505.5...” a...» z .
: i
3.4. i

.
n): I .yA-(I.
5.19....

Q
.5 . .7 ..
:. 5...... 25... u o

a

.
. ....
. 9.1,...

-.
x.i-|;..Q..£?
. I at
a ‘3‘...- .o.. .

I31... 5 ..\

1.94%...“ 53......“ .
.‘ . ’PllVI 4&1.
ix... 3:.

i.

I. r':
.17..

.. xx.
.33.... :
\3Is.
. n .«r» I .

g :lei. s . :A

naﬁhi due. .2.) in"...
5.4.. $45. 3 3. niftiff
.. . z... p. 11...? sh:
I .. . u.

t 1. :05". v0.09! .,
iaquiu 21.. HERE.“

y 1.1:]:
.. . .5)

94...... 313.55.. :4 .

421: .
‘1sz.-
..71 .0
2.10).

\él. Blit‘:

 

v5 {tifrt
1:, I. l. saylnov... . 1.3

 

lililllllllllil ;

r

This is to certify that the

thesis entitled

Adhesion, Processing and Interfacial Properties of

Preceramic Polymer Coatings on Carbon Fibers

presented by
Tauseef Muhammad Chaudhry

has been accepted towards fulﬁllment
of the requirements for

M.S. degree in Materials Science

 

Major professor

Date September 22, 1995

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

 

 

 

LIBRARY
Mlchlgan State
University

 

 

 

PLACE ll RETURN BOX to romovo this chockout from your rocord.
To AVOID FINES Mum on or bdoro duo duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ADHESION, PROCESSING AND INTERFACIAL PROPERTIES OF

PRECERAMIC POLYMER COATINGS ON CARBON FIBERS

Tauseef Muhammad Chaudhty

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

MASTER OF SCIENCE

Department of Materials Science and Mechanics

1995

ABSTRACT
ADHESION, PROCESSING AND INT ERFACIAL PROPERTIES OF
PRECERAMIC POLYMER COATINGS ON CARBON FIBERS
By
Tauseef Muhammad Chaudhry
Preceramic inorganic polymer coatings are applied to graphite ﬁbers by dip coating tows
of ﬁbers into hexane solution of the silsesquioxane terpolymer
—[MeHSiO]o,1[MeSi(OnPr)O]o_2[MeSi(O)1,5]o_7— to form silsesquioxane coatings typically
0.1 pm thick. The coatings are ' then pretreated by heating in nitrogen to high
temperatures (200 °C increments to 1000 °C). The amorphous polymeric coatings were
shown to convert to amorphous silicon oxycarbide coatings on heating to temperatures
above 600 °C. Single carbon ﬁbers coated with silicon oxycarbide coatings were
pyrolyzed and then encapsulated in a transparent epoxy matrix so that stress could be
applied and the adhesion and fracture behavior of the coatings could be observed. The
properties and adhesion of the coatings to the carbon ﬁbers change with the degree of
pyrolysis. Initially the adhesion is low and a mixed ﬁber-coating and coating-matrix
interfacial failure is observed. After thermal processing to 800 °C and above, a signiﬁcant
increase in adhesion with circumferential cracks in the coating are detected and the ﬁber-
coating interfacial failure is observed. The interface is observed as intact in the case of the
unstrained samples, whereas it is disrupted in the samples subjected to uniaxial tension. A
non-linear axisymmetric ﬁnite element model of this single ﬁber coated with material
having properties varying in a manner similar to the coating was used to analyze the

mechanical state of stress in the coating and at the interface.

DEDICATED
TO
Dr. Muhammad Iqbal, The Poet-Philosopher of the 20th Century
And

Professor‘Annemarie Schimmel, The Orientalist of Present Time

Who Profoundly Moved and Inspired Me.

iii

ACKNOWLEDGMENTS

I thank Allah, the Creator and the Cherisher of the universe for giving me the
ability to complete this work. I am thankful to my parents who always stressed the
importance of education as well as personal character.

My heart felt thanks are due to my adviser Professor Lawrence T. Drza] for giving
me the opportunity and for his guidance and encouragement towards understanding of
things and towards puriﬁcation of thought.

I would like to thank the friendly staff of Composite Materials and Structures
Center (CMSC). My thanks are due to Mike Rich and Brian Rook for their help and
suggestions in operating the equipments at CMSC. I am especially thankﬁil to Dr. Henjen
Ho for the helpful and informative discussions on the ﬁnite element model he developed
and for his friendship. I thank Dan Hook for his help in surface analysis. My thanks go to
Dr. Steve Rozeveld for the help in environmental scanning electron microscopy.

I am indebted to Dr. John Heckman of the Center for Electron Optics for his
assistance in the ultrarnicrotomy and transmission electron microscopy.

Finally, I would like to thank all my friends here at the Michigan State University

for their support and ﬁ'iendship.

iv

List of Tables

List of Figures

Chapter 1:

Chapter 2:

Chapter 3:

TABLE OF CONTENTS

INTRODUCTION

BACKGROUND

2.1 Composite Interphase
2.2 Theories of Adhesion

2.2.1 Adsorption and Wetting
2.2.2 Interdiffusion

2.2.3 Electrostatic Attraction
2.2.4 Chemical Bonding
2.2.5 Mechanical Adhesion

2.3 Adhesion Tests

2.3.1 Single Fiber Fragmentation Test
2.3.2 Microindentation Technique

2.4 Carbon Fibers

2.5

Polymer-Derived Ceramics

2.6 Inorganic Preceramic Polymer Coating

2.6.1 Synthesis

2.6.2 Pyrolytic Behavior of Bulk Silsesquioxane
2.6.3 Oxidation Resistance

2.6.4 Coating Method and Pyrolysis

2.7 Coating Characterization Methodology

3.1

EXPERIMENTAL

Materials

10
10
11
12
12
13
13
14
18
21
23
25
25
27
28
29
30

31

31

3.2 Adhesion Characterization
3.2.1 Single Fiber Fragmentation Test
3.2.2 Microindentation or ITS Test
3.3 Coating Uniformity and Thickness
3.4 Coating Chemistry and Composition Characterization

3.5 Ultramicrotomy and Transmission Electron Microscopy
Chapter 4: RESULTS

4.1 Coating Uniformity and Thickness
4.2 Coating Chemistry and Composition
4.3 Adhesion and Failure Modes
4.3.1 Single Fiber Fragmentation Test
4.3.2 ITS Test

4.4 Ultramicrotomy and Transmission Electron Microscopy

Chapter 5: FINITE ELEMENT MODEL and DISCUSSION
5.1 Material Properties
5.2 Non-Linear Finite Element Analysis of the Effect
of Coating Modulus (200 °C-6OO °C)
5.3 Non-Linear Finite Element Analysis of the Effect
of Coating Modulus (800 °C-1000 0C)
5.4 Effect of Coating Thickness

Chapter 6: CONCLUSIONS

Bibliography

33
33
37
39
42
43

48

48
56
59

69
74

87
89

91

93
96

99

101

Table 3.1

Table 4.1

Table 4.2

Table 4.3

Table 4.4

LIST OF TABLES

Properties of (a) Thornel P-55 carbon ﬁbers and (b) EPON 828
epoxy matrix cured with 14.5 wt.% of mPDA.

Surface elememtal composition of the coated carbon ﬁbers as a
function of processing temperature by X-ray Photoelectron

Spectroscopy.

Ratios of surface elements by X-ray Photoelectron Spectroscopy.

Critical aspect ratio and interfacial shear strength determined by
the single ﬁber fragmentation test.

Interfacial shear strength by the Microindentation with the interfacial
testing system (ITS).

32

57

57

61

71

Figure 2.1

Figure 2.2
Figure 2.3

Figure 2.4

Figure 2.5
Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

LIST OF FIGURES

Schematic representation of ﬁber-matrix interphase in a composite
material.

Schematic representation of the single ﬁber fragmentation process.

Schematic of Interfacial Testing System (ITS).

Major processing steps necessary to prepare the four types of
polymer—derived ceramic.

Chemical structure of silsesquioxane terpolymer.
Chemical structures of (a) EPON 828 epoxy and (b) curing agent.

(a) Silicon mold and (b) Loading jig used in the single ﬁber
fragmentation test.

Characteristic X-ray and Auger electron production.

Schematic representation of (a) a top view showing the cutting
angle to the ﬁber and (b) a ribbon formed by the sections (in
which the ﬁber is in an elliptical shape) in Ultramicrotomy
(diagram is not to scale).

Scanning electron micrograph of the surface of uncoated P—55
carbon ﬁbers showing axial striations and some dirt from the
manufacturing process.

Scanning electron micrograph of 200 °C heat treated coated
carbon ﬁbers showing the same axial striations and dirt as was

the case with the uncoated ﬁbers.

Environmental scanning electron micrograph of 400 °C heat
treated coated carbon ﬁbers.

Environmental scanning electron micrograph of 600 °C heat
treated coated carbon ﬁbers.

Scanning electron micrograph of 800 °C heat treated coated

15

19

24

26

34

35

4o

46

49

50

51

52

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12

Figure 4.13

Figure 4.14

Figure 4.15

Figure 4.16

Figure 4.17

Figure 4.18

carbon ﬁbers.

Scanning electron micrograph of 1000 °C heat treated coated
carbon ﬁbers.

Auger sputtering proﬁles of silsesquioxane coated ﬁbers
pretreated to 200 °C, 400 °C, 600 °C, 800 °C and 1000 °C.

Interfacial shear strength determined by the single ﬁber
fragmentation test.

Transmitted polarized light photoelastic stress pattern of a
typical uncoated P-55 carbon ﬁber under load in the single
ﬁber fragmenation test.

Transmitted polarized light photoelastic stress pattern of a
typical 200 °C treated coated carbon ﬁber under load in the
single ﬁber fragmenation test.

Transmitted polarized light photoelastic stress pattern of a
400 °C treated coated carbon ﬁber under load in the single
ﬁber fragmenation test.

Transmitted polarized light photoelastic stress pattern of a
600 °C treated coated carbon ﬁber under load in the single
ﬁber fragmenation test.

Transmitted polarized light photoelastic stress pattern of an
800 °C treated coated carbon ﬁber under load in the single
ﬁber fragmenation test.

Transmitted polarized light photoelastic stress pattern of a
1000 °C treated coated carbon ﬁber under load in the single

ﬁber fragmenation test.

Reﬂected light micrograph of 1000 °C treated coated ﬁber
with coating fractures indicated by vertical marks.

Interfacial shear strength determined by the microindentation
test with the Interfacial Testing System (ITS).

A typical micrograph of a specimen in the microindentation
(ITS) test showing an indented ﬁber (in the center).

Load versus displacement graph for the uncoated and coated

53

54

55

61

62

63

65

66

67

68

7O

71

72

Figure 4.19

Figure 4.20

Figure 4.21

Figure 4.22

Figure 4.23

Figure 4.24

Figure 4.25

Figure 4.26

Figure 4.27

carbon ﬁbers obtained from the ITS test.

Transmission electron micrograph of a typical ultramicrotomed
section containing an elliptical shaped ﬁber in the matrix. The
direction of the knife is shown by the arrow and it is perpendicular
to the interface.

Transmission electron micrograph of ultramicrotomed section of
200 °C treated coated ﬁber obtained from strained (above) and
unstrained (below) dog-bone coupons. The direction of the knife
is shown by the arrows.

Transmission electron micrograph of ultramicrotomed section of
400 °C treated coated ﬁber obtained from unstrained (above) and
strained (below) dog-bone coupons. The direction of the knife

is shown by the arrows.

Transmission electron micrograph of ultramicrotomed section of
600 °C treated coated ﬁber obtained from unstrained (above) and
strained (below) dog—bone coupons. The direction of the knife

is shown by the arrows.

Transmission electron micrograph of ultramicrotomed section of
800 °C treated coated ﬁber obtained from unstrained (above) and
strained (below) dog-bone coupons. The direction of the knife

is shown by the arrows.

Transmission electron micrograph of ultramicrotomed section of
1000 °C treated coated ﬁber obtained from strained (above) and
unstrained (below) dog-bone coupons. The direction of the knife
is shown by the arrows.

Transmission electron micrograph of an ultramicrotomed section
obtained from the strained dog-bone coupon of 200 °C treated
coated ﬁber showing the coating adhering to the matrix. The
direction of the knife is shown by the arrow.

Transmission electron micrograph of an ultramicrotomed section
obtained from the strained dog-bone coupon of 800 °C treated
coated ﬁber showing the coating adhering to the matrix. The
direction of the knife is shown by the arrow.

Typical graph showing the presence of silicon (adhering to the
matrix) in the Energy dispersive X-rayt analysis of ultramicrotomed

73

75

76

77

78

79

8O

82

83

Figure 4.28

Figure 5.1
Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

section from a coated ﬁber's strained dog-bone specimen.

Percentage of the coating adhering to the matrix in the
ultramicrotomed sections from strained dog-bone specimens of
200 °C to 1000 °C treated ﬁbers observed through the transmission
electron microscope.

Finite element model.
Stress-strain curve of the constituent materials.

(a) Fiber axial and (b) interphase shear stress distributions for
fragments with various EJEm ratios; (b) maximum ﬁber axial
stress as a function of E/Em. A 20 um debond length is assumed
at the ﬁber break.

(a) Fiber axial and (b) interphase shear stress distributions for
fragments with various interphase thicknesses for EJE... = 1.8.
A 20 um debond length is assumed at the ﬁber break.

(a) Fiber axial and (b) interphase shear stress distributions for
fragments with various interphase thicknesses for Fri/Em = 0.3.
A 20 um debond length is assumed at the ﬁber break.

84

85

88

90

92

95

97

 

 

INTRODUCTION, , - ...

 

 

The long term goals of the work are (1) to develop methods of modifying ﬁber
surfaces with well-deﬁned ceramic coatings made using precerarnic polymers, (2) to
identify (deﬁne) the ceramic coating and coating-ﬁber interface failure mechanisms at both
the atomic and nanostructural levels, and (3) to use the data obtained to identify the
fundamental material, processing and adhesion factors that control interface properties in
order to have the maximum beneﬁt of these coatings on Metal Matrix Composite (MMC)
and Ceramic Matrix Composite (CMC) mechanical properties.

The ﬁrst step in this process has been to establish a benchmark interface and to
clearly deﬁne its attendant properties. The characteristics of the benchmark interface
coating material, silicon oxycarbide, SiOxCy or black glass have been studied [1]. SiOxCy
was chosen because (1) SiOxCy is amorphous, (2) it is possible to prepare very well-
defmed materials, where the chemistry and the evolution of the material with time and
temperature are known in detail, and (3) SiO,.Cy is a matrix material used in commercial
composites. Results were presented which provided the experimental methods and the
characterization of the processable inorganic polymer precursor,
-[MeHSiO]o,l[MeSi(OnPr)O]o,2[MeSi(O)1.5]o_7—. In addition, it has been shown that these

coatings are effective in increasing the oxidation resistance of the carbon ﬁbers

themselves. Most of the associated changes in chemistry and oxidation resistance
improvement with conversion from inorganic polymer to ceramic coating take place in the
temperature regime from 600 °C to 1000 °C. During this process of conversion from
inorganic polymer to ceramic, the coating converts to an amorphous silicon oxycarbide,
there is a volumetric shrinkage associated with this conversion and the coating has the
properties of a ceramic instead of a polymer.

To determine whether the precerarnic inorganic polymer coating can provide not
only the thermal oxidative protection during both processing and use in MMCs or CMCs
but also the appropriate composite properties, it is desirable to know how and at what
point in the thermal processing cycle the coating-carbon ﬁber interface undergoes changes
that affect the interfacial adhesion and failure mode. Also, it is important to identify the
locus of interfacial failure i.e. between ﬁber and coating or between coating and matrix.
This work is directed at determining the interfacial changes and the locus of failure in
order to optimize both the coating chemistry and the conversion process.

Intense efforts to develop superior ﬁber-reinforced composites with exceptional
mechanical properties and long-term structural reliability have been undertaken during the
past decade. A considerable part of this effort has focused on developing basic principles
concerned with how stress from an applied load is distributed (1) within the composite
matrix materials; (2) through the ﬁbers themselves and especially, (3) at the ﬁber-matrix
interface where primary load transfer occurs.

Much of the work on load transfer at the interface has targeted the development of

fundamental structure-property relationships from ﬁber-matrix interface behavior. Several

groups have explored approaches based purely on mechanics with some basic assumptions
concerning adhesion of the ﬁber to the matrix [2-4]. Others have considered the idea that
.load transfer occurs solely at the interface via the chemical links at that interface [5,6].
Both methods fall short of accurately predicting composite performance. It is now
accepted that coupling these approaches to structure-property relationships provides a
more accurate method of predicting composite behavior [7,8]. Recent studies show that
small changes in the level of adhesion and the properties of the material in the interphase
region can have profound effects on composite mechanical properties [9,10].

In large part, these relationships were developed for rigid, reinforcing ﬁbers in
relatively compliant matrix materials, polymers. Less effort has been directed towards
deve10ping interfacial structure-property relationships for CMCs and MMCs. This
problem derives from the technical difﬁculties inherent in reproducibly processing CMCs
and MMCs. Additional problems arise because it is much more difﬁcult to characterize
interface properties in MMCs and CMCs by comparison with polymer composite systems.
As a consequence, detailed modeling studies on load transmission across the ﬁber-matrix
interfaces in CMCs and MMCs have not been forthcoming. Furthermore, little effort has
been made to develop general methods of tailoring ﬁber-matrix interfaces in these
materials to optimize performance as a function of application. The continued and
growing importance of ﬁber-reinforced CMCs and MMCs in both defense and industrial
applications requires that these problems be addressed and provides motivation for the

work reported here.

 

(ﬂﬂUﬂERZ

BACKGROUND,

 

 

To develop detailed structure—processing-property relationships, one must identify
key criteria that can be used; ﬁrst to assess a given interface, and then to justify any
resulting conclusions concerning long term mechanical properties. At the macroscopic
level, we must be concerned with interfacial failure mechanisms both within the ﬁber
coating and at the coating-matrix interface, either during processing or in use. At the
atomic scale, we must be concerned about interface chemistry.

Establishment of structure-property relationships between ﬁber-matrix adhesion
and composite behavior depends on both the level of adhesion and the interfacial failure
mode if composite properties are to be optimized for strength and toughness. High
strength interfaces by themselves produce a low degree of composite toughness by
allowing cracks to propagate across ﬁbers in a catastrophic manner. Very low values of
adhesion produce an increase in composite toughness but at the expense of static
mechanical properties. If a ceramic coating is placed at the interphase, the brittleness and
fracture behavior of the coating during composite processing and subsequent usage can
reduce the ﬁber strength and composite properties. Understanding of the ﬁber surface

chemistry, the development of interfacial bonds, the interdiffusion and resulting interphase

property gradient are all necessary information required before optimum interfaces can be
designed.

At the atomic level, quantiﬁcation of the elemental composition and identiﬁcation
of particular bonding arrangements present both at the ﬁber surface and in the interfacial
material, especially as both evolve with processing, may provide insight into
nanostructural events, e.g. failure modes. Auger, XPS and FTIR spectroscopic methods
can be used to quantify the structural and compositional changes that occur with
processing which can be used as the basis to quantify the factors that affect adhesion of an
applied coating or matrix.

Total encapsulation of the ﬁber in a polymer matrix has the advantages of both
placing the ﬁber under the same state of stress encountered in a CMC or MMC and also
allowing the interface to be interrogated and observed with methods found useful with the

ﬁber-matrix interphase in composite materials.

2.1 COMPOSITE INTERPHASE

The juncture that exists when the matrix and ﬁber come together in composite
materials is a three dimensional interphase rather than a two dimensional
interface/boundary. The complexity of this region can best be illustrated with the use of a
schematic model which allows the potential important characteristics of this region to be
identiﬁed as shown in Figure 2.1.

The interphase exists from some point in the ﬁber where the local properties begin

to change from the ﬁber bulk properties through the interface into the matrix where the

 

 

 
 
   
  
  
 
  

Bulk matrix
of different properties
, Sizings
Chemical,
and Mechanical. Adsorbcd material
Environments
Fiber chemistry
Fibertopograph’y
Fiber morphology
Bulk Fiber

 

Figure 2.1 Schematic representation of ﬁber-matrix interphase in a

composite material [12].

 

local properties again equal the bulk properties. Components of this region can be
identiﬁed. The ﬁber may have morphological variations near the ﬁber surface which are
not present in the bulk of the ﬁber. The surface area of the ﬁber can be much greater than
its geometrical value because of pores or cracks present on the surface. The atomic and
molecular composition of the ﬁber surface can be quite different from the bulk of the ﬁber.
Surface treatments can add surface chemical groups and remove the original surface
giving rise to chemically and structurally different region. Exposure to air before
composite processing can result in the adsorption of components which may alter or
eliminate certain beneﬁcial surface reactivity. Thermodynamic wetting of the ﬁber surface
by the matrix is a necessary condition for ﬁber-matrix adhesion and is determined by the
free energies of the components. Both chemical and physical bonds exist at the interface
and the number and type of each strongly inﬂuences the interaction between ﬁber and
matrix. The structure of the matrix near the ﬁber surface can be inﬂuenced. Changes in
reactivity due to adsorption of matrix components can alter the local morphology.
Untreated matrix components and impurities can diffuse into the interphase region altering
the local structure.

Each of these phenomena can vary in magnitude and can occur simultaneously in
the interphase region. Depending on the system the interphase itself can extend from a
few to a few hundred nanometers in depth. The structure of this region can have
profound effects on the performance of the composite in terms of its mechanical strength,

chemical and thermal durability. Likewise the exact nature of this region must be

understood if accurate life prediction models are to be developed. A need for a
multidisciplinary approach must be recognized because of the complexity of this region.

The complexity of the interphase region requires that a variety of characterization
methods be applied to fully understand the chemical and physical nature of this region.
The methods and the information which they provide are: (1) Surface Spectroscopies, i.e.
atomic and molecular information about the reinforcement surface; (2) Thermal
Desorption, the identiﬁcation and quantiﬁcation of the species able to volatilize at the
interphase during fabrication; (3) Surface Energies, the thermodynamic quantities
necessary to predict the wettability of the reinforcement surface and the interaction with
the matrix; (4) Mechanical response of the interphase to environment or the interfacial
shear strength; (5) Microtomy and Microscopy, i.e. the determination of the locus of
failure in the interphase region.

Ceramic materials are known for their high temperature resistance with the
drawback of low toughness and strength. Ceramic matrix composites are developed to
give high strength, high toughness and better damage tolerance, and a lower variability of
strength over monolithic ceramics. The ﬁber-matrix interface in ﬁber-reinforced ceramic
matrix composites controls the mechanical behavior of these materials. An extremely
strong bond does not allow for crack deﬂection or debonding at the ﬁber-matrix interface
therefore a crack propagating in the matrix simply passes through the ﬁbers undisturbed
resulting in brittle fracture. Conversely, an extremely weak interface leads to a low matrix
fracture stress and low ultimate strength, for as the composite is stressed, load is not

transferred efﬁciently from the matrix to the ﬁbers, thus the properties of the

reinforcement are not utilized. Therefore, interfacial forces must be controlled to produce
a composite material with good matrix failure stress and ultimate strength that also
exhibits gradual composite failure through effective ﬁber pull-out. Coating the ﬁber with a
suitable reinforcement can be used to control the interfacial forces.

Metal matrix composites are produced by two ways: molten metal processing and
powder metallurgy. The former route is the most severe from the point of view of
interfacial reactions because the liquid metal is very reactive to the ﬁber surface. Main
elements that contribute to the role of interphase in metallic matrix composites are: (1)
thermal residual stresses due to the mismatch in the thermal expansion coefﬁcients of ﬁber
and matrix, (2) chemical reaction between the ﬁber and matrix materials, (3) ﬁber surface
characteristics that may include large asperities and/or a certain roughness amplitude, and
(4) friction between the ﬁber and the matrix that may result in improved toughness due to
energy dissipated in ﬁber bridging and pullout. Extensive interface reaction can result in
degradation of the reinforcement, which is particularly a problem in continuous ﬁber
reinforced metal matrix composites, and it can also result in a modiﬁcation of the matrix
metallurgy and properties. A protective coating to provide thermo-oxidative stability to
the reinforcing ﬁber can help prevent degradation of the reinforcement by controlling

kinetics and diffusion.

10

2.2 THEORIES OF ADI-IESION
Bonding at an interface is due to adhesion between ﬁber and matrix. Adhesion can
be attributed to ﬁve main mechanisms which can occur at the interface either in isolation

or in combination to produce the bond. These mechanisms are brieﬂy discussed below.

2.2.1 Adsorption and Wetting

When two electrically neutral surfaces are brought sufficiently close together there
is a physical attraction which is best understood by considering the wetting of solid
surfaces by liquids. In the case of two solids being brought together the surface roughness
on a micro and atomic scale prevents the surfaces coming into contact except at isolated
points. In addition the surfaces are usually contaminated. Even if the contamination is
removed, and strong adhesion occurs at the contact points, the adhesion averaged over the
whole surface will be weak. For effective wetting of a ﬁber surface the liquid resin must
cover every hill and valley of the surface to displace all the air. Weak boundary layers
must be avoided also.

Wetting can be understood in terms of two simple equations. The Dupre equation
for the thermodynamic work of adhesion, WA of a liquid to a solid states that
WA: 71 + 72 + 712 (2.1)
where 'y; and 72 are the surface free energies of the liquid and solid respectively and 712
is the free energy of the liquid-solid interface. This equation can be related to the physical

situation of a liquid drop on a solid surface by using the Young equation. When the forces

11

are resolved in the horizontal direction at a point where the liquid touches the solid as well
as the vapor, then the Young's equation states
‘st = 'YSL + ’YLv 0089 (2.2)
where 'st. 75L and ﬂy are the surface free energies, or surface tensions of the solid-
vapor, solid-liquid and liquid-vapor interfaces respectively, and 9 is the contact angle. For
spontaneous wetting to occur 9 = 0°. Zisman introduced the concept of critical surface
tension of wetting. 7c such that only liquids with 'yLv < Yr: will spontaneously spread on the
solid. This is a useful parameter in considering the wetting of ﬁbers by resins.

A value for WA can be obtained by combining equations (2.1) and (2.2) and
putting y, = m , 72 = ﬂy and Y”; = 73L, namely,
WA: 'st + 'YLv— 'YSL (2.3)
WA represents a physical bond resulting from highly localized intermolecular dispersion
forces which, in the ideal situation, can give very strong adhesion between resin and
carbon or glass ﬁbers. However, this strong physical bond is usually not achieved (1)
because the ﬁber surface is contaminated so that the effective surface energy is much
smaller than that of the base solid, (2) because of the presence of entrapped air and other
gases at the solid surface, and (3) because of the occurrence of large shrinkage stresses
during the curing process which leads to displacements at the surface which can not be
healed.
2.2.2 Interdiffusion

It is possible to form a bond between two polymer surfaces by the interdiffusion of

the polymer molecules on one surface into the molecular network of the other surface.

12

The bond strength will depend on the amount of molecular entanglement and the number
of molecules involved. Interdiffusion may be promoted by the presence of solvents and
plasticising agents and the amount of diffusion will depend on the molecular conformation
and constituents involved and the ease of molecular motion. Interdiffusion of the
precerarnic polymer coating with the polymer matrix may account in part for the bonding
achieved when ﬁbers are pre-coated before incorporating into the matrix.
2.2.3 Electrostatic Attraction

Forces of attraction occur between two surfaces when one surface carries a net
positive charge and the other surface a net negative charge. The strength of the interface
will depend on the charge density and the potential difference between the two surfaces.
Although electrostatic attraction is unlikely to make a major contribution to the ﬁnal bond
strength of ﬁber-matrix composites, it could well have an important role in the way that
coupling agents are laid down on the surface of glass ﬁbers.
2.2.4 Chemical Bonding

This is of particular interest for ﬁber composite materials because it offers the main
explanation for the strength of the bond between ﬁber and matrix. A chemical bond is
formed between a chemical group on the ﬁber surface and a compatible chemical group in
the matrix. The strength of the bond depends on the number and type of bonds and
interface failure must involve bond breakage. The processes of bond formation and

breakage are in some form of thermally activated dynamic equilibrium.

13

2.2.5 Mechanical Adhesion

Some bonding may occur purely by the mechanical interlocking of two surfaces. A
resin which completely wets the ﬁber surface follows every detail of that surface. The
strength of this interface in tension is unlikely to be high unless there are a large number of
re-entrant angles on the ﬁber surface. The strength in shear may be very signiﬁcant and
depends on the degree of roughness. A quite separate factor which also relates to the
roughness of the ﬁber surface is the potential for an increased bond strength, for example,
chemical bonding because of the larger surface area which is available.

In addition to the simple geometrical aspects of mechanical adhesion, there are
many internal stresses in a composite material which develop during processing operations
and mechanical testing which affect the apparent strength of the ﬁber-matrix bond. Thus,
for example, resin shrinkage during curing of thermosetting polymers and differential
thermal expansion of the matrix and ﬁbers can produce tensile, compressive and shear

stresses at the interface depending on the geometry of the ﬁbers and the component.

2.3 ADHESION TESTS

The level of adhesion between ﬁber and matrix affects the mechanical properties of
a composite. Several methods have been proposed to relate structure and property to
composite ﬁber-matrix interfacial behavior [9,11,12]. The ﬁbers are considered to be the
load bearing constituent and the mechanism of load transfer at the ﬁber-matrix interphase
plays a major role in the mechanical and physical properties of the composite. In this

study, the Single Fiber Fragmentation and the Microindentation techniques are used for

14

the adhesion characterization. A description of these ﬁber-matrix adhesion tests is given
below.
2.3.1 Single Fiber Fragmentation Test

This technique was originally used by Kelly and Tyson [13], who observed a
multiple ﬁber fracturing phenomenon on application of a tensile force in a system
consisting of a low concentration of brittle ﬁbers embedded in a copper matrix. Drzal et
al. used graphite ﬁber/epoxy system to study the interfacial shear stress transfer
mechanism upon application of a tensile load to the single ﬁber coupon [7,14,15,16].

In this method a ﬁber is embedded in a matrix material such that the strain to
failure of the matrix is at least three times higher than that of the ﬁber. An axial stress
induced in the ﬁber by interfacial shear stress increases from the ﬁber ends until the ﬁber
fracture strength of is reached as shown in the Figure 2.2. At that point, the ﬁber
fractures at some site where the ﬁber stress is at a maximum, depending on ﬁber defect
distribution and probability. Continued application of stress to the specimen will result in
repetition of this fragmentation process until all remaining ﬁber lengths become so short
that the shear stress transfer along their lengths can no longer build up enough tensile
stresses to cause any further failures with increasing strain. This maximum ﬁnal
fragmentation length of the ﬁber is referred to as the critical length lc . The equilibrium of
forces gives
(of+dof)1tr2=of1rr2+21rrrdx

d0} =(2’t/r)dx

I x
I d6f=J (21/r)dx
0 0

15

 

 

 

 

 

 

 

 

l———1

 

 

V

 

 

 

 

-)l l, k—
G3>Cz>61
Ti
<-——>
o 4—( o+do

 

k——3§-—4

Figure 2.2 Schematic representation of the single ﬁber fragmentation process.

16

x
of, - om = (2/r) JO 1 dx

Where r is the radius of the ﬁber and cm is the stress at the ﬁber end which is equal to
zero as the ﬁber end can not take stress and of, is equal to 6f. If the matrix surrounding
the ﬁber is a perfectly plastic matrix, the shear stress 1: at the interface can be considered
to be constant along the short ﬁber critical length. Also, maximum ﬁber stress occurs at x
= I, / 2. Therefore the above equation can be written as

t: Ofd/Zlc (2-4)
where d is the diameter of a circular ﬁber. Because the ﬁber-matrix interface is placed
under shear, the calculated value of I should be an excellent predictor of the interfacial
shear strength. Because the tensile strength is characteristic of a particular ﬁber type, the
ratio of the critical length to ﬁber diameter is the important parametric combination. I, I d
is often called the critical aspect ratio.

In practice, there is a distribution of critical lengths. Any fragment with a length
slightly exceeding 1, will break in two, yielding at the conclusion of the experiment, a
random distribution of fragment lengths between lc / 2 and I, . A two parameter Weibull
distribution is used to ﬁt the data as shown in the following equation.
t=(0'f/2[3)l‘(1—1/0t) (2.5)
where on and B are the scale and shape parameters from the Weibull distribution, F is the
gamma function.

A signiﬁcant advantage of this technique is the observation of the failure process in

transmitted polarized light i.e. frictional debonding, interfacial crack growth parallel to the

17

ﬁber-matrix interface and matrix crack growth perpendicular to the ﬁber axis and the locus
of failure can be identiﬁed.

If the reinforcing ﬁber is replaced with a ﬁber having a ceramic coating, the same
stresses are developed except that they must be transmitted through the ceramic into the
ﬁber substrate. Since ceramics are usually stiff, brittle materials, fracture will take place
within the coupon. For the precerarnic inorganic polymer coatings studied here, both the
failure mode and the interfacial shear strength can be quantiﬁed. By applying a tensile
load to the adhesive, a shear stress is created which causes the oxide coating to fracture in
regular lengths. The measurements of these lengths I combined with the physical
properties of the coating (thickness t) allows the shear strength between the coating and
the ﬁber to be determined through the use of the following equation
t=(2+4t/d)tm—O'fd/21 (2.6)
where subscript m refers to matrix.

The fragmentation test can be used to evaluate the changes in adhesion during
processing as well as after fabrication. A limitation is that the fracture properties of the
encapsulating polymer determine whether or not the test can be carried to completion and
hence critical length can be achieved. If the test can be conducted, the critical length to
coating thickness ratio and the failure mode can be determined by in situ microscopy.

The high magniﬁcation required to see the events at the interfacial region can not
be accomplished by the conventional optical microscopy and the Scanning Electron
Microscope (SEM) can not penetrate into the single ﬁber specimen. Ultrarnicrotorning of

the specimen offers the possibility of examining the interphase directly under the high

18

magniﬁcation capabilities of the Transmission Electron Microscope (TEM). Features such

as ﬁber-coating failure, coating-matrix failure and coating-ﬁber adhesion can be identiﬁed

by this procedure.

2.3.2 Microindentation Technique

Another approach for the characterization of the shear strength of the interphase

was ﬁrst proposed by Mandel] and coworkers [17,18]. In this technique, single ﬁbers
perpendicular to a cut and polished surface of a regular high volume fraction composite are
compressively loaded to produce debonding and/or ﬁber slippage as shown in Figure 2.3.
In contrast to methods which use a simpliﬁed single ﬁber composite model system to
provide information on ﬁber-matrix adhesion, the microindentation method is an in .situ
interface test conducted on actual sections of real composites and therefore has the
advantage of reﬂecting actual processing conditions. It can allow determination of the
interface strength due to fatigue or environmental exposure or possibly monitor interface
properties of parts in service. If a carbon ﬁber with its ceramic coating is encapsulated in a
polymeric matrix and subjected to the same microindentation procedures as a reinforcing
ﬁber alone, the response of the system should be solely dependent on the interphase and its
mechanical properties.

The microindentation method is conducted on a specially constructed apparatus
called an interfacial testing system (ITS) as shown in the Figure 2.3. The measurement of
the force-displacement curve which is measured during the indentation tests can identify
the failure load and hence the adhesion or interfacial shear strength. Closer inspection of

the force-displacement curve can also be used to quantify the properties of the interphase

19

83% maﬁa. 3835 cc eueaeﬁm 3 2:5

51550 «one - > X

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

§=mﬁoo 1
_ _ e938“ €8u§>
- ._
. 6:350 mg .N
cases 33 5%“me
«use: 295m
5.5065
033.80. —.0
O O O
«58.5 32>

 

ease: 83>

 

 

 

 

 

 

20

as well as to delineate the failure mode. Although one data point is determined for each
ﬁber tested, sample manipulation and automation procedures make data collection very
efﬁcient. This method is very good for making relative measurements within the same
system and would also be ideal in a quality control environment.

Interfacial shear strength is calculated through the use of a ﬁnite element analysis
of the stressed area for each ﬁber under examination. The microindentation test is done
on. individual selected ﬁber which is surrounded by neighboring ﬁbers located at various
distances and distributed in a variety of arrangements, which range from a hexagonal array
to random and dispersed conﬁgurations. The diameter of the tested ﬁber and the distance
to the nearest neighbor ﬁber are recorded for each test and a simpliﬁed axisymmetric ﬁnite
element model (FEM) is used. This model includes the ﬁber, surrounding matrix, and
average composite properties beyond the matrix. Maximum interfacial shear stress occurs
at a distance that is a fraction of a ﬁber diameter below the free surface for most cases. It
is assumed that debonding initiates at this point due to shear stress. The maximum shear
stress along the interface is insensitive to probe stiffness as long as the contact area does
not approach the interface. The maximum shear stress is signiﬁcantly affected by the
relative elastic constants of the ﬁber and matrix as well as the thickness of the matrix layer.
In practice, the elastic constants are known to within a good approximation under most
conditions. In the ﬁnite element model, the thickness of the matrix layer is an
approximation and is intended to represent the local ﬁber arrangement and spacing. The
actual matrix layer is always nonsymmetrical, with signiﬁcantly different distances to

neighboring ﬁbers in different directions. The effect of actual local ﬁber arrangement can

21

not be included in a simple data reduction scheme. The most important aspect of ﬁber
arrangement is the distance to the nearest neighbor ﬁber.

Calculations of interfacial shear strength assume either a maximum interfacial shear
stress criterion or a maximum radial tensile stress criterion, ignoring other stress
components and residual stresses in both cases. The interfacial shear strength (1:, ) is
calculated from:

Ti = 0m (Tmax / 8r )FEM (2.7)

where om is the average compressive stress applied to the ﬁber end at debonding and (IM/
6;)ng is the ratio of the maximum shear stress to applied stress. Calculations of the

interfacial tensile strength or, for tensile radial stress at the surface is given by:

0:1 = 0m (Gr/max/ Er )FEM (2.8)

where (or/ml 5f )FEM is the ratio of the maximum radial tensile stress at the surface to the

applied ﬁber pressure resulting from the ﬁnite element analysis. In all cases, the FEM

results are calculated for the ratio of matrix layer thickness to ﬁber diameter of 0.4.

2.4 CARBON FIBERS

Continuous ﬁlament carbon ﬁbers are produced by pyrolyzing, or decomposing by
heating, ﬁbers that contain enough carbon so that the resultant carbon ﬁber is both
physically and economically attractive. Carbon ﬁbers are derived from three major
feedstock or precursor sources: rayon, polyacrylonitrile, and petroleum pitch. Rayon
precursors are derived from cellulosic materials and were one of the earliest precursors

used to make carbon ﬁbers. Their advantages were that they were well characterized and

22

readily available; their most signiﬁcant disadvantage was a relatively high weight loss, or
low conversion yield to carbon ﬁber. Polyacrylonitrile (PAN) precursors are the basis for
the majority of carbon ﬁbers commercially available today.

The carbon ﬁbers used in this study are pitch based. Pitch precursors are based on
petroleum asphalt, coal tar, and polyvinyl chloride. Pitches are relatively low in cost and
high in carbon yield. Their most signiﬁcant drawback is nonuniformity from batch to
batch. Sizings (coatings) and ﬁnishes are often applied to ﬁber bundles (tows) to improve
their handling characteristics. They must be formulated to adhere to but not to interfere
with the performance of the composite. A sizing may incidentally react chemically with
the surface. The ﬁbers are also surface treated to chemically modify the nascent ﬁber
surface that alters the surface chemical composition, such as oxidation, acid-base
reactions, and etching, without a deliberate coating of the surface. A surface treatment
may incidentally result in the formation of a coating.

High strength, high modulus carbon ﬁbers are about 7 to 12 pm in diameter and
consist of small crystallites of “turbostratic” graphite, one of the allotropic forms of
carbon. In a graphite single crystal the carbon atoms are arranged in hexagonal arrays
which are stacked on top of each other in a regular ABAB . . . sequence. The atoms in the
layer or basal planes are held together by very strong covalent bonds and there are weak
van der Waal forces between the layers. This means that the basic crystal units are highly
anisotropic. To obtain high modulus and strength the layer planes of the graphite have to

be aligned parallel to the axis of the ﬁber. The pitch based carbon ﬁbers have tensile

23

modulus as high as 100x 106 psi (724 GPa), tensile strength of 325,000 psi (2.24 GPa) and
have a density of 2.0 g/cc.
Carbon ﬁbers are most widely used in polymer-, ceramic-, and metal-matrix

composite materials.

2.5 POLYMER-DERIVED CERAMICS

The coating material used in this study is a precerarnic polymer. The most widely
recognized ceramic products produced from polymers are ﬁbers. However, other three
uses of polymer-derived ceramics include coatings for oxidation protection, matrices for
ceramic ﬁber-reinforced composites and nonfugitive binders for ceramics prepared ﬁ'om
powders.

Conversion of the polymer to a stable ceramic phase (that is, pyrolysis), is done by
heating in a nonreactive atmosphere such as argon, helium, or nitrogen to about 1000°C or
higher. Pyrolysis is accompanied by substantial shrinkage caused by weight loss and
densiﬁcation. Densities of precerarnic polymers are typically 1.1 to 1.2 g/cc, while
polymer chars show densities of 2.3 to 2.6 g/cc. Shrinkage in ﬁber processing is of only
minor consequence, but polymer shrinkage in the preparation of coatings can be a major
concern.

The major processing steps necessary to prepare the four types of polymer-derived
ceramics are shown in Figure 2.4.

Polymer-derived coatings have been reported to be effective in protecting carbon

and metal substrates against oxidation and wear. Another advantage of polymer—derived

24

.888 8>top¢oaboq me 899 .88 05 manna 8 8380: 38m wﬁmmoooa Home:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 88E

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0888 £888
anagram o
A in- ..
‘1‘ _ .
~35... 8a :5 n 8.6qu
v 8 cam—BE
huh—8.».— omﬁubu _ 2:8 2’83
11 -. l -L
930 aux—o5 ||1.l v.50 _ Baa unbear—
moﬁxo 63532 _
,rT - :2. n -
8:58
232 m o
25 _ 2E a 2 2 Eu 25
838 a Shaman
e858 3:2 283 Eu use cam
523 eéeaeez 58 33858 Ea

 

 

 

. huge.— 35808...”

 

 

25

coatings is their ease of application. The coatings can be applied by spraying, brushing or

dipping.

2.6 INORGANIC PRECERAMIC POLYMER COATING

The ﬁrst step in developing a detailed understanding of the fundamental factors
that control interface properties and the mechanics of failure is to establish a benchmark
interface and composite system, and clearly deﬁne its attendant properties. In this study,
we have chosen to work with interfaces prepared using silsesquioxane terpolymer
-[MeHSiO]o,1[Me(nPrO)SiO]o_2[MeSi(O)1_5]o_7- because it is soluble in hexane, which is
volatile and permits very facile coating of the ﬁbers; high temperature stability in air; and
because there is already considerable literature on the utility of silsesquioxanes as
precursors to silicon oxycarbide, SiO,,Cy [19-27]. The chemical structure of this
silsesquioxane terpolymer is shown in Figure 2.5. Moreover, -[MeSi(O)1,5],,-,
polymethylsilsesquioxane has been used as a precerarnic polymer for fabrication of silicon
carbide powders[21], “black glass” (70% SiOz /20% SiC/ 10% C) composite matrices for
carbon ﬁbers [20] and for the fabrication of black glass ﬁbers [28].
2.6.1 Synthesis

The above silsesquioxane terpolymer is synthesized via reactions (2.9) and (2.10):
[29,19]. The copolymer is produced by catalytic redistribution of -[MeHSiO],,- oligomer

using dirnethyltitanocene, szTiMez as the catalyst precursor.

 

-[MeHSiO],,- CpgTiMezl 20°C _ MeSiH3 + -[MeHSiO]o,3[MeSi(O)l,5]o,7- (2.9)

26

 

 

“<3 Me
_. _ .. _‘ .. /Si"0\3i/ _
Me O 0/ \
\S//o / Me 0
MC Me I l 0 SI / M
I l o / \O Si/ 6
S' o s \ /Si\o77K
-— 1 — 1—0 s
/l \O O\ Sl—O-d—
| | Me |
H 0 Me M
e
L T or — \ J 02 _ " 07

 

 

 

 

 

 

n-Pr

Figure 2.5 Chemical structure of silsesquioxane terpolymer

27

Following catalytic redistribution, the resulting copolymer, -[MeHSiO]o,3[MeSi(O)1510,7- is
reacted with n-propyl alcohol as follows to give the required product.

-[MeHSiO]o,3[MeSi(O)1,5]o,7- + n PrOH szTiMezl 20°Cl-H2 ‘

 

-[MeHSiO]o.1lMe(nPrO)SiO]o_2[MeSi(O)1,5]o,7- (2.10)

The exact ratio of monomer units was determined by proton NMR as described elsewhere
[19].
2.6.2 Pyrolytic Behavior of Bulk Silsesquioxane

The pyrolytic behavior of bulk silsesquioxane provide the basis for establishing the
reaction processes that occur as the n-propoxy derivative,
-[MeHSiO]o,1[Me(nPrO)SiO]o,2[MeSi(O)[,5]o,7-, transforms to SiOxCy on heating.

Diffuse Reﬂectance Infrared Fourier Transform (DRIFTS) and solid state Magic
Angle Spinning Nuclear Magnetic Resonance (MAS NMR) show that on heating to
900°C-1000°C in increments of 200 °C (one hour holds), -[MeHSiO]o,3[MeSi(O)1.5]o_7-
undergoes well deﬁned changes in structure [1]. The copolymer is stable to 200 °C;
however on heating to 400 °C it loses the starting monomer units and transforms to a
polymer consisting solely of —[MeSi(O)1_5],,- or ”T” monomer units. This clear material is
relatively stable on heating for one hour at 600 °C. The MAS NMR indicates a slight
redistribution of the T groups to form a material with a composition of
-[(CH2)2SiO]o,1[Si02]o,1[MeSi(O)1_5]o_g-. At 800 °C, redistribution is more pronounced
giving a product with a composition of -[(CH2)2SiO]o,2[Si02]o,3[MeSi(O)15105-. In this
instance, some of the Si-C bonds are cleaved and free carbon is formed. The process is

much more extensive at 1000 °C. Heating silsesquioxanes to temperatures above 1300 °C

28

results in the formation of SiC; and SiC [21,22]. A Thermogravirnetric Analysis (TGA) of
this terpolymer shows a ceramic yield of 86% at 1000 °C. The high ceramic yield is
important because it indicates that during composite processing, the interface material will
contribute minimally to offgassing. Two main features of TGA are 6-7% mass loss that
occurs between 400 °C to 600 °C and the 11-12% mass loss at 600 °C to 800 °C. The ﬁrst
mass loss likely corresponds to elimination of -[MeHSiO]- units. The second mass loss
likely to correspond to hydrocarbon materials, perhaps through fragmentation of the
propoxy groups. The terpolymer does not undergo any mineralization processes up to
600 °C. The 800 °C and 1000 °C DRIFT spectra contrast greatly with the spectra seen at
lower temperatures. The sharp IR absorption seen for molecular vibrations give way to
the broad, featureless bands common for amorphous SiO,,Cy materials [25,27,30]. The
mineralization occurs above 600 °C. The polymer undergoes volumetric shrinkage and

becomes more dense and probably thinner as it transforms into ceramic.

2.6.3 Oxidation Resistance

The carbon ﬁbers are not stable above about 600 °C in air. One of the purposes of
the coating on carbon ﬁbers is to prevent the ﬁber from rapid oxidization. The resistance
to air oxidation should also be a measure of resistance to chemical damage during
composite processing, e.g. during inﬁltration of molten metals into woven preforms. The
oxidation studies carried out on P-55 carbon ﬁber show that 0.1 pm thick coating increase

ﬁber oxidation resistance by 75 °C to 100 °C relative to the uncoated ﬁber [1]. This

29

increased oxidation resistance may be sufﬁcient to warrant efforts to make graphite ﬁber
aluminum matrix composite materials.
2.6.4 Coating Method and Pyrolysis

The procedures for coating the carbon ﬁbers begin with removing the carbon ﬁber,
sizing by heating tows of ﬁbers in a sealed quartz tube to 900 °C in the presence of
nitrogen for one hour in a single zone Lindberg tube furnace equipped with a Eurotherm
programmable temperature controller. Tows of the ﬁbers 60 to 80 cm in length are
wrapped lengthwise on a 20x5x0.6 cm graphite rack and placed inside a ported quartz
tube. The quartz tube is then sealed with a cap, inserted into the furnace, connected to a
supply of dry, compressed nitrogen and then heated at 20 °C/minute to 900 °C followed by
a 1 hour hold. The samples are then allowed to furnace cool. The resulting desized tows
are then coated with the silsesquioxane as follows.

The rack containing the desized tows is dipped into a 2.5 wt.% solution of the
silsesquioxane in hexane, allowed to stand for 1-2 minute, removed and allowed to dry for
a period of 5 minutes in air. The dried ﬁbers are again placed in the ported quartz tube
and ﬂushed and dried in a ﬂow of nitrogen for a minimum of 10 minutes to ensure removal
of any remaining solvent. The system is then heated (20 °C/minute/N2) to selected
temperatures (200 °C, 400 °C, 600 °C, 800 °C, 1000 °C) and held at temperature for 1
hour followed by furnace cooling. The pyrolyzed, coated tows are unloaded at room

temperature and then stored in air prior to an analysis.

30

2.7 COATING CHARACTERIZATION METHODOLOGY

The characterization of the ﬁber coatings during thermal processing can be
followed through single ﬁber measurements. Coating and/or interfacial coating/ﬁber
failure due to mismatches in coefﬁcient of thermal expansion, residual processing stresses,
etc. can be identiﬁed at the nano— or microstructural level. Both adhesion strengths and
failure modes can be quantiﬁed using embedded single ﬁber methods developed for
polymers as well as using microindentation methods (ITS) to test the in-situ ﬁber-matrix
interface on real ﬁber-coating-matrix systems of interest if the effect of the polymer matrix
is taken into account. A complicating factor is the expected large change of ﬁber coating
mechanical properties associated with the conversion from inorganic polymer coating to
ceramic. By the selective use of single ﬁber specimens in transparent organic and
inorganic matrices, stress states can be induced that duplicate those encountered in real
composites during and after high temperature processing. A 3 dimensional model of a
single ﬁber in the matrix can be used to quantify the stress states and optimize the
processing steps.

The methodology can be coupled with the application of ﬁber coatings of known
composition and thickness as an evaluation and optimization tool for their development.
It is expected that determination of this information will lead to a protocol for

optimization of the processing, as well as the properties of the ﬁber coatings.

 

 

CHAPTER3

EXPERIMENTAL

3.1 MATERIALS

The carbon ﬁbers used in this study are AMOCO’s ThornelrM P-55. Some
properties of the ﬁber are given in Table 3.1(a). The ﬁbers were coated with the
precerarnic silsesquioxane polymer and were pyrolyzed. Tows of the Thornel P-55
graphite ﬁbers, 60—80 cm in length, were wrapped lengthwise on a 20 cm x 5 cm x 0.6 cm
graphite rack and placed inside a ported quartz tube. (1) The quartz tube was then sealed
with a cap, inserted into the furnace, connected to a supply of dry, compressed nitrogen
and then heated at 20 °C/minute to 900 °C followed by a 1 hour hold. The samples were
then allowed to furnace cool. (2) The resulting desized tows were dipped into a 2.5 wt.%
solution of the silsesquioxane in hexane, allowed to stand for 1-2 minutes, removed and
allowed to dry for a period of 5 minutes in air. (3) The dried ﬁbers were again placed in
the ported quartz tube and ﬂushed (and dried) in a ﬂow of nitrogen for a minimum of 10
minutes to ensure removal of any remaining solvents. (4) The system was then heated (20
°C/minute in nitrogen) to selected temperatures (200 °C, 400 °C, 600 °C, 800 °C and 1000
°C) and held at temperature for 1 hour followed by furnace cooling. The pyrolyzed tows
were carefully handled at room temperature to prevent damage to the coating by

inadvertent ﬂexing of the ﬁbers.

31

32

(a) Thornel P-55 carbon ﬁber

 

Diameter 10 um to 13 pm
Tensile Strength 1.9 GPa
Tensile Modulus 380 GPa
Density 2.0 g/ml

 

 

 

 

 

 

 

(b) Epon 828 Epoxy

Tensile Modulus 3.6 GPa
Tensile Strength 89.6 MPa
Shear Modulus 1.24 GPa

 

 

 

 

 

 

 

Table 3.1 Properties of (a) Thornel P-55 carbon ﬁbers and (b) EPON 828
epoxy matrix cured with 14.5 wt.% of mPDA.

33

The polymer matrix used in this study is the EPON® 828 epoxy resin (Shell
Chemical Company, Houston, Texas) cured with meta-phenylene diamine (mPDA)
(Aldrich Chemical Company). The 828 epoxy resin is an undiluted diglycidyl ether of
Bisphenol A. mPDA is in ﬂake form. The chemical structures of the epoxy and the curing
agent are shown in Figure 3.1. The epoxy was mixed with 14.5 wt.% mPDA, degassed
under vacuum for 2-5 minutes and air-cured for two hours at 75 °C and two hours at 125

°C in an oven. The resulting matrix has the properties given in Table 3.1(b).

3.2 ADHESION CHARACTERIZATION

The adhesion characterization was done by (i) The Single Fiber Fragmentation
Test and (ii) The Microindentation or the ITS (Interfacial Testing System) test.
Experimental procedures for these tests are given below.
3.2.] Single Fiber Fragmentation Test

A description of this method is given in section 2.3.1. The fabrication of the dog-
bone specimen and alignment of the ﬁber was accomplished by the aid of silicon RTV-664
[31] eight cavity mold of 1 inch long gage length, about 1/8 inch wide and 1/ 16 inch deep
as shown in Figure 3.2 (a). Single ﬁbers of approximately six inches in length were
selected from the ﬁber bundle by hand. The ﬁbers were separated with a ﬂowing air
stream and handled only at ends. Once selected, a ﬁber was mounted in the mold cavity
and held in place with a small amount of rubber cement at the ends of the sprue of the
mold. The rubber cement does not come in contact with the cavity which contains the

gage length section of the ﬁber.

34

0 CH3 0
/ \ | / \
CHZ—CH-CHz-O- O -C— O —O-CH2-CH-—CH2
|
. CH3

(a) Diglycidyl Ether of Bisphenol-A

NH;

(b) meta -Phenylene diamine

NH2

Figure 3.1 Chemical structures of (a) EPON 828 epoxy and (b) curing agent.

35

 

 

 

 

Figure 3.2 (a) Silicon mold and (b) Loading jig used in the Single Fiber
Fragmentation Test.

36

The specimens were fabricated with the Epon 828 epoxy resin cured with 14.5
wt.% meta-Phenylenediamine (mPDA). To prepare the matrix, the appropriate amounts
of the resin and the curing agent were heated separately in an oven for the curing agent to
melt. The resin and the curing agent were then thoroughly mixed and degassed prior to
pouring in the mold cavities.

The prepared resin was poured into the mold cavities in which the ﬁbers were
previously mounted with the help of a pipette to direct the resin ﬂow to all parts of the
cavity and to a level just above the height of the mold surface. The assembly was then
transferred to an oven where a curing cycle of two hours at 75 °C and two hour at 125 °C
was completed.

Once fabricated, the single ﬁber dog-bone coupons were then tested by applying a
tensile load axially which is transmitted to the ﬁber by shear at the ﬁber-matrix interphase
with the help of a loading jig as shown in the Figure 3.2 (b). The loading jig can be ﬁxed
on an optical microscope and the ﬁber fracturing phenomena was observed under an
optical microscope. The ﬁber axial stress rises from the ends until the ﬁber fracture
occurred. Continued application of stress to the coupon resulted in repetition of this
process until all remaining ﬁber lengths were shorter than the “critical length”. The ﬁber
critical length (1,) and hence the interfacial shear strength were determined with the help of
a Weibull distribution computer program. The interfacial shear strength is a good
estimator of the adhesion between ﬁber and matrix. The transparent nature of the matrix
allows to see transmitted polarized light photoelastic stress patterns in‘ the dog-bone

coupon under stress and different failure modes can also be seen, as photo-elasticity is the

37

science which deals with the effects of stress upon light traversing transparent materials.
This can be accomplished by placing the specimen between two sheets of polarizing
material that have been oriented so that their axes of polarization are perpendicular and
viewing the specimen against a source of even diffuse light, such as a light table. A
coupon with no defects will show a relatively smooth graduation of intensity from the
edges to the center of the gage section. This pattern represents stresses induced by the
different coefﬁcients of thermal expansion of the matrix, the silicon mold and the graphite
ﬁber during the cure cycle of the resin. Specimens with defects in or near the ﬁber will
show distinct stress patterns at these points. These stress patterns need the help of an
optical microscope in order to be seen. Even though the coupon is transparent, the
surfaces of the coupon are seldom optically ﬂat. A ﬂuid of the same index of refraction as
the 828 matrix when introduced between the coupon and a cover glass placed on the
coupon greatly reduces image blurring due to the irregular surfaces of the coupon and
insures repeatable magniﬁcation of the coupon. The cover glass can be used on the top
surface as well as on the bottom surface of the coupon. The surface tension of the ﬂuid is
sufﬁcient to hold the lower cover glass in place.
3.2.2 Microindentation or Inerfacial Testing System (ITS) Test

A description of this test is given in the section 2.3.2. The ITS test is conducted
on the stage of an optical microscope. A diamond-tipped probe, which is mounted on the
objective lens holder of the microscope, is used to push single ﬁbers from their
surrounding matrix. Initiation of ﬁber debonding results in a load drop that is sensed by a

load cell attached to the specimen holder. The parameter of interest in this test is

38

interfacial shear strength which is a derived rather than a directly measured value. To
prepare a specimen for testing, many ﬁbers were embedded in the Epon 828 epoxy matrix
and then cut into small pieces which were then embedded in a metallographic specimen
mount with the ﬁbers normal to the specimen surface. This specimen is a ring of about 1
inch in diameter and 3/4 inch in height. Grinding and polishing the surface were done
using conventional metallographic techniques. A good specimen should be free from any
debonding of the ﬁbers. The specimen was placed on the microscope stage and the
operator begins the test by specifying the testing variables from a series of menus on the
computer screen. The operator then selects a good ﬁber to be tested by moving the
motorized stage with linear motion in all three axes (Figure 2.3). The most important
aspect of ﬁber arrangement is the distance to the nearest neighbor ﬁber. The test is begun
by moving the ﬁber to be tested directly beneath the diamond-tipped indenter. A 10 um
diameter diamond indenter was used. The stage then rises at a pre-selected rate until the
indenter contacts the ﬁber. In automatic operation, the stage continues to rise until
debond is attained, as measured by the pre-selected load drop criterion. In semi-automatic
operation, the operator can interact at any stage of the test by stopping the continuation of
load and can look at the degree of debond. A graph of load versus stage displacement is
shown on the computer monitor as the test runs. The operator can change the debond
criterion. The semi-automatic operation and 120° debond criteria were used to determine

the interfacial shear strength in all the tests.

39

3.3 COATING UNIFORMITY and THICKNESS

Scanning Electron Microscope (SEM) and Environmental Scanning Electron
Microscope (ESEM) were used to observe the surface uniformity / topography of the
ﬁbers. The surface analytical technique of Auger Electron Spectroscopy (AES) was used
to determine the thickness of the ﬁber coatings and how they changed with processing.

Topography of the ﬁber surface was characterized by a JEOL ISM—T330 SEM
(Scanning Electron Microscope) and by an ElectroScan ESEM (Evironmental Scanning
Electron Microscope) model 2020.

The ESEM has the advantage of examining specimens in gas/water vapor upto 50
torr at temperatures of -160 °C to +1000 °C and positive ions formed in the process
neutralize the negative surface charge that can build up on an insulating specimen, and
coatings to prevent charging are not required.

The AES technique is based on the similar process as the XPS technique, except
that in AES the surface is irradiated with a beam of electrons. The incident electrons
ionize atoms of the surface creating vacancies in their inner electron shell. When an
electron is ejected from an inner shell of an atom the resultant vacancy is soon ﬁlled by an
electron from one of the outer shells (Figure 3.3). The energy released may appear either
as an X-ray photon, the characteristic X-rays used in electron probe microanalysis and in
X-ray ﬂuorescence spectroscopy, or be transferred to another electron which is ejected
from the atom (Auger electron) with an energy, EA, determined by the three energy levels
concerned, where

EA=E1—E2—E3 (3.1)

40

or Auger electron

 

 

////P'//// E.
hv=E1-E2

Wher photon

—O———O——E2

Incident electron
E > E1

 

 

Figure 3.3 Characteristic X-ray and Auger electron production.

41

E1 and E2 are the binding energies of the atom in the singly ionized state while E3 is that of
the doubly ionized state. The Auger electrons are detected by an electron spectrometer.
Auger transitions are typically denoted by three capital letters such as KLL, LMM, MNN
etc. The letter on the left refers to the electron shell in which the initial vacancy occurred;
the middle letter refers to the shell from which an electron comes to ﬁll the initial vacancy;
and the letter on the right refers to the shell from which Auger electron is emitted. For
lower energy events leading to a photon or electron with energy less than 2 keV the Auger
process is dominant with more than 95% of ionizations leading to the ejection of Auger
electron. A primary electron beam of energy between 5 and 10 keV is employed to
achieve the Auger effect.

The AES technique, similar to the XPS technique, only examines the electrons that
originate within the tens of angstroms of the top surface region. The principle advantage
of the AES over XPS technique is the ability to focus and scan the probing electron beam,
and obtain information on the spatial distribution of surface elements at high
magniﬁcations. AES has the capability to map elements and it is less sensitive to the
chemical environment of the elements than the XPS technique.

The AES technique is well suited to composition-depth proﬁling and hence for the
analysis of thin ﬁlrrrs and surface coatings. The bombardment of a target surface is
accomplished with inert argon ions. A surface atom or molecule becomes sputtered if the
energy transferred to it is greater than its surface binding energy and the imparted

momentum has a component normal to the surface.

42

The coated ﬁbers were mounted on a stage horizontally and clamped from both
ends and then analyzed by AES for the determination of coating thickness by determining
the composition proﬁle as the ion beam sputtered through the coating. Individual ﬁbers
were selected for the sputtering at magniﬁcations between 5000X-7000X and 2 x 2 mm
raster area was used for all the ﬁbers. The sputter rate calibration of the ion gun was
determined (for 8102) to be 106 °A / minute. 10 kV electron beam energy was used. All
the ABS analyses were carried out using Perkin-Elmer Physical Electronics (PHI) model

660 Scanning Auger Multiprobe.

3.4 COATING CHEMISTRY and COMPOSITION CHARACTERIZATION

X-ray Photoelectron Spectroscopy (XPS) surface analysis was used to determine
the compositional change in the 0.1 pm thick coatings as they were being thermally
processed from 200 °C to 1000 °C with increments of 200 °C. XPS is also known as
Electron Spectroscopy for Chemical Analysis (ESCA), though the title XPS is more
scientiﬁcally correct than ESCA.

In XPS, the sample is illuminated with X-rays under ultra-high vacuum conditions
which excite photoelectrons from the surface [32]. Mg Kor X-rays (1253.6 eV) or Al Ka
X-rays (1486.6eV) 'are commonly used. The resultant photoelectrons have a kinetic
energy (Bk) which is related to the X-ray energy (hv) and the binding energy (E) of each
core electron (which is characteristic of the individual atom to which it is bound) by the
Einstein relation [33]:

ER = hv - Eb (3.2)

I 43

The photoemission is observed by measuring the energy spectrum of the emitted
photoelectrons by an electron spectrometer. The photoelectron energy is dependent on
the precise chemical conﬁguration of the surface atoms. The emitted photoelectrons can
escape from tens of angtroms of the solid surface.

The silsesquioxane coated carbon ﬁbers heat treated at 200 °C, 400 °C, 600 °C,
800 °C and 1000 °C temperatures were mounted on a stage horizontally and clamped from
both ends and then analyzed by XPS to see the change in composition of the coating as a
function of heat treatment temperature. Although the mean free path of the
photoelectrons was only 1-2 nm, the composition measures were taken as an indication of
the bulk composition of the coating. XPS analyses were conducted on a Perkin-Elmer
PHI 5400 ESCA system using an Al Koc source. Spectra were collected at a base
pressure of approximately 10‘9 torr and electron take-off angle of 65° using a position
sensitive detector (PSD) on a 180° hemispherical analyzer set at 89.45 eV pass energy for
the survey scans (0-1200 eV) and 35.75 eV for the narrow scans of the elemental regions

used for composition analysis.

3.5 ULTRAMICROTOMY and TRANSMISSION ELECTRON MICROSCOPY
To identify the nature of the interfacial region and the locus of interfacial failure
(i.e. coating-ﬁber or coating-matrix) we need to look at the region of interest in the single
ﬁber dog-bone coupon. The high magniﬁcation required to see the events at the
interfacial region can not be accomplished by the conventional optical microscopy and the

SEM can not penetrate into the single ﬁber specimen. Ultrarnicrotoming of the specimen

44

offers the possibility of examining / analyzing the interphase directly under the high
magniﬁcation capabilities of the Transmission Electron Microscope (TEM) along with the
energy dispersive X-ray analysis.

The TEM samples must be thin enough to transmit sufﬁcient electrons to form an
image. The samples must also be stable under the electron beam and in a high vacuum.
The TEM is designed to provide the information from the interactions of electrons (elastic,
inelastic) as they pass through the specimen. This is most commonly manifested as a two
dimensional projection of those events transduced to a light signal viewed on a ﬂuorescent
screen. Increasing the sample thickness increases the beam absorption and reduces the
image resolution and sample stability. The practical sample thickness is limited to 100 nm
for a 100 kV electron beam.

Malis et al. [34] have reviewed the ultrarnicrotorning techniques for materials
science which cover metallic and ceramic materials. Following is a procedure for ultra-
thin microtomy of ﬁber-epoxy matrix interface of the single ﬁber specimen.

Small pieces (~ 2-4 mm) were cut from the dog-bone specimen and were
embedded in a polymer matrix in order to facilitate grip of the sample in the sample
holder. A rectangular block of 5 x 10 mm is typical. The block is then hand trimmed into
a trapezoid of dimensions no longer than 0.25 mm. A thin layer of the epoxy matrix is
there over the ﬁber. This layer was then further out by using a glass knife to make it
smooth and to avoid unnecessary use of the diamond knife which is required for the ultra-
thin microtorning. A 55° (mounting angle) diamond knife was used. The knife was

mounted at 4° (to the vertical axis) in the knife holder towards the block. The knife was

45

set at a cutting angle of 15-20° to the ﬁber as shown in Figure 3.4 (a) in order to get an
artifact free elliptical shaped ﬁber in the matrix (Figure 3.4 (b)). A solution of de-ionized
water with low concentration of acetone (i.e. ~ 1 drop of acetone per 20 ml of the water)
was used to ﬁll the diamond knife boat with the help of a syringe. Acetone reduces the
contact angle of the water to produce better wetting of the knife edge. Initially, the boat
was ﬁlled to the brink of overﬂow and then excess water was drawn off to form a concave
ﬂuid surface behind the diamond knife. This procedure ensures good wetting of the knife
edge. The knife was advanced manually towards the block face until reﬂection of the
block face off the water meniscus was observed. The knife was then slowly advanced
towards the block at 1 pm steps until cutting begins. Several 1 pm sections were cut to
remove the trimming artifacts and prepare a clean smooth face for the ultra-thin
sectioning. Motorized sample advancing was turned on and a cutting speed of 0.3 mm/sec
was selected. The longer side of the trapezoid was set to be the lower side which ﬁrst
comes in contact with the knife. This procedure helps identify the cutting direction of the
knife. The thickness of the sections was set at 60-100 nm to obtain silver-gold colored
sections.

The sections remain attached and form ribbons (Figure 3.4 (b)) that ﬂoat on the
water. To manipulate the ribbons, an eyelash applicator was prepared. An eyelash
applicator is simply an eyelash mounted on a small round wooden stick using a drop of
white correction ﬂuid or nail polish. The ribbons were assembled with the help of the

eyelash applicator for the pick up by a TEM grid.

46

Matrix with a ﬁber embedded \7/
T l Knife’s edge

 

 

 

/7\

en»: t
k s‘ ‘ .

       

 

(b)

Figure 3.4 Schematic representation of (a) a top view showing the cutting
angle to the ﬁber and (b) a ribbon formed by the sections (in
which the ﬁber is in an elliptical shape) in ultramicrotomy
(diagram is not to scale).

47

To 'collect the sections, the knife is receded and the ribbons are arranged away
from the knife edge with the help of the eyelash applicator. The TEM grids used were
200 or 300 mesh forrnvar coated copper grids stabilized with carbon. A grid is picked up
by a tweezer and under the microscope of the microtome it is placed lightly over the top
of ribbons and is then removed from the boat. The sections remain attached to the grid.
The grid is then drained on a ﬁlter paper and stored for the examination under TEM.

An RMC MT -7 Ultramicrotome was used to carry out the ultrarnicrotorning
process.

The sections were examined under a JEOL-lOOCX TEM at an accelerating voltage
of 100 kV. The energy dispersive X-ray analyzer attached to the TEM was used to

identify the material adhering to the side of the matrix in the sections.

 

CHAPTER 4

RESULTS

 

 

4.1 COATING UNIFORMITY and THICKNESS

Surfaces of the ﬁbers were examined with the Scanning Electron Microscope
(SEM) and Environmental Scanning Electron Microscope (ESEM). Auger sputter
experiments were used to determine the thickness of the ﬁber coatings and how they
changed with processing. Figures 4.1, 4.2, 4.3, 4.4, 4.5 and 4.6 show the SEM
micrograph of uncoated P—55 carbon ﬁber, SEM micrograph of the ﬁber coated and heat
treated at 200 °C, ESEM micrograph of the ﬁber coated and heat treated at 400 °C,
ESEM micrograph of the ﬁber coated and heat treated at 600 °C, SEM micrograph of the
ﬁber coated and heat treated at 800 °C and SEM micrograph of the ﬁber coated and heat
treated at 1000 °C respectively. Figure 4.1 shows the extrusion marks (striations) running
along the length of the ﬁber and “dirt” from the manufacturing process in the uncoated
ThornelTM P—55 carbon ﬁber. The ﬁber diameter range was observed to be from 10 um to
13 um. Figures 4.2 to 4.6 show the same extrusion marks and “dirt” which were the
features of the uncoated ﬁber with the coating present on the ﬁbers. These ﬁgures show
that the coatings are continuous and reasonably thin.

The Auger sputtering results of the coated ﬁbers are illustrated in Figure 4.7.

Peak-to-Peak intensity is plotted against sputter time. The coating contents (i.e. oxygen

48

49

 

i:Uthm’l‘l) .

X2, 999

 

Figure 4.1 Scanning electron micrograph of the surface of uncoated P-55
carbon ﬁbers showing axial striations and some dirt from the

manufacturing process.

Figure 4.2

 

50

, If

W f!
{19th 9392’s
;' /

Scanning electron micrograph of 200 °C heat treated
coated carbon ﬁbers showing the same axial striations and dirt
as was the case with the uncoated ﬁbers. So, the coating is

uniform.

51

 

Figure 4.3 Environmental scanning electron micrograph of 400 °C heat

treated coated carbon ﬁber.

52

 

Figure 4.4 Environmental scanning electron micrograph of 600 °C heat

treated coated carbon ﬁber.

53

 

Figure 4.5 Scanning electron micrograph of 800 °C heat treated coated

carbon ﬁber.

54

 

Figure 4.6 Scanning electron micrograph of 1000 °C heat treated coated

carbon ﬁbers.

[I'm-01% 9w"

"ll 10 Nils."

55

 

 

  

 

I 10

OO4OOIOOOO

   

H N. "1 l1". KM”

O -. N w ‘ 'J' 'i V 1. u.
0 I I D 1

 

 

m C N
CARBON -

 

 

 

 

 

 

10
q o
s O -
r ‘ ,
y - .'
- l
‘ ’ o
I
Q
3
2
l
0
' a
{F D
, 0 1?
1p 1)
. 0 0
0 1»
7 9 1’
4L 1»
‘ 4b 0
4L 4»
5 1+ 0
1r ‘P
4
v 4»
3 .H
2
1
O

  
 
   

 

—
O

—o- —O-—-o—._¢.+“;_. . . .
—+—._....._.__‘ "+H—1p

 

'1'! lo "“380”
o -‘ ‘V H A u- n V o u

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ro ‘ i ‘f
1 0
, . 8 mac
4) 1
' 0 (b
1» 1
.. 7‘ 1 CARBON t
- 4
3. r
t l
if s .
2 4 V
i ‘ 4y
I: OXYGEN 1*
J
2 ‘f 0
r 0 SILICON} W
{t
o A : 4 c . t i c ‘ - ‘ ~ 4 i *
O 5 10 IS 20 25 I) 35 0 '5
WWII "IL-in.

Sputter Rate = 106 'A/min

Figure 4.7 Auger sputtering proﬁles of silsesquioxane coated ﬁbers

pretreated to 200 °C,400 °C, 600 °C, 800 °C and 1000 °C.

 

 

ll

+—+<—-

44—‘0-1'44-4-04-9 {+4

56

and silicon) are seen to decrease with the sputter time until they reach a constant level.
The silicon content of the coated ﬁbers, however, does not decrease much with the sputter
time as compared with the oxygen content. The carbon content increases with the sputter
time. The sputter time for the coating contents to reach a constant level in the coated
ﬁbers treated at 200 °C, 400 °C, 600 °C and 800 °C is about 10 minutes with a calibrated
sputter rate of 10.6 urn/minute. Hence the coating thickness can then be determined by
multiplying the sputter time with the sputter rate. The coating thickness determined from
this analysis is approximately 0.1 urn.

After 800 °C thermal treatment, the sputtering proﬁle changes signiﬁcantly for the
1000 °C case. The length of time necessary to sputter through the coating to the
underlying ﬁber increases signiﬁcantly. This can not be interpreted as due to a thicker
coating. The conversion from the preceramic inorganic polymer has been completed in
the 600 °C-800 °C range and the sputtering rate for the ceramic would be expected to be
much lower than for the polymer. Indeed, with the weight loss of almost 14%, it would
also be expected that the coating would be somewhat thinner and signiﬁcantly more dense

than its initial value.

4.2 COATING CHEMISTRY and COMPOSITION
X-ray Photoelectron Spectroscopy (XPS) was used to quantify the silicon, carbon
and oxygen contents of the surface coatings as a function of processing temperature.

Table 4.1 shows the surface elemental composition of carbon, Oxygen and silicon for the

coated ﬁbers heat treated at 200 °C, 400 °C, 600 °C, 800 °C and 1000 °C. Table 4.2

57

Table 4.1 Surface elemental composition (atomic%) of the coated carbon

ﬁber as a function of processing temperature by X-ray
Photoelectron Spectroscopy.

 

 

 

Element 200 °C 400 °C 600 °C 800 ”C 1000 °C
Carbon 44 42 41 30 27
Oxygen 35 37 40 50 49
Silicon 22 20 19 21 24

 

Table 4.2 Ratios of surface elements by X-ray Photoelectron

Spectroscopy.

 

 

 

Ratio 200 °C 400 °C 600 "C 800 °C 1000 °C
O:C 0.80 0.89 0.97 1.67 1.82
O:Si 1.62 1.83 2.13 2.38 2.04
Si:C 0.49 0.49 0.45 0.70 0.89

 

58

shows the ratios of surface elements of the coated ﬁbers processed at 200 °C, 400 °C, 600
°C, 800 °C and 1000 °C. As seen in Table 4.1 and Table 4.2, the relative amount of silicon
remains constant between 19 at.% and 24 at.% over the entire range of processing
temperatures, whereas the carbon content declines from about 44 at.% to 27 at.% with a
coincident increase in oxygen content from 35 at.% to 49 at.%. The trends are more
clearly seen by comparing the ratios of the elements in Table 4.2. O:C, O:Si and Si:C
ratios are constant through 400 °C and start to Change slightly at 600 °C. Since XPS is a
surface technique and the pyrolytic conversion is rate dependent to some degree, the
change at 600 °C may result from the surface sensitivity of the data. The 1000 °C values
of 24 at.% silicon, 49 at.% oxygen and 27 at.% carbon produce an O:Si ratio of about 2
which can be interpreted to suggest that SiOz and free carbon are formed, rather than
SiOXCy.

Both the 600 °C and the 1000 °C pyrolyzed ﬁbers appear to offer signiﬁcantly
improved oxidation resistance over the uncoated ﬁbers [1]. Tows of both ﬁbers survive 3
hours heating in air to 800 °C. Qualitatively, it was concluded that the 600 °C processed
ﬁbers survive better than the 1000 °C processed ﬁber coating based on the degree of
surface irregularity resulting from air oxidation.

The 600 °C processed material consists of a coating of intact silsesquioxane,
consisting primarily of -[MeSi(O)L5]- units with all the reactive monomer units,
-[MeHSiO]-, eliminated. In addition, the -[MeSi(OnPr)O]- component is likely to have
decomposed to propene and new -[MeSi(O)1,5]- units. The resulting material is quite

uniform and exhibits the well-organized high temperature stability of polymethyl

59

silsesquioxane. In contrast, the 1000 °C material has rnineralized and consists of the
oxycarbide mixed with SiOz and free carbon. The free carbon is present as amorphous or
graphitic 10-100 nm sized particles. If the amounts of free carbon are signiﬁcant (10-15
wt.%), as expected, then one can view the 1000 °C material as containing the equivalent
of open pores, as the free carbon will oxidize rapidly under the reaction conditions. The
only reason one would not expect immediate oxidation of the ﬁber is because oxidation
and gas transport through the resulting 10-100 nm sized pores are likely to be mass
transport limited. Furthermore, some but not all the porosity generated can be expected
to be interconnected, thus limiting the points at which the resulting pores penetrate to the
graphite ﬁber below.

It can be said that it is possible to make uniform, adherent
-[MeHSiO]o,.[MeSi(OnPr)O]o,2[MeSi(O)l,5]o.7- silsesquioxane coatings on graphite ﬁbers
[1]. These coatings evolve chemically, on pyrolysis, to produce 0.1 pm thick coatings
which increase ﬁber oxidation resistance by 75 °C-100 °C relative to uncoated ﬁbers. The
important questions to be answered are what are the properties of the coating and how
does it adhere to the ﬁber surface under processing conditions encountered during metal
matrix composites and ceramic matrix composites fabrication as well as under the

mechanical loads expected during application.

4.3 ADHESION and FAILURE MODES
Adhesion measurements were made using the Single Fiber Fragmentation

Test (SFFT) and the Microindentation test or the ITS (Interfacial Testing System) test.

60

4.3.1 Single Fiber Fragmentation Test

The SFFT results for the uncoated and the coated ﬁber treated at 200 °C, 400 °C,
600 °C, 800 °C and 1000 °C are shown in Table 4.3 and a graphical representation of
these results is given in Figure 4.8. The interfacial shear strength (IFSS) can be computed
from the relationship given in equation (Eq.) (2.5) for the uncoated ﬁber and either Eq.
(2.5) or Eq. (2.6) for the coated ﬁber depending on whether the coating acts as an integral
part of the ﬁber (Eq. (2.5)) or as a separate layer independently of the ﬁber (Eq. (2.6)).
The measurable parameter in either case is the ﬁber fragment length-to-diameter ratio
(lg/d). Eq. 2.5 was used to calculate the interfacial shear strength of the coated ﬁbers.

Table 4.3 shows that in the Single Fiber Fragmentation Test, the uncoated ﬁber
gives a value of the critical length-to-diameter ratio (lc/d) of around 59. A transmitted
polarized light micrograph of a typical P-55 carbon ﬁber under load in this single-ﬁber
coupon is shown in Figure 4.9. The epoxy matrix used to fabricate the coupons is
photoelastically active. The light areas are regions of high interfacial stress. The pattern
displayed is typical for a system with moderate adhesion and an interfacial failure mode.
An interfacial shear strength for this ﬁber is about 22.21 MPa.

Once the ﬁber is coated with the 0.1 pm silsesquioxane coating which has been
heat treated at 200 °C, the critical aspect ratio (lc/d) increases, indicating that the coating
has reduced the stress transfer efficiency between the coating and the matrix. A value of
interfacial shear strength of 9.33 MPa is calculated for this coated ﬁber. A transmitted
polarized micrograph of a typical P-55 carbon ﬁber having the 200 °C treated coating

under load is shown in Figure 4.10. The stress pattern is very weak, reﬂecting the lower

61

Table 4.3 Critical aspect ratio and interfacial shear strength (IFSS)
determined by the single ﬁber fragmentation test.

Uncoated
200
400

600
800
1000

 

 

 

 

 

Interfacial Shear Strength (MPa)

 

 

 

Fiber Treatment

Figure 4.8 Interfacial Shear Strength Determined by the Single Fiber
Fragmentation Test.

62

UNCOATED

 

Figure 4.9 Transmitted polarized light photoelastic stress pattern of a
typical uncoated P—55 carbon ﬁber under load in the single ﬁber

fragmentation test.

63

 

50 1.1m

Figure 4.10 Transmitted polarized light photoelastic stress pattern of a

typical 200 °C treated coated carbon ﬁber under load in the

single ﬁber fragmentation test.

64

stresses generated at the interface. Failure appears to be interfacial between the coating
and the epoxy matrix. The loss in interfacial shear strength is probably due to the low
modulus of the precerarnic inorganic coating. The ﬁnite element analysis discussed later
concludes that adhesion as measured by the interfacial shear strength depends directly on
the modulus of the coating.

Pyrolysis of the ﬁber coatings to 400 °C and 600 °C produces a slight upward
trend in the adhesion as measured by this test. However, the error associated with the
measurement is larger than the average change from 200 °C to 600 °C, making conclusions
difﬁcult. It would be reasonable to expect that the higher thermal treatments produce a
more cross-linked material having a higher modulus and hence the interfacial shear
strength would increase. The photoelastic stress patterns of the 400 °C and the 600 °C
treated ﬁbers are shown in Figure 4.11 and Figure 4.12 respectively. These are not
substantially different from that of the 200 °C sample.

The interfacial behavior changes substantially after the pyrolysis at 800 °C. The
fragmentation length-to-diameter ratio decreases substantially to the level seen for the
uncoated ﬁber ( i.e. 53) and remains low (i.e. 59) after the 1000 °C pyrolysis. This
corresponds to a much higher degree of adhesion, as the values of the interfacial shear
strength of the 800 °C and the 1000 °C treated ﬁbers are 22.60 MPa and 21.28 MPa
respectively which are very close to that of the uncoated ﬁber. The photoelastic stress
patterns for the 800 °C and the 1000 °C treated coated ﬁbers (Figures 4.13 and 4.14
respectively) are also signiﬁcantly different from those of the coated ﬁbers treated up to

600 °C. Initially at low strains when a ﬁber has broken, the photoelastic stress patterns are

65

 

 

Figure 4.11 Transmitted polarized light photoelastic stress pattern of a
400 °C treated coated carbon ﬁber under load in the single

ﬁber fragmentation test.

66

100 um

 

Figure 4.12 Transmitted polarized light photoelastic stress pattern of a
600 °C treated coated carbon ﬁber under load in the single

ﬁber fragmentation test.

67

 

Figure 4.13 Transmitted polarized light photoelastic stress pattern of an
800 °C treated coated carbon ﬁber under load in the single

ﬁber fragmentation test.

68

10000

 

Figure 4.14 Transmitted polarized light photoelastic stress pattern of a
1000 °C treated coated carbon ﬁber under load in the single

ﬁber fragmentation test.

69

intense and uniform. At higher strains, however, regular variations in intensity appear
perpendicular to the ﬁber axis. Observation of the coating at higher magnifcations
indicates the presence of circumferential cracks in the region near the ﬁber ends which
were not detected in the coatings pyrolyzed at temperatures upto 600 °C. The
photoelastic stress patterns and the appearance of cracks in the coating were observed in
both the 800 °C and the 1000 °C treated coatings. The cracks are more readily observed
in reﬂected light micrographs of the ﬁbers as shown in Figure 4.15 (Lines have been
added above the ﬁber to direct the readers’ attention to the location of the coating cracks).
4.3.2 ITS Test

The ITS results of uncoated and the coated ﬁbers are shown in Table 4.4 and a
graphical representation is shown in Figure 4.16. A typical micrograph of a specimen in
the ITS test showing an indented ﬁber is given in Figure 4.17. The interfacial shear
strength of the uncoated ﬁber (i.e. 20.78 MPa) obtained from the ITS test is about the
same as that of the fragmentation test in spite of the fact that the two adhesion
measurement methods differ substantially in the principle of their operation. The
interfacial shear strength values for the coated ﬁbers treated at 200 °C, 400 °C and 600 °C
are decreased to about 6.8 MPa from the value of 20.8 MPa for the uncoated ﬁber. An
increasing trend in the interfacial shear strength is there in the 800 °C and the 1000 °C
treated coated ﬁbers as the values are 7.1 MPa and 10.8 MPa respectively. The trend of
results of the ITS test are similar to those obtained from the fragmentation test. The load
versus displacement plot shown in Figure 4.18 obtained from the ITS test shows a

continuous increase in the load until the debond criterion is met in the case of the uncoated

70

 

Figure 4.15 Reﬂected light micrograph of 1000 °C treated ﬁber with coating

fractures indicated by vertical marks.

Table 4.4 Interfacial shear strength (MPa) by the microindentation with the

71

interfacial testing system (ITS).

 

 

 

 

 

 

 

 

 

Uncoated
Carbon 200 °C 400 °C 600 °C 800 °C 1000 ”C
20.78 :1: 0.57 6.87 :l: 0.66 6.82 i 0.72 6.57 :t 0.63 7.09 i 0.70 10.80 :1: 1.06
25

 

Interfacial Shear Strength (MPa)

 

 

 

 

 

Fiber Treatment

 

Figure 4.16 Interfacial Shear Strength Determined by the
Microindentation with the Interfacial Testing System.

 

72

 

Figure 4.17 A typical micrograph of a specimen in the microindentation

(ITS) test showing an indented ﬁber (in the center).

73

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6.
5 3 //
4: /
Uncoated /
LOAD 3
(gram) 2 1 / /,/Q t:d
1 ’Z //
1 1
0 d er.

0020406081112141618
DISPLACEMENT (micron)

Figure 4.18 Load versus displacement graph of the uncoated and coated

carbon ﬁbers obtained from the ITS test.

74

ﬁbers, whereas, in the case of the coated ﬁbers, a sudden drop in load can be attributed to

the low adhesion level between the ﬁber and the matrix.

4.4 ULTRAMICROTOMY and TRANSMISSION ELECTRON MICROSCOPY
Ultramicrotomy and transmission electron microscopy was done to explore the
nature of the interfacial region and to identify the interfacial failure (i.e. ﬁber-coating
failure or coating-matrix failure) in the single ﬁber fragmentation test’s dog-bone coupons.
Transmission electron micrograph of a typical ultramicrotomed section containing
an elliptical ﬁber in the matrix is shown in Figure 4.19. The knife direction is shown by
the arrow and it is perpendicular to the interface. Note that the ﬁber is not in the form of
a smooth unperforated layer, but it consists of small broken pieces of the graphitic
structure. We want to look at the interface between the ﬁber and the matrix. The results
of the ultramicrotomy of the 200 °C, 400 °C, 600 °C, 800 °C and 1000 °C treated coated
ﬁbers for both unstrained and strained dog-bone coupons when observed through
transmission electron microscope are shown in Figures 4.20, 4.21, 4.22, 4.23 and 4.24
respectively. Arrows are drawn in the ﬁgures to show the direction of the knife. The
direction of the knife is always perpendicular to the interface. In these ﬁgures, the
unstrained samples of all the treated ﬁbers show that the ﬁber is very well attached to the
matrix. In other words, the interface in the unstrained coupons is intact. Whereas, in the
strained samples, the ﬁber is not well attached to the matrix. This represents a disrupted
interface because the sample has been subjected to uniaxial tension which caused an

interfacial failure.

75

 

Figure 4.19 Transmission electron micrograph of a typical ultramicrotomed
section containing an elliptical shaped ﬁber in the matrix. The
direction of the knife is shown by the arrow and it is
perpendicular to the interface.

76

jib-[ii

#4

 

Figure 4.20 Transmission electron micrographs of ultramicrotomed section
of 200 °C treated coated ﬁber obtained from strained (above)
and unstrained (below) dog-bone coupons. The direction of the

knife is shown by the arrows.

77

 

Figure 4.21 Transmission electron micrographs of ulﬁamicrotomed section
of 400 °C treated coated ﬁber obtained from unstrained (above)
and strained (below) dog-bone coupons. The direction of the

knife is shown by the arrows.

78

 

Figure 4.22 Transmission electron micrographs of ultramicrotomed section
of 600 °C treated coated ﬁber obtained from unstrained (above)
and strained (below) dog-bone coupons. The direction of the
knife is shown by the arrows.

79

 

Figure 4.23 Transmission electron micrographs of ultramicrotomed section
of 800 °C treated coated ﬁber obtained from unstrained (above)
and strained (below) dog-bone coupons. The direction of the

knife is shown by the arrows.

80

' U NSTRAIN ED ”M

comm

 

Figure 4.24 Transmission electron micrographs of ultramicrotomed section
of 1000 °C treated coated ﬁber obtained from strained (above)
and unstrained (below) dog-bone coupons. The direction of the

knife is shown by the arrows.

81

In case of the sections obtained from the strained coupons (of 200 °C to 1000 °C
treated coated ﬁbers), it was observed that a little part from the ﬁber side (about 0.1 pm)
is adhering to the matrix at different spots all around the ﬁber-matrix interface. Figure
4.25 and Figure 4.26 show the transmission electron micrographs of ultramicrotomed
ssection in which the part from the ﬁber side is adhering to the matrix in the case of 200
°C and 800 °C treated strained samples respectively. The direction of the knife is shown
by the arrow in the micrographs. Since the thickness of the coating is about 0.1]1m, the
adhering material is the coating of the ﬁber. Energy dispersive X-ray analysis was done to
verify that the material adhering to the matrix side is the coating, i.e., to verify that it
contains the coating element (i.e. silicon) in it. Silicon was detected in each of the 200 °C,
400 °C, 600 °C, 800 °C and 1000 °C treated coated and strained ﬁber’s ultramicrotomed
section. Figure 4.27 shows the typical silicon peak obtained in the X-ray analysis of a
coated ﬁber’s ultramicrotomed section. An attempt was made to closely observe the
ultramicrotomed sections of the strained dog-bone coupons of the 200 °C, 400 °C, 600 °C,
800 °C and 1000 °C treated coated ﬁbers in order to qualitatively determine the amount of
the coating adhering to the matrix. The results of this observation are shown in Figure
4.28. The 200 °C sample shows about 50%, the 400 °C and 600 °C show about 60%,
whereas the 800 °C and the 1000 °C samples show about 75% coating adhering to the
matrix. This can be attributed to qualitatively represent the type of interfacial failure i.e.
whether the failure is between ﬁber and coating or between coating and matrix. The 800
°C and the 1000 °C treated strained samples have a greater percent (i.e. 75%) of ﬁber-

coating failure. Therefore, the ﬁber-coating failure is dominated in the 800 °C and the

82

 

Figure 4.25 Transmission electron micrograph of an ultramicrotomed section
obtained from the strained dog-bone coupon of 200 °C treated
coated ﬁber showing the coating adhering to the matrix. The
direction of the knife is shown by the arrow.

83

 

Figure 4.26 Transmission electron micrograph of an ultramicrotomed section
obtained from the strained dog-bone coupon of 800 °C treated
coated ﬁber showing the coating adhering to the matrix. The
direction of the knife is shown by the arrow.

84

 

 

 

Remaining:

   

 

 

 

 

SILICON

.....

 

 

Figure 4.27 Typical graph showing the presence of silicon (adhering to the
matrix) obtained in the energy dispersive X-ray analysis of
ultramicrotomed section from a coated ﬁber’ 8 strained dog-bone

specimen.

85

 

 

PERCENTAGE

 

 

200 C 400 C 600 C 800 C 1000 C

FIBER TREATIVIENT

Figure 4.28 Percentage of the coating adhering to the matrix in the
ultramicrotomed sections from strained dog-bone specimens
of 200 °C to 1000 °C treated ﬁbers observed through the

transmission electron microscope.

86

1000 °C treated coated ﬁbers. The 200 °C, the 400 °C and the 600 °C treated strained
samples can be said to have been failed by both the coating-matrix failure and the ﬁber-

coating failure because the percent coating adhering to the matrix in these samples is 50%-

60%.

 

  

CHAPTER 5

 
 

f. FINITE ELEMENT MODEL and DISCUSSION

 

When the ﬁber heat treatment temperature increases, the precerarnic coating on the
ﬁber converts from a polymer to a more brittle ceramic. The modulus (and perhaps the
strength) of the coating increases as the heat treatment temperature gets higher. In
addition, as a result of the mass loss due to the elimination of the starting monomer units
at high temperature, the coating becomes thinner with increasing temperature. Although,
in this case, we are not able to measure directly the coating mechanical properties, the
effect of the increase in the temperature of the coating heat treatment on the interphase
shear strength can be evaluated through the effect of coating modulus and thickness on the
coating stress transfer of the fragmented ﬁber. A non-linear ﬁnite element analysis was
used to examine the inﬂuence of the coating parameters on the coating stress ﬁelds of a
single ﬁber fragmentation specimen.

A schematic diagram of a ﬁber fragment from the single ﬁber fragmentation
specimen is shown in Figure 5.1. The fragment is composed of three concentric cylinders
representing the ﬁber, the coating, and the matrix. Because of axisymmetry of the ﬁber
fragment, only a quadrant of the plane containing the cylindrical axis has been analyzed.
The aspect ratio (ﬁber lengthzﬁber diameter) for the fragment is 126.5. The outer matrix

diameter of the model is 10.8 times the ﬁber diameter, large enough to ensure that the

87

E
63.25 of -—*ll*—

]¢—L/2

Figure 5.1

88

 

 

 

 

 

 

 

 

 

 

 

 

 

 

z (axisymmetry)
4 applied
displacement
r 4 4 A
45 matrix
/:
5:
ﬂ: coating
/:
/.
g:
45 ﬁber
g5 (df=8um)
/:
/:
/.
/~
/:
/.
é!
/ r VII/‘2‘}??? > r (symmetry)
l 4 .1
I 5.4 df
Finite Element Model.

 

 

 

Note: drawing not to scale

89

coating stress-strain transfer is not disturbed [by this boundary edge. The ﬁber diameter
used in the model is 8 pm. Along the asymmetric and axisymmetric (ﬁber center line)
axes, roller constraints are applied. A gap length of 2 pm is used in the model to account
for the recoil of the ﬁber when ﬁber break occurs. It was found that the coating
debonding occurs immediately after ﬁber fracture because of the presence of stress '
singularities at the ﬁber breaks [35,36]. Thus a debond crack of 20 pm in length along the
ﬁber-coating interface was included in the ﬁnite element model. A relative displacement
of 8 pm is applied at the upper edge of the model. The effect of the coating thickness on
interfacial behavior is evaluated for thicknesses t, = 0.0125 df, 0.05125 df, 0.125 (If and
0.25 df, where t, and df are the coating thickness and ﬁber diameter respectively. In the
system under investigation here with coating thickness of 0.1 um, the range of coating

thickness in the ﬁnite element analysis includes this thin layer plus coating layers that are 4,

10 and 20 times as thick.

5.1 MATERIAL PROPERTIES

The carbon ﬁber used in this numerical analysis is linearly elastic and transversely
isotropic. The matrix Epon 828 epoxy with 14.5 wt.% of mPDA (which is the
experimental matrix used in this study), is isotropic and non-linear elastic-plastic. For
simplicity, the matrix is assumed to behave perfectly plastically after 6% strain. The
material behaviors of the ﬁber and the matrix are shown in Figure 5.2 and are well

documented in the literature [37,38].

9O

i—s
d
.J
.u‘
.1
a—q

3;.

O\
o
F
ﬁber axral O—e curve
B
D)
S.
><
' (,3
('0
O
t:
2
9
L

 

00
O

 

 

 

 

 

4:.
O
l -

 

stress (MPa)

I

 

 

 

 

 

 

 

.. K;- """""""""""""""" * """"""""""""

/ fragment is created at 1% matrix Strain
0 I, 1 ’
0

 

 

 

 

 

l r i L 1 4

i
0.01 0.02 0.03 0.04 0.05 0.06
strain

 

Figure 5.2 Stress-strain curve of the constituent materials.

91

From experimental observation, it was found that the ﬁber began to fracture at
approximately 0.8%-1% coupon strain in the single ﬁber fragmentation test. In the model,
the fragment is assumed to be created at 1% matrix strain at which point the matrix
modulus is lower than it was in the initial non-strained state. The coating property effect
is evaluated by varying the coating modulus from that of a compliant polymer coating, e. g.
E,/ E, = 0.7, to that of a brittle precerarnic polymer coating, e.g. Bil Em = 1.5, and to that
of a more brittle ceramic coating, such as E,/ Em = 7.5 and E,/ Em = 30, where E, and Em

are the initial coating and matrix moduli respectively.

5.2 NON-LINEAR FINITE ELEMENT ANALYSIS of the EFFECT of COATING
MODULUS (200 °C-600 °C)

The ﬁber axial stress distributions for E,/ Em = 0.7, 1.5, 7.5 and 30 are shown in
Figure 5.3(a). It is found that the maximum ﬁber axial stress (an indicator of stress
transfer efﬁciency) increases about 2% when the coatingzmatrix stiffness ratio increases
from 0.7 to 7.5. The ﬁber ineffective length (the length within which the ﬁber axial stress
recovers 95% of its far ﬁeld value) decreases 6.3% when E,/ 13., increases from 0.7 to 7.5.
The decrease in the ﬁber ineffective length with the increasing coating modulus in the
numerical analysis corresponds closely to the decrease in the ﬁber critical length with
increasing coating heat treatment temperature from 200 °C to 600 °C in the experimental
observation. From the experimental results, it is found that the ﬁber critical length
decreases 18% when the ﬁber heat treatment temperature increase from 200 °C to 600 °C.

Thus the numerical and experimental results show that as the property of the coating on

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m jiT T I I I rﬁ I I I I I I
A 3500 ‘ '
to r. - ..
E 3000 W
v r- : \ .,
3 2500 g-
g 2000" "’3 -- ———71
7: —o—E./E =0? 3“ i a
'— 15m ‘ m \
5 ----D---E./E =1.5 \\ -
.- 1am I m 'j‘ ;
£- —o---Ei/Em=7.5 . \ g. r
500 MAME/E =30 A “dbl=20|.tm
0 ° III 1 r r—I) r 1_‘|
01020304050607080
IN,
(a)
5
22’
i?
33.
5
8
1...
(b)
60 I T I I ITj j
’3
a.
E
E
a
8
D
E
E-
2
.E
0
01020304050607080
(C) 21d,

Figure 5.3 (a) Fiber axial and (c) interphase shear stress distributions for
fragments with various E, / Em ratios; (b) maximum ﬁber axial
stress as a function of E, / Em. A 20 um debond length is
assurned at the ﬁber break.

93

the ﬁber improves or as the heat treatment temperature gets higher, the ﬁber critical length
becomes shorter and hence the stress transfer efﬁciency becomes higher.

However, increasing the coating stiffness does not always increase the stress
transfer efﬁciency. At a certain value of E,/ Em, the maximum ﬁber axial stress starts to
decrease with increasing Ei/ Em as shown in Figure 5.3(b). In this numerical study, the
maximum ﬁber axial stress for El Em = 30 is less than that for E,/ Em = 7.5 although such

a high ratio is not expected to be realistic in this experimental system.

5.3 NON-LINEAR FINITE ELEMENT ANALYSIS of the EFFECT of COATING
MODULUS (800 °C-1000 0C)

The experimental results show that there is an abrupt decrease in the critical length
when the heat treatment temperature changes from 600 °C to 800 °C. This abrupt change
in critical length, however, is not reﬂected in the ﬁnite element analysis. As discussed
earlier, the conversion of the coating from polymer to ceramic takes place above 600 °C.
Apparently this polymer to ceramic conversion results in a dramatic change in the ﬁber-
coating adhesion as well as the coating properties. However, the level of adhesion and the
chemical bond strength have not been incorporated into the ﬁnite element model.
Nonetheless, the mismatch between the ﬁnite element result and the experimental
observation indicates that a signiﬁcant phase change of the coating must have occurred
between 600 °C and 800 °C.

The ﬁnite element analysis suggests that for ﬁber coatings treated at 200 °C, 400

°C and 600 °C, the bonding between the ﬁber and the polymer coating is approximately

94

the same and is not as strong as the bonding between the ﬁber and the more ceramic
coating treated at 800 °C and 1000 °C. This conclusion is supported by the ITS test
results. The coating-matrix stiffness ratio is relatively low (less than 7.5). Thus increasing
the heat treatment temperature (hence the coating stiffness) increases the stress transfer
efﬁciency (Figure 5.3(b)) or decreases the critical length (Table 4.3). For the ﬁber
coating treated above 600 °C, the bonding between the ﬁber and the ceramic coating is
much stronger and the debond length is accordingly shorter. As a result, the stress
transfer efﬁciency is much higher and the critical length is much shorter. There is an
increase in the coating stiffness as the coating undergoes a phase change from polymer to
ceramic. The coating-matrix stiffness ratio is relatively high (above 7.5) but an increase in
coating stiffness will result in a decrease in the stress transfer efﬁciency (Figure 5.3(b)).
This might explain the increase in the critical length when the coating heat treatment
temperature increases from 800 °C to 1000 0C. In addition, the coating tensile stress
increases when the coating stiffness becomes higher, which in turn can cause tensile failure
of the coating. Hence, the coating will debond as a result of the shear stress
concentrations at the coating fractures. This conclusion is supported by the experimental
observation of the irregular photoelastic stress patterns and the appearance of the
circumferential cracks in the 1000 °C heat treated coating (Figure 4.15). Therefore, for
the thermal treatment of the ﬁber coating, a balance between the oxidation resistance
(bonding quality) and the material properties has to be achieved.

The shear stress distributions for E, / Em = 0.7, 1.5, 7.5 and 30 are shown in

Figure 5.3(c). It is found that, near the tip of the debond crack, the shear stress

95

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4000 - E . .7 , I
A 3500
a _ =‘
...A' =—-A=—‘ AL:- \
E 3000_ V _.,\\,_\
a 2500 \éii
g 2000 010 933‘
'-' —O—t.= .
g 1500 ' um \.\
c} t=0.41p.m \._
E 1000 —o t-lO {3
1.: - ' “m \§
0 *— l
0 10 20 30 2’0 50 6O 70 80
(a) ‘
60
’3 —O—ti=0.10um j
E 50 D -°ti=0.4l|.tm
E40 —0---ti=l.0|.tm
'0 --A--ti=2.0um
E 30
g 20
:e:‘ .
2 10
.5 . dbli=l2t)prn
0 - - . ‘ .
0 10 20 304050in 80
(b) zld

Figure 5.4 (a) Fiber axial and (b) interphase shear stress distributions for

fragments with various interphase thicknesses for Ei/Em = 1.8. A

20 um debond length is assumed at the ﬁber break.

96

distributions oscillate with the stress oscillation being greater for the high Iii/Em case. The
shear stress distributions for all cases reach the same plateau as a result of matrix
plasticity. It implies that the matrix behavior is a dominant parameter in the coating shear

stress transfer.

5.4 EFFECT of COATING THICKNESS

The effect of coating thickness on the coating stress transfer was evaluated for E,/
Em = 1.8 and E, / Em = 0.3. The coating thicknesses t, studied are 0.0125 df, 0.05125 df,
0.125 df and 0.25 df. The ﬁber axial and coating shear stress distributions for E, / Em = 1.8
are shown in Figure 5.4. When the coating thickness increases from 0.0125 df to 0.25 df
(an increase of l900%), the ﬁber axial stress increases 6% while the shear stress plateau
increases 23%. For E, / Em = 0.3, when the coating thickness increases from 0.0125 df to
0.25 df, the ﬁber axial stress increases 4% (Figure 5.5(a)) while the shear stress plateau
increases 24% (Figure 5.5(b)). Thus the stress transfer efﬁciency increases with the
increase in coating thickness at the expense of increasing the likelihood of shear failure.
The stress transfer efﬁciency increases slightly with a thicker coating for a compliant
polymer coating (E, / Em < 1) as well as for the more brittle preceramic polymer coating
(E, / Em > 1). If the change in the coating thickness is not large, the change in the coating
stress distribution is marginal. Nevertheless, the coating stress transfer efﬁciency depends
on several factors such as the debond length, the frictional stress between debonded

surfaces, the ﬁber modulus, and the coating modulus and thickness. Overall, the dominant

97

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m : T I I l I l
A 3500 ‘
E 3000
g 1500 —o—ri-0.10um \X’é -
._ ---D---ti=0.4lp.m \\ ~
g 1000 —°"'ti=1-Ol4m ."\ d

500 -. .- _ »dbl=20

O A ti-2.0pm 1 r :1 i lint-i)
0 10 20 3O 40 50 60 7O 80
mi,
(a)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

interphase shear stress (MPa)

 

 

 

 

 

 

0 . ' i r i r r 1 r
O 10 20 3O 4O 50 60 7O 80
Z/df

 

35.3

Figure 5.5 (a) Fiber axial and (b) interphase shear stress distributions for
fragments with various interphase thicknesses for Ei/E"l = 0.3. A

20 um debond length is assumed at the ﬁber break.

98

factor for the coating behavior is the material response of the matrix and the phase of the

coating (polymer or ceramic).

 

CONCLUSIONS

   

 

The silsesquioxane coating applied to the graphite ﬁbers by dip coating were
shown to form silicon oxycarbide coatings 0.1 tun thick. After pyrolysis in nitrogen to
high temperatures (200 °C increments to 1000 °C), amorphous polymeric coatings were
shown to convert to amorphous silicon oxycarbide coatings on heating to temperatures
above 600 OC.

Single carbon ﬁbers coated with silicon oxycarbide coatings encapsulated in a
transparent epoxy matrix exhibited differences in adhesion and fracture behavior as the
coating mechanical properties changed with the degree of pyrolysis. Initially the adhesion
of the coating in the precerarnic inorganic polymer state was low and interfacial failure
was observed. After pyrolysis to 800 °C and above, a signiﬁcant increase in adhesion was
detected but the failure mode remained interfacial. Circumferential cracks were detected
in the coating. In terms of the locus of interfacial failure, the ﬁber-coating failure was
dominated in the 800 °C and 1000 °C treated strained samples and a mixed ﬁber-coating
and coating-matrix failure was observed in the samples pyrolyzed up to 600 °C. The
interface is observed as intact in the case of the unstrained single ﬁber samples, whereas it

is disrupted in the samples subjected to uniaxial tension.

99

100

A non-linear axisymmetric ﬁnite element model of the single ﬁber coated material
having properties varying in a manner similar to. the coating was used to analyze the
mechanical state of stress in the coating and at the interface. It was shown that changes in
coating thickness had little effect on the critical length-to-diameter ratio. Increase in
coating modulus in the model produced interfacial behavior which correlated well with the

observed experimental adhesion and interfacial failure results.

BIBLIOGRAPHY

BIBLIOGRAPHY

M. Harris, T.M. Chaudhry, L.T. Drzal and RM. Laine, “Silicon oxycarbide coatings
on graphite ﬁbers”, Part 1. Chemistry, processing and oxidation resistance, Materials
Science and Engineering A, 195 (1995) p 223-236.

Rosen, B.W., A simple Precedure for Experimental Determination of the
Longitudinal Shear Modulus of Unidirectional Composites, J. Comp. Mater. 6, 552-
4 (1972).

J.M. Whitney and LT. Drzal, Axisymmetric Stress Distribution Around An Isolated
Fiber Fragment, Toughened Composites, AS 771/! STP 93 7, 179-96 (1987)

L]. Broutman and RD. Agarwhal, A theoretical study of the effect of an interfacial
layer on the properties of composites, Polym. Eng. Sci, 14 (1974) 581-588.

E. Plueddermann, Silane Coupling Agents, Plenum, New York, 1982.

E. Fitzer, K-H. Geigl, W. Huttner and R. Weiss, Carbon, 18, (1980) 389.

LT. Drzal, M.J. Rich, M.F. Koenig and PF. Lloyd, Adhesion of graphite ﬁbers to
epoxy matrices. II The effect of ﬁber ﬁnish, .1. Adhes, 16 (1983) 133-152.

L.T. Drzal, The effect of polymeric matrix mechanical properties on ﬁber-matrix
interfacial shear strength, Mater. Sci. Eng. A, 126 (1990) 289-293.

L.T. Drzal and M. Madhukar, “Fibre-Matrix Adhesion and its Relationship to

Composite Mechanical Properties”, J. Mater. Sci., 28, (1993) p 569-610.

101

10.

ll.

12.

13.

14.

15.

16.

17.

102

L.T. Drzal, M. Madhukar and MC. Waterbury, Adhesion to carbon ﬁber surfaces:
surface chemical and energetic effects, Composite Struct., 26 (1993).

Drzal, LT, and P]. Herrera-Franco, Composite F iber-Matrix Bond Tests,
Engineered Materials Handbook, 3, 391 (1991).

PI. Herrera-Franco and L.T.Drzal, “Comparison of Methods for the Measurement of
Fibre-Matrix Adhesion in Composites”, Composites, Vol. 23, Number 1, 1992, 2-27.
A. Kelly and W.R. Tyson, “Tensile Properties of Fiber-Reinforced Metals:
Copper/Tungsten and Copper/Molybdenum”, J. Mech. Phys. Solids, Vol. 13, 1965, p
329-3 50.

L.T. Drzal, M.J. Rich, JD. Camping, and W.J. Park, Interfacial Shear Strength and
Failure Mechanisms in Graphite Fiber Compsites, Paper 20-C, 35th Annual Technical
Conference, Reinforced Plastics/Composites Institute, The Society of the Plastics
Industry, 1980.

L.T. Drzal, M.J. Rich and PF. Lloyd, Adhesion of Graphite Fibers to Epoxy
Matrices: I. The Role of Fiber Surface Treatment, J. Adhes., Vol 16, 1983, p 1-30.
L.T. Drzal, M.J. Rich and M. Koenig, Adhesion of Graphite Fibers to Epoxy
Matrices: III. The Effect of Hygrothermal Exposure, J. Adhes., Vol 18, 1985, p 49-
72.

J .F. Mandell, J .-H. Chen, and F .J . McGarry, “A Microdebonding Test for In-Situ
Fiber-Matrix Bond and Moisture Effects”, Research Report R80-l, Department of
Materials Science and Engineering, Massachusetts Institute of Technology, Feb.

1980.

18.

19.

20.

21.

22.

23.

24.

25.

103

Mandell, J .F ., Grande, D.H., Tsiang, T.-H. and McGarry, F .J ., “Modiﬁed
Microdebonding Test for Direct In Situ Fiber/Matrix Bond Strength Determination in

Fiber Composites”, Composite Materials: Testing and Design (Seventh Conference),

ASTM STP 893, J.M. Whitney, Ed., American Society for Testing and Materials,

Philadelphia, 1986, pp 87-108.

(a) M. Harris, in partial ﬁrlﬁlhnent of Master’s Thesis, Department of Materials
Science and Engineering, University of Michigan, (1994). (b) M. Harris, C. Scotto,
R.M. Laine, unpublished work.

R. Baney in “Ultrastructure Processing of Ceramics, Glasses, and Composites”,
edited by LL. Hench and DR. Ulrich, Wiley-Interscience, (1984) p 245-255.

D.A. White, S.M. Oleff, R.D. Boyer, P.A. Budringer, J .R.F ox, “Preparation of Silicon
Carbide ﬁom Organosilicon Gels: 1, Synthesis and Characterization of Precursor
Gels”, Adv. Cer. Mat, (1987) 2, 45-52.

D.A. White, S.M. Oleﬁ', J .R. Fox, “ Preparation of Silicon Carbide from
Organosilicon Gels: H, Gel Pyrolysis and SiC Characterization”, Adv. Cer. Mat,
(1987) 2, 53.

RI. Hurwitz, L. Hyatt, J. Gorecki, L. D’Amore: “Silsesquioxanes as Precursors to
Ceramic Composites”, Ceram. Eng. Proc., (1987) 8, 732-43.

H. Zhang, C.G. Pantano, “Synthesis and Characterization of Silicon Oxycarbide
Glasses”, J. Am. Ceram. 73, 958-63 (1990).

K. Kamiya, T. Yoko, K. Tanaka, M. Takeuchi, “Thermal Evolution of Gels from

CH3Si(OC2H5)3 by the Sol-Gel Method”, J. Non-cryst. Sol., ( 1990) 121, 182-87.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

104

GM. Renlund, S. Prochazka, R.H. Doremus, “Silicon Oxycarbide Glasses: Part 1,
Preparation and Chemistry”, J. Mater. Res. Soc. (1991) 6, 2716.

GM. Renlund, S. Prochazka and RH. Doremus, “Silicon Oxycarbide Glasses: Part
11, Structure and Properties”, J. Mater. Res. Soc. (1991) 6, 2723.

K. Kamiya, O. Makoto, T. Yoko, J. Noncryst. Sol.; 83, 208 (1986).

J .A. Rahn, R.M. Laine, Z-F. Zhang, “The Catalytic Synthesis of Inorganic Polymers
for High Temperature Applications and as Ceramic Precursors” in “Polymer Based
Molecular Composites”, Mater. Res. Soc. Sym. Proc., Vol. 171, D. Schaefer, W.
Rhine, eds, (1990) p. 31-38.

R.M. Laine, K.A. Youngdahl, F. Babonneau, J.F. Harrod, M.L. Hoppe, J.A. Rahn,
“Synthesis and High Temperature Chemistry of Methylsilsesquioxane Polymers
Produced by Titaniuim Catalyzed Redistribution of Methylhydridooligo- and
polysiloxanes”, Chem. Mater. (1990) 2, 464-472.

General Electric Company, silicon products department, Waterford, NY. 12188.
Walls, J .M. (Ed), “Methods of Surface Analysis”, Cambridge University Press,
Cambridge, 1989.

Einstein, A. (1905), Annalen der Physik, 17, 132.

Malis, TE and D. Steele, “Ultramicrotomy for Materials Science”, in Workshop on
Specimen Preparation for TEM Materials 11, R. Anderson, Ed., Materials Research
Society Symposium Proceedings, 199, Pittsburgh, PA, 315 (1990).

H. Ho and LT. Drzal, Non-linear numerical study of the single ﬁber fragmentation

test, part I: test mechanics, Compos. Eng, in press.

105

36. H. Ho and LT. Drzal, Non-linear numerical study of the single ﬁber fragmentation
test, part II: a parametric study, Campos. Eng, in press.

37. V3. Gupta, L.T. Drzal, C.Y.-C. Lee and M.J. Rich, The temperature-dependence
of some mechanical properties of a cured epoxy resin system, Polym. Eng. Sci., 25

(1985) 812-823.

h:

"‘11101111111111“