysis of carbon and carbon ﬁbers for
composites”, Journal of Electron Spectroscopy and Related Phenomena, v 81, p
319, 1996.

Goldstein, J. I., Newbury, D. E., Echlin, P., Joy, D. C., Romig, A. D. Jr., Lyman,
C. E., Fiori, C. and Lifshin, E., Scanning Electron Microscopy and X-Ray

Microanalysis, 2 “d edition, Plenum Press, 1992.

Drzal, L. T. and Schalek, R. L., “High-pressure Electron Beam Imaging”,
Advanced Materials and Processes, ASM International, v 152, n 1, July 1997.

85

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

Lang, R. W., Manson J. A. and Hertzberg, R. W., Polymer Engineering and
Science, v 22, p 982, 1982.

Haj ji, P., Cavaille, J. Y., Favier, V., Gauthier, C. and Vigier, G., “Tensile
Behavior of Nanocomposites from Latex and Cellulose Whiskers”, Polymer
Composites, v 17, n 4, August 1996.

Yosomiya, R., Fujisawa, T. and Morimoto, K., “Carbon Fiber having Isocyanate
groups on the Surface”, Polymer Bulletin, v 12, p 523, 1984.

On, Y-C., Zhu, J. and F eng, Y-P., “Interfacial Design of the Nonpolar Polyolefm
Ternary Composite with High Strength, High Toughness, and High Modulus”,
Journal of Applied Polymer Science, v 59, p 287-294, 1996.

Turi, E. A., Thermal Characterizaton of Polymeric Materials, v 1 & 2, 2Ind edition,
Academic Press, 1997.

Cullity, B. p., “Elements of X-Ray Diffraction”, 2 "1 edition, 1978.

Material Safety Data Sheet, Pyrograf III VGCF, Applied Sciences Incorporated,
Ohio, USA, April 19, 1996.

Callister, W. D., Jr., Materials Science and Engineering, third edition, John Wiley,
p 528, 1994.

Huttinger, K. J ., Zielke, U. and Hoffman, W. P., “Surface-Oxidized Carbon Fibers
I : Surface Structure and Chemistry”, Carbon, v 34, n 8, p 983-998, 1996.

Koyama, T., Endo, M. and Onuma, Y., “Carbon Fibers Obtained by Thermal
Decomposition of Vaporized Hydrocarbon”, Japanese Journal of Applied Physics,
v11, 11 4, April 1972.

Endo, M., Takeuchi, K., Kobori, K., Takahashi, K., Kroto, H. W. and Sarkar, A.,
“Pyrolytic Carbon Nanotubes ﬁ'om Vapor-Grown Carbon Fibers”, Carbon, v 33, n
7, p 873-881, 1995.

Iwashita, N. and Ingaki, M., “Relations Between Structural Parameters Obtained
By X-Ray Powder Diffraction Of Various Carbon Materials”, Carbon, v 31, n 7,
p1107-1113, 1993.

Darmstadt, H., Summchen, L., Ting, J-M., Roland, U., Kaliaguine, S. and Roy,

C., “Effects of Surface Treatment on the Bulk Chemistry and Structure of Vapor
Grown Carbon Fibers”, Carbon, v 35, II 10-11, p 1581-1585, 1997.

86

81.

82.

83.

84.

Endo, M., Koyama, T. and Hishiyama, Y., “Structure and Properties of
Graphitized Carbon Fiber”, Japanese Journal of Applied Physics, v 13, n 12,
December 1974.

Halpin, J. C., Primer on Composite Materials : Analysis, Technomic Publishing
Co., 1984.

Dresselhaus, M. S., Dresselhaus, G., Sugihara, K., Spain, I. L. and Goldberg, H.
A., Graphite Fibers and Filaments, Springer Series in Materials Science : 5,

Springer Verlag, 1988.

Alig, B., “Evolution of Pyrograf 111 TM”, Energeia, Center for Applied Energy
Research Newsletter, University of Kentucky, v. 17, n. 1, 1996.

87

"11111111111111