- -‘. ‘ ‘v V ‘ ’ V :

u' ‘ .nvv'r- . H

£ - . . v-.I'
l ‘..,‘“ ‘3’” ‘- . - -; ,- 1‘
,_ f\ . ).

‘. 4. v. ‘ ‘

o ' "L ¢
3"’?)

.. JA'a {’93

I» p “.1?“

_’,_ '|_
tuh

:4‘4 1
‘ 3! .

5391:}, I
'53-. .v‘ ’
giulgm.

.v’

m7.
. ,h ‘

’3'

“no
.«o.
:5 >9

RF
..-.

'3

mm?"
36.2. : r

4‘5353 ; :E7Eiii

u...

. - ..".
‘fl'l r

aw“ Mg.- m

wk}

:1

~
.

,-
‘

' ‘ 2:1. ‘5

V Lid"

é«;.k:‘::au
‘ if»???
‘ flip. .'

'1

:3 $1.-

 

THESIS

IIIIIIHHHHlllllllI’UIIIUIJINIIWIJIIMIIHHIIHHI

- 31293 01050 8798
LIBHAHY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled
Effects of Nickel, Niobium and Titanium Fiber-Coatings

On the Interfacial Properties of NiAl-A1203 Composites

presented by

Richard L. Schalek

has been accepted towards fulfillment
of the requirements for

Ph.D. Materials Science

 

 

degree in

    

Major professor

Date August 20, 1996

 

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

PLACE It RETURN BOX to remove We checkout from your record.
TO AVOID FINES return on or before due due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

#1

MSU Ie An Affirmetive Action/Emmi Opportunity lnetitwon
Wan-9.1

EFFECTS OF NICKEL, NIOBIUM, AND TITANIUM FIBER COATINGS ON THE
INTERFACIAL PROPERTIES OF NiAl-A1203 COMPOSITES

by
Richard L. Schalek

A DISSERTATION
Submitted to
Michigan State University
in partial f ulfillment of the requirements
for the degree of

DOCTOR of PHILOSOPHY

Department of Materials Science and Mechanics
1996

ABSTRACT

EFFECTS OF NICKEL, NIOBIUM, AND TITANIUM FIBER COATINGS ON THE
INTERFACIAL PROPERTIES OF NiAl-A1203 COMPOSITES

By

Richard L. Schalek

An assessment of the effects of nickel, titanium, and niobium fiber coatings on the
interfacial behavior of diffusion bonded NiAl-A1203 composites has been investigated.
The three coating materials were chosen to provide a broad spectrum of coating-fiber
chemical activity and to improve the normally poor hi gh-temperature fiber-matrix
adhesion. As-extracted fiber fragment distributions indicate a uniform distribution of
fragments for the weakly chemically reactive uncoated and niobium-coated fibers, while
the nickel and titanium coatings, representing highly reactive coatings, consisted only of
short fragments. The effects of the fiber coatings on the fiber surface morphology were
documented using scanning electron microscopy and atomic force microscopy. Small
morphological changes occurred for the uncoated fiber composite, while the coated fibers
showed significant increases in surface roughness.

The longitudinal thermal strain of small volume fraction, aligned composites (5-
and 7-vol.%) were measured by dilatometry from room temperature to 1450 K. From
these data, the nickel- and niobium-coated fiber composites showed distinct knees in the
heating half-cycle, accompanied by hysteresis and residual compressive strains. This

behavior indicates that compressive creep strains accumulated during the high

temperature expansion, and hence, good fiber-matrix load transfer occurs. In contrast to
this behavior, the uncoated fiber composite showed only a slight knee and small
hysteresis strain, and no residual compressive strain, indicating poor hi gh-temperature
load transfer. The titanium-coated fiber composite displayed an intermediate behavior.
Of the three fiber coating materials investigated, niobium emerges as the most favorable

for increasing high-temperature fiber-matrix load transfer.

C0pyn’ght by
Richard L Schalek
1996

ACKNOWLEDGMENTS

The graduate study environment permits the forging of many lasting relationships. I have
been fortunate to develop friendships with a number of people from a variety of different
departments. I thank all of the following for being friends. From the Materials Science
and Mechanics department Luiwen Chen, Zhewie Cai, Seungho Nam, Tony Ozzello, and
Michelle Hlvacek. From the Chemical Engineering Department Jay Kalantar, Shri Iyer,
and Brent Jones, and from the Mechanical Engineering Department, Todd Rachel.
Finally, from the Physics and Astronomy Department Dr. R.M. Ronningen and Dr. N.O.
Birge, and from the Center for Electron Optics Dr. Karen Klomparens.

The technical demands required of any scientific project are many. The personnel that
train the students and provide a creative, problem solving environment, are sometimes
under appreciated for their contributions to the work. With this in mind I thank the
"Physic's Shop Guys", Tom Palazzoa, Tom Hudson , and Jim Muns for freely and
willingly sharing with me their time, expertise and problem solving skills. The
Composite Materials and Structures Center stands as an exemplary entity for conducting
research. Thanks are required for the director of the center, Dr. LT. Drzal, for providing
and maintaining the facility. Also thanks to the manager of the facility, Mike Rich, for all
of the technical help, walks in the woods, golf outings, and general good-guy behavior. I
would also like to thank Brian Rook, for all his help and, in particular, for making a gold
crown for my wisdom tooth. Two other people provided a great working atmosphere
include Dr. Kevin Hook and Dr. Pedro Herrera.

I would also like to thank my advisory committee, consisting of Dr. MJ. Crimp, Dr. L.T.
Drzal, Dr. D.S. Grummon, Dr. W. C. McHarris, and Dr. H. Schock for providing

guidance and providing critical suggestions. I would particularly like to thank them for
their patience while I completed this work. In addition, I thank my advisor Dr. D.S.
Grummon for providing interesting conversations and for providing financial support.

I would also like to thank my mother, who, lacking formal education, taught me the

virtues of education, responsibility, loyalty, and commitment.

Finally, I would like to thank my wife, Dr. Diandra Leslie-Pelecky for her constant

companionship, love, and devotion.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... xii

LIST OF FIGURES ......................................................................................................... xiii

GLOSSARY OF SYMBOLS AND ABBREVIATIONS ............................................. xviii

CHAPTER ONE

INTRODUCTION .......................................................................................................... 1
1.1 Introductory Comments ..................................................................................... 1
1.2 Problem Statement. ........................................................................................... 1
1.3 Proposed Solution .............................................................................................. 2
1.4 Thesis Organization ........................................................................................... 4

CHAPTER TWO

REVIEW OF THE LITERATURE ................................................................................ 5

2.1. Thermodynamic and Chemical Aspects of the Fiber-Matrix Interface Region..5

2.1.1 Thermodynamics of Metal-Ceramic Interface Formation ........................... 5

2.2 Fiber-Matrix Adhesion Mechanisms ................................................................. 8

2.2.1 Mechanical Bonding .................................................................................... 8
2.2.2 Chemical Bonding ....................................................................................... 10
2.2.2.1 Classifications of Metal-Cerarnic Interfaces ........................................ 11
2.3 Fiber-Coating-Matrix Design Considerations in NiAl IMCs and MMCs ......... 12
2.3.1 Review of the Mechanical and Physicochemical Properties of NiAl ......... 12
23.1.1 Crystal Structure and Thermophysical Properties of NiAl ................... 13
2.3.1.2 Mechanical Properties of NiAl .............................................................. 16
2.3.1.3 Ternary Alloying Effects in NiAl .......................................................... 19

VII

2.3.2 Oxidation and Thennochemical Behavior of Monolithic NiAl

and NiAl Composites .................................................................................. 21
2.3.2.1 Fiber—Matrix Chemical Compatibility in NiAl Composites .................. 21
2.3.2.2 Oxidation Behavior of NiAl and NiAl Composites .............................. 23
2.3.2.3 The Effects of Thermal Cycling on the Mechanical Properties

of Composites ....................................................................................... 24

2.3.3 Coating Fibers to Fabricate Engineered Interfaces ..................................... 25
23.3.1 Theoretical Studies of the Effects of Fiber Coatings on

the Mechanical Behavior of Composites .............................................. 26

2.3.3.2 Experimental Investigations of Fiber Coated NiAl Composites ........... 28

2.4 Mechanical and Therrnomechanical Behavior of NiAl and NiAl Composites. 32

2.4.1 Mechanical Properties of NiAl-A1203 Composite Properties ..................... 32
2.4.2 Thermal Expansion Behavior of IMCs and MMCs ................................... 35
2.4.3 Residual Stresses and Thermal Strains in IMCs and MMCs ...................... 37
CHAPTER THREE
EXPERIMENTAL PROCEDURES .............................................................................. 45
3.1 Description of Materials .................................................................................... 45
3.1.1 FF and Saphikon Fibers .............................................................................. 45
3.1.2 NiAl Matrix Material .................................................................................. 45
3.1.3 Coating Materials ........................................................................................ 47
3.2 Fiber Coating and Diffusion Bonding Procedures ........................................... 47
3.2.1 Fiber Mounting Procedures ......................................................................... 47
3.2.2 Fiber Coating Deposition Process ............................................................... 47
3.2.3 The Diffusion Bonding Procedure .............................................................. 49
3.2.3.1 NiAl Diffusion Bonding Plate Fabrication ............................................ 49
3.2.3.2 Fiber Lay-up Methods ........................................................................... 51
3.2.3.3 Diffusion Bonding Process .................................................................... 56
3.3 Composite Thermal Cycling and Post Testing Sample Preparation ................. 56

viii

3.3.1 Composite Thermal Cycling in the Temperature Range of

373 and 1173 K. .......................................................................................... 56

3.3.2 Composite Thermal Cycling in the Temperature Range of
373 and 1423 K. .......................................................................................... 57
3.3.3 Electroetching Procedures ........................................................................... 59

3.4 Post-Compositing Fiber Extraction and Fiber Fragment

Length Determination ....................................................................................... 59
3.5 Fiber-Matrix Interface and Fiber Surface Morphology Characterization ......... 62
3.5.1 Scanning Electron Microscopy ................................................................... 62

3.5.2 Fiber Surface Morphology Characterization Using Atomic
Force Microscopy ....................................................................................... 64
3.6 Therrnomechanical Testing Procedures ............................................................ 64
3.6.1 Dilatometry: Physical Description and Measurement Technique .............. 64
3.6.2 Calculation of the Instantaneous CTE. ........................................................ 64
3.6.3 Procedure for Fracturing of NiAl-Saphikon Composites ............................ 64

CHAPTER FOUR

RESULTS AND DISCUSSION .................................................................................... 68
4.1 Surface Morphology of Uncoated and Coated Fibers ....................................... 68
4.1.1 FP Fibers ..................................................................................................... 68
4.1.2 Saphikon Fibers ........................................................................................... 68
4.2 As-Fabricated Composite Microstructure ......................................................... 75
4.3 Extracted-Fiber Surface Morphology ................................................................ 80
4.3.1 Uncoated Fibers ......................................................................................... 80
4.3.2 Nickel-Coated Fibers ................................................................................. 81
4.3.3 Titanium-Coated Fibers ............................................................................. 81
4.3.4 Niobium-Coated Fibers ............................................................................. 87
4.4 Surface Roughness Measurements of Extracted Fibers ................................. 91

ix

4.4.1 Discussion of the Correlations between the Fiber Surface
Roughness, Residual Thermal Stresses, and the Interfacial

Mechanical Properties ............................................................................... 92
4.5 Interfacial Chemical Analysis of As-Consolidated Composites .................... 98
4.5.1 EDX Analysis of As-Composited Interfaces ............................................ 98
4.5.2 Discussion of Possible Fiber-Coating-Matrix Reactions ......................... 101
4.6 Oxidation Behavior and Interfacial Characterization of
NiAl-Saphikon
Composites ..................................................................................................... 107
4.6.1 Results of NiAl-Saphikon Composites Thermally Cycled
Between 373 K and 1173 K. ..................................................................... 107
4.6.2 NiAl-Saphikon Composites Thermally Cycled 75, 150,
and 300 Times Between 373 and 1423 K. ................................................. 107
4.6.2.1 Composites with Exposed Fiber Ends ................................................ 107
4.6.2.2 Composites with Completely Embedded Fibers ................................ 111
4.6.3 Discussion of the Oxidation Behavior of Thermally Cycled
NiAl-Saphikon Composites ....................................................................... 116
4.7 Assessment of Consolidation Induced Fiber Damage for
Uncoated and Coated Saphikon Fibers .......................................................... 123
4.7.1 SEM Fractographic Study of Extracted-Fiber Fragments .......................... 123
4.7.2 Extracted-Fiber Fragment Distributions .................................................... 126
4.7.3 Correlation of the Fiber Fragment Distributions with
the Strengthening Ability .......................................................................... 132
4.8 Fracture Behavior of NiAl-Saphikon Composites ......................................... 135
4.8.1 Room-Temperature Composite Fracture Behavior .................................... 135
4.8.2 High-Temperature Composite Fracture Behavior ...................................... 144
4.8.3 Discussion of the Fracture Behavior of Thermally
Cycled Dilatometry Composites ............................................................... 153
4.9 Thermal Expansion Behavior of Low-Volume Fraction Composites ............ 159

4.9.1 Thermal Expansion Behavior of Monolithic NiAl,
NiAl-(Saphikon), and NiAl-(Ni-Saphikon) Composites ........................... 159

4.9.2 Thermal Expansion Response of NiAl-(Uncoated- and
Coated-Saphikon) Composites as a Function of Number of Cycles ......... 167

X

4.9.2.1 7-Vol.% NiAl-Saphikon Composites ................................................. 167

4.9.2.2 7-Vol.% NiAl—(Ni-Saphikon) Composite ........................................... 172
4.9.2.3 7-Vol.% NiAl-(Nb-Saphikon) Composite .......................................... 177
4.9.2.4 7-Vol.% NiAl-(Ti-Saphikon) Composites ......................................... 177
4.9.3 Discussion of the Thermal Strain Curves and the Conesponding
Instantaneous CI‘E Curves ........................................................................ 180
4.9.3.1 General Discussion of Composite Thermal Expansion Behavior ...... 180

4.9.3.2 Discussion of the Thermal Cycling Strain Response of
Uncoated- and Coated-A1203 Fiber Composites ............................... 191

4.9.3.3 Discussion of the Effects of the Fiber Coatings on
the Instantaneous CT E. ...................................................................... 198

4.9.4 Thermal Expansion Behavior of NiAl-(Uncoated- and
Coated-Saphikon) Composites at Temperatures Near the
Knee Temperature ................................................................................. 204

4.9.5 Disscusion of the Effects of heating Rate on the Thermal
Expansion Behavior of Uncoated, Niobium- and

Nickel-Coated Saphikon Composites ................................................... 205

4.10 Axial Matrix Stress Analysis ........................................................................ 214

4.11 Summary of Results and Discussion ............................................................ 218
CHAPTER FIVE

CONCLUSIONS ........................................................................................................... 223

BIBLIOGRAPHY .......................................................................................................... 225

LIST OF TABLES

Table 1 Interface energy, Go, and crystallographic relationships of

niobium-A1203 joints ......................................................................................... 8
Table 2 Metal-ceramic interface classification scheme ................................................. 12
Table 3 Summary of creep parameters for NiAl ............................................................ 19
Table 4 Reinforcement-NiAl chemical compatibility .................................................... 22
Table 5 Property summary of NiAl and NiAl-A1203 composites ................................. 33
Table 6 Summary of models used to predict dimensional changes
in IMCs and MMCs ......................................................................................... 38
Table 7 Interfacial relaxation processes in MMCs ........................................................ 42
Table 8 Properties of A1203 fibers ................................................................................. 46
Table 9 Table detailing the diffusion bonding processing parameters .......................... 57
Table 10 Gibbs free energy for nickel, niobium, and titanium reactions
with A1203 ...................................................................................................... 86
Table 11 Calculated roughness values for uncoated and coated extracted fibers .......... 97
Table 12 Calculated fiber-matrix interfacial stresses ..................................................... 100
Table 13 Summary of extracted-fiber fragment distributions ........................................ 133
Table 14 Calculated percent strengthening capability for fragmented fiber lengths ..... 134
Table 15 Summary of extracted fiber fragment aspect ratios ........................................ 137
Table 16 Thoma] expansion parameters for composites as a function of cycle ........... 170
Table 17 Summary of heating half-cycle instantaneous CTE observations .................. 202
Table 18 Thermal expansion parameters for composites tested at
difierent heating rates ..................................................................................... 211
Table 19 Summary of the structure-property relationships ........................................... 222

xii

LIST OF FIGURES

Figure 1 Strength versus (a) the work of adhesion and (b) the oxygen

affinity of metal ................................................................................................ 9
Figure 2 The crystal structure (primitive cubic CsCl (B2)) of NiAl .............................. 14
Figure 3 The Ni-Al equilibrium phase diagram ............................................................. 15
Figure 4 The flow stress of NiAl for a number of compositions ................................... 17
Figure 5 Correlation between the periodic table and the alloying behavior of NiAl ..... 20
Figure 6 Radial stress versus CTE and compliant layer thickness ................................ 29
Figure 7 Circumferential stress versus the CT E and compliant layer thickness ............ 30
Figure 8 Longitudinal stress versus the CTE and the thickness of a compliant

layer .................................................................................................................. 31
Figure 9 Tensile response for NiAl (a, b) and NiAl-A1203 composite (c, d) ................ 34

Figure 10 A schematic illustration of the origin and types of in-plane
residual stresses ............................................................................................. 39

Figure 11 The stress field of a spherical particle of SiC: (a) elastic
and (b) elastic-plastic .................................................................................... 40

Figure 12 A schematic illustration of the origin of the longitudinal residual stresses 44

Figure 13 Schematic of fibers mounted onto an aluminum frame ................................. 48
Figure 14 Schematic of the e-gun, fibers, and film thickness monitor .......................... 50
Figure 15 Electrodischarge machine plunger tool used to fabricate fiber channels ...... 52
Figure 16 SEM micrograph illustrating EDMed fiber channels in NiAl ....................... 53
Figure 17 FP fiber composite lay-up diagram for exposed and unexposed fibers ......... 54
Figure 18 Schematic diagram of the Saphikon fiber composite lay-up process. ........... 55

xiii

Figure 19 Optical micrograph of a typical NiAl-A1203 dilatometry composite ........... 58
Figure 20 Schematic illustration of the hi gh-temperature thermal cycling apparatus 60
Figure 21 Schematic diagram of the electroetching procedure used to expose fibers 61

Figure 22 Schematic of the Orton Model IOES-l dilatometer ...................................... 65
Figure 23 Flow diagram of the experimental procedures used in this investigation ..... 67
Figure 24 SEM micrograph illustrating the FP fiber "cobblestone" surface

morphology ................................................................................................... 69
Figure 25 SEM micrographs of a Ni-ooated FP fiber .................................................... 70
Figure 26 SEM micrograph of the cross-sectional of a Ni-coated FP fiber ................... 71
Figure 27 SEM micrographs of an as-received Saphikon fiber ..................................... 72
Figure 28 SEM micrographs of a Ti-coated Saphikon fiber .......................................... 73
Figure 29 SEM micrograph of a fractured Ti—coated Saphikon fiber ............................ 74
Figure 30 SEM micrograph of a transverse section of a NiAl-uncoated FP fiber ......... 76
Figure 31 An optical micrograph of a cross section of a NiAl-

(Ni-Saphikon) composite ................................................................................ 77

Figure 32 Optical micrograph of a longitudinal section of a NiAl-

Saphikon composite ...................................................................................... 78
Figure 33 SEM micrograph of an as-received FP fiber after submersion in etchant ..... 79
Figure 34 SEM micrographs of an uncoated FP fiber showing grain growth ............... 82
Figure 35 SEM micrographs of an uncoated FP fiber consolidated with

a high stress .................................................................................................... 83
Figure 36 SEM micrograph of an extracted uncoated Saphikon fiber ........................... 84
Figure 37 SEM micrographs of a nickel-coated (a) PP fiber, (b) a

Saphikon fiber ............................................................................................... 85
Figure 38 SEM micrographs of a titanium-coated (a) PP and (b), (c)

Saphikon fibers ............................................................................................. 88
Figure 39 SEM micrographs of Nb-coated (a) PP, and (b)

Saphikon fibers ............................................................................................. 90
Figure 40 Typical AFM line scan illustrating the undulating surface of

a Saphikon fiber ............................................................................................ 93
Figure 41 AFM scans of (a) uncoated, (b) Ni-, (c) Nb-, and (d) Ti-coated

Saphikon fiber ............................................................................................... 94

xiv

Figure 42 A histogram comparing the calculated frictional stresses ............................. 99
Figure 43 (a) X-ray line scans and (b) EDX spectrum of a Ti-coated

FP fiber interface ........................................................................................... 102
Figure 44 SEM micrograph of a NiAl-(Nb—coated Saphikon) interface ........................ 104
Figure 45 SEM micrograph of a thermallycycled NiAl-Saphikon

fractured composite ....................................................................................... 109
Figure 46 SEM micrographs of an uncoated NiAl-Saphikon fracture surface .............. 110
Figure 47 A SEM micrograph of an exposed uncoated fiber thermally

cycled 75 times .............................................................................................. 112
Figure 48 A SEM micrograph of an exposed Ni-coated fiber thermally

cycled 75 times .............................................................................................. 113
Figure 49 SEM micrograph of exposed Ti-coated fiber thermally

cycled 75 times .............................................................................................. 114
Figure 50 (a) A SEM micrograph and (b) a EDX spectrum of a

Ti-coated fiber ............................................................................................... 115
Figure 51 A SEM micrograph of an exposed Ni-coated fiber

thermally cycled 150 times ........................................................................... 117
Figure 52 A SEM micrograph of a Ni-coated fiber thermally cycled

150 times ....................................................................................................... 1 18
Figure 53 A SEM micrograph of a Ti-coated fiber thermally cycled 150 times ........... 119
Figure 54 SEM micrographs of an uncoated fiber thermally cycled 300 times ............ 120
Figure 55 SEA micrographs of a Ti-coated fiber thermally cycled 300 times .............. 121
Figure 56 A SEM fractograph of a uncoated embedded fiber thermally

cycled 300 times ............................................................................................ 122
Figure 57 A schematic illustrating the possible oxygen diffusion paths

in composites ................................................................................................. 124
Figure 58 A schematic of oxide scale formation in composites with

fiber ends exposed ......................................................................................... 125
Figure 59 Fractographs of (a) uncoated, (b) and (c) Nb-, ((1) Ni-, and

(e) Ti-coated fibers ......................................................................................... 127

Figure 60 Histogram of the extracted-fiber length distribution. .................................... 131

Figure 61 Percentage of the number of fiber fragments vs. the fiber
fragment aspect ratio ..................................................................................... 130

XV

Figure 62 SEM micrographs of (a) uncoated and (b) Ni-coated

NiAl-FP composites ...................................................................................... 139
Figure 63 SEM fractograph of a 7-vol.% uncoated Saphikon composite ...................... 140
Figure 64 SEM fractographs of a 7 vol.% NiAl—(uncoated-Saphikon) composite ......... 141
Figure 65 A SEM fractograph of a 7-vol.% NiAl-(Nb-Saphikon) composite ............... 142
Figure 66 SEM fractographs of a 7 vol.% NiAl-(Nb-Saphikon) composite .................. 143
Figure 67 A SEM fractograph of a 7-vol.% NiAl-(Ni-Saphikon) composite ................ 145
Figure 68 SEM micrographs of a 7 vol.% NiAl-(Ni-Saphikon) composite ................... 146
Figure 69 SFJVI micrograph of a Ti-coated Saphikon fiber composite .......................... 147
Figure 70 SEM fractographs of a uncoated Saphikon-NiAl composite ......................... 149
Figure 71 SEM micrographs of a NiAl-(Nb-Saphikon) composite ............................... 150
Figure 72 SEM micrographs of a Ni-coated Saphikon~NiAl composite ....................... 151
Figure 73 SEM fractographs of an 7-vol.% NiAl-(Ti-Saphikon) composite ................. 154
Figure 74 Optical micrograph of a 7 vol.% NiAl-(Ni-Saphikon) composite ................. 156
Figure 75 Optical micrograph of a 7-vol.% NiAl-(Ni-Saphikon)

composite after cycling ................................................................................. 157
Figure 76 Axial strain vs. temperature curve for monolithic NiAl ................................ 160
Figure 77 A plot of the instantaneous and theoretical CT E for monolithic NiAl .......... 161
Figure 78 The axial strain vs. temperature plot for a 7 vol.%

NiAl-Saphikon composite ............................................................................. 162
Figure 79 A schematic of the thermal strain vs. temperature plot ................................. 164
Figure 80 (a) heating and (b) cooling CT E plots of a 7-vol.%

Saphikon-NiAl composite ............................................................................. 165
Figure 81 The axial strain vs. temperature for a 5-vol.%

NiAl-(Ni-A1203) composite .......................................................................... 166
Figure 82 (a) heating and (b) cooling CTEs of a 5-vol.%

NiAl—(Ni-Al203) composite .......................................................................... 168
Figure 83 A plot of thermal strain vs. temperature for a 7-vol.%

NiAl-A1203 composite ................................................................................... 169
Figure 84 Plots of the CTE for a 7-vol.% NiAl-Saphikon composite .......................... 173
Figure 85 Plots of the CTE for a 7-vol.% NiAl-Saphikon composite ........................... 174

xvi

Figure 86 The thermal strain vs. temperature for a 7-vol.%

NiAl-(Ni-Al203) composite .......................................................................... 175
Figure 87 Plots of the CT E for a 7 vol.% NiAl-(Ni-A1203) composite ......................... 176
Figure 88 The thermal strain vs. temperature for a 7-vol.%

NiAl-(Nb-A1203) composite ......................................................................... 178
Figure 89 Plots of the CT E for a 7 vol.% NiAl-(Nb-A1203) composite ........................ 179
Figure 90 The thermal strain for a 7- and 5—vol.% NiAl—(Ti-A1203) composite ........... 181
Figure 91 Plots of the CTE for a 7 vol.% NiAl-(Ti-A1203) composite ......................... 182
Figure 92 Plots of the CTE for a 5 vol.% NiAl—(Ti-A1203) composite ......................... 183
Figure 93 Changes in (a) specimen strain and (b) matrix stress

during thermal cycling .................................................................................. 185
Figure 94 The fiber tensile stress, 0, and shear stress, 1:, along the

fiber-matrix interface .................................................................................... 190
Figure 95 The heating half -cycle knee temperature versus the

number of thermal cycles .............................................................................. 192
Figure 96 The average CTE as a function of thermal cycle ........................................... 196
Figure 97 The strain of (a) uncoated, (b) Ti, (c) Ni, and

(d) Nb-coated fiber composites. .................................................................... 206
Figure 98 The strain for (a) uncoated, (b) Nb-, and (c) Ni-coated fiber composites ..... 211

Figure 99 The knee temperature for uncoated, Ni-, and Nb-coated fiber composites 215

Figure 100 Matrix stress for the a 7—vol.% NiAl-Saphikon composite ......................... 218
Figure 101 Matrix stress for a Ni-coated fiber composite ............................................. 219
Figure 102 Matrix stress for a Nb-coated fiber composite ............................................ 220

xvii

GLOSSARY OF SYMBOLS AND ABBREVIATIONS

Symbols (with units)

A
0i

once-couc-

.énm

o.

8 '9 m-r ”70¢ =3"; 36""???Wfi5, 3100-9? on

effective roughness amplitude (run)

average coefficient of thermal expansion (CTE, K‘l); i stands for fiber (1) or
matrix (m)

Burgers vector (nm)

load transfer parameter (unitless)

proportionality constant for the interfacial shear strength (m'l)
crack flaw size (nm)

diffusion coefficient (mz-S'l)

preexponential term (m2-S'1)

elastic modulus (Pa)

hardening SIOpe (matrix modulus above the yield point (Pa)

strain rate (3'1)

strain; i stands for fiber (1) or matrix (m)

shear modulus (Pa)

strain energy release rate (J-m‘z)

interface energy (J-m ‘2)

surface energy; i stands for ceramic (c), metal (m), or metal—ceramic (mc) (J-m‘z)
constant (Pa)

Boltzmann's constant (J-K‘l)

mode I critical stress intensity factor (Pa-m‘05)

intrinsic matrix shear constant (Pa)
fiber length (mm)

critical fiber length (um)

strain rate sensitivity exponent (unitless)
coefficient of kinetic friction (unitless)
true plastic strain exponent (unitless)

Poissons ratio (unitless)
activation energy (eV)

gas constant (J-mol'l-K'l)

fiber radius (um)

creep stress exponent (unitless)
radial stress (Pa)

hoop (circumferential) stress (Pa)

xviii

oz axial (longitudinal) stress (Pa)

Orargh fiber surface roughness induced radial stress (Pa)
om matrix stress (Pa)

0; fiber stress (Pa)

oc composite stress (Pa)

5: current matrix yield stress (Pa)

t time (seconds)

r shear strength (Pa)

17,, matrix yielding shear stress (Pa)

AT T " Tlower (K)

Vi vo ume fraction; i stands for fiber (1) or matrix (m) (unitless)
x distance (m)

Wad work of adhesion (J -m'2)

Abbreviations

AFM atomic force microscopy

BDTT brittle-to-ductile transition temperature
CTE coefficient of thermal expansion

EDM electrodischarge machine

EDX energy dispersive x-ray

LVDT linear variable differential transformer
MMC metal matrix composites

IMC interrnetallic matrix composites

ROM rule-of-mixtures

SEM scanning electron microscopy

TEM transmission electron microscopy

xix

CHAPTER ONE
INTRODUCTION

1.1 Introductory Comments

Operational conditions for projected aerospace, energy conversion, and advanced
transportation designs demand materials with high temperature structural properties
substantially exceeding those of currently available superalloys. The ordered
intermetallic alloys Ti-Al, Ni-Al, and Fe-Al show particular promise for producing the
next significant advancement. In particular, NiAl -- by virtue of its high thermal
conductivity, low density, excellent oxidation resistance and high low-to-intermediate
temperature strength -- has received the most attention for applications in the temperature
range of 300 K to 1473 K. Unfortunately, before NiAl can be used in high-temperature
structural engineering applications, its high temperature strength, creep resistance, and
low temperature fracture toughness must be significantly improved. The metallurgical
processing routes being pursued for remedying these inadequacies include traditional
monolithic alloying, and compositing with advanced particulate reinforcements.

1.2 Problem Statement

Though NiAl possesses many attractive properties, deficiencies in the mechanical
properties include low high-temperature strength, lack of creep strength resistance, and
absence of low-temperature fracture toughness (~5 MPa/m°~5). One route to eliminate (or

minimize) these inadequacies is by compositing the matrix with an advanced
1

2
reinforcement. The appropriate reinforcement phase must meet a stringent requirement

of high temperature strength, combined with both mechanical and chemical matrix
compatibility. Mechanical compatibility includes matching the reinforcement-matrix
coefficients of thermal expansion (CTE) and maintaining adequate high-temperature
reinforcement-matrix adhesion, while promoting low temperature fracture toughness.
Interfacial chemical reactions occurring both during composite fabrication, and under
long-term thermal cycling service conditions, severely limit the types of acceptable
reinforcement materials. The requirement for chemical compatibility is, thus, the most
restrictive prerequisite for potential reinforcements.

Maintaining hi gh-temperature chemical compatibility suggests the use of an inert
reinforcement such as A1203 fibers or whiskers. Two drawbacks of A1203
reinforcements, however, are a large CT E mismatch (~6 x 1045K'1) with NiAl, and the
lack of hi gh-temperature adhesion due to the absence of reinforcement-matrix chemical
reactivity. The CT E mismatch generates interfacial residual stresses when cooling from
consolidation temperature to room temperature, causing a combination of radial,
circumferential, and longitudinal stresses within the matrix. These stresses may cause
interfacial debonding and matrix cracking, and effectively increase the low-temperature

adhesion, further reducing the fracture toughness of the composite.

1.3 Proposed Solution
Central to the NiAl composite design problem is appropriate control of the
interface properties to obtain both high-temperature interfacial adhesion and low-
temperature fracture toughness, while maintaining both thermochemical and
thermomechanical interfacial stabilities. This suggests that the fiber-matrix interface

should have the characteristics of a ceramic matrix composite at room temperature (a

3
weak fiber-matrix interface), but have those of a metal matrix composite at high

temperatures (a relatively strong interface). One potential solution is the creation of a
graded interface through the application of a suitable reinforcement coating material. For
example, the coating material may act as a transient reactivity layer to improve high
temperature adhesion and long term chemical stability, or may simultaneously provide a
ductile interface that increases the fracture toughness of the composite. Upon processing,
the coatings form a graded interphase that provides a smoother transfer of stress from
fiber to matrix.

The selection of fiber-coating materials must be commensurate with acceptable
alloying practices for the NiAl matrix and available thermodynamic data on the possible
fiber-coating reactions must be considered. In light Of these requirements, the coating
materials Nb, Ti, and Ni were selected for detailed study. Niobium is attractive for
several reasons. For instance, the Nb-Ale3 hi gh-temperature reaction produces a very
thin (monolayer) reaction zone and has a CTE close to that of the fiber [Morozuni, 1981;
Pierik, 1990]. In addition, niobium is ductile at room temperature and several
investigations have been directed toward understanding the alloying behavior of Nb and
NiAl [Cotton, 1993]. Titanium, on the other hand, is highly reactive with A1203, forming
extended reaction zones, and producing a strong chemieal bond between the fiber and
matrix [Clark, 1984]. The hi gh-temperature reaction, Ni-Al203, provides chemical
bonding and fiber wetting, and readily dissolves into the matrix upon consolidation.

This study investigates the properties of niobium-, titanium-, and nickel-coated
polycrystalline and single-crystal A1203 continuous fibers composited with NiAl. The
three primary areas investigated are: (l) interfacial thermochemical stability, (2)
thermomechanical behavior, and (3) hi gh-tempeiature fiber-matrix bonding.
Investigation and characterization of the interfacial thermochemical stability, including
the oxidation behavior, was approached by thermally cycling model composites in

laboratory air. Qualitative and quantitative fiber surface morphology studies are reported

4
and described. The thermochemical behavior of low volume fraction composites (5-7

vol.%) is investigated using dilatometry to measure thermal strain of composites as a
function of temperature (300 K - 1425 K). The instantaneous CT E and the matrix
stresses may be calculated from these data, which provide qualitative information on the
high temperature interfacial shear strength. The thermochemical studies also provide
qualitative information on interfacial metallurgical stability. Results of process-induced

fiber damage, and its relationship to the effectiveness of load transfer, are also included.

1.4 Thesis Organization

This introduction has defined the engineering problem and suggested a proposed
solution. The second chapter presents a literature review beginning with a description of
the thermodynamic and chemical aspects of the fiber-matrix interface. This is followed
by a brief description of fiber-matrix adhesion mechanisms and an examination of fiber
coating design considerations. The next section reviews the mechanical and
thermomechanical behavior of NiAl and NiAl composites. Chapter Three describes the
experimental details and procedures used in acquiring the various data.

The results and discussion are presented in Chapter Four, with the fiber surface
morphology and composite microstructure discussed first. Composite thermal cycling,
thermochemical stability and composite fracture behavior are then discussed. Finally, the
thermal expansion behavior of Saphikon composites and stress analysis results are

presented. Conclusions are enumerated in Chapter Five.

CHAPTER 2
REVIEW OF THE LITERATURE

2.1. Thermodynamic and Chemical Aspects of the Fiber-Matrix Interface Region

Composite technology exploits the stiffness and strength of the reinforcement
phase through dispersion and control of the load transfer from the matrix to the
reinforcement. The role the reinforcement-matrix interface plays in strengthening the
material is, therefore, crucial to understanding the mechanical behavior of a composite
[Metcalfe, 1974]. As a background to the complex physicO-chemical interactions at the
reinforcement-matrix interface that ultimately determine the mechanical response of the
composite, the interfacial chemical thermodynamics will be discussed first.

2.1.] Thermodynamics of Metal-Ceramic Interface Formation The formation
of a metal-ceramic bond yields energy when intimate contact (distances approaching 10
nm) is established between the metal and ceramic surfaces. The simplest description of
this interface formation considers the differences in the metal and ceramic surface
energies. Surface energies for liquid metals generally range from 0.072 .l-m'2 for Cs to
3.24 .l-m‘2 for Re [Murr, 1975]. Surface energies for solid metals are generally less than
these values, and anisotropy effects corresponding to different crystallographic
orientations must be considered. Experimental measurements of the surface energies Of
ceramics are less abundant, but the known values are generally lower than those for

metals, and range from 0.06 .l-m'2 for liquid 8203 to 1.19 .l-m‘2 for TiC (0.905 J-m'2 for
solid A1203) [Kingery, 1976].

6
The thermodynamic criterion for a liquid metal to wet a solid ceramic requires

that the surface energy of the ceramic be greater than that of the metal [I-Iowe, 1993].
Since this is generally not the case, wetting of ceramic fibers by liquid metals rarely
occurs unless a chemical reaction takes place. Unfortunately, fiber-matrix reactions may
compromise the fiber mechanical properties and, in most cases, should be limited. On the
other hand, inadequate wetting produces fiber-matrix interfacial voids that act as stress
raisers and potentially degrade the composite mechanical properties. These interrelated
and opposing effects make choosing appropriate fiber reinforcements for a given matrix
material difficult.

The Dupré equation puts these ideas into a thermodynamic and mathematical

basis by describing the change in free energy, AG, during the formation of a metal-

ceramic interface,

AG=ym+yc+yu, 2—1

where 7m and Ye are the surface energies of the metal and ceramic, respectively, and fine
is the metal-ceramic interfacial energy [Adamson, 1960]. For practical cases, when
chemical bonding occurs at the interface, and interfacial separation produces metal or

ceramic plasticity, the Dupré equation can be rewritten as

y_+yc-ym 5W“, 2-2
where Wad is the work of adhesion. This equation indicates improvement in metal-
ceramic bonding (adhesion) requires a decrease in the interfacial energy, fine.

A general observation regarding 7mg can be made for metal-A1203 systems
[Nicholas, 1989]: in chemical bonding, the interfacial energy between various metals and
N203 increases roughly with the melting point of the metal, and appears to depend

equally on the interaction of the metal with the metallic or nonmetallic element of the

7
ceramic. Furthermore, a qualitative estimate of the wetting characteristics of ceramics by

metals can be obtained by using the semi-empirical relationship, between the ionicin and
the electronegativities of the component elements [Pauling, 1948]. For example, A1203
is about 63% ionic, MN is 40%, Si3N4 is 30% and SiC is only 12% ionic. In general,
highly ionic ceramics, such as Ale3, are difficult to wet and are chemically stable since
their electrons are tightly bound. Ceramics with more metallic bonding character, such as
SiC, tend to have a higher surface energy and are more easily wetted, and are more

chemically reactive than the ionically bonded ceramics.

Evaluation and comparison of studies correlating the interfacial strength and the
chemical bonding of metal-ceramic systems are difficult because the structure and extent
of reactions at the interface are often not well characterized, and different testing
techniques are employed. With this proviso, several recent studies performed on well
characterized interfaces in metal-A1203 systems allow some insight into bond-strength
relationships [Klomp, 1987]. Table 1 gives the dependence of the interface strength,
expressed in terms of a critical energy release rate Go, on the crystallographic orientation
for single and polycrystalline bcc Nb diffusion bonded to hop Al203 under high vacuum
conditions [Elssner, 1985]. The interface fracture energy is highest when the metal -
ceramic interface parallels the close packed planes —(110) II (0001)— and directions
—[110] II [1120]-— and decreases for other orientations. The interfacial energy is lowest
in close packed planes making Wad higher. In the case of polycrystalline samples, the
fracture energy is lower because of the high defect density, large number of slip systems,
and poor atomic matching across the interface provide crack initiation sites.

The room temperature tensile strength versus the work of adhesion is compared
for various metal-A1203 systems is shown in Figure 1a [Klomp, 1987]. These data
indicate that the tensile strength scales with the melting point of the metal. A similar
trend, displayed by the data in Figure 1b, show the tensile strength of several pure molten

metals- and Ni alloy-Al203 interfaces versus the oxygen affinity of the metals or alloying

Table 1 Interface energy, To, and crystallographic relationships for Nb-A1203 joints.

 

 

 

mycrysmifine N5 Single Crystal
(g.s. z 370 pm) (110) Plane

3%? 1"¢=26Jm‘2 I‘¢>150.lm'2
_ [1101Nbu [1&0]st

Sapphire rc = 55 J m-2 R =74 J m"
(1100) [0011Nbll [OOOIISapphire

 

 

Tc=1101m'2
[001le ll [lliOISappriim

 

 

 

 

 

elements [Crispin, 1976]. These data emphasize that the chemical affinity of the
elements across the metal-ceramic interface may provide a useful estimate of the strength

of a metal-ceramic interface.

2.2 Fiber-Matrix Bonding Mechanisms
. The composite system, combined with the consolidation process (e.g., squeeze
casting, diffusion bonding), determines the type and extent of the fiber-matrix bonding;
however generalities apply to all systems. Mechanical and chemical bonding are the two
main categories of interfacial bonding and will be discussed below [Chawla, 1987].
2.2.] Mechanical Bonding A purely mechanical bond implies that all chemical
sources of bonding are absent. The origins of mechanical bonding can be either a

 

Tensile Strength (We)
a s § § § § §

 

 

 

 

0 1000 2000
“Lina/m ’

(a)

 

    

 

Tensile Strength (MPa)
. s s g t s

 

CI
l l

200 400 600 Foo
; kJ/(g-atom oxygen)

0»

°—sfi

8

Figure 1 Strength versus (a) the work of adhesion and (b) the oxygen affinity of metal.

10
mechanical interlocking, resulting from inegularities on the fiber surface, or from

frictional effects arising from stresses produced by differences between the coefficient of
thermal expansion of the matrix and the fiber. In addition to providing a higher debond
stress and increasing the strength of the composite, increasing the fiber surface roughness
can increase the fracture toughness of the composite by increasing the amount of energy
required for fiber pull-out [Paransarthy, 1992]. The fiber surface roughness can change
by surface faceting transitions that take place during consolidation, such as occurs in the
MgO—NiO system [I-Iandwerker, 1990]. Other fiber surface toughening processes include
dissolution of the fiber into the matrix [I-Iowe, 1993]. Also, fibers may be naturally
rough.

2.2.2 Chemical Bonding Chemical bonding involves transfer and interactions
of electrons on an atomic scale [I-Iowe, 1993]. Chemical bonding can be classified
according to four types [Metcalfe, 1974]: (l) dissolution and wetting bond; (2) reaction
bond; (3) exchange-reaction bond; and (4) oxide bond. For the case of dissolution and
wetting bonds, interaction forces develop when the atoms of the constituents approach
within a few atomic diameters of each other. This attraction produces wetting and
dissolution. An example of this bonding condition is niobium reinforced aluminum. On
the other hand, reaction bonding requires the diffusion controlled transfer of atoms from
one or more of the constituents to the interface reaction site and, hence, depends on the
reaction kinetics for interphase growth. The exchange-reaction bond is a special case of
the reaction bond and proceeds by two sequential, distinguishable, reactions (though from
a practical viewpoint the reactions may appear indistinguishable). An example of this
type of reaction is a titanium-aluminum alloy reinforced with boron, which forms a
boride containing both titanium and aluminum, followed by an exchange between the
titanium in the matrix and the aluminum in the diboride [Klein, 1969]. Finally, in
contrast to the previous types Of bonding mechanisms, the oxide type bond classification

occurs when an oxide reinforcement is present. Though no new principles of bonding are

l 1
involved with oxide bonds, and the three former bond types are applicable, developments

in the fabrication and cost of oxide reinforcements, such as single crystal and polycrystal
A1203 fibers, combined with extensive industrial use, makes this classification subset
sensible [Weddell, 1990].

2.2.2.1 Classifications of Metal- Ceramic Interfaces To help
minimize confusion between different metaloceramic systems, four interface
classifications, based on the type of the chemical bonding and interphase formation, are
summarized in Table 2 [Reddy, 1989]. Class I consists of a pure metal oxide at the
interface and is represented as A0x IMOyIM , with the interface between the ceramic and
the metal being a thin layer of the lowest valent metal oxide, MOy. This type of interface
is quite common and examples include oxide glass or ceramic INidNi and
Al 203|Cu20 + Cu . Class II interfaces have ternary oxides at the interface. This system

is represented as A0, IAMOll [M and occurs only when a compound of the type

+y
AMOx+y exists in the ternary system A-M-O. Typical examples of these interfaces are
ziO,|Niz:O,|Ni , Al 2O,|CuA10,|Cu, and Al ,O,|NiAr 2O,|Ni.

Class III represents the formation of a solid solution in the ceramic near the
interface. This system is represented as A0, IAOII — MoylM . In this case no new phase
is formed, but a concentration gradient of M0,. in the ceramic exists near the interface.
An example of this class is the formation of MgO-NiO solid solution at the interface.
Finally, Class IV represents the formation of a solid solution in the metal, and is
represented as AOll IM(A,O). Certain reactive metals react with oxide ceramics,
resulting in metallic solid solution without forming a new phase at the interface. An
example of this class is Al203-Nb, where Nb has a high solubility for both Al and 02 and
the dissolved components have low thermodynamic activity.

12

Table 2 Metal-ceramic interface classification scheme (adopted from Reddy, 1989).

 

2.3 Fiber-Coating-Matrix Design Considerations in NiAl IMCs and MMCs '

In addition to the thermodynamic issues previously addressed, a number of other
factors must be considered when choosing an appropriate fiber coating material for a
NiAl-A1203 composite. Among these factors are the metallurgical and mechanical
properties of NiAl, and those of the fiber coating. In this section, a review of the main
issues regarding tailoring the interfacial properties of NiAl-A1203 composites is given.
First a survey of relevant properties of NiAl and A1203 are presented, followed by
chemical compatibility and metallurgical stability requirements for NiAl-(coated-A1203)
composites. Finally, some theoretical and experimental examples of coated fiber
composites systems are given.

2.3.] Review of the Mechanical and Physicochemical Properties of MA] The
physical and mechanical properties of NiAl have been published in two review articles by
Miracle and Noebe[Miracle, 1993; Noebe, 1993]. In addition to these reviews, three
papers detailing the potential uses of NiAl in the aerospace industry have recently been
reported [Darolia, 1993; Walston, 1993; Darolia, 1991]. These cited works provide an

13
excellent summary of the current metallurgical understanding of both single crystal and

polycrystal NiAl.

2.3.1.1 Crystal Structure and Thermophysical Properties of MA! NiAl is
a Hume-Rothery B-phase electron compound with a valence electron/atom ratio of 3/2,
that crystallizes in the primitive cubic 82 structure (space group Prn3m, CsCl prototype)
as shown in Figure 2. Stoichiometric Ni-SO at.% Al melts congruently at 1911 K and, as
Figure 3 indicates, has a wide phase field consisting of a BZ-ordered intermetallic over
the composition range of 65 -76 wt.% (45 to almost 60 at.%) Ni [Okamoto, 1993]. The
room temperature lattice constant of stoichiometric Ni-50Al is 0.2887 nm and the lattice
parameter of NiAl on either side of stoichiometry can be described by a linear

relationship of the form,

a°(nm) = 0.299839- 0.000222 * (x), for x = 50-60 at.%Ni or

a°(nm) =0.266819-0.000438 *(x), forx = 45- 50at.%Ni 2-4
[Noebe, 1993]. Using x-ray measurements, Cooper concluded that no ionic bonding
occurs in NiAl, but that d-band filling by Al-to-Ni charge transfer occurs and results in a
mixture of covalent and metallic bonds [Cooper, 1963].

Another thennophysical property closely related to the bonding structure of the
material is the coefficient of thermal expansion. The atomistic cause of thermal
expansion is explained by the changing curvature of the interatomic potential energy
function. In general, strongly bonded solids expand at lower rates than weakly bonded
solids [Speyer, 1994] and the coefficient of thermal expansion decreases as the melting
point of the material increases. The thermal expansion behavior of polycrystalline NiAl
has been studied by a number of investigators and is found to be a strong function of
temperature, but not of composition [Clark, 1984]. The thermal expansion coefficient for
NiAl is about 30% less than that for pure Ni, but is very similar to that of Ni-base
superalloys [Darolia, 1991] and can be described by

14

 

 

 

 

 

 

 

 

   

. Nickel Atoms 0 Aluminum Atoms

Figure 2 The crystal structure (primitive cubic CsCl (32)) of NiAl.

15

.aewae 8% 55.53 2.2 a: m 2:5

 

 
   

 

 

 

 

 

 

 

 

 

 

 

 

.z .232 .82».— ..393 _<
8. 8 8 2 S 9. 3 8 8 2 e
, 8.
(N N A v
t a I'
n all I“ N
. 8.
case
033 .. 8.
03...
r 82
. 83
:2.
Unan— fi
8:
03: 3:5
. 83
082
.1 b 4 h < b b d h # h b- u
2: «a 8 e. a. 8 a. 8 8 2 . 8a

.232 .59.».— sass.

 

J.

D.

16

a(K") - 1.1602x10" + 4.08531x10'9T - 1. 58368x10'”T2
+4.18374x10"‘r3 2-5
in the temperature range of 300 K to 1500 K [Clark, 1984].

2.3.1.2 Mechanical Properties of MA! The mechanical properties of NiAl
have been investigated in numerous ways with, an unusual variety in results reported by
different investigators. This lack of agreement may be attributed to differences in grain
size, interstitial contents, texture, and chemistry of the material being tested. Only the
major mechanical properties will be summarized here. Further details may be found in
the reports by Noebe, et al. and Miracle [Noebe, 1993; Miracle, 1993].

The Young's modulus of NiAl was found to be sensitive to processing technique,
because of induced <111> texture, but not to composition [Khadjikar, 1990; Rusovic,
1979]. The reported modulus for extruded NiAl powder can be described by the
following equation [I-Iellman, 1990],

E(GN — m") = 249.3 -0. 072T + 3x10‘5r’. 245
Other temperature dependent mechanical properties include the shear modulus, G, and
the Poisson's ratio which are described by
G(GN - m" ) = 76.6- 0017T and
2-7
v: 0307+ 215x10'5T, 23
respectively [I-Iellman, 1990].

The flow properties of NiAl are highly sensitive to variables such as composition,
strain rate, grain size, cooling rate, and specimen fabrication. Like bcc metals, the flow
stress of NiAl, shown in Figure 4 for a number of compositions, exhibits a strong
temperature dependence at low absolute temperatures, attributable to a large Peierls stress
[Pascoe, 1968]. Though the yield stress is sensitive to metallurgical processing

properties, all studies agree that increasing the testing temperature decreases the yield

17

 

 

1m l I 1' U I I U
1m .Nl43Al rGrain size = 50 um I .,

 

1400

1000

Yield Stress (MPa)

§§§§

 

 

 

 

 

o l l l l l l
0 200 400 600 800 1000 1200 1400
Temperature (K)

Figure 4 The flow stress of NiAl for a number Of compositions.

18
stress and, that for stoichiometric compositions, the yield stress is independent of grain

size [Nagpal, 1990]. The insensitivity of the yield stress to the grain size is apparently
due to the lack of dislocation pile-ups providing grain boundary strengthening and is
more sensitive to the lattice resistance to dislocation motion [Baker, 1991].

At low temperatures, NiAl deforms predominantly by <100>{011} slip (and
occasionally by slip on {001}) Consequently, only three independent slip systems are
available for deformation [Groves and Kelly, 1963]. Since no cross-slip is known to
occur, two additional independent slip systems are required to satisfy von Mises criterion
for extensive, uniform crack free deformation of a NiAl polycrystal. This lack of an
adequate number of slip systems is the principle origin of the low temperature brittleness.
At elevated temperatures, dislocation climb and glide mechanisms are activated and cause
increases in both the ductility and the fracture stress of NiAl. [Bowman, 1992]. This
behavior is illustrated by the brittle-to—ductile transition temperature (BDTT) occurring in
the range of 550-700 K [Noebe, 1993].

The hi gh-temperature creep behavior of NiAl follows that for metals and alloys in
which the second stage, or steady state creep being expressed by the Dorn formulation
[Meyers and Chawla, 1984]. A summary of the stress exponents and activation energies
determined by various investigators for creep of NiAl is presented in Table 3.

The results of these investigations indicate that, with a stress exponent ~ 5 - 7, the
high temperature creep in NiAl is climb controlled. Strategies for improving the creep
resistance of NiAl include solid solution strengthening [Rudy, 1986], precipitation
strengthening [Polvani, 1976], dispersion strengthening [Whittenbergen 1990], and
composite reinforcement [Bowman, 1992]. Though improving creep resistance is vital
for producing the necessary mechanical response of the material, generally these
improvements must be balanced with a decline in other properties, such as strength or

oxidation resistance.

2.3.1.3 Ternary Alloying Efi'ects in NiAl

19

As mentioned above, studies of

alloying NiAl with a ternary element have been performed in an effort to understand the

effects of enhancing the strength or creep resistance by second phase precipitation, solid

solution hardening or grain refinement. When choosing a fiber coating material, the

alloying behavior between the fiber coating and NiAl matrix must be considered. Cotton

Table 3 Summary of creep parameters for NiAl.

 

 

 

 

 

 

 

 

 

 

 

 

AI Iat%5 Grain II'emp. (K) 11 Q (kJ-mol -1) Ref. ‘
Sham)
W. —3'9_- Wi- 1400 6.0 -77. T1 'Wiiiuenberger,
1988
W. .6 ”T -20 F'__—'_‘1100-1400 T. —_§14 Whittenberger,
1987
‘10— 12 T1 - 1300 6 3T— ‘Whineniaerger,
1990
TF3?” W1 -11 Tm. -4.6 3% Yan,l9fi_
30 500 1173 i7 -- Rudy,l
. 30.4 1000 10737—1730— ’7.0-3.3 730390 Vandervoort,
1966

 

 

 

 

 

 

 

 

et al. [1993] have shown that the alloying behavior of NiAl appears to correlate with

group number in the periodic table. Figure 5 illustrates an NiAl alloying behavior

classification scheme of the periodic table: Type-A (Groups 1118, IVB, and VB), where

the solubility of the third element is usually low and at least one ternary intermetallic

phase occurs; Type-B (Group VIB), in which the third phase forms a pseudobinary

eutectic with NiAl; and Type C (Groups VIIB,and VIIIB), in which the third element

exhibits a large solubility and may form an isostructural BZ intermetallic. Both Ti and

Nb fall into the Type-A category, and commonly form either a Heusler or Laves, phase

20

.32 .6 6:93 M.28.... 2.. as use 6.8.5.. 9.. 563.3 88.88 m use...

ESE-:9 2:2 5.3 3.238 wine—3.. D

.32 5 3.328 a»... "O Eh I

2.3.... %.........e.......... 00.....22 a. 2.»... g
2:30 33.....— oESuEBE— reckon. .4 RFC. I

 

=.

a0

«a:

 

 

 

 

a: 3. r. a. E 3
.6 a... P. 5. .N >
:N .6 .z o _ 6m
m: a. [ =5 IIL a; a; m> .5. a:

«3:3... .89... xi... .6 8.5 8...:

21
with a solute preference for the Al site [Polvani, 1976; Stoloff, 1984]. The solubility of

Type-A elements is usually limited at ambient temperatures, and precipitation is expected
for alloying levels above about one percent. Nickel belongs to Type-C, where the ternary
additions show a large solubility in NiAl and, generally, form 32 compounds.

2.3.2 Oxidation and Thermochemical Behavior of Monolithic NiAl and NiAl
Composites Two essential properties required of hi gh-temperature composite materials
are resistance to cyclic oxidation, and good interfacial metallurgical stability. Gains in
improving composite properties, such as stiffness and strength, will be of limited value if
oxidation becomes the limiting factor in determining the upper use temperature. NiAl
has excellent cyclic oxidation resistance, due to the easy formation of strictly Al203
protective oxide layers occurring after a brief transient stage of metastable A1203 phase
formation [Doychak, 1989; Pettit, 1967].

2.3.2.] F iber-Matrix Chemical Compatibility in MA] Composites The fiber-
matrix chemical reactions that continuously occur during thermal cycling of composite
systems present major concerns for long-term stability. Thermodynamic analyses of
fiber-matrix combinations help screen different reinforcement materials for their
chemical compatibility with a given matrix. A reinforcement material-matrix
combination can be considered to be chemically incompatible for either of the following
reasons: (1) the reinforcement material is reduced by an element of the intermetallic
matrix, thus releasing an element of the reinforcement material or, (2) two—product
compounds form at unit activity [Misra, 1989]. If neither process occurs, then either the
product compounds are formed at reduced activities, or the elements of the reinforcement
material are dissolved in the matrix. In such cases, the magnitudes of the equilibrium
activities of the reaction products, at a specified temperature, determine whether a
reinforcement is compatible with the matrix. For screening purposes, an activity value of

0.001 was chosen by Misra, and the ceramic reinforcements in Table 4 were determined

22
to be chemically compatible with NiAl at the expected use temperature between 1273 and

1573 K.

Recently, Misra, et al. have discussed the requirements of the ideal fiber-matrix
interface in NiAl-A1203 composites [Misra, 1992]l . They concluded that a strong
chemical fiber-matrix bond is required to increase interface shear strength and to
withstand radial stresses experienced during heating. In addition, the ideal coating would
be ductile at room temperature and would be thick enough to effectively reduce the

interfacial stresses and confer some ductility to the interface.

Table 4 Reinforcement-NiAl chemical compatibility table.

 

* Borderline situations

Pierik, et al. also reported on NiAl-fiber interfacial compatibilities by studying
different types of ceramic reinforcements, including Sumitomo and Nexte1480 A1203
fibers [Pierik, 1990]. Since the compositions of Nextel 480 and Sumitomo fibers consist
of Al 203 grains with ~15 wt.% of Si02 located at the grain boundaries the glassy phase
boundary reacts with the NiAl matrix during consolidation. The reaction products

include mullite (Al58i2013) and isolated grains of Ni, Si, and A1 at the fiber-matrix

 

lBorderlinesituations

23
interface. No mechanical property information was reported, so the effects of the

reaction products on the bond strength cannot be assessed. These results, however, do
suggest that alumina fibers with a significant amount of second phase are not likely to be
considered candidate reinforcements.

2.3.2.2 Oxidation Behavior of NW and NiAl Composites Among the
various environmental effects, thermal cycling to temperatures greater than 0.5 Tm in
corrosive atmospheres provides one of the most severe tests to determine if a composite
system is potentially useful. Studies of monolithic NiAl oxidation have characterized
scale-growth mechanisms and kinetics, scale microstructures and morphologies, and
oxide-substrate adherence [Doychak, 1989]. These studies indicate that the oxidation
mechanisms in NiAl are different from those of other alumina formers because of the
strong tendency to form strictly A1203 at high temperatures. In addition, a transient stage
of oxidation at intermediate temperatures results in the formation of metastable A1203
phases such as 0-A1203 and y-A1203 [Doychak and Rhule, 1989].

Only a few studies of the oxidation behavior of NiAl composites have been
performed. So as a starting point, considerations of reinforcement-induced changes in the
oxidation resistance of other composite systems will be presented. In most cases, the
reinforcement increases the oxidation rate of the composite by providing a fast diffusion
path along the interface, or by disrupting the formation of a protective interfacial scale.
Surface-related interfacial problems are more severe for continuous fiber composites
when the interface emerges at a free surface, and thus acts as a long-range path for
transport of oxygen into the matrix interior. As an example, titanium alloys reinforced
with SiC monofilaments suffer progressive interfacial oxidation after prolonged periods
at relatively modest temperatures (573 - 873 K) [l-Iartley, 1988]. These temperatures are
too low to cause internal fiber-matrix attack, but oxygen egress from the free surface is
rapid and causes degradation of the interfacial structure, and hence of the composite

mechanical properties.

24
In contrast to the oxidation induced degradation behavior, some research groups

report either no decrease in oxidation resistance, or an increase in oxidation resistance.
For example, Wei studied the isothermal oxidation behavior of Al-SiC monofilament
composites up to 673 K and found no change in the oxidation behavior due to the
presence of the fiber [Wei, 1990]. These results were explained on the basis of a high
degree of fiber-matrix adhesion, combined with a large amount of matrix plasticity which
blocked a fast diffusion path along the interface. An improvement in composite
oxidation resistance is illustrated by Stott and Ashby, who report an increase in the
oxidation resistance for 3 um SiC particles in a Ni matrix (with a volume fraction > 10%,
and exposure to pure oxygen at 1373 K) [Stott and Ashby, 1978]. The SiC particles
apparently impede the transport of Ni2+ ions through the growing NiO layer, producing
an increase the composite oxidation resistance.

Doychak, et al. reported on the isothermal and cyclic oxidation behavior of
aligned, continuous fiber, NiAl-Saphikon and NiAl-(Ni-Saphikon) composites [Doychak,
1992]. Extensive interfacial scale formation for both the composite and the bulk
materials were seen in the isothermal experiments. The thermal cycling experiments
revealed oxide growth rate constants for the composited materials consistent with
alumina formers . Composites reinforced with fibers coated with ~1um nickel had rate
constants 2-10 times higher than the uncoated fibers. This increased rate constant was
attributed to Kirkendall interfacial porosity and suggests that the oxidation behavior of
coated fiber composites may be substantially different than uncoated fiber composites.

2.3.2.3 The Efi‘ects of Thermal Cycling on the Mechanical Properties of
Composites The CTE mismatch, A01 , for IMCs and MMCs range from 4 x 10 ‘6 K'l
(W/Ti system) to 24.6 x 10 '5 K'l (C/Al system). This mismatch produces a thermal
strain of 10'3 or greater, which is on the order of most matrix yield strains, and may thus
damage the composite through nucleation and growth of interfacial micro-damage.
Thermal cycling damage has been reported in several metal matrix composite systems:

25
tungsten fiber/copper [Yoda, 78; Yoda, 79], A1203 fiber (FP)/magnesium [Bhatt 1983],

and SiC/titanium [Park, 1983]. The thermal cycling damage reported in these
investigations include microvoids or cracks at the interfaces, which leads to interface
debonding. This micro—damage results in degradation of mechanical properties, as shown
by the results of Grimes et al. [19T7] where the ultimate tensile strength of continuous
boron fiber/6061 Al composites, that were cycled between room temperature and 693 K
for various numbers of cycles, was reduced by ~15%.

Micro-damage, which has been observed in several continuous fiber MMC
systems [Write, 1975], Often results in permanent dimensional changes. Although the
majority of investigations report that the major dimensional changes occur along the fiber
axis, Kyono, et al. [1986] observed dimensional changes in a continuous graphite
fiber/5056 Al composite in the transverse direction. This transverse dimensional change
appears to have been caused by complete debonding of the matrix-fiber interfaces during
the early stages of cycling, followed by transverse swelling.

2.3.3 Coating Fibers to Fabricate Engineered Interfaces Several deposition
techniques are available to coat long fibers and, to a lesser extent, to coat short fibers and
particulate reinforcements. These techniques may be divided into chemical vapor
deposition (CVD) [Alam, 1990] and physical vapor deposition (PVD), which includes
electron beam evaporation, ion beam sputtering, and magnetron sputtering [Everett,
1991; Cai, 199?]. Other procedures that are used to coat fibers include spraying and
plating techniques [Everett, 1991].

Among the possible roles of the fiber coating, the encouragement of wetting and
the production of a reaction banier are the most common. Promotion of wetting is
usually brought about through the stimulation or modification of some local reaction
[Delannay, 1988]. Various other fiber-coating wetting agents have been explored,

particularly for carbon fibers in aluminum and magnesium [Amateau, 1976; Katman,

1987].

26
From a thermodynamic point of view, highly stable compounds emerge as strong

contenders for barrier coating materials. In practice, most barrier coatings only slow
down the interfacial reactions, rather than providing thermodynamically stable interfaces
[Pierson, 1978]. In addition to thermodynamic stability, or at least reduced reaction
kinetics, the barrier must impair transport of reactants through the coating [Kieschke,
1991]. Some coatings are designed to protect the fiber surface from damage (or
oxidation) prior to incorporation into the matrix. An example of this is the application of
a 50 nm SiC layer on C fibers to protect them from oxidation [V incent, 1989].

2.3.3.1 Theoretical Studies of the Efi'ects of Fiber Coatings on the Mechanical
Behavior of Composites Theoretical analysis of the effects of fiber coatings on the
axial tensile stresses and strengths of MMCs has been investigated by Xia et al. [Xia,
1993]. The analytical investigation by Xia et al. focused on composites composed of
aligned fibers in which the central core of the coated fiber was surrounded by a reaction
zone, and the matrix layer was exposed to a longitudinal applied stress. Two conditions
were analyzed: low interfacial bond strength, and high interfacial bond strength. For low
interface strengths, with debonding occurring before matrix yielding, the composite yield
stress increased with increasing debonding strength. If matrix yielding took place before
debonding (high interfacial strength), the composite yielding stress became constant until
debonding. Furthermore, with increasing reaction zone thickness, debonding and
yielding occur at a lower stress level regardless of the interfacial strength. These results
indicate that interfacial debonding relaxes the stress concentration in fibers, and matrix
yielding increases the stress concentration.

Theocaris and Dcmakos also investigated the effects of fiber coating thickness
and modulus (assumed to be proportional to the yield stress) on the room temperature
elastic properties of composites [Theocaris, 1992]. The types of fiber coating systems
considered include coatings, not exceeding 20% of the fiber radius, with a modulus

greater or less than the matrix. In the case of a coating stiffer than the matrix, the

27
composite longitudinal modulus was greater than the transverse modulus. For a coating

less stiff than the matrix, a fall in composite longitudinal modulus was seen as coating
thickness increased. Although the moduli of the uncoated fiber composites are slightly
higher than those of the matrix, the composite moduli always vary between the respective
moduli for the matrix and the less stiffer coating.

Other researchers have used finite element analysis to study the role of the
mechanical properties of the coating material in reducing the thermal residual stresses, by
applying a compensating, or compliant, coating to the fiber [Arnold, 1990 and 1992;
Morel and Saravanos, 1990]. This approach suggests that the two most important fiber
coating parameters for reducing the matrix and in-plane stresses are the coating-matrix
CT E mismatch, and the thickness of the coating. Of particular interest is the reduction of
the tensile hoop stresses, which are believed to be responsible for generating radial matrix
cracks. The calculated results, presented in Figures 6 to 8, indicate that the compliant
layer indeed reduces the radial and circumferential stresses within the fiber and the
matrix, but increases the longitudinal stresses. In addition, the mechanical properties of
the compliant layer, such as yield stress, strain hardening slope, and stiffness, dominate
the reduction of the interfacial stress state and have little, if any, effect on the fiber or
matrix stress state.

An analytical study on the reduction of thermal stresses in MMC by Jansson and
Leckie [1992] produced results similar to that of Amold et al. [Arnold, 1992]. In the
study by Jansson and Leckie, the CT E of the interface layer was determined to be the
most important parameter for reducing the matrix stresses. If the CTE of the interlayer
material is significantly higher (greater than 2 times) than either the matrix or fiber
(compensating layer), substantial reductions in the matrix stresses may be realized. For
most interiayer materials, the longitudinal matrix stress is the least affected, while the

radial stress is the most affected. Reduction of the hoop stress, close to the fiber can be

28
reduced substantially, suggesting that coatings may be successful in preventing cracking

in composites.

The theoretical studies mentioned above are all limited to room temperature
analysis. In addition to the temperature dependence shortcoming, a number of issues still
remain to be incorporated into the formalism, particularly the effects of fiber-fiber
interaction, spacing and volume fraction.

Warwick and Clync have presented an analytical model predicting the principal
stresses in coated fiber composites undergoing thermomechanical loading [Warwick and
Clyne, 1991]. A general trend of large residual stresses (300-500 MPa) generated at the
fiber-matrix interface leads to interfacial matrix plasticity both during heating and cooling
cycles. Metcalfe established a theoretical framework for understanding the effect of the
reaction layer thickness, and properties on the fracture behavior of Ti-B filament
composites [Metcalfe, 1974]. In this system, a fiber-matrix interfacial reaction results in
the growth of a Ti32 layer. The mechanical properties of the composites were analyzed
by assuming that the flaw size for crack initiation is proportional to the reaction layer
thickness and that cracks do not initiate in the matrix. When the reaction layer is < 1pm
thick, the size and quantity of reaction-layer flaws is less than the existing population in
the fiber and, consequently, fracture is controlled by the fiber. As the reaction layer
thickens, a critical point is reached where the stress intensity associated with the reaction-
zone crack exceeds that in the fiber and results in failure within the reaction zone.

2.3.3.2 Experimental Investigations of Fiber-Coated NiAl Composites Only a
few studies have investigated the effects of fiber coatings on the mechanical properties of
NiAl-A1203 composites. As an example, Misra et al. studied the properties of a NiAl-
(molybdenum-Saphikon) composite and found that, though the coating did not degrade
the fiber tensile strength, the coating did not protect the fiber from damage during
consolidation [Misra, 1993]. Furthermore, room temperature tensile tests revealed

debonding at the fiber-molybdenum interface, indicating that the Mo-NiAl bond is

29

 

 

 

 

 

 

 

 

 

 

 

 

 

 

70
L
O
0.100
.-.....-.. 0.075
0.050
0 Compliant
La er
1; ~70 /
E .1...
l-
O
-210
(I. e
um

Figure 6 Radial stress versus CTE and compliant layer thickness.

30

 

[Compliant
600 "' f Layer f

     

 

 

 

500 *-
é Matrix
0° ' mi L
3
0.100
0.075
................... 0.050

 

 

 

 

 

 

Figure 7 Circumferential stress versus the CTE and compliant layer thickness.

31

 

Compliant
Layer

Matrix

0.200
0.100
.......... 0,075
0.050

 

 

 

 

 

 

 

 

Figure 8 Longitudinal stress versus the CT E and the thickness of a compliant layer.

32
stronger than the Mo-A1203 bond. Additional evidence of low Mo-A1203 bond strength

was obtained when high-temperature tensile tests revealed composite modulus and tensile
strength that were less than bulk NiAl material. These results indicate that no fiber-
matrix load transfer occurs. In contrast to these results, Bowman and Noebe reported
prelinrinary fiber push-out results on nickel-coated A1203 fibers composited with NiAl
and indicated an increase in the interfacial shear strength [Bowman, 1990]. They
attribute this increase to the nickel coating reacting with the A1203 fiber and forming a

chemical bond.

2.4 Mechanical and Thermomechanical Behavior of NiAl and NiAl Composites

Significant efforts, reviewed by Amato and Yang, have centered on enhancing the
mechanical properties of NiAl through grain refinement, macro- and micro-alloying, and
incorporating a second phase [Amato, 1994]. The purpose of the following section is to
summarize some of the studies investigating the mechanical and thermomechanical
properties of NiAl-based composites. In addition to these discussions, the thermal
expansion behavior and residual stresses in IMCs and MMCs will be addressed.

2.4.] Mechanical Properties of NiAl-A120 3 Composites The mechanical
behavior of a composite can be classified according to either bulk or interfacial
properties. An evaluation of the interfacial shear strength, as measured by a fiber pushout
test, of continuous fiber reinforced NiAl-Saphikon composites by Noebe et al. found that
the NiAl-A1203 debond shear strength varied between 25-100 MPa, and that the bond
strength was purely frictional [Noebe, 1990]. Although improvement in the low-
temperature fracture toughness was achieved, little or no increase in the high-temperature
tensile strength was observed. Subsequent investigations indicated an interfacial shear

strength of 87 MPa at room temperature [Moose, 1990] and that organic binders applied

33
to the NiAl powder prior to processing the composites resulted in a weak bond, 50- 100

MPa, compared to the bond strength of virgin NiAl powder, which was 280 MPa
[Bowman, 1992]. Bowman also found that low-temperature fracture toughness was
improved by compositing, but that hi gh-temperature toughness was nil, due to the
brittleness and lack of interfacial bonding of the Saphikon fibers. Thermal cycling
experiments showed that a weak interfacial bond resulted in reduced oxidation resistance
[Bowman, 1992] and suggested that fiber coating(s) may be necessary to provide
adequate oxidation resistance.

Bowman et al. report that the mechanical properties of monolithic NiAl are
inferior to 25 vol.% NiAl-A1203 composites at high temperatures [Bowman, 1995].
Typical tensile curves acquired at 300 K and 1200 K are shown in Figure 9, with the
corresponding measured properties tabulated in Table 5. These results indicate that the
hi gh-temperature behavior of the composites are inferior to the high-temperature behavior
of the monolithic NiAl. Furthermore, both the composite strength (flow and fracture) and
failure strain were less than that of the monolithic material. A combination of poor fiber-
matrix bonding, and degradation of fiber properties, appears to be responsible for the lack

of mechanical property improvement at high temperatures.

Table 5 Property summary of NiAl and NiAl-A1203 composites [Bowman, 1995].

 

 

ngp FEAT NiAl-A1263

E(GPa) [Gym/[133) I gf(%) Er(GPa) lEn(GPa) I gx(MPa) l a;
—300 —241 I 120 I 0.4 262 170 73.6 0.17
2 0 >20

—276] —|103]—

 

 

 

 

 

 

 

34

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

000 400
300 - 300 ._
g g 1:1.
L L' ”n
g 200 £200 :
Ill) - 100 ..
E1
0 J 1 0 r 1
0.0 0.1 02 0.3 0.0 0.1 02 0.3
su-un (5) Strain (95)
(I) (c)
400 400
300 — 300 —
l 100 — g 100 -
100 100
0 l l o 1
0.0 10 20 30 0.0 s 10 15
Strain ('5) Shin (‘5)

(b) (‘0

Figure 9 Tensile response for NiAl (a, b) and NiAl-A1203 composite (c, d).

“pans:
on the 1
than
Solidifi
mimn
numen'

TYSOn

35

NiAl composited with 10 pm diameter A1203 (FP) fibers has been less thoroughly
investigated: Anton and Shah focused on the mechanical properties of hot isostatically
pressed NiAl-FF composites (Anton, 1992]. The bend tests obtained in this study showed
a substantial increase in tensile strength at room temperature, but only a slight increase at
1473 K. Together, these studies indicate that, though acceptable mechanical behavior
may be achieved in one area, deficiencies in the total composite behavior may exist,
implying that the fiber-matrix interface must be properly tailored to achieve full
realization of the required mechanical properties.

2.4.2 Thermal Expansion Behavior of IMCs and WCs The thermal
expansion behavior of IMCs and MMCs can provide a substantial amount of information
on the temperature dependence of the composite stresses and the level of fiber-matrix
adhesion. A three-part study on the thermal expansion behavior of directionally
solidified eutectics and monolithic 6061 Al provides a basis for understanding the
micromechanics associated with expanding composites [Gannong, 1974]. (Note that the
numerical results initially reported were incorrect and were, subsequently, corrected by
Tyson ['l‘yson, 1975].) The study by Garmong identified and modeled the temperature
dependent behavior of several effects typically observed in composite thermal expansion.
These effects include heating and cooling hysteresis, plastic Shakedownl , and a
cumulative net strain. Furthermore, when the composites were thermally cycled, the
mechanical properties tended to degrade through four basic types of damage:
reinforcement failure, specimen shape instability, matrix plastic deformation, and
physiochernical instability.

Another one-dimensional model, presented by Yoda et al., includes an expression
describing the fiber-matrix interface, which is presumed to consist of a fiber, a thin

interface region and matrix [Y oda, 1978]. The model limits the fiber to purely elastic

 

1 'I'hetermshakedownisusedintheanalysisofplasticallydeforming structures toindicateapproachtoa
system of residual stresses sufficient to inhibit further plastic flow.

36
deformation; allows the matrix to deform as an elastic/plastic body; and the allows the

interface to slide by Newtonian viscous flow. Comparison of the model predictions to
experimental results for a continuous W/Cu composite confirmed the predicted
dependence change in plastic strain on the fiber length, fiber diameter, and the volume
fraction. Wakashima and co-workers, on the other hand, assume an elastic perfectly
plastic matrix, with interfacial sliding, and predict that the net strain per cycle as the fiber
volume fraction increases, in agreement with data for the W/Cu system [Wakashima
1974].

As an example of a typical thermal expansion study, Warwick and Clyne
investigated the thermal fatigue of Mg-l lwt%Li/20vol%SiC whisker composites.

During thermal cycling, a progressive increase in composite length (strain ratchettingz)
occurred [Warwick, 1990]. Though an exact explanation of this behavior remains to be
made, the response is probably the result of asymmetrical stress relaxation processes
occurring during the heating and cooling cycle. Other studies correlate the thermal
expansion curves to the axial tensile behavior of the composite and indicate that the
plastic matrix accommodation occurring during thermal cycling controls the mechanical
response of the composite Wakashima, 1975].

Unlike the cases involving continuous fibers, the modeling of dimensional
changes in short fiber MMCs requires a three-dimensional analysis. Taya and Mori
proposed such a 3D model to predict the dimensional change of short fiber and particulate
reinforced metal matrix composites, using an Eshelby-type model in which the average
matrix stress is monitored at intervals during the thermal cycle, and the possibility of both
creep and plastic deformation of the matrix considered (Taya, 1989]. Table 6
summarizes the numerous models and the assumed deformation modes of the matrix and

interfaces. One common conclusion of these models (assuming perfect bonding) is the

 

2 Progressive length changes in aligned fibrous composites is referred to as strain ratchetting.

37
prediction of increases in the net plastic strain per cycle with increasing fiber volume

fraction, and fiber aspect ratio.

2.4.3 Residual Stresses in MMCs and IMCs The thermomechanical
response of either a MMC or IMC is a complicated interaction between the fiber-matrix
load transfer, the yield and creep behavior of the matrix, and the thermally induced
interfacial residual stresses. Two recent publications by Clyne and Withers [1993] and
Taya and Arsenault [1989] of the thermomechanical behaviors of MMC's illustrate the
veracity of this statement. Since fabrication of MMCs and IMCs occurs at high
temperatures, the resulting composites contain a significant amount of macroscopic
elastic stresses that are produced by differential thermal contraction strains after cooling
to ambient temperatures. These residual stresses are difficult to control, as they arise
from the inherent fiber-matrix CTE mismatch and can continuously regenerate during
thermal cycling.

To illustrate the origins of the residual stresses, Figure 10 shows a schematic
illustration of the in-plane (circumferential and radial) stresses occurring in continuously
reinforced MMCs. This schematic indicates that, upon cooldown from consolidation, the
circumferential (or hoop stresses) are tensile, while the radial stresses are compressive.
(In this discussion, as in most MMCs or IMCs, the CTE of the fiber is assumed to be less
than the matrix.) The magnitude of these stresses can be large and many times are greater
than the yield stress of the matrix. As an example, the interfacial stress state about a
spherical SiC inclusion in a titanium matrix is shown in Figure 11 [Clyne, 1993]. Two
cases are illustrated. The first case illustrates purely elastic behavior and displays the
magnitudes of the calculated stresses. The second case, which is more realistic, includes
a plastic zone around the fiber. The differences between the two cases include a decrease
in the magnitude of the stresses, along with a change in the sign of the circumferential

Stl'BSS.

38

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00:03.5 0:: :53: :0 000—2 8:09.82:

 

8.880
8830:: 30:080.
mom: 8:02:87. 0:: 3:85:03 0:83
. :0 .wfifi 0:980 3:0:0233D 80.8.0me 0:08am .3 :0 85F
£838 :3:
Z300": 0:083:00 0:: a 0:83 >
39 .9»:— -BI._ _ -3 :23 80002580 00:00am: .0:8% .3 :0 0 «P.
0:89:00 0:02: Er 0:8.
.8: a .2: 2-0230 388 team team! a:
883::
0:89:00 80:00:: 0:83
52 as... 2.0%“ .3... :20 05 920 sewn €2.39
0:89:00 i1
£2 :3... 053:8 as caste sea:
:3 :0 «60> :05? :23 3:080:25: 80:85 -0umafl .3 :0 80>
8:89:00
08: :8: 088265 008.:
.8080 32.0625..— 05 0:028:00 82D -0w8F 80:50
889$
00:80:31: 80u8=aa< 0:89:00 8me 8:. :3“ 000—2

 

.0022 0:: 0.03: E 880:0 3:280:50 “0:008 3 :08 20.00.: :0 8m 0 030:.

 

39

 

.8805 33:8: 0:295 :0 8%: 0:: Euro 05 :0 8:885 0:805» < OH 053":

 

8.38 :— 0..- 880.50
3:333:80 :50...
“008050 0:08.553 3:...— .5035:

5:... 55:. 0.... he... .3: 5.5 .8. :

                 
   

  

    

@— 8: n 83

«5x11577757

 

0:59—0:53. 0.52:0;th

./ / _///m
/m m

 

2.000...—

 

G— Sr: 58055—0 0— sh:

BEER—E03.
0:388... u h 2

oz

    

35808.:— H h. ~< £0.55. 30: %//%

yy/mmy yy
: i/ ,/

8w? 5::

 

 

 

 

 

23982.0 3V 23 830 3 05 .o 225; 55.3” a .o 22,. 863 2E : Ema

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

41
A schematic diagram depicting the origin of residual stresses in the axial direction

(parallel to the longitudinal direction of the fiber) is shown in Figure 12. In this case the
thermal expansion behavior and the residual stress state of an unbonded and perfectly
bonded composite are depicted during one complete thermal cycle. For the unbonded
composite case, the relative expansion differences between the two constituents appear at
different stages of the temperature cycle, but no stresses develop in either constituent.
When perfect bonding exists, interactions between the constituents cause changes in
thermal expansion and, hence, create stresses in each of the constituents. From the
diagram, the matrix is in residual tension at room temperature, while the fibers are in
residual compression. The evolution of these stresses is shown at different stages during
thermal cycling and indicates a stress free state and a temperature when the signs of the
stresses reverse. Upon cooling the original state of the stresses is recovered.

Residual stresses are traditionally categorized into three types according to the
length scale over which they act [Bourke, 1993]. Type I stresses act over distances
measured in millimeters and are often referred to as macroscopic. These stresses arise
when different cooling rates occur within a specimen (i.e., large temperature gradients
present over an extended area). Examples of Type I stress generation are forming and
welding operations. Type II stresses act over lengths comparable to the microstructure
(typically 1-100 um), while Type III stresses exist at an atomic scale, such as dislocation
pileups, and vary over individual grains. Type-II thermal stresses in MMCs and IMCs
result from a mismatch of CPR: or intergranular interactions between the fiber and
matrix. Type I stress may be minimized by cooling the composite slowly, but Type II
stresses are inherent and always produce stresses. In continuous fiber composites, Type
II stresses are generally dominant [Bourke, 1993].

The fiber-matrix interfacial stresses behave dynamically during thermal

excursions and relaxation processes take place to lower the internal strain energy. Table

42
7 lists several mechanisms that reduce the load carried by the reinforcement, and hence

can be classified as relaxation processes [Clyne, 1993]:

Table 7 Interfacial relaxation processes in MMCs.

 

 

 

 

 

 

 

jetsamphic Processes messes
Inclusion I-Jracture 5islocation motion
Interfacial debonding/sliding Dislocation rearrangement
Matrix cavitation Diffusion
_ Structural transformation
Recrystallization

 

 

 

 

The mechanisms on the left represent catastrophic processes and ultimately lead to
composite failure, while the right hand mechanisms are less detrimental to composite
behavior, but do reduce the composite load-bearing capacity. For any matrix-
reinforcement system, the relative incidence of these processes is a complex function of
variables such as temperature, inclusion aspect ratio, the applied load, and the rate at
which loads are applied.

The effect of the residual stresses on the thermomechanical behavior of well-
bonded MMCs is convincingly illustrated by the yield stress asymmetry displayed in
tensile-compression testing [DeSilva, 1969]. This asymmetry results when the fiber-
matrix CT E mismatch prestresses the matrix in residual tension and thus lowers the
macroscopic yield stress during tensile testing. Relaxation of these stresses occurs
through dislocation motion and rearrangement, diffusion, creep, structural

transformations and recrystallization [Clyne, 1993] that occur as the temperature

43
increases; all of these processes reduce the load transferred to the reinforcing phase.

Arsenault and Taya investigated the magnitude of the thermally induced residual stresses
in short fiber SiC-Al metal matrix composites by determining the difference in the yield
stresses (A0,) between tension and compression [ArsenaulL 1987]. A three—dimensional
theoretical model, based on the Eshelby method, predicted residual stresses, the
difference in yield stress, A0,, and matrix stress values to be less than the experimentally
determined values. The yield stress and the work-hardening rate in the compressive
stress-strain curve, however, were well predicted by the model.

Larsson [1978] studied the influence of thermal stresses on the mechanical
properties of tungsten fiber reinforced stainless steel metal matrix composites using
thermal expansion behavior. Two methods, with terms including plasticity and creep
effects, were derived for calculating the internal stresses appearing in fibrous composites.
The analysis showed that thermal stresses were of such magnitude that fiber damage and
matrix fatigue were possible when the composite was subjected to heating and cooling
cycles. Calculations using higher fiber volume fractions accelerated the amount of
thermal cycling damage, while for lower volume fractions, higher loads are imposed on
each fiber. For example, the compressive fiber stress at room temperature for Vf = 0.25 is
about 10 % higher than for Vf= 0.75. A similar effect can be seen on the maximum fiber
tensile stress around 800 K when the composite is heated. The experimental analysis also
indicated that the cooling and heating rates do not affect the expansion behavior

significantly.

3&5» 3.32 gasses. as so EMS 2: a 8:335 seasons. < 2 03mm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8:83am

 

8.5.»:
.5520

2

CHAPTER THREE
EXPERIMENTAL PROCEDURES

3.1 Description of Materials
3.1.1 PP and Saphikon Fibers Polycrystalline FP (Dow Chemical)

and c-axis single-crystal Saphikon " (Saphikon, Inc.) A1203 fibers, having the properties
described in Table 8, were used as reinforcements for a NiAl matrix. Tows of the FP
fibers were cut from a large spool, then ultrasonically washed in acetone for 2 hours to
remove any loose dirt, oil and contaminants from the fiber surface. The fibers were
covered and allowed to dry at ambient temperature for a minimum of 5 hours. The FP
fibers have a reported composition of > 99 wt.% A1203 with minor (0.01-0. 1%) impurity
elements of Mg, Si, Ca, Zr, and Fe [Pysher and Tressler, 1992]. The Saphikon fibers were
received with a ElS-LV hydroxypropyl methylcellulose (Dow Methocel T") sizing applied
by the manufacturer to minimize surface damage. The sizing was removed by soaking
the fibers in ice water for 1 hour and rinsing with deionized water according to the
procedure outlined by Traumbauer, et al. [1992]. The fibers were subsequently allowed
to dry for a minimum of 24 h. The cleaned and dried fibers were wrapped in acetone—
washed aluminum foil and stored in a vacuum dessicator until used.

3.1.2 NiAl Matrix Material The extruded NiAl material used in this study
was provided by Ron Noebe from the NASA-Lewis Research Center in Cleveland,
Ohio. The composition of the material was Ni-49.5 at.%Al, as determined by EDX
analysis, with no intentionally doped impurities.

45

 

 

 

 

 

 

Bab _ Give mw _ _ _

swam 02 02 353; Ev v.3 aoxmamm
mo _ om _ 8A_ we _omm.o QB _ pawlm _ E _ E

2:: A53 59 EB 39

cam 58620 mom? 6:903 25 2:82 3382

:65 EE $3 Eb 836m 325 88m same. .BE

 

 

 

 

 

 

.23: moxie seas: w 05¢

 

47
3.1.3 Coating Materials The fiber coating materials consisted of niobium,

titanium, and nickel. Both the nickel and titanium were in the form of 6 mm diameter rod
with 99.95 purity and were purchased from Aesar Chemical Co. These rods were cut to
alength 125 mm for use in a specially designed down-firing Thennionics electron-gun
(see Figure 3.2.2). The dimensions of the niobium (donated by Teledyne Wah Chang
Albany) were 75 mm x 100 mm x 3.17 mm with a nominal composition of 0.2 wt.% Ta,
0.05 wt.% W, and 0.02 wt.% Zr. Approximately 0.01 wt.% N, C, and 0 were also
present. Niobium rectangles, 5 x 75 mm, were cut from the sheet an used as feed stock in
the electron gun. The coating materials were lightly polished with 340 grit SiC paper to
remove the surface oxidation layer, and other gross impurities, then ultrasonically washed

in acetone for ~5 minutes.

3.2 Fiber Coating and Diffusion Bonding Procedures

3.2.1 Fiber Mounting Procedures The ends of 20.3 cm (8 in.) long fibers were
taped onto a aluminum frame. To ensure that the fibers remained in place during the
deposition process, the fiber ends were lightly coated with carbon paint as schematically
depicted in Figure 13. The carbon paint was allowed to dry for at least 12 hours before
connecting the aluminum frame to a mechanical feedthrough inside the vacuum chamber.
This connection allowed the fiber frame to be rotated during the coating process. The
geometry of the fiber coating deposition is schematically depicted in Figure 14.

3.2.2 Fiber Coating Deposition Process A custom designed rod fed, water-
cooled, Thennionics 3 kV electron-beam evaporation system was used to deposit the
coatings which fired downward at an angle of 30' to the vertical. The coating deposition
rate was controlled by adjusting the electron-beam emission current, the length of rod

protruding from the hearth, and the rod feed rate. Since deposition rates greater than 1

.ofisc 8:585? 5 9:0 @8558 33: .8 oumEonom ma 05mm

0 ’—
1 ‘
8m”

99a. 245-225

 

2.3. 89.5

 

 

 

 

 

 

.. m6

 

 

 

 

smack—39¢ 73.528:—
ufiufie 8 95950

Box—E .3 iii—mam 2:95 555:5?

h}:

'r‘n'

49
nm-S‘l generally resulted in large pieces of material spewing from the melt pool, rates

less than 0.8 nm-S'l, as measured by a Leybold Inficon film thickness monitor
periodically moved into position above the fibers, were utilized. The fiber holder was
rotated 45' every 3 minutes to expose different areas of the fiber to the evaporated flux of
atoms. This procedure resulted in an approximately uniform fiber coating thickness.
Typical evaporation times of 6 hours produced coating thicknesses of ~l—2 pm. The
vacuum chamber was pumped to a base pressure of ~1x10‘5 Pa, but increased to ~4x10’5

Pa during the depositions.

3.2.3 The Diffusion Bonding Procedure

3.2.3.1 NiAl Difiits‘ion Bonding Plate Fabrication Diffusion bonding plates
of NiAl were sectioned to dimensions of 1.5 x 1.5 x 0.2 cm and 0.7 x 2.7 x 0.2 cm
using a low speed diamond saw. The plates were polished using a sequential grinding
procedure. Initially, one side of the plates was mechanically polished with SiC papers
ranging from 240 to 600 grit. The plates were further polished using alumina grit
(600 grit, 5 run, and 0.3 pm) on polishing wheels covered with syvelt cloths.
Similarly, the other side of the plates were then polished to a surface finish of 5 pm.
These plates were then glued (cyanoacrylate adhesive), with the 0.3 pm surface down,
to a rectangular steel bar with parallel sides. The NiAl plates were then machined on
a surface grinder equipped with a 600 grit SiC impregnated grinding wheel. To
remove the plates after grinding, the steel bar was ultrasonically cleaned in acetone
until the adhesive holding the samples dissolved. This procedure produced diffusion
bonding plates with parallel sides.

Continued fabrication of the NiAl diffusion bonding plates consisted of using

a Materials Science Ltd. Model MK-II electro-discharge machine (EDM), with

Mechanical Feedthrough

 

‘—-Turbol'ump

Evaporated Flux
Shutter

Evaporant Material

Feedthrough
Mechanism

 

Water Cooling Lines

p
g:
E;
.. ->_p no t
t 1‘ '
J~\.Uh

)1. 9.).

‘ Feed Material
Ni, 11, and Nb

      

 

 

 

 

 

 

 

Film Thickness
Monitor

Vacuum Chamber

 

 

Coo-pump

 

Figure 14 Schematic of the e-gun, fibers, and film thickness monitor.

51
settings of R = 3, C = 3, and V = 4, and custom plunger tool, as shown in Figure 15,

to machine ~230 um diameter grooves approximately 190 um apart into the plates.
Figure 16 shows a portion of an EDMed bonding plate where it can be noted that the
grooves do not extend to the end of the plate. This design ensures that the fiber ends
are not directly exposed to the environment. After machining, the plates were
scrubbed with a nylon brush dipped in a household cleaner ultrasonically washed in
warm water for 3 minutes, acetone for 15 minutes, and hexanes for 15 minutes. After
the first and second washings, the plates were dried with a hot air gun, and allowed to
dry naturally after the last washing.

3.2.3.2 Fiber Lay-up Methods The fiber lay-up procedure for the FP-fiber
composites began with the positioning of ~10 fibers on the polished side of one of the
bonding plates using a small amount of double-sided tape as shown in Figure 17, to
ensure that the fibers remained in place. The polished side of the second bonding plate
was then placed on top of the first plate and pressed against the fibers. The resulting
specimens were approximately 9 mm wide, 18 mm long and 2 mm thick.

To fabricate the NiAl-Saphikon composites, two-ply rectangular specimens
(25 mm x 6 mm x 1mm) consisting of either 5 or 7 vol% were fabricated in a manner
similar to the foil-fiber-foil technique [Mackay, 1991]. The fiber volume fraction was
determined by counting the number of fibers, measuring the length of the fibers, and
assuming a fiber diameter of 140 am. The fiber lay-up procedure was performed
under a low power optical microscope and proceeded by placing individual fibers into
each groove until all of the grooves were filled. To keep the fibers in place during
handling, an extremely small amount of cyanoacrylate was placed on the fiber ends
using a sharpened toothpick. The three components of the composite were then
positioned together as shown in Figure 18. To prevent the N iAl plates from bonding
to the TZM compression platens, boron nitride coated alumina plates were placed

between the platens and the composite.

52

Attaches to EDM
(Anode)

  
   
  

Copper Tubing

/

‘—

Brass holder with
alignment screws

Stalnlas steel shim stock,

~125 pm (0.005")
thickness \

Lmfil

Figure 15 Electrodischarge machine plunger tool used to fabricate fiber channels.

 

Figure 16 SEM micrograph illustrating EDMed fiber channels in NiAl.

.EB: 3830:: one 08098 5.. Emma. a??— 85888 “on: an 5 853m

23: hr.—

EEoi

 

. . 5
usage-93:8

 

 

IA
’

 

 

\

3.8a... 5.5:. .52

55

.388.— mzrna— 85388 coca maximum 05 Co Eflwfie ouaaoaom mg paw—n.

 

#
i _

Alli. sargigse

 

 

 

son-.6

Emma“ s
/ \V
L“ a...

V Z

 

 

 

56
3.2.3.3 The Difi‘usion Bonding Process Table 3-2 shows the diffusion

bonding processing schedules carried out at a pressure of ~5x10'3 Pa in a Centorr
vacuum furnace coupled to a MTS 810 system. The heating rate was controlled at
~36 K—min‘l from room temperature to the maximum temperature, while the cooling
rate was ~34 K-min’l from the maximum temperature to 723 K, when the furnace
power was turned off and the composite cooled. To minimize slipping during the
heating stage of the process, a preload of ~0.5 MPa was applied to the couple. Once
the maximum processing temperature was attained and held for 5 minutes, the full
stress was then slowly applied over a period of 5 minutes. Just as the cooling cycle
began, the load was decreased to ~05 MPa to minimize composite fracture during

cooldown. Figure 19 is an optical micrograph of a typical NiAl-Saphikon composite.

33 Composite Thermal Cycling and Post-Testing Sample Preparation

3.3.1 Composite Thermal Cycling in the Temperature Range of 3 73 and 11 73 K
Composites with uncoated FF and uncoated Saphikon fibers composites were
consolidated and placed in a stainless steel wire gauze envelope, sewn together with
alumnel wire. The envelope was then suspended by alumnel wire attached to a motor
driven pulley system which lowered the specimens to the approximate center of a tube
furnace held at a constant temperature of 1173 K. After soaking for 5 minutes the
specimens were raised into a forced-air cooling chamber. The temperatures and cycling
limit were controlled by a CN8600 Series Process Controller from Omega Engineering,
Inc. The high-temperature and cooling soak times were 5 minutes. After cooling, the
specimen temperature was approximately 373 K as measured by a K-type thermocouple

attached to the specimen.

57
3.3.2 Composite Thermal Cycling in the Temperature Range of 373 to 1423 K

NiAl-Saphikon composites were fabricated as described in section 3.2 and with fibers
exposed to the atmosphere. The small hot zone of the heating lamps required
composite dimensions of approximately 7 mm x 5 mm x 2 mm. As schematically
depicted in Figure 20, the composites were positioned at the center of the exit hole of
the heaters and were surrounded by Kaowool" thermal insulation. A K-type
thermocouple was spot-welded directly to the specimen for accurate determination of
the specimen temperature. The thermal cycling experiments took place in ambient
atmosphere with direct radiant heat from two parabolically focused infrared spot
heaters (Research Inc.) outfitted with 750 Q/Cl quartz lamps. The cycling
temperatures and times were controlled by a CN8600 Series Process Controller
(Omega Engineering, Inc) connected to a 140 volt Variac and cycled between 373K
to 1423 K with a hi gh-temperature soak time of 5 minutes. The composites were
allowed to cool naturally and were thermally cycled 75, 150 and 300 times with

approximate moling rate of 569 K-min.‘l and a soak time of 5 minutes.

Table 9 Table detailing the diffusion bonding processing parameters.

 

 

 

 

 

 

 

 

s rme ll'emp. m “i omments
Name (hr) (K) ~ (MPa) l
D81 3 1623 0.82 20 Incomplete matrix
bonding
D82 4 1573 0.80 20 Lack of matrix
infiltration
DB3 4 1673 0.85 15 Lack of matrix
infiltration
DB4 4 1673 0.85 20 Fibers show
_ substantial Creep
DBS 4 1723 0.88 5 Complete bonding
for FP fibers;
minimal amount of
matrix deformation
DB6 4 1748 0.90 5 Saphikon Only

 

 

 

 

 

 

 

 

 

1 inch

 

 

 

Figure 19 Optical micrograph of a typical NiAl-A1203 dilatometry composite.

59
3.3.3 Electroetching Procedures Matrix removal was accomplished by

electroetching the surfaces transverse to the fiber direction as illustrated in Figure 21.
Initially, the composite was coated with Microshield lacquer to protect all surfaces
except one end of the composite. The specimen was then suspended into a beaker
filled with 10% perchloric acid and 90% acetic acid which was continuously stirred
by a magnetic stirring bar. The applied voltage was kept approximately constant at
12 volts, while the current varied (depending on the specimen dimensions) from 0.15

to 0.35 amps. Approximately 200 am of the matrix was removed so that the fiber

surfaces could be examined.

3.4 Post-Compositing Fiber Extraction and Fiber Fragment Length Determination

Coated and uncoated Saphikon composites were prepared as described in
Section 3.2.3.2. The diffusion bonding conditions were the same as for the
dilatometry composites described above. The composite length was 17 mm and with
10 fibers per specimen. The loss of some fiber fragments during filtering and
transferring was unavoidable, so to minimize these losses, two composites of the
coated fibers were tested, while three uncoated fiber composites were tested.

Fiber length distributions for as-composited and tensile tested composites
were determined by measuring the lengths of fibers etched from the composites using
a heated (340 K) bath of 50% water, 33% nitric acid, and 17% hydrochloric acid.
Higher temperatures were initially tried; however, the boiling etchant caused the
specimen to bounce, potentially fracturing the fibers and distorting the fiber length
distribution. Typical dissolution times were 5 h. After allowing the etchant to cool,
the mixture was filtered through a piece of Whatrnan filter paper. When the filtering

was complete, fibers caught in the filter paper and left in the beaker were carefully

Power Input
Air

Filtered Cooling

   

 

my

“viva“..wvaco. '4
a‘o’o’o‘o‘o’q’o‘o’c’o‘.‘
e o e e v o
19.9.5353
c.0400

e c o
v e o v e o
”gauge“

ct.

0‘0...
o-........

Infrared Spot Heater

Composite Specimen
Insulated with Kaowool

 

Aluminum Parabolic
Focussing Cone

'a'l.’.‘.’!.’a’.‘lx

23!.‘Ia‘fiflan‘lx'.

a‘.‘.‘.‘.‘

9

Figure 20 Schematic illustration of the high-temperature thermal cycling apparatus.

 

Infrared Spot Heater

Power Input
Filtered Cooling
Air

   

61

 

   
  

Mun-lieu Lacquer

NiAl Ctr-pasta
MW Stirrlag Bar

 

Etched NiAl

Figure 21 Schematic diagram of the electroetching procedure used to expose fibers.

62
rinsed with water to remove the remaining etchant. All of the fibers in the beaker

were then transferred to filter paper and dried.

The extracted fibers were carefully place into a 60 mm x 3 mm polycarbonate
petri dish lined with a 10 squares per millimeter grid. The fibers were then imaged
with an Olympus optical microscope outfitted with a Polaroid camera. Photos were
then taken at a 60 x nominal magnification until the entire set of fiber fragments were
photographed. The grid served as one calibration source, while a 10 mm rule was
periodically placed in a frame for additional scale confirmation. The fibers in the
photographs were then measured using a millimeter ruler with an accuracy of 0.5 mm.

Multiplying this value by the appropriate conversion factor provides the fiber length.

3.5 Fiber-Matrix Interface and Fiber Surface Morphology Characterization

3.5.1 Scanning Electron Microscopy The fiber surface morphological and
interfacial chemical characterization was performed using a Hitachi S-2500C scanning
electron microscope (SEM) equipped with a Link energy dispersive x-ray (EDX) analysis
system operating at energies of 5-15 ch. For both uncoated fibers and coated fibers
extracted from the matrix after consolidation by the process described in Figure 16. A
gold coat ~ 15 nm thick was sputtered onto the fiber surface to minimize charging effects.

3.5.2 Fiber Surface Morphology Characterization Using Atomic Force
Microscopy

Saphikon fiber surface scans were performed in height mode on a Digital Nanoscope III
atomic force microscope (AFM) with the scan range and the resolution (number of data
points per line) held constant. Control and composited fibers approximately 3 - 5 mm
long were extracted from NiAl composites and placed on a steel sample holder coated

with double-sided tape. On a captured image, the section command was used to display

63
the cross-sectional profile. Linear roughness measurements were made on the reference

centerline. The RMS (Standard Deviation) is calculated as

2
“km 3-,

N

 

where law is the average of the Z values between the scan length, Zi is the current 2
value, and N is the number of points. The mean roughness, R8, is the mean value of the

roughness curve relative to the center line and is calculated as

1 L
R. — Eflflxmx 3.2

where f(x) is the roughness curve relative to the center line and L is the length of the
roughness curve. On the same images, the roughness command was used to calculate the
area roughness for the entire image. Analogous relations for the area roughness

parameters were calculated using:

(Zi - Zav )2
RMS = N , and 3—3

L.‘-:

ffl(x.y)ldxdy. 34

 

.1...
1" L4.

where Zavg is the average of the Z values within the given area, Zi is the current Z value,

and N is the number of points within the given area.

64
3.6 Therrnomechanical Testing Procedures

3.6.1 Dilatometry: Physical Description and Measurement Technique As
shown in Figure 22, the Orton Model IOES-l dilatometer has a horizontal furnace design
which utilizes a fused silica pushrod to transmit sample length changes to a linear
variable differential transformer (LVDT). The heating rates of 2.54 K-min. '1, 3.75 K-
rrrin'l , and 4.54 K-rnin:l were controlled by varying the starting position of the variable
transformer which supplies the power to the furnace. The composite temperature was
measured using a platinum/platinum-10% rhodium thermocouple placed directly on top
of the composite. The outputs for both the temperature and the LVDT were connected to
an Omega A-D board on a dedicated IBM XT 80086 computer. Data was taken at 1
minute intervals for both the heating and cooling segments of each cycle. The cooling
segment of the cycle commenced when the furnace was automatically turned off after
reaching the maximum temperature. During cooling, the rate was not controlled, since
the dilatometer is not capable of decreasing the temperature, so the initial cooling rate

was very fast and decreased at lower temperatures.

3.6.2 Calculation of the Instantaneous CT E Instantaneous CT E
data (a = dec/d'l') were calculated by fitting a straight line through 15 strain versus
temperature points and assigning the slope of the line to the central temperature data
point. A linear least square program was written and executed with a tolerance of 0.05 .

The rule-of-mixtures (ROM) instantaneous CT E was calculated from

V V
__a E +a[EI t 35

a“ " Emvm + E,v, ’

where the subscripts m and f represent the matrix and fiber, respectively, E is the

temperature dependent elastic modulus and V is the volume fraction [Schapery, 1968].

65

.esaaafi 782 .082 8:0 as .6 8523 «a same

 

 

 

 

 

 

 

 

 

 

 

Asia-cm no coin—lam £382
S>A

 

— aleIadada1.1adadadc1a1a1aialalc1a aIaIIl10101odaaalalal01‘a ddddd
as 0 aa- a as. a can a can a sea a sea a as. 0 ace 0 can a sac

. O O O . . . . . . .
.C.:f.O.‘...................'.O.U...U..1

a a a a aaaaaaaaaaaaaaaaaaaaaa a a;
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ”coca-aoaoaaaca

 

 

 

: _
2.5. 5.3.5

 

66
3.6.3 Procedure jbr Fracturing NiAl-Saphikon Composites Fabricated

composites, and the gripping ends of steel tongs, were placed in a Ney Model 6-525
muffle furnace at a temperature of 1323 K for 15 minutes. One end of the composites
were grasped, while still in the furnace, with the tongs and fractured in bending mode
either transverse or parallel to the fiber direction. The low temperature fracture
specimens were produced by fracturing in bending mode at ambient temperatures.
Fracture specimens were coated with ~15 nm of gold and analyzed in the SEM.

The experimental procedures are summarized and the three areas of investigation

are schematically described by Figure 23.

67

douawumgfi £5 E 09.: mos—030a Econ—conga on. 00 page 30E MN 95?".

 

OOOOOOOOOOOOOOOOOO
. . . . PL P b . h b P . bl . . . A

 

 

 

 

 

 

 

 

 

 

 

”mum”. 3.508%
was.” Banach—Eek?» ”33.;
”$5.”. Essa—SEEP was;
......; hoe—9:25.302

 

 

 

 

OOOOOOO I I I ‘ ‘ ‘

 

 

 

 

«832—800 “0 Simeon
cesium HEB...
.

 

 

 

     

   

I... .0. 0 I'll. II...
«can». .m"gamma”.mammmummm.u.name were?
can.” u u ounce-accomuamuuauna 0 use.” sea.” 0a.. a cum

     

    

     

    

”cums” ...... ”Mme"

                

 

 

— 8:89:00 .0

8:2on ”3&0 anus...

 

 

 

aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

DhDDDDDDID bbbbbbbbbbbbbbbbbbbbbbbbbb

8:3an «wane sea w
u. 832: 83.828 .

 

 

 

 

  
   

.n.” gdé .u.“
mm augmaez ”moan—was— mm
academic

aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
...................................
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaa
cccccccccccccccccc

 

are:

aaaaaaaaa
aaaaaaaaa
aaaaaaaaa

 

 

 

 

.00....IOJ COIN-”C”.-
IIIIII

 

 

 

IIIIIIIIII
.. O

 

.....
It... II‘ICII.“ IIIIIIIIIIIIIIIIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

ill’lli}

 

 

 

moan-no.5
0250:0800-»<
.

 

 

 

 

 

_§ea€oo «3:38 _
.
a as .32 .2
a? 22m .80

 

 

 

 

 

 

CHAPTER FOUR
RESULTS AND DISCUSSION

4.1 Surface Morphologies of Uncoated and Coated Fibers

4.1.1 FP Fibers The SEM micrograph in Figure 24 illustrates that the as:
received FP fiber surface morphology consists of a cobblestone texture with no clearly
defined grain boundaries. E-beam deposition of the nickel, niobium, and titanium
coatings onto the FP fibers resulted in the conformally coated surfaces shown in Figure
25. All of the coatings appeared to have good adhesion to the fiber, and coating cracks
were rarely seen. Figure 26 shows a cross-sectional view of a fractured Ni-coated FP
fiber and illustrates a ductile coating with a thickness of ~1um.

4.1.2 Saphikon Fibers The surface morphology of a washed and uncoated
Saphikon fiber is shown in Figure 27. The low magnification view shown by the
micrograph in Figure 27a indicates a smooth surface, while the higher magnification view
in Figure 27b indicates a periodic ridge structure. The small ridges on the fiber surface
are oriented perpendicular to the c-axis and are presumed to be a consequence of the fiber
growth process. As a typical example of a coated Saphikon fiber, Figure 28a shows the
surface morphology of a titanium coated fiber. The numerous large panicles scattered
throughout the surface of the coating correspond to ”spitting” events that occur when melt
pool instabilities are caused by entrapped gases within the feed material, too high an e-
beam current, and feed material extending beyond the crucible. The higher magnification
micrograph in Figure 28b shows submicron grain-like regions. A crossectional view of a

fractured Ti-Coated Saphikon fiber is shown in Figure 29 and illustrates a coating
68

  

69

 

.-

.1‘.4. acts). 1...:

7? 4.... . s (is... Lune. {as
. a

I.‘ fly“ .I‘... §.- . \
...uQ... ensues. ... (is.
”rein“; £33.... M9? .90....

.vu- -
. . .Hr acetic... H:
........,s.....e v...
., a #5. ix: a... . 2;.
. Q E.“
$31K... .. T .r
in}... t?» .36.:
We?;.‘0~r~‘.b...§ . wt. . fl.
.f, ... a . . _
.alu.. .N‘ ’ kc—‘rtn

Figure 24 SEM micrograph illustrating the FP fiber "cobblestone" surface morphology.

70

 

Figure 25 SEM micrographs of a Ni-coated FP fiber.

71

 

Figure 26 SEM micrograph of the cross-sectional of a Ni-coated FP fiber.

72

“was

 

(b)

Figure 27 SEM micrographs of an as-received Saphikon fiber.

 

(b)

ted Saphikon fiber.

i

hsofaT

Figure28 SEM

74

 

Figure 29 SEM micrograph of a fractured Ti-coated Saphikon fiber.

75
thickness between 1 and 2 pm. No macroscopic cracks were observed in any of the fiber

coatings, and adequate fiber coating adhesion was achieved. These results suggest that

the Saphikon fiber coatings were adequate for this study.

4.2 As-Fabricated Composite Microstructure

Figure 30 shows a transverse SEM micrograph of NiAl-(uncoated-FP) composite
processed for 4 hours at 1723 K under an applied stress of 5 MPa. The FP fiber exhibited
a substantial amount of creep, under all processing conditions, resulting in ellipsoidal
cross-sections. A conservative calculation assuming an initial circular cross-section and
constancy of fiber cross-sectional area indicates a transverse creep strain on the order of
25% for the fiber shown in Figure 30. Table 10 indicates other diffusion bonding
conditions tried and lists comments. Lower temperatures and higher stresses, in general,
lead to incomplete matrix infiltration and lack of pore closure. Since diffusion bonding is
a uniaxial process, mass flow around the fiber is required to completely infiltrate the
fiber.

Figure 31 shows a transverse optical micrograph of an as-fabricated
NiAl/Saphikon composite indicating almost complete matrix infiltration and good fiber
alignment. No tangential matrix cracks were observed, in any of the composites, though
a sparse number of radial matrix cracks occurred in the Ni- and Ti-coated fiber
composites. In contrast to the FP fibers, no detectable creep of the Saphikon fibers
occurred. A polished longitudinal NiAl-Saphikon specimen, processed under an applied
stress of 7 MPa, at a temperature of 1723 K for a time of 4 hours, is shown in Figure 32,
and indicates the presence of some voids, but mostly macroscopically well-bonded

interface regions. No composite curvature in the thickness direction was noted, and

76

 

Figure 30 SEM micrograph of a transverse section of a NiAl-uncoated FP fiber.

 

Figure 31 An optical micrograph of a cross section of a NiAl—(Ni-Saphikon) composite.

i

.
.--a-t.$srl.~. .‘ '
. . _ .. ‘ '

v- _.-._-... . ‘
r-.' f'?““'"'"

.‘\ .-

_. L
- .. - —- “-H.
. , ‘

 

Figure 32 Optical micrograph of a longitudinal section of a NiAl-Saphikon composite.

 

Figure 33 SEM micrograph of an as-received FP fiber after submersion in etchant.

80
consistent with other composite processing schemes, the consolidated fibers are often

coincident with grain boundary triple points.

4.3 Extracted-Fiber Surface Morphology

4.3.1 Uncoated Fibers Figure 33 is an SEM micrograph of an as-received
FP fiber subjected to the same etching treatment as composited fibers. The absence of
pitting or surface morphology alterations illustrates that the etching procedure has no
observable effect on the fiber. In contrast to this result, Figure 34 illustrates many
changes in the surface morphology of an uncoated FP fiber processed for 4 hours at 1723
K with an applied stress of 5 MPa. In general, the effects of the consolidation were
dominated by extensive grain growth (~300 % increase), though fiber fracture in the
longitudinal and transverse directions occasionally occurred. The SFNI micrograph in
Figure 34b reveals that the grain boundaries remain cohesive and pitting is not observed.
(Note the particulate material dotting the fiber surface corresponds to NiAl particles that
were not completely dissolved during the etching process.) As illustrated in Figure 35,
lower processing temperatures, combined with higher applied stresses, resulted in severe
grain boundary decohesion and surface pitting. As a trend, the severity of the induced
fiber surface damage depends primarily on the consolidation temperature and the applied
stress, with the processing time exerting a lesser influence.

The fiber surface morphology of an extracted uncoated Saphikon fiber processed
for 4 hours at 1748 K and an applied stress of 5 MPa is shown in Figure 36. One
difference between the as-received fiber surface morphology and the as-processed surface
morphology is the appearance of thermally grooved grain boundary imprints, with widths
ranging from 1.2 pm to 0.33 inn, and an average height of ~287 nm. In addition,

numerous pores, ~ 350 nm in diameter, are randomly distributed on the fiber surface. As

81
in the FP fiber case, the exact cause(s) of the pitting is unknown. (To determine if the

etchant affected the fiber surface morphology, cleaned as-received fibers were also
subjected to the matrix etchant dissolution process, and no fiber surface morphology
alterations were observed.) Similar features have been reported by other workers
investigating A1203-NiAl composites fabricated by the powder cloth technique and by
binderiess techniques [Draper, 1994; Asthana, 1995].

4.3.2 Nickel-Coated Fibers The surface morphology of an extracted nickel-
coated FP fiber, consolidated for 4 hours at 1748 K and an applied stress of 5 MPa, is
shown in Figure 37a. Along with the increased grain size, a significant amount of
intergranular cracking has occurred. A small amount of pitting is also apparent, though
no obvious gross chemical reaction has occurred.

Figure 37b shows the fiber surface morphology of a Ni-coated Saphikon fiber
processed for 4 hours at 1748 K and a stress of 5 MPa. The micrograph displays a
random distribution of smooth and rough regions decorated with submicron pits.
According to Table 11, a spinel reaction product forms at high temperature from nickel—
A1203 reactions. (A large negative value of AGO is indicative of a reaction with a large
thermodynamic driving force.) The EDX spectrum, shown in the inset of Figure 37c,
contains a nickel signal, which indicates the possibility of a reaction product, though the
signal may correspond to small amounts of matrix material adhering to the fiber surface.
Since the consolidation temperature was 20 K above the melting point of nickel , and,
since only a finite amount of nickel was available for reaction, the Al203-nickel chemical
reactions probably occurred during the early stages of the consolidation process.

4.3.3 Titanium-Coated Fibers The SEM micrograph shown in Figure 38a
illustrates how the chemical reactions between the titanium coating and A1203 affects the
surface morphology of a FP fiber processed for 4 hours at 1748 K with an applied stress
of 5 MPa. The titanium-A1203 chemical reaction(s) are so extensive that the grain

boundary structure of the fiber is difficult to discern. Large pits appear both in the grain

 

(J ~’.‘.‘A>‘-‘;~‘.:i-"' ‘
(3 i 0 '31.! . sV,‘}‘,”

L‘ C.
q, \fok p D. x .o‘
”lie-"fiwa .~. . ~\ k P
. I - .7 . .‘Y,‘

 
 
  

Figure 34 SEM micrographs of an uncoated FP fiber showing grain growth.

 

Figure 35 SEM micrographs of an uncoated FP fiber consolidated with a high stress.

 

Figure 36 SEM micrograph of an extracted uncoated Saphikon fiber.

_—-—.-—

 

 

Figure 37 SEM micrographs of a nickel~coated (a) FP fiber, (b) a Saphikon fiber.

86
interior and at the grain boundaries, while smaller pits reside inside the larger pits. The

consolidation temperature was 125 K less than the melting point of titanium (1623 K).
Similar behavior was seen in fibers composited under different conditions, but in all cases
significant fiber-coating reactions occurred. A general survey of the various diffusion

bonding parameters indicates that the processing temperature was the most significant

parameter in determining the surface morphology, while the applied stress and time were

less important.

Table 10 Gibbs free energy for nickel, niobium, and titanium reactions with A1203.

 

 

 

 

 

 

 

 

 

 

 

Reaction AGRO (1000 - 1400K) (kJ) 156!— ’

(1/3)A1203 + Ti = TiO + (2/3) A] 3.49 (1300K) Bafin,
1973

_(2/3)Aizo3 + Ti = T102 + (43) Al 133.sT( 1300K) Barin,
1973

8T1+ A1203 =3T10,6700 33+fi13m -1807——.8+0.0167Tln(T) Kang,
_-0.367 T_ 1990

(1/3)A1203 + Nb = o + (23) Al 1327.5 - 0.01735 T Howe,
1993

Ni + (1/2)02 + (137111203 = -60.09 + 0.0044 T Lenev,
Nio-(4/3)Arzo3 1965

 

 

 

A SEM micrograph of the titanium-coated Saphikon fiber surface morphology
(consolidated for 4 hours at 1748 K under an applied stress of 5 MPa) is shown in Figure
38b and reveals a topography with relatively smooth regions surrounded by protrusions.
Figure 38c is an SEM micrograph of a smooth region of the fiber surface which shows
the presence of numerous submicron pits. Similar to the nickel-coating case, the
consolidation temperature was above the melting point of titanium (~55 K). This

suggests that the reaction product formation and roughening occurs during the early

512

ill

FL

pt:
0b

ma

U18

fibt
mo

Gib

87
stages of composite consolidation and terminates quickly since the titanium coating is

consumed. Table 11 lists three possible reaction paths between titanium and A1203. The
third reaction (8Ti + A1203 = 3Tio_570033 + 2Ti3Al), showing the formation of titanium
oxide and titanium aluminide intermetallic reaction products, is the most probable
because it has the largest negative free energy change [Wang, 1993]. Though the EDX
spectrum of the fiber surface shown in the inset of Figure 38b does not indicate the
presence of titanium, the reaction products may have been dissolved by the etchant.

Also, fibers from other processing conditions did show titanium present on the fiber
surface. The other two reactions have slightly positive Gibbs free energy, but trace levels
of impurities and higher temperatures, may make these reactions thermodynamically
possible.

4.3.4 Niobium-Coated Fibers In contrast to the titanium-coated and nickel-
coated fibers, a slightly different morphology develops for processed niobium-coated FP
fibers. The SEM micrograph in Figure 39a illustrates a typical surface morphology of a
niobium-coated FP fiber processed for 4 hours at 1723 with an applied stress of 5 MPa.
Fiber surface impressions, or possibly chemical reaction sites, ~0.25 am in diameter, are
present in the grain interiors, and some grain boundary decohesion is also present One
observation seen in this system, but not seen in the nickel and titanium system was the
large amount of residual niobium remaining on the fiber surface. This fiber coating
remnant is a result of the etching not attacking the niobium vigorously. Since the melting
point of niobium is 2750 K (1027 K greater than the processing temperature), the coating
material did not liquefy, but should have been sufficiently soft for creep and flow of the
material to occur.

Figure 39b shows the fiber surface morphology of a niobium-coated Saphikon
fiber. The small undulations shown by the micrograph in Figure 39b indicates that the
niobium coating caused significant chemical dissolution of the A1203 fibers. Though the

Gibbs free energy of reaction in Table 11 is positive, experimental reports indicate that

 

(b)

Figure 38 SEM micrographs of a titanium-coated (a) FF and (b), (c) Saphikon fibers.

 

(C)
Figure 38 (cont'd).

‘e'\
11

 

 

(b)

Figure 39 SEM micrographs of Nb—coated (a) FF, and (b) Saphikon fibers.

let

P61
U111
0011

The

00113

91
niobium reacts with A1203 at high temperatures by dissolving the aluminum and oxygen

atoms into solution with the niobium [Klomp, 1987; Ohuchi, 1991; Morozumi, 1981;
Turwitt, 1985], and, hence, no reaction zone forms [Burger, 1989]. In our case, the
intimate contact between the niobium coating and the A1203 fiber, coupled with the high
vacuum of the diffusion bonding process, promotes a favorable condition for a redox
dissolution reaction [l-lowe, 1993]. Fiber facetting is also observed in Figure 16c;
however, at this time the mechanism(s) for this observation is not understood, but could
be a consequence of the fiber manufacturing process. Another investigator studied the
interfacial behavior of single-crystal A1203 fibers composited with niobium, at higher
temperatures and pressures, but for shorter times, and fiber facetting was not observed

[Petrich, 1993].

4.4 Surface Roughness Measurements on Extracted Fibers

Since energy dissipation associated with frictional sliding of the fiber against the
matrix, is a major contributor to the toughness of composite, quantifying the roughness of
the interface topography is important. AFM microscopy was used to quantify the surface
roughness of the extracted Saphikon fiber fragments. A typical line profile of an as-
received and cleaned Saphikon fiber is plotted in Figure 40. Only small, somewhat
periodic undulations are observed. A typical line profile of the uncoated fiber shown in
Figure 41a, measured away from the grain boundary ridges, indicates only small
undulations. As shown in Figure 41b, the nickel coatings produced a roughness curve
consisting of large amplitude, long wavelength deviations separated by smooth regions.
The niobium-coated fiber line profile shown in Figure 41c has large, smooth, deviations.
In contrast to these three profiles, the titanium-coated fiber profile shown in Figure 41d

consists of a continuous spectrum of small amplitude, short wavelength, regions joined

92
by a small number of large amplitude asperities. All of these observations are

qualitatively consistent with the SEM results of Figures 36 through 39.

The linear and area surface roughness values are reported in Table 12. These data
indicate that the diffusion bonding process decreased the surface roughness of uncoated
Saphikon fibers from a mean roughness, Ra, value of 1.6lnm to 0.33 nm. This
comparison, combined with the surface morphologies shown in Figure 27 and Figure 36,
suggests that fiber dissolution occurred during the consolidation process. Though
thermodynamic calculations nrle out NiAl-A1203 reactions below 1583 K [Misra, 1986],
small amounts of interfacial impurities combined with a processing temperature of 1748
K may make such reactions favorable [Calow, 1972]. Table 12 also indicates that the
mean roughness of the titanium-coated fiber is ~490 times greater than the extracted
uncoated fiber, while the nickel- and niobium-coated fibers are ~150 times rougher.
These results are consistent with the SEM results and, in particular, indicates that
increasing the surface roughness increases the fiber coating-fiber chemical reactivity.

4.4.1 Discussion of the Correlations between Fiber Surface Roughness, Residual
Thermal Stresses, and Interfacial Mechanical Properties The interfacial parameters
contributing to composite mechanical response include fiber-matrix adhesion, residual
clamping stresses, and frictional stresses. Interfacial friction stresses occur when, after
debonding, fiber sliding causes interfacial irregularities to move against each other, while
the residual thermal stresses arise from fiber-matrix CTE mismatch. These stresses affect
the load carrying ability and pull-out behavior of the fibers and, hence, the composite
toughness. For both types of stress, the frictional sliding forces produce an interfacial
stress given by mammal = no, (where it is the coefficient of friction between the sliding
surfaces and a, is the radial stress). For an undisplaced fiber, 0,, is caused by residual
clamping stresses only, while for a partially debonded fiber, contributions from both

types of stresses occurs.

 

10—

 

Diaplacement (run)
>
i,
<1
i

 

 

 

Figure 40 Typical AFM line scan illustrating the undulating surface of a Saphikon fiber.

 

 

 

 

 

 

 

 

 

 

20.0 .
E Uacoated
: 10 ............ ..
a 0 \Lrfl M— : +~A _ .‘fi
g .10... ..........

.204) 1 .
0 2.5 5.0 7.5 10
Line Length (run)
30 :
Ni—eoamd

.
. .
q-...a--..-..........., .................... r. .......................................
. s
u .
I i
c a

 

 

 

 

 

 

Displacement (nm)
e
I
>
E:
F
P
‘4‘?-
'i’

 

 

.10 .-
40 .i
o 2.5 5.0 7.5 to run
U" Lastl- (ll-m)
10 1
Nb-coated

 

 

 

 

 

 

Displacement (nm)

 

 

 

 

 

 

 

 

 

”E

5

E

i

3 ;

g. .20 L 1 i

0 2.5 5.0 7.5 10 pm

Lille Lastl- (ll-III)

Figure 41 AFM scans of (a) uncoated, (b) Ni-, (c) Nb—, and (d) Ti-coated Saphikon fiber.

95
To assess the effects of the fiber surface roughness on the frictional stress, a

model developed by Kerans and Parthasarathy will be used [Parthasarathy, 1994]. In this

model, the radial frictional stress is related to the roughness according to:

(A)
= — , 4-1

0rough I

where
A - k(E‘—) , 4-2
Vf
and k = _.__E£!I__—_. 4.3
E,(1+Vm)+Em(l+ v,)

Here A is the effective roughness amplitude, r is the fiber radius, E; the elastic modulus
of either the fiber or matrix, and Vfand Vm the Poisson's ratio of either the fiber (1) or
matrix (m). (Note that this expression rigorously applies only after complete debonding
of the fiber when the relative fiber-matrix displacement is large compared to the half
period of the roughness and . It would be expected, however, that the roughness would
play a role even when the displacements are smaller than the half period of the roughness,
although the relative importance should decrease.)

An exact analysis of the thermal stress problem in IMCs usually requires a
rigorous numerical solution because the constituent phases, particularly the matrix, can
behave as an elastic/plastic/creep body under thermal loading. However, in this section
the main consideration will be an analysis based on an elastic body which provides an
upper bound on the actual thermal stress in the composite. Following the derivation of
the thermal residual stresses in a fiber embedded in an infinite matrix, the radial
component of the residual stress can be written as [T aya, 1989]:

(o, + GmX2Gm +321“)

0,: G,+G_+A,_ (a,-am)AT 4-3

 

where

V
Az—EL 44

'“ (1+ Vm)(l-2Vm)

and Gm is the shear modulus of the matrix. Following the analysis by Taya and
Arsenault [1989], the axial residual stress given by is:

V
0.. magi“ 4,,

C

where
AT>0, Aa- a, -a,,

and Ec is the rule of mixtures composite modulus.

Using equations 41, 4-3, and 45, and the material constants in Table 5 and Table
10, together with the average and maximum fiber surface roughnesses for each type of
fiber in Table 12, the frictional, radial and axial stresses were calculated and tabulated in
Table 13. The histogram in Figure 42a compares the frictional stresses calculated from
the average surface roughness of the uncoated and coated fibers. The calculated frictional
stress of 0.108 MPa for the uncoated fiber is too small to be apparent on this plot. In
contrast, the nickel- and niobium-coated fibers produce frictional stresses of ~20 MPa,
while the roughness of the titanium-coated fiber produces a frictional stress more than
twice this value. Using the maximum roughness values, the frictional stresses were
calculated and plotted in Figure 249b. A similar trend to Figure 249a is observed;
however, the frictional stresses are now above 100 MPa (except for the uncoated fiber).
Frictional stresses of this magnitude approach the NiAl tensile yield stress, which is

temperature dependent and ranges from 75 - 600 MPa [Noebe, 1993].

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sm." «.814 8:... u v.8“ oar" N84 38o

MNM oak .83 u 88 2w 9; _ -8522

88 mwNa o. N? _ N80.u 3N4 n83. _ 38o

ox: RMN e. 8 New 5: m. .5353.

o. No." 3.3 v on 8? N. Na. $3. 38

a .80 E o .8 .x m a. am 0.82 .332

8.9." 8.9. 8.3 _ 81 8. on 8.3

m. ON de 2 QM mm. o Rd 388:

8. on 8. Ba 2 a 38.1 8.2" Base
1 i1. 18.. :1 Ga NN1.N 1 wow 1

1p can

  

805.501 3.5.

Mat—E. :3— each...—

 

 

ERE 080838 00800 new 02000:: ~01. mos—m3 mmocnwaoa 0823030 2 203.

98
The analysis by Kerans and Parthasarathy indicates that the critical parameter

influencing the frictional stress is the ratio of the roughness amplitude to the fiber
diameter. This implies that the maximum frictional stress associated with displacing a 10
um diameter FP fiber would be greater than the stress associated with displacing a 140
um diameter Saphikon fiber with the same roughness amplitude. Though no roughness
measurements were made on the FP fibers, the SEM micrographs indicate that grain
growth and surface roughening produce roughness amplitudes greater than that measured
for the Saphikon fibers. This suggests that the frictional stresses during fiber pull-out
willbe greater for the smaller diameter FP fiber and, hence, may improve composite
toughness.

The thermal stresses tabulated in Table 13 represent ideal values, since no
allowances were made for an interphase region, such as a coating or reaction zone, and no
plasticity terms were included. The limitations to this calculation are obvious since a
value of 17.5 GPa for the thermal residual radial stress is 100 times greater than the room
temperature yield stress of the matrix. On the other hand, a value 194 MPa for the axial
residual stress is relatively close to the experimentally measured NiAl yield stress of 125
MPa [Noebe, 1993; Dutta, 1993]. A more accurate technique, based on the thermal
expansion behavior of a composite, which takes into account the elastic/plastic/creep

properties of the composite, will be presented in section 4-10.

4.5 Interfacial Chemical Analysis of As-Consolidated Composites
4.5.1 EDX Analysis of As-Composited Interfaces Figure 43a shows the
aluminum, nickel, and titanium x-ray line scans across a titanium-coated FP fiber-matrix

interface processed at 1673 K (the melting point of titanium is 1693 K) under an applied

 

Frictional Stress (MPa)
8

LLMLLJIIIJJLAIIILJIIJJJ

  
  

 

 

 

0
lines ted N iekel Titanium Niobium
Fiber Costing
(a)
500 b r r l

J l l I l l I 1 I I 1 l 1 l l 1 J A 1

100

Maximum Frictional Stress (MPa)

 

 

Uncoated N iekel Titanium Niobium
Fiber Costing

(b)

Figure 42 A histogram comparing the calculated frictional stresses.

100
stress of 15 MPa for a time of 4 hours. The full-width half-maximum of the titanium x-

ray peak is ~O.75 pm, and the signal disappears ~1.5 pm from the fiber interface.

In Figure 250b, the energy dispersive x-ray spectrum, acquired with the beam placed on
the matrix ~l pm from the fiber-matrix interface indicates a strong nickel peak at 7.48
keV, an aluminum peak at 1.48 keV, and a titanium peak at 4.51 keV. The peak at 2.12
keV originates from the thin gold film sputtered onto the sample to minimize charging
effects.

The spatial resolution of SEM x-ray analysis, 1 pm, makes an exact determination
of the diffusion of titanium into either the matrix or the fiber difficult; however, Weber et
al., performed a TEM investigation which indicated that limited titanium diffusion occurs
in both polycrystalline alumina (~0.8 um) and Saphikon fibers (~O.2 um) [Weber, 1993]
composited with Ti3Al processed at 1373 K for 2 hours. In the present case, the diffusion
bonding temperature was ~20 K less than the melting point of titanium. This higher
processing temperature increases the diffusivity of titanium into the matrix and the
reactivity of titanium with A1203. Since the original thickness of the titanium-coating
was ~l pm, the detection of titanium near the interface suggest the presence of reaction

products with the fiber or matrix or both.

Table 12 Calculated fiber-matrix interfacial stresses.

 

 

 

 

 

 

 

 

 

 

 

FrEr Frictional Stress Frictional Stress Axial Stress 531;] l
Coating (MPa), (Ra) (MPa), (Rum) (MPa) Stress
fi (GPa)
Uncoated 0.198; 4.34
Nickel 16.55 143. l 194 17.5
Titanium 52.79 883.8
iobium 17.77? 164.2 ‘

 

 

101
In contrast to composites processed at the lower temperatures, several attempts at

characterizing the fiber-matrix interfaces in 7-vol.% NiAl-(Ti-Saphikon) composites
consolidated at 1748 K resulted in a titanium signal localized at the interface with little, if
any, signal originating from the matrix. Since the diffusion bonding temperature was ~50
K greater than the melting point of titanium, the fiber coating liquefied and reacted
vigorously with the fiber. The liquid titanium phase, combined with the applied stress of
the bonding process, can be expected to be extruded along the interface and to diffuse
into the matrix material. This processing condition is anticipated to cause a transient
condition that terminates during the composite consolidation procedure.

Figure 44 shows an SEM micrograph with nickel, aluminum, and niobium x-ray
line scans of a Nb-coated Saphikon fiber, diffusion bonded for 4 hours at 1723 K (a
homologous temperature of 0.63) with an applied stress of 10 MPa. Here the full width
half maximum (FW HM) of the niobium peak is ~lttm. EDX analysis of Ni-coated fibers
was impractical due to the large nickel content of the matrix.

4.5.2 Discussion of Possible Fiber-Coating and Coating-Matrix Reactions
Engineering the microstructure at the fiber-matrix interface is essential for controlling
the mechanical behavior of the composite. When considering possible scenarios of the
interfacial processes that can occur during the diffusion bonding process, reactions
between the fiber and the coating, along with reactions between the matrix and coating,
must be considered. Some consequences of the fiber-coating reactions, like fiber surface
roughness and surface chemistry, have been previously described. These concepts will be
discussed further, and the coating-matrix reactions elaborated on.

Metal-A1203 reaction products forming a reaction zone between the fiber and the
matrix represent the severest case of metal-ceramic reactivity. The thickness of the
reaction zone plays a crucial role in altering the strength of a continuous fiber metal

matrix composite. For instance, if the reaction zone is thicker than a critical dimension,

 

 

Relative
Counts

 

 

 

 

 

Al
I I I T fit
1 2 3 4 5 6
E

lav
(keV)

(b)

Figure 43 (a) X-ray line scans and (b) EDX spectrum of a Ti-coated FP fiber interface.

103
the cracks (radial or circumferential notches) formed in the reaction zone upon loading

will effectively increase the intrinsic flaw population of the fiber and induce fracture.
This behavior will reduce the strength of the composite. If, however, the reaction zone
product is less than the critical value, the intrinsic defect population on the fiber dictates
the strength of the fiber. The reaction zone in this case may act to increase the fiber
matrix adhesion and, consequently, the strength of the composite. The critical thickness
is a function of several parameters, including the modulus of the fiber, the stress
distribution around the crack, the crack root radius, and the strength of the fiber
[Metcalfe, 1972].

Though chemical reactions may be thermodynamically favorable, the kinetics of
the reaction(s) must be investigated before proper interpretation of reaction zone effects
can be made. Most studies of metal-ceramic reaction zone thickness support a parabolic

growth rate model which can be expressed as,
x2 = 4Dt 4-6

where x is the reaction zone thickness, D is the diffusion coefficient, and t is the time
['faya, 1989]. The temperature dependence of the reaction zone thickness is contained in
the diffusion coefficient of Equation 46 Rewriting this equation, the explicit

temperature dependence is contained in an exponential term as,

x = CDthexp{%}, 47

where C is constant, Do is the preexponential term, Q is the activation energy controlling
the reaction zone growth, k is Boltzmann's constant, and T is the temperature. This
analysis illustrates the importance of the kinetics, and helps predict the metallurgical

stability of the interface. For the case of coated fibers, the thickness of the coating

104

 

 

Figure 44 SEM micrograph of a NiAl-(Nb-coated Saphikon) interface.

105
determines the ultimate extent of the reaction zone. Once the coating is consumed by

either the matrix or the fiber, the reaction terminates. Thus, a transient reactivity is
established during the early stages of consolidation and does not persist into the lifetime
of the composite. This approach provides a measure of interfacial metallurgical stability.

One of the most important classes of reaction at metal-ceramic interfaces is the
reduction-oxidation (redox) reaction with an accompanying dissolution of one or both of
the ceramic elements. The Ti-A1203 system serves as a model system with a strong
interfacial chemical reaction (i.e., the kinetic rate constant and the change in Gibbs free
energy is very large), where a reduction of the Al-O bonds occurs [Morozumi, 1981;
Turwitt, 1985]. Tressler et al. and Wang et al., have determined that the titanium reduces
A1203 to form a complex reaction zone consisting of two distinct zones; an inner zone
adjacent to the A1203 consisting of a TiO phase (Class I interface) and an outer zone of
Ti 3A] [T ressler, 1973; Wang, 1993]. The growth of the reaction zone was found to
follow a parabolic rate law and that aluminum diffusion through the reaction zone was the
rate determining step.

The alloying effects of the fiber coating materials and the NiAl matrix has been
summarized in section 2.3.1.3. Assuming that the titanium coating and the NiAl matrix
behave as a diffusion couple, the diffusion depth, x, of titanium atoms into the NiAl
matrix can be estimated using Equation 46 with D = 1 x 10-10 cm2/s and a time of one
hour (this time is substantially less than the diffusion bonding time, but should be
adequate for an order of magnitude calculation). Using these values a diffusion distance
is found to be 12 pm. The titanium diffusion into the NiAl lattice gives rise to an
additional degree of order in which the Al and Ti atoms occupy ordered sites on the
aluminum sublattice [Stoloff, 1984]. The resulting NizAlTi Heusler phase (8'), which has
better creep resistant properties than NiAl, may form a two phase alloy of NiAl (B) and B'
at the fiber matrix interface. This interphase region would be expected to exhibit an

appreciably greater resistance to creep than NiAl since the interactions of dislocations

106
with semicoherent interfaces and an intrinsically different glide behavior of the two

phases will involve processes that severely reduce the overall dislocation mobility.
Though the solubility of titanium in NiAl can be appreciable (~10 at. %) at high
temperatures (>1100 K), the solubility is limited to ~l at. % at ambient temperatures and
precipitation is expected when the compositions exceed this value.

The nickel-A1203 combination has a lower thermodynamic driving force for
reaction than the titanium-A1203 system, but still a significant reaction occurred as
evidenced by the change in fiber surface morphology shown in Figure 37. The most
probable reaction product is the nickel aluminate spinel (NiA1204) which forms a ternary
oxide (Class II interface) at the interface between Ni and a-A1203 [Williams, 1993;
Draper, 1994; Asthana, 1995]. Though the nickel-A1203 reaction is relatively well
understood, the effect of the spinel interphase on the mechanical integrity of the Ni-
Al203 interface is not [W asynczuk, 1987]. Some investigators claim that spinel
formation leads to strong interfaces [Klomp, 1985; Ohuchi, 1991], whereas others
consider it to have a weakening effect [Williams, 1993; Draper, 1994]. The discrepancies
between these studies indicates that different processing parameters and testing
configurations lead to disparate results.

When compositing NiAl with nickel-coated fibers, the nickel atoms simply
dissolve into the matrix and occupy aluminum lattice sites. The Ni-Al phase diagram
indicates a wide 8 phase field which can accommodate up to 60 at.% nickel. The
relatively small composition change at the interface is not expected to alter the
mechanical properties of the matrix since the dependence of the yield stress and
hardening rate for compositions near stoichiometry are weak [Noebe, 1993].

In most ceramic-metal systems only a few metals can bond to ceramics without
forming extended reaction layers at the interface. The Nb-A1203 system is one of the few
that serve as a model system [Howe, 1993; Petrich, 1993]. In this case, the reduction-

oxidation reaction results in the dissolution of both the O (interstitially) and Al

107
(subtitutionally) into the niobium, forming a solid solution of niobium with oxygen and

aluminum, but no reaction products [Ohuchi and Kohyarna, 1991]. This Class IV
interface, thus produces a very narrow ( one to two monolayers thick) fiber-niobium
interface. The dissolution of the oxygen into the niobium serves as a potent solid solution
strengthener. At high temperatures the solubility of niobium in NiAl approaches 10-at. %,
but at room temperatures the solubility is generally less than 2-at. %. Dissolution of the
niobium into NiAl produces a Laves phase (NiAlN b) with stress exponents in the 24

range.

4.6 Oxidation Behavior and Interfacial Characterization of NiAl-Saphikon
Composites

4.6.1 Results of NiAl-Saphikon Composites Ilrennally Cycled Between 373 K and
1173 K Figure 253 shows the fracture surface of a NiAl composite with (a)
exposed and (b) embedded fibers cycled to 1173 K for 1000 cycles. The NiAl-Saphikon
composites did not exhibit any fiber-matrix interface oxidation, but did exhibit extensive
growth of oxide needles on the grain boundaries. The oxide needle protrusions are more
clearly seen in Figure 45 to extend outward from the grains. This oxide scale
morphology is frequently observed in NiAl and is attributed to either 0- or b-A1203
metastable oxides [Doychak, 1989]. No dense oxide morphologies characteristic of or-
A1203 were observed. (Note that no electroetched specimens were investigated in this
temperature regime.) Because no interfacial oxidation occurred or mature oxide phases
in this temperature regime, the maximum thermal cycling temperature was increased to
1423 K.
4.6.2 NiAl-Saphikon Composites Thermally Cycled 75, 150, and 300 Times

Between 373 and 1423 K

4.6.2.1 Composites with Exposed Fiber Ends Figure 46 shows a SEM
micrograph of an uncoated exposed fiber thermally cycled 75 times. Continuous, thick
(>1 um), and adherent oxide scales have formed (A) at the matrix-fiber triple points,

108
while other regions have scales partially wrapped around the fiber circumference and then

extending down the length of the fiber (B). The micrograph also shows that the fiber
surface has some large gouges or pitted regions scattered on the surface. The origin of
the gouges is unknown, but may be generated either during the initial stages of the
consolidation process or by the oxide scale generating surface stresses which cause
chipping of the fiber surface. Furthermore, this micrograph indicates that the etchant
does not attack the fiber surface since etch pits were not observed. As shown by the SEM
micrograph in Figure 48, similar oxide scale formations were seen in the nickel-coated
fiber surface (A), and the fiber surface morphology is uniformly rough when compared to
the surface of the uncoated fiber shown in Figure 47.

Figure 49a shows an SEM micrograph of an exposed titanium-coated fiber
thermally cycled 75 times. Similar to the uncoated fiber, the titanium-coated fiber shows
localized oxide formation along the matrix-fiber triple points (A) and scale growth along
the circumference of the fiber (B). Figure 49b is a higher magnification SEM micrograph
that shows a rough fiber surface morphology decorated with submicron pits and a general
tendency of the roughness features oriented along the c-axis of the fiber. A rough surface
morphology is expected because of the hi gh-temperature titanium-Ale3 chemical
reaction described in Table 11; however, some features, may be an artifact of the etchant
attacking the Ti-A1203 reaction products. The increased fiber surface roughness
combined with Ti-A1203 and Ti-NiAl chemical reactions should promote fiber-matrix
adhesion.

Increasing the number of thermal cycles to 150 produces oxide scale development
on the circumference of the fiber end (A) as shown by the SEM micrograph of a titanium-
coated fiber in Figure 50a. In addition, a band of oxide scale (B) bridging the scales
formed along the matrix-fiber triple points. The EDX spectrum plotted in Figure 50b
obtained from the fiber surface indicates a strong titanium signal at 4.5 keV along with

the aluminum and oxygen signals at 1.49 keV and 0.52 keV, respectively.

109

 

.
. 7
: s.
t
. o

. I
. 6%”.

Figure 45 SEM micrograph of a thennally-cycled NiAl-Saphikon fractured composite.

110

 

Figure 46 SEM micrographs of an uncoated NiAl-Saphikon fracture surface.

1 l 1
This indicates that some of the titanium remains localized at the fiber-matrix interface

after 150 cycles. Qualitatively similar oxide scale development was also observed for the
uncoated and nickel-coated fibers as illustrated by the SEM micrograph of the nickel-
coated fiber in Figure 51. Due to the extensive amount of interfacial oxidation for the
exposed fiber end composites, thermal cycling was terminated after reaching 150 cycles.
4.6.2.2 Composites with Completely Embedded Fibers Embedded fiber
composites, thermally cycled 75 and 150 times, were examined and, in general, no
interfacial oxide scale was found. The SEM micrograph in Figure 52 shows a small
amount of oxide scale formation on a nickel-coated fiber. Since this was the only oxide
seen on any of the fibers tested, it is assumed that it resulted because of incomplete
consolidation. An SEM micrograph of an embedded titanium-coated fiber surface
thermally cycled 150 times is shown in Figure 53 and, though no oxide scales were
present, titanium-rich areas exist at the triple point region (A). Examination of
composites thermally cycled 300 times did not reveal any oxide scale formation. These
results indicate that embedded fibers are sufficiently protected from oxidation processes.
Another aspect of the thermally cycled composites investigated was the fiber
surface morphology after undergoing 300 cycles. The surface morphology of an
uncoated fiber is shown by the micrographs in Figure 54. These micrographs indicate
surface undulations, some gouging (left side of the fiber) and, in general, a rougher
appearance than the surface morphology of the exposed fiber end shown in Figure 47.
The SEM micrographs shown in Figure 55 illustrate the rough surface morphology of a
titanium-coated fiber thermally cycled 300 times. Comparing this with the surface
morphology of the exposed titanium-coated fiber end shown in Figure 50a several

differences are seen.

112

 

Figure 47 A SEM micrograph of an exposed uncoated fiber thermally cycled 75 times.

113

 

Figure 48 A SEM micrograph of an exposed Ni-coated fiber thermally cycled 75 times.

114

 

Figure 49 SEM micrograph of exposed Ti-coated fiber thermally cycled 75 times.

115

 

 

ii

 

 

 

 

Figure 50 (a) A SEM micrograph and (b) a EDX spectrum of a Ti-coated fiber.

1 16
To assist in determining the possible oxygen routes to the fiber-matrix interface of

embedded fibers, a composite thermally cycled 300 times was fractured and examined.
Figure 56 is a SEM fractograph of the interface and grain boundary regions of an
uncoated fiber. Though no interfacial oxide scale is present, extensive grain boundary
oxidation (identified by the whisker formation) indicates that oxygen diffusional paths are
present at the grain boundaries during thermal cycling.

4.6.3 Discussion of the Oxidation Behavior of Thermally Cycled NiAl-Saphikon
Composites The above results indicate that the two thermal cycling temperature
regimes produced different oxide scales at different locations in the exposed fiber
composites. For instance, after thermally cycling to 1173 K for 250 hours, the NiAl-
Saphikon exposed-end fiber specimens did not exhibit any fiber—matrix
interfaceoxidation. Only metastable 6-A1203 and 0-A1203 oxide scales lining the grain
boundaries were observed in this temperature regime. The 1173 K thermal cycling
experiments indicate that exposing NiAl-Saphikon composites to these temperatures for
250 hours is not sufficient to cause a-A1203 formation either at the grain boundaries or
the interface. Also, the presence of metastable intergranular oxide scales suggest that
oxygen diffusion along the grain boundaries is relatively rapid. These results are
consistent with other observations reporting that metastable oxide scales transform to a-
A1203 after an incubation period that is dependent on time, temperature, and other
oxidation variables (percent of oxygen in the atmosphere). As an example, Doychak et
al. observed only a-A1203 when oxidizing NiAl single crystals at 1373 K for 1 hour, but
O-Al 203 for 0.5 h at 1373 K [Doychak, 1989].

Increasing the maximum cycling temperature by 250 K to 1423 K, while
decreasing the total soak time, produced dramatic differences in the interfacial oxidation
behavior. In particular, mature a-Al 203 interfacial oxidation forms in composites with
exposed fiber ends. Grain boundary oxide scales seem to be absent or at least show no

maturation. To provide some physical insight into the interfacial oxidation process, the

117

 

Figure 51 A SEM micrograph of an exposed Ni-coated fiber thermally cycled 150 times.

118

 

Figure 52 A SEM micrograph of a Ni-coated fiber thermally cycled 150 times.

119

 

Figure 53 A SEM micrograph of a Ti-coated fiber thermally cycled 150 times.

120

  
 

Figure 54 SEM micrographs of an uncoated fiber thermally cycled 300 times.

121

 

(b)

Figure 55 SEM micrographs of a Ti-coated fiber thermally cycled 300 times.

122

 

Figure 56 A SEM f ractograph of a uncoated embedded fiber thermally cycled 300 times.

123
schematic shown in Figure 58 shows a likely sequence of events that occurred during one

thermal cycle [Doychak, 1992].

Due to the CI‘ E mismatch between NiAl and A1203, the matrix expands during
specimen heating to a greater extent (~6.5%) than the fibers. The fiber-matrix adhesion is
incapable of sufficiently constraining matrix expansion, therefore a cylinder of matrix is
exposed to the oxidizing environment, and an oxide scale is formed. During cooldown,
the matrix wants to contract back to its original position, but the cylinder of newly-
fonned oxide impedes the replacement Providing the oxidized matrix can slide back
around the fiber, the oxide acts as a wedge, putting the matrix in tension and causing
crack propagation along the interface.

Though only a small number of thermal cycles were performed, it is clear that
composites with uncoated or coated fibers exposed directly to the environment quickly
form interfacial oxide scales. Thus, though the fiber coatings may enhance fiber-matrix
adhesion, the increase is not sufficient enough to overcome the oxidation processes
associated with the differential strains and the interfacial imperfections (e. g., grain
boundaries, voids, etc.). Examination of the fiber surface morphology after thermal

cycling seems to suggest that wear occurs during the thermal excursions.

4.7 Assessment of Consolidation Induced Fiber Damage for Uncoated and Coated
Saphikon Fibers

4. 7.] SEM F ractographic Study of Extracted-F iber Fragments The SEM
fractograph of an uncoated fiber fragment shown in Figure 59a indicates that the fracture
initiated at the fiber surface (denoted by the arrow). Extensive examination of fiber
fragments failed to reveal preferential crack initiation at the grain boundary ridges or the
pores. In all cases, the rhombohedral planes, {1012}, were the primary cleavage plane
and rarely was basal plane, {0001} cleavage observed.

Fiber-Matrix
Interface

Fibers

Lattice
Diffusion

Matrix Cracks

 

 

 

Matrix Expansion
Exposes Virgin Metal
to Atmosphere

Figure 57 A schematic illustrating the possible oxygen diffusion paths in composites.

125

Matrix Fiber End

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fiber Earl

Matrix

 

Figure 58 A schematic of oxide scale formation in composites with fiber ends exposed.

126
Figure 59b and c show SFJVI fractographs of niobium-coated fibers. In general, as

shown in Figure 59b, the fracture cleavage originated at the fiber surface and propagated
along rhombohedral planes in a manner similar to that observed for the uncoated fibers.
Occasionally, a combination of basal and rhombohedral cleavage fracture occurred as
shown by the SEM fractograph in Figure 59c. Though fiber facetting resulted in acute
edges along the fiber length, these areas did not appear to be preferred crack initiation
sites.

Figure 59d and e are SEM fractographs of extracted nickel- and titanium-coated
fibers, respectively. Similar to the uncoated and niobium-coated fibers, the fracture
initiation sites occur on the fiber surface. Furthermore, longitudinal cracks along the c-
axis were occasionally seen.

The propensity f or rhombohedral cleavage, combined with identification of the
basal plane as the easy-cleavage plane above 973 K [Iwasa, 1984], suggests that fracture
propagation occurs during a low temperature segment of the consolidation process. Two
possible causes of low temperature fracture propagation are the presence of large residual
radial and hoop compressive stresses (~100 MPa), and large tensile axial stresses (~100
MPa), arising during the cool down segment [Bowman, 1994]. Another factor, which
may produce bending stresses at low and high temperatures and, hence, promote fiber
fracture, is the nonuniform loading of the composite because of platen misalignment or
matrix plate machining inaccuracies. The importance of these effects in producing fiber
breakage is difficult to estimate, but the bending stress contributions were minimized by
following careful experimental procedures.

4.7.2 Extracted-Fiber Fragment Distributions A histogram of the extracted-
fiber fragment length distribution is plotted in Figure 60, while a summary of the
distribution is presented in Table 14. These data indicate that no f nil-length fibers
survived the consolidation process and that the average extracted-fiber fragment lengths

were substantially less than the original 25 mm. Furthermore, ~70% of the fiber lengths

127

 

(b)

Figure 59 Fractographs of (a) uncoated, (b) and (c) Nb—, (d) Ni-, and (e) Ti-coated fibers.

128

 

Figure 59 (cond't).

129

 

(6)

Figure 59 (con't).

130
for the nickel- and titanium-coated fibers were in the 0-5 mm range, compared to ~35%

-40% for the uncoated and niobium coated fibers, respectively. Also, only 5% of the
nickel- and titanium-coated fibers were longer than 10 mm, while ~ 30% of the uncoated
and niobium-coated fibers fragments were found in this range. Examination of Table 14
identifies two fiber fragment distributions. The first distribution consists of uncoated
and niobium-coated fibers having approximately equal numbers of short, intermediate,
and long fragments. This distribution is represented by weak interfacial chemical
reactions and no reaction product formation. The second distribution represents the more
reactive nickel and titanium fiber coatings which tends to produce a gradual increase in
the number of fiber fragments with decreasing lengths as the fiber fragment length
decreases.

Clear correlations between fiber surface parameters and composite strength,
modulus, or toughness are difficult to obtain, but fiber-matrix chemical reactions
generally affect the composite mechanical response in two ways. First, increasing the
surface flaw population of inherently brittle material such as A1203 decreases the fiber
statistical strength [Curtin, 1991]. Secondly, fiber damage during the consolidation
process reduces the fiber aspect ratio, l/d (fragment length / fiber diameter) which
decreases the effectiveness of the fiber-matrix load transfer.

One approach of assessing the effect of fiber surface damage on the tensile
strength of the fiber is provided by using

I
2

a; = (4.)(5123 , 43
1.12 we I
where Ef is the fiber modulus, Gc’ is the strain energy release rate of the fiber, given by
(Kc)2/Ef, and c is the flaw size [Ochiai, 1981; Ochiai, 1988]. Using the average
extracted-fiber strengths of 1.2 and 1.6 GPa reported by Draper and Locci and a critical

stress intensity factor, KIC , for the rhombohedral f racture of 2.38 MPa-m05, the critical

131

 

 

d J m j q d d 4‘ A d d 1 1 4 d d d T II III I Ill
1.
IIIII;
.\\\\\\\\\\\\t
I
n n m " .\\\\\\\\\\\\\\\.
i .m .m b .m
a. I n I
F F F \\\\\\\\\\\\\\\\\\\\\
Tl
d I I a I .
d .w \\\\\\\\\\\\\\\\\\VV.
t a
I u u t
m a u a a a a a a a a.
c o \\\\\r\l\l\l\l\l
I c ..e b .... .
_I II II
“U NH NH TI '1’
.\\\\\\\\\\\\\\\
TI
'8 g a Ina-aaaaaaa.
.l/IIII/l/llI/I/I.
\\\\\\\\\\\\\\\\\\\\s
it
.aaaa-aaa-ua-a-alalal.
till)?
‘\\\\\\\\\\\\.
T
.a-aaaua-alI-aaa.
I’llIllIII/Illil/Ililllllfilliifi
\\\\\
.a-a-I-aluulaa.
r”””””’ly’.
\\\\\\\\N\\\\§§\\\\\s
can.aa.-aeolian-.IaI-I-III-Iau.
.II’ll/Ill!(Illllltlltlllllla
\\\\\\\\\\\\\ \§\\\\\\\\S.
a-aaaaa.aaa-aa.lea-aa-anoIaaaIlla-IlaIaaI-Iallaaala.
.II/ll7727777277777733737337377777777727
.\\\\\ \\\\\\\\\
vl
sea-a-aan.anno.u-aaan.auna-annual-anu-ulna-lunaI-a-I-lloaaal.
Ill?!I(III!!!IllII’ll3IIVIIllIVIII7I??/7¥I¥?7177¥73777777l€
s\\\\\\\\\\\\\\\
l.
aa-aannan-au-announce-nasal..-nan-Ila...
IIIIIIIIIIIIIIIIIIIIIIIIIIIIII’ll/Illllllllllllllla
.\\\\\\N\\\\\\\\\\\\\\\\\\\\\\\\\\\\N\\\N\t
VI.
.3
b L _ bl.— » p _ _ _ p p _ _ _ _ . _ P1— . _ “KKK“

 

u

Fiber Fragment Length (mm)

0123 456 7 89101112131415

«0.. 5 0 5 0

l

1
59.3 sea .35 8 28.3.—

Figure 60 Histogram of the extracted-fiber length distribution.

132
flaw size for fiber failure at the measured strength is found to range from 0.9 pm to 0.4

pm [Draper, 1994; lwasa, 1984]. The analysis of the fiber surface morphologies
indicates that the majority of the coated fiber surface flaws are on the order of or less than
these values. This observation suggests that the coated fibers have an in situ strength less
than 1.2 GPa which is ~50% decrease from the reported as-received fiber strength of 2.4
GPa [Draper, 1994].

4. 7.3 Correlation Between the Fiber Fragment Distributions and the Composite
Strengthening Ability The next concern is to determine how effective the
fragmented (both coated and uncoated) fibers are in transfen'ing load when compared to
the original continuous state. To evaluate this, a critical fiber length, lc, was calculated

using [Broutmann, 1967]

or

— 4-9
21:

L =

df
where df is the fiber diameter (~13O pm), OflS the fiber fracture stress, and 1: is the matrix
shear stress (33) or interfacial bond strength (whichever is lower). The value used for 1:
is crucial because interfacial debonding occurs if the interfacial shear stress reaches the
debond stress before either matrix yielding or fiber fracture occurs. If it is assumed that
1:” equals half the tensile yield stress, then 13,5 for the NiAl matrix is ~60 MPa at room
temperature and will be even lower at higher temperatures. Using an extracted fiber
strength of 1.5 GPa (~ 50% room temperature fiber strength), results in a predicted fiber
critical length of 1.6 mm. From Figure 256, ~82 % of the fiber fragments are greater than
2 mm, suggesting, that for all coatings, many fibers can effectively contribute to the

composite strength.

133

 

 

 

 

 

“cos—ween—
Elmmwra 3m 8.28.4 8L 8.» 3E .92
LED om o.m fl - S

 

 

 

 

 

 

 

  

Secs—mam Sci .88. .0 2.85m

 

 

.mcogeEfl—u Beams: eoncéofimbxo Co $885 9 2me

 

134
The effects of the fiber lengths on the composite strength is given by (assuming

perfect fiber-matrix bonding) [Chawla, 1987]

(1- v,). 410

Using the average fiber lengths listed in Table 14, a fiber critical length of 2 mm, a 7%
volume fraction, a NiAl flow stress of 160 MPa, and a 347 MPa for a continuously
reinforced composite, the percent strengthening capabilities of composites with
consolidation induced fiber breakage are listed in Table 15. These values indicate that
the uncoated, nickel-coated, and niobium-coated fiber composites will have 90% or more
of the strengthening capability of the continuos fiber composite, while the titanium-
coated fiber with an average fragment length of 3.6 mm was 87%. Decreasing the
average fiber length 50% and then 100% shows a significant degradation in the

strengthening ability of the nickel- and titanium-coated composites.

Table 14 Calculated percent strengthening capability for fragmented fiber lengths.

Coated

 

Alterations in the composite mechanical properties due to consolidation induced
fiber aspect ratio changes can be also analyzed using the frequency of fiber fragments vs.
the fiber aspect ratio shown in Figure 61 and summarized in Table 16. The generally

accepted guideline for achieving rule of mixture predictions requires the fibers to have an

135
aspect ratio of at least 25 [Clyne, 1993]. Table 15 shows that approximately 75% of the

uncoated fiber fragments and 65 % of the niobium-coated fiber fragments have an aspect
ratio greater than 25, while the nickel- and titanium-coated fibers show only 50% of the
fragments having an aspect ratio greater than 25. Assuming perfect fiber-matrix bonding,
this result suggests that composites fabricated with uncoated and niobium coated fibers
should show slight deviations from ROM predictions, while almost one-half of the
nickel- and titanium-coated fiber fragments may not be fully loaded and, consequently,

may not strengthen the matrix as effectively as niobium coated fibers.

4.8 Fracture Behavior of NiAl-Saphikon Composites

4.8.1 Room-Temperature Composite Fracture Behavior Room temperature
fracture specimens of NiAl-FF and NiAl-Saphikon composites were investigated using
scanning electron microscopy. Figure 62a and 4—62b shows the brittle fracture surface of
NiAl-(uncoated-FP) and NiAl-(Ni-FP) composite, respectively. The fiber-matrix
interfaces surrounding both types of fibers appear cohesive and no distinctions between
coated and uncoated fibers are apparent.

The SEM fractograph in Figure 63 shows the room temperature fracture surface of
a 7 vol.% NiAl-(uncoated-Saphikon) composite dilatometry specimen thermally cycled 3
times from room temperature to 1460 K. In general, the brittle fracture surface showing
transgranular cleavage corresponds to that for a typical bcc metal. Also, the fiber-matrix
interfaces appear cohesive, and no interfacial gapping is present Further investigation of
the fracture surface is shown in Figure 64a and illustrates an interface along a

longitudinal section of the fiber. The fiber surface appears relatively smooth.

136

 

25 l I l T l l l l i

: E33 Niobium-coated Fiber
.. I Uncoated Fiber

8

_
0r
UI'YV

.1 .
. i

1 l l l 1 1 l I l A l l .l l l l l L

 

Percent of Total Fiber Fragments

 

   

mm M E‘

20 30 40 50 60 70 80 90
Fiber-Fragment Aspect Ratio (1M)

(20

 

 

 

25 _ T I I I I f I r r 4

§ Ea Titanium-coated Fibers *

i so - E I Nickel-coated Fibers .1
h ’ E q
'6 15 *’ _
.‘E ’ é a
In. g I
‘3 ' ~
i- lob .
a- ' . .
9 5 g ..
= i: : i .1
5 s ' 5 2 E: a
r I I: : : 1 = :o 't

0’ 8.31
0 10 20 30 40 50 60 70 80 90

FlbepFragrnent Aspect Ratio (l/d)
(b)

Figure 61 Percentage of the number of fiber fragments vs. the fiber fragment aspect ratio.

137

 

 

 

 

 

 

 

 

 

 

Se.
MN a n8 NN « Q8 2 a NLPN on a For. 882 .NE
IE m.m mo WE 8. - 8
4.3 a: me We 8 - pm
3N mm _.mm WNN a. - mN
. RN- va. . IFNM m.NN - -88 -

 

 

 

       

 

880.02 _

flees—ween— aBE .309. 0 E83;

   

 

 

 

23m 8093
Seaman 82E

 

 

 

 

80:8 Son—mm EoEwmc has: 36838 .0 5885 3 033.

138
The high magnification fractograph of a fiber-matrix interface in Fr gure 64b shows a

small interfacial void (A), but otherwise, intimate contact between the fiber and matrix is
apparent. The angular step morphology indicates that brittle cleavage fracture occurred
near the interface, though a small amount of quasi-cleavage, with tear ridges, is seen at B.
Occasionally, grain boundary oxidation, identified by a continuous rough surface
morphology, was seen on grains located near the edge of the composite.

The fracture surface of a 7-vol.% dilatometry specimen composited with niobium-
coated Saphikon fibers and thermally cycled 4 times from room temperature to 1450 K is
shown in the SEM micrograph of Figure 65. Trans granular brittle cleavage fracture is
clearly evident (A). The absence of matrix plate interfaces indicates that complete
consolidation of the matrix plates occurred. Also shown in this micrograph is grain
boundary oxidation (B). Figure 66a shows a rough fiber-channel surface morphology
oriented parallel to the fiber axis. A closer inspection of the fiber-matrix interface is
shown in Figure 66b and illustrates intimate contact between the fiber and the matrix. In
addition, the matrix material next to the fiber is devoid of cleavage steps and appears to
consist of slip lines oriented at approximately 45° to each other (A). EDX analysis of the
interfacial region gave a strong niobium signal, verifying the presence of the coating
material after four thermal cycles.

Figure 67 is a SEM micrograph of a fiber-matrix interfacial region from a 7-vol.%
NiAl-(Ni-Saphikon) composite, thermally cycled from room temperature to
approximately 1423 K for four cycles and then fractured. The micrograph shows intimate
fiber-matrix contact and slight evidence of radial cracking (A). There is also a small
amount of oxide scale present near the interface (B). Figure 68a and b are SEM
micrographs of the interface along a longitudinal section of a fiber. The left-hand side of
the micrograph shows matrix material (A) and a small ridge of matrix material (B)
adhering to the fiber surface. The right-hand side of the micrograph shows the formation

of a mature oxide scale (appears to be a dense alumina and not a metastable oxide) on a

139

 

(b)

Figure 62 SEM micrographs of (a) uncoated and (b) Ni-coated NiAl-FP composites.

 

Figure 63 SEM fractograph of a 7-vol.% uncoated Saphikon composite.

I41

 

Figure 64 SEM fractographs of a 7 vol.% NiAl-(uncoated-Saphikon) composite.

142

 

Figure 65 A SEM fractograph of a 7-vol.% NiAl-(Nb—Saphikon) composite.

143

5 J”: ./”’//9
.l', ’f,./ ’..’A/
_ ‘1 g/ .5516 35

“6“ .
\. 7’

av

 

Figure 66 SEM fractographs of a Nbcoated Saphikon fiber composite.

144
grain boundary (C). A high magnification view of the fiber surface is shown in Figure

68b. The dimpled surface morphology was not present in the previous investigations of
fiber surface morphology, but presumably are the result of fiber-nickel coating reactions
that occur during the consolidation of the composite.

Qualitative information about the interfacial mechanical behavior information is
shown in Figure 69, where a room-temperature fracture surface of a titanium-coated
Saphikon fiber composite reveals a brittle interphase, as evidenced by the numerous small
cracks, labeled 'A'. Furthermore, the intimate fiber-matrix contact suggests a strong
interfacial bond. The interphase region shown was ~5 pm thick, well beyond the distance
of the x-ray signal, before cleavage fracture occurred (B). Figure 69 illustrates the
roughness of the fiber trough for a fractured titanium-coated Saphikon fiber dilatometry
composite thermally cycled two times. In this specimen no grain boundary oxidation was
observed on the fracture surface.

4.8.2 High-Temperature Composite Fracture Behavior Figure 70a shows an
SEM fractograph of a NiAl—(uncoated Saphikon) dilatometry specimen thermally cycled
4 times and fractured at ~750 K. In general, the matrix exhibited transgranular brittle
cleavage fracture with most fiber-matrix interfaces appearing cohesive, though
occasionally voided regions (A), presumably originating, from incomplete consolidation,
are seen. These voided regions act as stress raisers upon thermal loading and may
accelerate fiber-matrix decohesion. Furthermore, interfiber matrix cracking with a path
along the mating region of the middle matrix plate and an outside matrix plate is evident
(B). Figure 70b is a SEM fractograph or the fiber-matrix interfacial region illustrating a
large circumferential crack and gapping around the fiber. Contrary to the room
temperature fracture surface results shown in Figure 64, a survey of the fracture surfaces

did no reveal any grain boundary oxidation.

145

 

Figure 67 A SEM fractograph of a 7-vol.% NiAl-(Ni-Saphikon) composite.

...
we

.11).. «rural. .hr

Kit r . I
if...:.l.:.<.rv:.«v r

 

Figure 68 SEM micrographs of a 7 vol.% NiAl-(Ni-Saphikon) composite.

147

 

Figure 69 SEM micrograph of a Ti-coated Saphikon fiber composite.

148
Figure 71 shows both high and low magnification SEM fractograph of a 7—vol.%

NiAl—(Nb-Saphikon) dilatometryl specimen thermally cycled 8 times. Most of the
micrograph (A) indicates transgranular fracture similar to that seen in the previous
nricrographs, while the left-center part (B) indicates intergranular brittle fracture. The
intergranular fracture is a result of oxidation embrittling the grain boundaries, providing
easy crack propagation through the composite. Though grain boundary oxidation was
seen, oxidation through the thickness of the composite was never observed. The
micrograph of a fiber in the region without oxidation shows a small amount of fiber-
matrix interfacial decohesion, but other fibers seem to be bonded well enough to fail
when the matrix failed. In general, the composite displays a well bonded fiber-matrix
interface, with a transgranular fracture surface. In the high magnification micrograph
shown in Figure 71b the fiber-matrix interface shows good fiber-matrix adhesion, though
one small void is present at the point marked 'A'.

The fracture surface of a 7 vol.% NiAl-(Ni-Saphikon) dilatometry composite
thermally cycled 3 times to ~1473 K is shown by the SEM micrographs in Figure 72.
The fiber ends, which were flush with the fracture surface, indicate a large amount of
fiber-matrix adhesion existed and resulted in little fiber pull-out (though fiber ends
extending beyond the matrix may shatter during composite fracture). Figure 72b shows
an isolated fiber surrounded by unoxidized (A) and oxidized (B) regions, with a crack
through a nearby oxidized grain to the fiber-matrix interface. A closer examination of the
fiber surface and trough is shown in the micrograph of Fr gure 72c. This micrograph
shows a rough fiber surface and dimple imprints in the fiber trough. The top region of
the fiber shows some interfacial gapping (A), while Figure 72d shows an unidentified
interfacial phase (A) and its relation to the fiber surface contours. Numerous attempts at
determining the type of phase using EDX analysis failed due to the height of the regions

surrounding the interface.

149

 

Figure 70 SEM f ractographs of a uncoated Saphikon-NiAl composite.

150

 

Figure 71 SEM micrographs of a NiAl-(Nb—Saphikon) composite.

151

 

Figure 72 SEM fractographs of a Ni-coated Saphikon-NiAl composite.

152

 

(d)

Figure 72 (con't).

153
Similar fracture surface behavior is seen in the SEM micrographs shown in Figure

73, of a 5 vol.% NiAl-(Ti-Saphikon) dilatometry composite thermally cycled four times.
A typical brittle transgranular fracture surface is seen, with fiber-matrix interfaces
showing intimate adhesion. Figure 73c shows a random roughness of the fiber channel
surface morphology. A close examination of the fiber channel surface did not reveal a
preferred direction of roughness, suggesting that the roughening did not result from a
thermal cycling wear process.

To further understand the effects of thermal cycling on a dilatometric composite,
the exterior surface of a 7 vol.% NiAl-(Ni-Saphikon) composite in the as-composited
state and thermally cycled 4 times were examined using optically microscopy. No
macroscopic cracks are present in the as-consolidated composite shown in Figure 74.
After 4 thermal cycles (to maximum temperatures of 1423 K) the grain boundary
structure becomes delineated by a, presumed, oxide phase, as shown in Figure 75a (A).
Furthermore cracks transverse to the fiber direction are also observed in Figure 75b (A).
These cracks did not appear to penetrate the composite thickness or cause delamination.
No cracks parallel to the fiber direction were seen.

4.8.3 Discussion of the Fracture Behavior of Thermally Cycled Dilatometry
CompositesAlteration of the interfacial mechanical properties via application of fiber
coatings potentially modifies composite fracture toughness in either a positive or negative
manner. The fracture surfaces of the thermally cycled composites provides a number of
observations related to either the different diffusion bonding process or the effects of the
fiber coatings on the mechanical properties of the interface. Consistently observed
among the fracture surfaces (both low- and hi gh-temperature) were low energy
transgranular cleavage fracture modes. This behavior is common among bcc (or variants
of bcc) metals, with the cleavage planes of the {100} type, but sometimes along {110}
[Reed-Hill, 1973].

154

 

Figure 73 SEM fractographs of an 7-vol.% NiAl-(Ti-Saphikon) composite.

155

 

(C)

Figure 73 (con't).

.2838 232832132 839, N a 0o flames... .880 8 28E

 

 

(b)

Figure 75 Optical micrograph of a 7-vol.% NiAl-(Ni-Saphikon) composite after cycling.

158
The relatively large NiAl grain size, which resulted from the high consolidation

temperature and long consolidation time, exacerbates this brittle behavior, since the
DBTT increases with grain size [Schulson, 1983]. Since little evidence of ductile fracture
was observed, it appears that both temperature regimes were below the DBTT.

Comparing the uncoated and nickel-coated fiber composites showed essentially
the same interfacial fracture behavior. Dramatic differences in the fiber roughness were
observed, though no indication of surface wear was apparent. In contrast to these results,
evidence for changes in the interfacial mechanical properties was observed by
investigating the fracture surfaces of titanium-coated fiber composites. As shown in
Figure 69, the interfacial region fractures in a distinctively different mode than the NiAl
matrix. This observation provides evidence that the titanium coated fiber produces an
interphase region with the matrix that possesses different mechanical properties than the
matrix. In particular, the numerous cracks in the interphase indicates enhanced brittleness
and, perhaps, increased modulus. Since NiAl is brittle at room temperature, the niobium-
coated fiber composites is the provided the most promising behavior of the three coating
materials, because of the interfacial ductility was increased found

Observations of intergranular oxidation formation was unexpected, since NiAl is
generally considered to be highly oxidation resistant. Two factors responsible for this
oxidation behavior are the internal matrix stresses, caused by the presence of the fibers,
and the strain incompatibilities between adjacent grains. These two effects are coupled,
since the thermal stresses generated by the fibers during cooling cause high tensile
stresses in the grains and at the grain boundaries. These tensile stresses are difficult to be
accommodate, since only three independent slip systems operate, causing microcraclc
formation. Microcracks formed near the surface of the composite provide fast oxygen
diffusion paths, and lead to intergranular oxide formation. As intergranular oxides are
nucleated and grow they cause additional stresses at the grain boundaries. Such stresses

facilitate nucleation and continuous development of oxides at the grain boundaries.

159
4.9 Thermal Expansion Behavior of Low-Volume Fraction Composites

4.9.1 Thermal Expansion Behavior of Monolithic NiAl, NiAl-(Saphikon), and
NiAl-(Ni-Saphikon) Composites The heating and cooling half -cycle of axial strain
versus temperature for monolithic NiAl, plotted in Figure 76, displays the expected
behavior of a bulk elastic material. The plot is characterized by the lack of a hysteresis
between the heating and cooling curves, and no net residual strain at the end of the
complete cycle. A small nonlinear deviation, observed at temperatures above 1450 K,
suggests that the small spring force exerted by the actuator caused yielding of the
specimen. Though the heating curve is smooth, noise from the frictional resistance
between the sample and the quartz tube is apparent in the cooling curve. Figure ‘77 is a
plot of the experimental and theoretical instantaneous CTE versus temperature of the
NiAl heating curve. Good agreement between the two is evident.

The plot in Figure 78 is the thermal response of a 7 vol.% NiAl-(A1203) fiber
composite following post-fabrication cooling. The features differentiating this curve
from the monolithic case, shown in Figure 76, include a knee (A) at ~1178 K in the
heating half-cycle, a hysteresis strain between the heating and cooling half -cycles and a
knee (B) at ~954 K on the cooling half-cycle. These features illustrate the fundamental
differences in thermal response between composited and monolithic N iAl.

To facilitate clarity in describing and understanding the above, and the following,
thermal strain versus temperature curves, the schematic shown in Figure 79 defines
several terms that will be referred to in the remainder of the thesis. From this diagram the
heating and cooling half -cycle knees are defined, while two strain values, the hysteresis

strain and the residual compressive strain, Asp, are identified.

I60

 

.22 25:80:. 000 9:00 23800an .m> 586 302. or Ezwi

:6: 82 ea:—

d d 4 — 4 4 J

05 9.3200th
:3 89

   

q d a d _‘ d + * 1—J + 1 d 14

cevccn
4 d e

  

      

m...
.d

3

u

3

w

w

_ m.
We

. m.
u
1m.—

 

 

161

 

 

 

 

20 UUTITI1IIIIITIIIIIU'IUU
'- '1
A
fl ,— u
U
M .
“? 15 _.
9 ..
fl
x d
V I
a p ‘
010' "
m I- d
5 - ..
0 __ .
0
fl . ..
s 5- ._
E
a v- u:
H
m " N
a . .-
~
I- .1
lililllilillJJlllJlllJ

 

0
300 500 700 900 1100 1300 1500
Temperature (K)

Figure 77 A plot of the instantaneous and theoretical CT E for monolithic NiAl.

162

 

2 T i r 1 fr r i r r r l I r r i r I r i T r r

' NiAl-Al O ‘

in 2 3 I «It

. Composite (7 -Vol.%) / .
= 1 e 5 '- A ,/ '1' ‘—
e- i- 143 -I
0
h " n-
H _.I
W - 4
_ 1- //-'/ d
.2 A"
N 1 ~ riff": d
< b £5.43": T
«a if B
e: - .. r
O _ .1
3 . //
Q d
m 0 o 5 — - —4

0 r r L l r r r i r r L i r r r i r r r i r r r

 

 

 

300 500 700 900 1100 1300 1500
Temperature (K)

Figure 78 The axial strain vs. temperature plot for a 7 vol.% NiAloSaphikon composite.

163
To further illustrate the effects of fiber reinforcement on the thermal expansion

behavior of the composite, the instantaneous heating and cooling CTE for both the
monolithic and the 7 vol.% NiAl-(Al203) composite, along with the instantaneous CT E
for the Saphikon fiber, are shown in Figure 80a and 80b, respectively. As the plots
illustrate, the instantaneous CI‘ E of the 7 vol.% NiAl-(A1203) composite agrees well with
the rule-of-mixtures (ROM) prediction at low temperatures, but deviates at ~600 K and
begins to approach the monolithic value. At ~1100 K the instantaneous CTE decreases
dramatically and approaches the fiber CT E at ~1300 K, after which it increases toward
the ROM value. The behavior of the cooling half -cycle instantaneous CT E shown in
Figure 288b, and indicates a distinctively different behavior from the heating curve CT E.
Comparing the CT E of the heating and cooling half-cycle indicates a discontinuous jump
at ~1450 K. As the composite cools the average value approaches the ROM prediction.

Figure 81 shows the axial strain versus temperature of a 5—vol.% NiAl-(Ni-Al203)
composite. Qualitatively, this curve is similar to the uncoated fiber composite curve
shown in Figure 78. The heating and cooling knee temperatures occur at, A at ~1255 K
and B at ~998 K, respectively. Though similar in form, several aspects differentiate the
thermal response of the two types of composites. First, measuring the change in strain
between the heating and cooling half -cycles, at a temperature midway between T313“
and T Aknee. the uncoated fiber composite has a hysteresis strain of -0.038 and the nickel -
coated fiber has a differential strain of -0.062, indicating a 60% increase in hysteresis
strain occurs for the nickel-coated fiber composite. Furthermore, the average CTEs
(calculated from a linear fit), above the knee temperature for the uncoated and nickel
coated fibers, are 14.0 x 10 '5 K'1 and 13.4 x 10 '6K'1, respectively.

The instantaneous CTE for the 5-vol.% NiAl-(Ni-A1203) strain data are shown in
Figure 82. On average, the ROM prediction is followed at temperatures below 1100 K.
On nearing, and subsequently passing through the knee temperature, an increase in the

CTE is observed, followed by a large decrease, with a minimum at ~1300 K. Beyond the

 

164

 

 

n Heating half-cycle knee

.5

g Heating A / Cooling
m /

.3

5 A = hysteresis strain

Coo 'ng half-cycle knee
Temperature
Residual
Strain

Figure 79 A schematic of the thermal strain vs.temperature plot.

165

 

 

 

 

 

4
.1
q
A 1
".'
3‘ ‘
o ‘
3': ‘
V ‘
E3 «
U -
g b '.O‘ 4
,-o
-0
8 P -o‘ 4
'--e‘
a 0‘ 4
d
3 ' ‘
g 5- _
B p q
—
r 4
~ -l
P 4
0 A141 AllLAlllllALLl‘lL

30 0 500 700 900 1 100 1300 1500
Temperature (K)

(a)

 

20 ‘rfimT'vvavvr‘Ivvv'vaITff

_
(In
TV ' W Yfi T

10

V

Instantaneous CTE (x 10‘ K)

‘TY'V
ALLLlLLL4llllllALLL

 

 

 

0 L L L l L PL 1 A l A l A A l I L L L l A A A

3 0 0 500 700 900 1 100 1300 1500
Temperature (K)

(b)

Figure 80 (a) heating and (b) cooling CT E plots of a 7-vol.% Saphikon-NiAl composite.

166

 

 

2 firI I I I T Ier I l I I I IfI I l I I I
105'-
.E ,.
g _
m ,.
'3 l-
x 1-
<
H h
E
O I-
U
x. ..
0
=- r
005— d

 

 

oiiiijiilitiliniliiilinr

3 0 0 500 700 900 1 l 00 1300 1500
Temperature (K)

 

Figure 81 The axial strain vs. temperature for a S-vol.% NiAl-(Ni-Ale3) composite.

167
minimum, the CTE increases toward the ROM value again, and then finally decreases.

(Note that the CT Es at low temperatures are equal for the heating and cooling half-

cycles.)

4.9.2 Thermal Expansion Response of NiAl-( Uncoated- and Coated-Saphikon)
Composites as a Function of the Number of Cycles

4.9.2.1 7-Vol.% NiAl-Saphikon Composites Three thermal cycles of a 7-vol.%
NiAl-Saphikon composite, plotted in Figure 83 illustrate several effects. (Note that the
first cycle corresponds to the data presented in Figure 78.) In contrast to cycle #1, cycles
#2 and #3 show a less distinct knee in the heating half ~cycle. For the three cycles, no net
residual strain accumulation, Aep, was measured. The heating-cooling half -cycle
hysteresis strain indicates that plastic yielding occurred during the heating half-cycle and
was due to the presence of the fibers. Furthermore, the knee and small slope change of
the heating half -cycle suggest that fiber-matrix adhesion was poor. The thermal
expansion parameters for these plots (and other composites) are tabulated in Table 17.
The data from Table 17 indicates that the knee temperature of the heating half -cycle
increases after the first cycle and then decreases. On the other hand, the knee temperature
during the cooling cycle appears to be constant for the first two cycles and then increases
for the third cycle.

Figure 84 shows the instantaneous CI'E plots for each heating half -cycle. Each
cycle has the same qualitative shape, exhibiting a close match to the ROM predictions,
followed by a gradual increase until the knee temperature is reached, where after a
decrease in the CPR occurs. The instantaneous CT E plot shown in Figure 84a illustrates
that the CTE of the first cycle reaches a value smaller than the CTE of the fiber. The
instantaneous CT Es for the cooling half -cycles are shown in Figure 85. The general
shape of the curve corresponds to a CTE value that is approximately equal to that for

monolithic NiAl, followed by an increase above this value, and then a decrease to a value

168

 

 

 

 

207 [1" I 711711
b 4
b d
A d
-
O 1
3 .
"e -
— d
5
q
0 .4
a .
3 .4
5‘ ‘
I: ~ .
£25- -
It .
~
l. -
oLAllllAlllAllA‘1lLllJAL

 

3 0 0 500 700 900 1 1 00 1300 1500
Temperature (K)

(a)

 

m V V 1 Y I V ‘ T I V ‘ V I Y I I 1 T ‘l V I I

 

Instantaneous CTE (x 10" K")

 

 

 

5 F —+
1
o A I l l l L l l l l I l L l 1 l 1 L l 1 I l l
3 0 0 500 700 900 1 1 00 1300 1500
Temperature (K)

(b)

Figure 82 (a) heating and (b) cooling CI‘ Es of a 5—vol.% NiAl-(Ni-A1203) composite.

 

69

1

63888 834332 @436 a n8 EBEKES .m> 583 35.85 Co .03 < mm 0.5mm

O: 9.39.0959
66m— 82 a: n — 8a 85 8m 9 e n

 

 

 

1J4—‘fi d u — d u - 4+4 dl— q d A — q d J mOG.
1 1
1 e
n a
e
. -meu
3
. J u
.. L 19
- . V
I 1— u.
1 n w
T r s
I J 1’
l imam m.
1 i
n [a
1 r
_ bF _ p h p — p _ b h FL _ _ p a _ _ p .L We”

 

 

 

170

Table 16 Thermal expansion parameters for composites as a function of cycle.

 

 

Cycle Knee KLnee
# Temperature Temperature

Heating (K) Cooling (K)

L 1 1148‘? ""5"f_9 :10
2 115815 945110

3 11741: 5 1014:: 10

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ k 7-vol.% NiAl-tNl-Alzgl) Composite
Cycle Knee Knee Hysteresm Ae Avg. Avg. “
# Temperature Temperature Strain p (x106) (x106)
Heating (K) Cooling (K) Below Above
Tm TL.
l 112 :t: - -O.(B7 12.8 1 .2
2 1175:;5 | - l 0.1375 0.0729 13.2 83—5 '—
_‘3 1 122815 I - 1 02120 | -0.l978 1 .1 .1
7-VOI.26 NIAI-(Nb-Al &) Composite
Cyce Knee Knee Hysteresis A8 Avg.CTE Avg. '
# Temperature Temperature Strain " (x106) (x 10-6)
Heating (K) (3001111809 Below Above
T ._..~.» T .. .‘A
1 :1: ‘7‘ :1: 1 0.1"“ 1.07: 1. '1'.

 

 

 

—2 11255” I 9963:315 |-0.1146 0.0779 12.8 f 8.9
—+_73 f 1228:: | 969:30 |-0.06Z_+_-0'—|—_—§.0521 12. | 8.8

 

 

 

 

 

 

 

7-vol.% NiAl-(Ti-Ale3) Composite
ce ’ ee Knee Hysteresis Ae Avg " ’ Avg.
# Temperature Temperature Strain p (x10‘6) (x1045)
Heating (K) Cooling (K) Below Above
a 0.1'." 1 .4 - "—
[ - | - ] 0.0581 ] -0. 3 I 13.4 J -

 

623
3 1 10:14:10 [00466 1 ~41 L E1 l -
L 4 | - | 1101¢8 [c.0349] ~o | 13.6 | -

lh-L

171

Table 16 (cont'd).

 

 

5-vol.% NiAl-('l'i-Klig) Composite
Hysteresrs

      

 

 

       

 

 

 

- . . 12.4
’2 | - 1 1008110 [0.0511 [0.0338 | 133 l |
I—‘a | - | - | 0.0409 [0.1125 1 13.5 | |
| 4 | — | 954310 [ 0.0511 ] 0.0112 | 13.5 | l

 

172
approaching the ROM prediction. At temperatures below ~900 K, the curve becomes

extremely erratic due to the noisy data in the low-temperature cooling half -cycle. A
distinctive difference between the CPR of the heating half-cycle and the cooling half -
cycle is the discontinuity in CTE occurring at the maximum temperature. This
discontinuity is a consequence of the composite response changing from a plastic
deformation mode to an elastic deformation mode as the reversal in temperature occurs.

4.9.2.2 7-Vol.% NiAl-(Ni-Saphikon) Composite The thermal cycle plots of a
7-vol.% Ni-coated Saphikon fiber composite shown in Figure 86 differ from the
previously described uncoated fiber composite. The thermal strain in the nickel-coated
fiber composites exhibit distinct heating half-cycle knees, followed by a large change in
CPR (given by the slope of the strain vs. temperature curve). Upon cooling, a wide
hysteresis and a net compressive residual strain develops. Furthermore, the knee in the
cooling half -cycle was absent for all cycles. The behavior of the 7-vol.% Ni-coated fiber
composite also contrasts with the thermal expansion response of the 5 vol.% nickel -
coated fiber composite, as shown in Figure 81, where both a heating and cooling half-
cycle knee occurred. These results suggest that the thermal response of a composite is a
function of the volume fraction.

The heating half-cycle instantaneous CT E plots for each thermal cycle are shown
in Figure 87. These plots are similar to the instantaneous CT E of the 5 vol.% NiAl-(Ni-
Saphikon) composite shown in Figure 82, and similar to the behavior of uncoated fiber
composite shown in Figure 80. At the lower temperatures the CT Es follow ROM
predictions and deviate toward the NiAl CI‘E for temperatures approaching the knee
temperature. At the knee temperature, a dramatic decrease in the instantaneous CT E is
seen, followed by an increase toward the ROM prediction and then a decrease.

Furthermore, the CTE of the second heating half-cycle dips below the fiber CT E value.

173

00m—

82

85888 eezfiamafiz 88.61. a .8 Eb 2.. e no: we 2:3”.

SC

3: Bacchus-Huh
2: _ 8. .2. 3m

eon

 

1‘ 1 l q
f

 

4 4 ‘ r-t ‘ 1 i ‘ 4 4 4 - ‘ 1‘ ‘ q 1

D h h

P 1P p by P P F P‘ b

‘

h

4

 

1?

I)

 

 

0
an

'8
in

(r)! ,0! x) 3.13 "waninflflll

6C
a: 933.13;
82 32 3: 8a .2. 3...

 

‘+l‘qi4‘lq“““ dd“‘ q 441‘

 

L7 by 1' 1- b h b h h D F - 1’ D FL p D ’ ~ L ’ b

A ALLA

A

 

 

eon

(,3 .01 113.13 moo-mum!

sn—

3

Cc 9.5.9.0..th

oo-— 80 .2. .3. can

 

4JTA

 

 

1 4 I d 1 1

1‘4l‘dldad4a

l .
.0

1 l
C
at

It
at

 

 

h>r>-5P>_>>>herb a

( ,3! ,0! 1) 31,3 enoemueuul

174

8:838 88388.22 88.6: a 5.. Eb 2.. e 82; mm 8:3

3V

3: 9.3.51.th

 

82 82 so: 8. .2. 8» can
.l..11....%14.e.......o
. 1
v L
1 1“
v A
.

v A
. .
I 1.:
Y a
.
v a
1 . in.
. c .
1 n
v a
v 1
L»—Fp»_>>>-I>»->.-.pu a

 

 

 

(.31 ,or 1031.3 mower-mar

A3

a: 5.2.9.5.

 

8.: SM- 00-— 8. 86 80 can
4444.144111+11<q<11.4110
1
v 1
v A
.
1 in
v 4
w
v A
. 1
1 1.:
v 1
Y
r
v A
I 1m—
. 1
. L
r A
v A
y p r p r > p p r r p _ #1» > p pl» p _ . . p 9"

 

 

 

(1')! ,0! 3) 31.3 ".mmflll

3

O: 9.39.05th
32 ee- — 03 2:. 3m

 

I I Y Y Y I I I Y Y t T j 1 T I

I

 

‘ i 4 q 1 d l‘ - 1 1 ‘ d 1 d ‘ ‘ 14 ‘ 1‘ 1‘ ‘ ll 1‘

P D b -b Prb F>L1h D F D _ D hhlpllybh

A

I l L A l

A L A A l A A

 

 

can

0
u

'3
d

('3' ’01 x) 31;) snoellflllnslu

175

26888 EON—«12:32 88.6.3-5 e .2 2382.58 .m> Ewan 35.55 2:. cm Ezwfi

CC unannomfioh
8mg 8: ca—— :3 :2 can can

 

ITIr

 

1

1 l 1
fl
1S rerxv mama

In
1-1
urea

LllllllLLLJl

IIIrlTIIIIIIIIIIIII'IIIT]

 

 

 

h b p — b p _ — p h h _ n p n — n p n — n n b m."

176

85858 ao~_<-_zv-_<_z 839 a. a he who 2: a 82m 5 2:9...

6V

 

6: season-ash
32 .8— 2: q 8. :2. .3 can
1 41 1 - d 4 ‘ d. ‘1 14 4 d l 1 1 1 ‘ ‘ d d ‘1 4 .

I l L l A L
n

a...

 

o
(.3! ,0! 3) 31.3 ”mm-l

 

 

 

ADV bLbnbLb-bbbbLbbhberFFba AMV

Cc 3.3-club

 

 

 

 

 

 

Scot-cairn.
82 82 :2 8a 8.. 8m 3m 82 82 2.: 8. .2. 8m eon
‘W1“““‘q““d‘d‘.“‘ o D
' A r44‘.1‘“d“4fiqd“d‘d‘d+“ ‘
V
n '
s .a-
n m n m
a m
a u
.. m
o. w .—
8 m
)
x m
M x
o M
2 x n. y
r X
l\ v n ..I
r A 1‘
. 1
L
Phhb‘hppbb-h-DP’.-hb. a 'b'.-Pb._’b._P’.-1Pbp_hp.Lg

 

 

 

 

177
4.9.2.3 7-Vol.% NiAl-(Nb-Saphikon) Composite The thermal expansion

response for three complete cycles of a 7-vol.% NiAl-(Nb-Saphikon) composite is plotted
in Figure 88. These curves indicate thermal responses similar to the nickel-coated fiber
composite curves shown in Figure 83. In particular, as shown by the data in Table 17, the
average CTE below the knee temperature in the heating half -cycles were found to be
approximately the same after the first cycle. In addition, for all three cycles, the average
CT E of the niobium-coated fiber composite above the knee was less than below the knee
and decreased to an approximate constant value of 8.9 x 10“5 K'1 after the first cycle.
From Table 17, the knee temperature value increases from 1115 K to 1178 K as the
number of complete cycles increases. For all three cycles the cooling half -cycle data was
noisy, but in general, a net residual strain occurred.

Figure 89 show plots of the instantaneous CTE for each heating half -cycle of the
7-vol.% NiAl-(Nb-Saphikon) composite. Below the knee temperatures, the thermal
response of the composite approximately follows the ROM prediction. In contrast to the
thermal expansion behavior of the uncoated and nickel-coated fiber composite, when
approaching the knee temperature no rise in the CTE is occurs. Above the knee
temperature the CT E steeply decreased to values below the CTE of the fiber, followed by
an increase to a values less than the ROM predictions.

4.9.2.4 7-Vol.% NiAl-(Ti-Saphikon) Composites The thermal strain plot for
four complete cycles of a 7-vol.% and 5-vol.% NiAl~(Ti-Saphikon) composite are plotted
in Figure 90a and b, respectively. These composites behave differently than the ones
previously discussed. The absence of knees in the heating half -cycle is the most notable
difference. Also, as the number of cycles increases, the net residual strain and the

hysteresis strain decrease, while a knee in the cooling half-cycle begins to appear.

 

178

6:89:00 RON—$92132 95.9-5 a hot. EBEKES .m> 58% 39:05 of. ww Eswfi

GE 0.5.9.2.th
63— 82 :e n _ 8a 8h 8m 9 e n

JIJI d 411 a d — * d 4 — d :1 d — 411T? + d mOGI

 

 

m...

l L 1 l

—

v-t
mans [crxv manna

T I I I I I I I I I I I I I ‘l' I I T I l I I T I l 1 I I I

lllllll+llll4ll
In
0
F1

 

 

 

179

.2_§58 209.3522 9.32 n a 8.. Eb 2: a so: am 2:5

8v

3: flaistloh
8n— .8— 00 u u a: 02. 8n 0 o n

 

 

‘i‘i‘lq W* ‘ d ‘ 1 It ‘ 4 1 d ‘ 4 4 I d ‘ 1‘ 1 c
I A
r n
v . II
1 A .
. .. w
V A
. m
1
. m
I m
w a
. n
v I
0
v
9
.l
. m.
(
t
r
r

 

 

 

bib—bibbbbhtbb>PblLi>b$hP.°N

3v 3

 

 

'0

 

 

 

 

Sam 86— ee-— g 86 8» OOH 8m— 3 OO- § 86 gm Gen
.l.a....r«.....di...:+e .rlat......a...........e
V I
V A
u A I-

. . u
1.. 1mm
T' A m
r' a
w 4 C
.I . m
I lean
. r a
T n.
y L I.
V I .’
I rn—X
r .. (hr
v

. A

r L
>>nbp>.ppbbb>bppp>>->>pa psrP>FP>>.-p.ph>L>->Pa

 

 

 

e
—

Q
—
(,3! ,0! x) 31;) snoammsul

180
Table 17 indicates that the average CTE of the heating half -cycle for either the 7-

vol.% or the 5-vol.% composites is approximately constant at 13.5 x 10‘6 K4, which is
larger than either the nickel- or niobium-coated fiber composites, but less than the
uncoated fiber composite. Data from Table 17 also indicates that the knee temperature
for the cooling half cycle of the S-vol.% composite increases as the number of cycles
increases; however, the trend for the 7—vol.% composite is not clear.

The instantaneous CT E plots, for each heating half-cycle of the 7-vol.% and 5-
vol.% NiAl-(Ti-A1203) composite, are shown in Figure 91 and 92, respectively. Except
for the first cycle, of the 5-vol.% composite where a decrease in CT E is seen after ~1200
K, the CTE, on average, tends to be slightly higher than the ROM predictions. A similar
behavior is seen for the 7-vol.% composite, though no decrease in CTE is seen during the
first cycle (this, however, may be a consequence of the relatively low maximum
temperature (~1000K)).

4.9.3 Discussion of the Thermal Strain Curves and the Corresponding
Instantaneous CT E Curves

4.9.3.] General Discussion of Composite Thermal Expansion Behavior The
coefficient of thermal expansion of monolithic materials is a simple function of
temperature, not related to any internal mechanical def orrnation or external variable such
as heating or cooling rate (assuming that uniform temperature is maintained). The fibers
influence the axial thermal strain and matrix stress as schematically depicted in Figure 93.
The assumptions used to predict this ideal behavior include: (a) that the fiber remains
completely elastic throughout the entire temperature range, (b) perfect fiber-matrix
bonding, and (c) the effects of transverse stresses are small (for metal matrix composites
this is generally the case since the Poisson ratio of ceramic fibers is low (~0.2)).
Furthermore, the composite is assumed to be metallurgically stable and the fibers are well
separated.

The material parameters and physical processes determine the longitudinal strain

response of a unidirectionally reinforced composite can be described using the model of

 

181

 

2°5T—‘FI III III IIf III III
I I I I l

e
.e

   

  

IITTIITf

 

Percent Axial Strain

lllllll‘llLJlllllLLllllllllll

 

.0.5 L1 1 l l l j I l 1 l l l l l l 1 L14 1 l 1

3 0 0 500 700 900 l 1 00 1300 1500
Temperature (K)

 

(a)

 

  
    

 

20 TV I Y I *r I T T I I rT I

l. 4

P d

2- a

r 4

.5 - .

: ~ .

a ' ‘

~105— ‘

.2 F ‘

I .- .

< . .
H

I ' ..

o 1.. _

2 - -

o _ .
a.

P. -I

0. ..

.4

q

A 11 l l l l 1 I 1 A 111 1 l 1 1 I.T

 

 

 

5

g 0 0 500 700 900 l 1 00 1300
Temperature (K)

(b)
Figure 90 The thermal strain for a 7— and 5—vol.% NiAl-(Ti-Al 203) composite.

182

259.58 20216-22 s4? 5 a 5. Eb 2.. .o no: a 2:9...

.3 8v

3.: {sink—30h

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3c 23.-ell;
8.2 —
8n 3: 8. 8.. 8... 2:. 8.... 82 2.: 8a 8.. 8... 3n
1 1 1 fl 1 1 1 d 1 1 1 d 11 1 1 d 1 1 1 d 1 1 1 ° 1 1 1 - 1 1T1 ‘ 1 1 1 q 1 1 1 d 1 1 1 d 1 11‘ c
w A
1 .4
m- r A
m . .
n I r n
m . .
o T
u r .
.- D f A
3 H. 19-
n .
l
o 7
9 r
n— 1."... I m—
r
.
bubbbP»hbr¥b\>/>h>r>~?>r
’1‘ 8 r P h b n h h - “WWb r b > .P h r h bL b b 8
G: 339.06.5h G: 93.-sunfish
8.... 8n. 2:. 8. 8.. 8... 2:... 8m. 82 2:. 8. 8.. 8... can
1 1 11 d 1 l1 1 .— 1 1 1 q 1 1 1 d1 1 1 1 q 1 1 1 c
I L 1 1 1 d 1 1 1 d *1 1 dr T1 ‘ 1 1+4 1 1 1
. n . .
I L m I J
I A ‘ Y A
I l n m I A
. . n n
u
U a
e
V A u
1 A .
1 1: m
. . 3 e—
)
h x
I
. 0
r A f
r 1 _ X.
I A m II\I n.
V A Y A
’ ’ b — p b . PL! t. P iP b ’ b _ D F P lP tP b b I A
o" F P F b I b h b F b b > D b b. I n b h u I b a

 

 

(.3! ,or 3) 31,3 monument]

(r)! ,0! x) 313 enoaununsul

183

23858 .moN 2.5.22 9.22 m a 5. Eb 2: .o 3o: 8 saw:

.3

O: 33...»:th
.2. com o o a

com- 894 oeqq Goa

 

a

TYWIII

 

Di D DrL D D D by D D D F

411—4414114d441‘1114144

D Di D .— D D D - V D D

 

 

.8

O: 9.39.2.5;
8.. 8... e on

32 §~ oo~— 8a

a
H m
in m
. m
i o
. m
. w
3
m,
. u
A o.
in. v.”

.o«

 

I

VIIIII

 

Lin-bth-pb-h

 

D D h h D b — b D b

1 1 C d 1 1 1 _ 1 11.114‘ 1 1T1“ 1 1 1 c

(,3! ,0! x) 31;) moamnnsul

 

 

ON

.3

O: «haunt—Sou.

cen— eo—u 30 8.. So. can

 

 

1d111‘111d111-111‘1111

.F D - D. D D b D D D P D D D W D D D h D D

 

 

 

D

33

.3

3.: 95.9.0935.

9N

8n— ec—— 8a 3.. 8.» can

 

1 11 1 14 1 1 1 d 1 1 1 11 1 1 .— 1 1 1

 

DL D b D D D b F D D D D D D - D D D F D

L l A .L A

on

G
d

ll
.1

 

 

 

on

(P)! ,01 1) 3.1.3 snoemmsul

(PX ,ol 1) 3D snoammsul

184
Garmong [Garmong, 1974]. In this description, with the absence of an externally applied

stress, stress equilibrium requires that,

afo-i-amezo, 4-11

and strain continuity across the fiber-matrix interface requires that,

4-12

where oi represents stress, ei represents total strain, Vi represents the volume fraction,

and i equals either f or m and refer to the fiber and the matrix, respectively. The total

composite strain, ac, is the sum of independently acting strains and is given by
8c = 3m + such-ear . 4'13
where

smock-I'd = sciatic + 91mm" scnqr 4‘14

Theses strains are defined mathematically by

T:
8;!!!“ = [aidT , 4-15
I
and the elastic strain is given by
8;“, = EL, 4-16

where i represents either the fiber or matrix.

185

      
  

F Heatin
; 7
E / Cooling

 

 

Tensile

 

Axial Matrix
Stress

Compressive

a
o
u
u
,-
.-
----
a
CI.
00"

 

Increasing Temperature

Figure 93 Changes in (a) specimen strain and (b) matn'x stress during thermal cycling.

186
The plastic strain experienced by the matrix can be described by the Ludwik relation

[Ludwik, 1909],

.. -_r_-n ] 417

where oys is the instantaneous yield strength of the matrix, and K and n are constants.
Since no straightforward formulation for creep under varying stress and temperature
conditions is available, the Dorn formulation describing steady state creep can be used to

predict the creep strain of the matrix,
wens-s)»: (m

where A is a constant and s is the stress exponent, G is the matrix shear modulus, b is the

Burgers vector, D0 is the frequency factor, Q is the activation energy for diffusion, k is

dt
the Boltzmann's constant, d—I: is the inverse heating rate, and R is the gas eonsan From

this description, the fiber strain induced by the load transfer from the matrix is called the
mechanical strain.

Equations 4-15 through 4-18 indicate the material parameters and strains affecting
the hi gh-temperature thermal expansion of composites. Experimentally, the composite
strain, ac, during a change in temperature, and in the absence of external loading can be

measured and is given by

sc=(%) =‘fafd'l'+rslm,“m_l 4-19

187
where of is the thermal expansion of the fiber. (An equivalent expression is valid for the

matrix if one assumes perfect fiber-matrix bonding.) The experimentally measured

composite CTE is defined as
ac =1! Al) 420
Jr\ 1 ,
and, hence,
a --a—[TzardT+ e 1 421

where the term in brackets represents the total composite strain (thermal plus mechanical)
that the fiber experiences during a temperature change. Though the CTE is
staightforwardly measured, the contributions from the individual strain terms cannot be
readily separated.

The ideal thermal strain versus temperature of a composite experiencing one
complete thermal cycle can now be qualitatively described by considering Figure 93a.
(This discussion of the axial strain and stress versus temperature schematic compliments
the discussion of the origin of longitudinal residual stresses in section 2.4.3.) Starting
from point A the composite expands elastically (linear region of the plot) as the sample is
heated to "B”. At B, the matrix begins to deform plastically, causing a knee to develop,
by either time dependent (creep) and/or time independent (yielding) deformation. Also,
the knee represents the region where a change in the strain-temperature slope occurs.
When cooling begins (C), the composite unloads elastically until “D” when the matrix
yields causes a knee in the cooling half-cycle. After arriving back at the starting
temperature no net dimensional change has occurred.

From the strain versus temperature schematic shown in Fr gure 93a, the ideal

matrix stress state can be qualitatively illustrated as shown in Figure 93b. This figure

188
illustrates that after post-fabrication cooldown, a longitudinal residual tensile stress (A),

equal to the matrix yield stress, is present in the matrix. On heating, the matrix stress
decreases linearly. The stress then passes through zero as the residual stresses relax and
goes through a region of elastic compression as the fiber inhibits matrix expansion. Upon
increasing temperature, the matrix stress increases to a value large enough to cause
compressive yielding (plastic deformation) at B. A period of progressive plastic flow
then follows with the matrix stress decreasing as the matrix relaxes. On cooling (from
C), the composite unloads elastically and the matrix stress continues to decrease and
passes through zero. With decreasing temperature the matrix contracts faster than the
fiber and tensile stresses accumulate in the matrix. At D the matrix yields in tension as
the residual stresses approach the matrix yield stress, before returning to “A”.

This description of the thermal strain versus temperature behavior allows the
thermal expansion quantities can be more completely defined:

Heating half-cycle knee: indicates the residual tensile matrix stress has been
relieved and compressive stresses are sufficiently large to exceed the temperature
dependent yield stress and cause the onset of compressive plastic deformation via creep
or flow processes.

Cooling half-cycle knee: indicates the accumulation of matrix tensile stresses
exceeds the matrix yield stress and causes plastic flow. (Generally this occurs at
temperatures where creep processes are not active.)

Hysteresis strain: is the relative measure of the differential strain between heating
and cooling half -cycles, generated by interfacial matrix flow processes occurring above
the heating half-cycle knee.

Residual strain, Asp: the compressive residual strain indicates that compressive
creep processes caused compressive creep strains to occur above the heating half -cycle

knee and is not fully recovered by the tensile stresses generated during cooldown.

189
Since the matrix has the important function of transferring the applied load to the

stronger and stiffer fiber, understanding matrix-fiber load transfer is central to
understanding the thermomechanical behavior of a composite. Matrix-fiber load transfer
occurs when interfacial shear strains, resulting from elastic moduli differences between
the fiber and the matrix, generate interfacial stresses. The magnitude and distribution of
these interfacial shear stresses then determines the strengthening capabilities of a
composite system. Following the approach by Myers and Chawla [Myers, 1984],
assuming a single fiber in an infinite matrix, the distribution of the fiber stress and the

interfacial shear stress is given by

 

 

 

l (El. 1
a, - life.2 Ill- ShaTz-lj, 4-22
2
and
b E r efi Sinhfl ('15) ‘x‘ 423
L2 we) ’

respectively. The variation of 5} and ‘t with x (the distance along the fiber) is shown in
Figure 94. In equations 422 and 423 Ef is the fiber modulus, If the fiber radius, and so

the composite strain. In addition, the load transfer parameter is defined by

1

21:0 I
l
E,A,ln(§)
rt

where Af is the fiber cross-sectional area, R is defined in Figure 94, and Gm is the matrix

190

 

 

 

 

   
   

 

 

 

 

 

     

 

 

. .... Mews.:2:3:s:s:2:s:3:3:2:::3:3:3:2:s:s:s:s:23322122123:
133113}if:5:5:5:5:3:5:3:3:325m52313332313:3:5:S:3:44333555525253
':.:-:- Ill/[IV 43:23:23:
-:-:-: /F_bey /:-:-:.:-:-:-
”3:133:22: Gapp°
’:':':':':l':f:f:f:f:3:[: --: :-:-:E13: : : : :3:3:3:3:3: 1:3:3:3:3:3:3
t;.;.-.-.-:-,-:.:-:-:-:-:-:.:-:.:o.-:-:-:-:-:-.1.:-:-:-:o:-:
l.:.::::::::[::1:2:2:Z:2:I:2:::3j:11:ICC‘ICiI'I'I'I'I'I'
L-:-:-:-:-:I:-:-:-:-:-:-:-:-:-:-:I:-:3:3:3:3:1:313:3:3:3:3.
l I l I
I l
I i i :
I l : : l I
I 3° : : 3° I
I I I I
ml I I '

 

 

o - — — o
. . .4“
Stress : :
l l
1

Distance I
along Fiber l
“r I

Figure 94 The fiber tensile stress, a, and shear stress, I, along the fiber-matrix interface.

 

191
shear modulus. Though this expression is approximate, in all analyses 8 is proportional

to 4.
Er

The shear stress given by equation 4-23 will be the smaller of the strength of the
fiber-matrix interface in shear, or the shear yield stress of the matrix. These two stresses,
along with relaxation mechanisms discussed in section 2.4.3, control the matrix-fiber load
transfer.

4.9.3.2 Discussion of the Thermal Cycling Strain Response of Uncoated- and
Coated-A120 3 Fiber Composites Comparing the thermal strain versus temperature
schematic in Fr gure 93a with the experimental thermal expansion plots of the 7-vol.%
NiAl-(Saphikon) and the S—vol.% NiAl-(Ni-A1203) composites shown in Figures 78 and
81, indicates close agreement. Similarly, the 7-vol.% nickel- and niobium-coated fiber
composites, shown in Figures 86 and 88, display the expected behavior during the heating
half -cycle, but the cooling half-cycles deviate from ideal predictions.

Quantitative and qualitative comparisons between the strain responses of the
uncoated, niobium- and nickel-coated fiber composites consist in analyzing the shift in
the heating half -cycle knee temperature, the change in slope (and concavity) above the
knee temperature, the hysteresis strain, the cooling half-cycle knee temperature, and the
net residual strain. The data in Table 17 summarizes these quantities and the following
discussion considers each in order.

Heating Half-cycle Knee Temperature: The histogram in Figure 95
illustrates a trend of increasing heating half -cycle knee temperature with increasing
number of cycles for the uncoated and the nickel-coated fiber composites. Increases in
the knee temperatures indicates that stress relief, coupled with plastic flow, occurs at
higher temperatures for each complete cycle. This trend suggests that after one complete
cycle the interfacial region has work hardened allowing larger residual tensile stresses in

the matrix after cooldown. Thus higher temperatures are required to relieve the

 

 

 

 

ersus the number of thermal cycles.

Figure 95 The heating ha] f -cycle knee temperature v

193
additional residual stresses and to cause the matrix to flow in compression. In addition,

the relative knee temperature changes for the uncoated and nickel-coated composite is
different. For instance, after the first thermal cycle, the nickel-coated fiber composite
knee temperature changes ~50 K per cycle, while the uncoated fiber composite shows
differences of only 10 K and 15 K. The larger knee temperature increases exhibited by
the nickel-coated fiber composite suggests that the interface microstructure (dislocation
density or organization), was different than the uncoated fiber composite after the first
thermal cycle. This probably results when the nickel coating increases the fiber-matrix
adhesion (load transfer) and causes more extensive interfacial work hardening than in the
uncoated fiber composite.

Figure 95 also indicates that the heating half-cycle knee temperature for the
niobium-coated fiber composite remains approximately constant for each thermal cycle.
Also, after the first thermal cycle, the niobium-coated fiber composite knee temperature is
slightly lower (AT = 30 K) than either the uncoated or nickel-coated fiber composites.
The previous discussions in Sections 4.5.1 and 4.8.1 indicates the presence of niobium at
the interface. This result combined with an almost constant knee temperature suggests
the niobium does not strengthen the interfacial region (at high temperatures) and has a
lower yield stress than the NiAl matrix. Also, it appears that the composite matrix shear
yield stress. Unfortunately, no information concerning the locus of failure can be gleaned
from dilatometry measurements.

In contrast to the uncoated, nickel-, and niobium-coated fiber composite results,
the thermal response of the titanium-coated fiber composites did not show a knee in the
heating half-cycle. Several possible reasons responsible for this behavior include a poor
fiber-matrix bonding (once the mechanical clamping stresses have been relieved), a high
interfacial yield strength, or the occurrence of interfacial damage. The increased surface
roughness and the expected Ti-Al203 and Ti-NiAl reactions, along with the visual

observations that the composites showed severe transverse cracking and brittleness after

194
thermal cycling, provide indirect evidence of good the fiber-matrix adhesion.

Additionally, an average CTE of approximately 13 x 10'5 K4, comparable to the average
ROM value (approximately 13.5 x 10'6 K4), suggests that significant matrix-fiber load
transfer is occurring throughout the temperature regime. This result is in slight
disagreement with the calculated instantaneous CT Es shown in Figures 91 and 92 which
indicate, on average, agreement with ROM values at low temperatures, but show
deviations toward the monolithic NiAl values at temperatures as low as 700 K The
reasons for the missing are unclear at this time.

Change in Slope: Another important quantity calculated from the heating
half -cycles is the average CI'E (the slope) above the knee temperature. This CTE
provides a method for qualitatively assessing differences in fiber-matrix adhesion. The
uncoated fiber composite shows only a small reduction in slope (ACI‘E = 0.5 x 10'6 K4)
above the knee temperature. Though the CTE is slightly reduced, it is still above the
predicted average ROM value of ~13.5 x 10'6 K4. This implies that matrix-fiber load
transfer takes place, but not in an amount sufficient enough to cause a large amount of
matrix yielding (indicated by a larger change in slope). This implication of a small
amount of fiber-matrix bonding is consistent with a mechanical bond (it may be possible
that hoop stresses may not be exhausted) and/or a weak chemical bond.

The histogram in Figure 96 illustrates that the average CTE of the nickel- and
niobium-coated fiber composites decreases as the cycle number increases. The fiduciary
line representing the average CTE of Al203 the values are close to the average CTE of
the fibers (8.8 x 10'6 K4) and substantially lower than the values reported in Table 17 for
the uncoated fiber composite. These results indicate effective matrix-fiber load transfer
in the coated fiber composites, suggesting a high level of fiber-matrix adhesion. This
increased adhesion causes deformation processes which work harden the interface region.
Progressive increases in the interfacial stresses continue until the stored energy, combined

with the applied temperature, activate stress relaxation processes (listed in Table 7) which

195
reduce the load transfer effectiveness. These relaxation processes also caused a mild

concavity in the nickel- and niobium-coated fiber composites, not observed in the
uncoated fiber composites. This concavity (which has a negative slope) is an indication
of activated creep.

Hysteresis Strain: From the data in Table 17 the nickel-coated fiber composite
has the largest hysteresis strain, while the niobium- and titanium-coated fiber composites
have intermediate values and the uncoated fiber composite the smallest. The hysteresis
strain is large for composites showing large changes in slope above the heating cycle
knee temperature. The titanium-coated fiber is the exception to this statement since no
heating half -cycle knee temperature was observed, but for all cycles showed a hysteresis
strain. This suggests that, though no knee was seen, a significant amount of deformation
(i.e., if no deformation occurs, the composite behaves like a monolithic material)
occurred during the heating half-cycle. From the amount of hysteresis strain, it appears
that the titanium-coated fiber composite has a high level of fiber-matrix adhesion.

Cooling Half-cycle Knee Temperature: Similar to the heating half -cycle, a
complex thermal response occurs during the cooling half-cycle. In the case of uncoated,
titanium- and niobium-coated fiber composites following an initially linear response
(elastic unloading) a knee appears. As described in section 4.9.3.1, this knee represents
the temperature where the matrix tensile stresses accumulate and rise above the yield
stress of the matrix. The knee temperatures shown in Table 17 do not indicate a
consistent trend, but all of the knee temperatures are in the temperature range of 940 to
1050 K except for the fourth cycle of the 5-vol.% NiAl-A1203 composite (1101 K). The
absence of a trend may partially be the result of noisy data, and missing or diffuse knees.
These observations suggest that the fiber coatings do not alter the interfacial yield stress
of the composites.

In contrast to the cooling half -cycle of the niobium- and titanium-coated fiber

composites, the thermal response of the nickel-coated fiber composite remains linear with

 

 

 

 

197
no knee being observed. The absence of the knee indicates that the interfacial stresses

accumulated above the knee are not fully recovered.

Net Residual Compressive Strain, Asp: Residual compressive strains result
when the high-temperature compressive stresses generate compressive strains which are
only partially recovered when the composite is cooled. The dimensional change of the
composites is referred to as strain ratchetting. All of the coated fiber composites
exhibited a net residual plastic strain after the first thermal cycle, while the uncoated fiber
composite showed no length changes. This result indicates that the coated fiber
composites generated a larger amount of compressive deformation at high temperatures
than the uncoated fiber composite. The net residual stresses combined with the change in
slope above the heating half-cycle knee are indications of increased fiber-matrix adhesion
for the coated fiber composites. Again, the titanium-coated fiber composite does not
follow the same behavior: though no knee is observed all other thermal expansion data
indicate a significant amount of fiber-matrix adhesion occurs.

Repeated cycling results in either additional accumulated residual strain (nickel-
coated fiber composite and to some extent the niobium—coated fiber composite) or in a
diminishing residual strain (as in the titanium-coated fiber composite). Continued
accumulation of residual compressive strains indicates the same level of fiber-matrix
adhesion persists for each complete thermal cycle and that plastic Shakedown is not
occurring. On the other hand, the titanium-coated fiber composites shows a consistent
decrease in residual stress as the number of cycles increases. The accumulation and
subsequent reduction in net residual stress is probably related to damage mechanisms
occurring at the interface and indicates that plastic Shakedown is occurring.

From the above discussion of thermal expansion parameters it is apparent that the
nickel, niobium, and titanium coatings increase the matrix-fiber load transfer at high
temperatures when compared to the uncoated fibers. (Mechanical clamping stresses

appear to dominate the load transfer at the low temperatures.) Furthermore, it appears

198
that the nickel and niobium coatings cause a significant amount deformation to occur

above the knee resulting in a strain ratchetting behavior. In contrast to this behavior, the
uncoated fiber composite showed a knee, which implies fiber-matrix adhesion causing
interfacial yielding, but not to the extent observed in the nickel- and niobium-coated fiber
composites. The behavior of the titanium-coated fiber composite is distinctive, though it
appears but it appears that adequate fiber-matrix adhesion is present (as evidenced by the
CTE obeying the ROM predictions), to produce residual strain accumulation. This may
be a result of the accumulation of interfacial damage during thermal cycling.

4.9.3.3 Discussion of the Efi‘ects of the Fiber Coatings on the Instantaneous CT E
The instantaneous CT E of unidirectional composites is governed by the fiber volume
fraction, along with the mechanical properties, and the thermal properties of the
composite constituents. Below the knee temperature elastic expansion occurs and the
instantaneous composite CTE should be bounded by the C'I'Es of the fiber and matrix in
this temperature regime. Above the knee temperature, where plastic deformations occurs,
these bounds may be violated. Violations of this type were observed in the instantaneous
CTE plots of the nickel- and niobium-coated fiber composites shown in Figures 87 and
89. To demonstrate some of the key material variables responsible for this effect, a
simple micromechanical model based on an inelastic rule-of-mixtures formulation is now
described [Urquhart, 1993]. (More sophisticated numerical methods have also been used
to determine the instantaneous thermal response; however, the same qualitative results are
obtained [Paley, 1992; Pindera, 1993].)

The analyses by Urquhart et al. assumes that the matrix phase behaves in an
elastoplastic (with linear work hardening) manner with the moduli of both constituents
invariant with temperature [Urquhart, 1993]. The expression for the instantaneous CT E of
the composite can be determined by starting from total the composite strain, which is a

sum of the fiber thermal, at, and mechanieal strains, email, where

e, = afAT , 4-25

and

£mechied=—’ 4‘26

and 01 is the longitudinal fiber stress (equivalent to the axial composite stress) and Bf is

the fiber modulus. The instantaneous CTE can then be written as

 

a““(T,a,) = a{ +—a—- —l-) a, +-1—(£g‘-) , 427
(71‘ Ef Ef dl‘
(a_-af),22..(§._§+) -01.:(;,__VL)
do: 6T EnErn 6T 1’ an Vm
where — = 1 1 V 428
(If _+__.._L
E! Em VII

and 01 is the longitudinal fiber stress, am is the OPE of the matrix, of is the CTE of the
fiber, Em is the modulus of the matrix, and 5,; is the hardening slope (modulus of the

matrix above the yield point). Furthermore, (7; is the current matrix yield stress, which

is a function temperature and matrix plastic strain. Now assuming that the elastic moduli
of the fiber and matrix are independent of temperature (which is not too bad of an

assumption for Al 203 or NiAl) and letting

a““=a,+(-L)(XL)%. 4.29

200

This expression indicates that (rimIt can be less than the CT E of the fiber since a??? < O.

From this analysis it is clear that there are several material factors influencing
composite instantaneous CTE. Clearly the effect of the matrix yield stress and the rate of
change of yield stress with temperature plays a major role in the high-temperature
thermal response of a composite. For example, if the interfacial stress reaches the matrix
yield stress at a temperature corresponding to a large gradient in yield stress distribution
(intermediate to high temperatures for NiAl), there will be a drastic change in the thermal
response. If, however, the interfacial stress reaches the matrix yield stress at a
temperature corresponding to a smaller gradient, the change in thermal response will be
more subtle. From Figure 4 the NiAl yield stress at temperatures near the reported knee
temperature (1000 -1200 K) is less than 100 MPa, with small gradients.

By comparing the instantaneous CTE for the composites with that of the matrix,
fiber, and ROM calculations, one can differentiate changes in composite behavior due to
mechanical effects, such as matrix inelasticity and fiber debonding. Some of the
similarities and differences among the uncoated and coated-fiber composites's heating
half -cycle instantaneous CTE behaviors are summarized in Table 18. The information
contained in Table 18 shows that all of the composites obeyed —at least on average— the
ROM predictions at low temperatures where the yield strength and modulus of the matrix
are high. In addition, the ACT E mismatch induced clamping stresses are large in this
temperature regime.

Comparing the plots of the uncoated and nickel-coated fiber composite
instantaneous CT Es shown in Figures 84 and 87, respectively, indicate subtle but distinct
differences. For instance, both composites show a deviation toward the monolithic NiAl
CT E. This deviation indicates a release of the ACT E mismatch induced clamping
stresses below the knee temperature. This behavior implies poor fiber-matrix bonding,

since mechanical bond controls the load transfer process. Closer inspections of Figures

201
84 and 87 indicate the approach of the deviations is quite different in the two composite

systems. In the uncoated fiber composite system, the instantaneous CTE gradually
deviates —starting at temperatures as low as 600 K— from ROM values. In contrast, the
nickel-coated fiber composite closely mimics the ROM predictions until ~ 1000 K, when
sharp deviations toward the NiAl CTE are observed. The differences in the deviations,
suggest that the nickel-coated fiber bonds more strongly with the matrix than the
uncoated fiber composite.

In addition to these comments, the maximum instantaneous Cl‘ Es are different.
For instance, the uncoated fiber composite has an instantaneous CI‘E slightly greater than
the NiAl CTE, while the nickel-coated fiber composite has an instantaneous CTE equal to
or just below the NiAl CTE. This discrepancy may be caused by a slight error in
calibration, or may be real and be caused by differing degrees of fiber-matrix adhesion.

If the latter is the case, this data indicates that the nickel coating provides a better fiber-
matrix bond than the uncoated fiber.

In contrast, the niobium-coated fiber composite shows no deviation toward NiAl
CT Es and, hence, indicates the presence of strong fiber-matrix bonding. The titanium-
coated fiber composites showed no sharp deviations indicating of the release of clamping
stresses, but did, in some cases, show a gradual CTE rise toward the NiAl values. (Note
that most numerical or analytical studies do not predict a deviation toward the matrix
CTE since perfect fiber-matrix bonding is assumed.)

The knee in the strain versus temperature curve represents a transition from ROM
to fiber dominated behavior. This transition corresponds to a temperature regime where
the matrix yield strength and modulus are low. These changes in matrix mechanical
properties, combined with the AC'I'B induced stresses, produce instantaneous CI‘Es that
can violate the constituent material bonds. As the information in Table 18 indicates, all
of the composites except for the titanium-coated fiber composites dropped below the

ROM values after passing through the knee temperature.

.865 “ES 93 288 05 he own—9a m38€=r

 

 

 

 

 

 

 

 

 

 

l o: o: 36.8w 4:036:85 8% 83:82.
mood. we» 8» on we» E3322
awed. 8% 8» 93% dead a $322
muod. o: 8» Beam £3 81 @9825

ATM 0.2 5 “BE m0~a¢~ 20¢

a 8383th 2:82.th 85H mEBEKEoH

REM 05 .8? measom 5am 283 ms 22 33

ego—m owEo>< 631 98 20m 1.0 cog—05 e338. c2838 3 8:?> 20m 9580 59E

 

 

 

 

.mcoamtomno m8 msooasqsmfi Bermuufin wanna: Co mane—gm E 2an

 

203
In addition, the nickel- and niobium-coated fiber composite instantaneous CTE dropped

below the A1203 CTE. The violation of these bounds indicates persistent fiber-matrix
adhesion above the knee temperature, and that the interfacial yield stress has been
exceeded (reeall section 4.7.3). (Note that these results suggest that the locus of failure
occurs in the matrix and not at the fiber-matrix interface. Since dilatometry measures
average properties, some of the local properties may be smoothed out and a definitive
determination of the locus of failure may be impossible to locate.)

Though not explicitly expressed in the instantaneous CTE formalism shown
above, the effect of matrix work hardening on the composite thermal response also affects
the shape of the instantaneous CT E curve. For instance, a low work hardening rate
material flows easily upon yielding (assuming perfect fiber-matrix bonding). The easy
flow properties cause the instantaneous CT E to decrease more rapidly. Using the second
and third cycle instantaneous CTEs, the average slopes of the linear region immediately
following the knee temperature are recorded in Table 18. The reason that the first cycle
was not included in the average was that the first cycle slopes for the uncoated, nickel-,
and niobium-coated composites were approximately the same, 005 x 10‘5K'1.

In the case of the uncoated fiber composite, the steepness of the slope decreased
significantly after the first cycle. The change in slope after the first cycle suggests that,
assuming perfect fiber-matrix bonding, the interface was strain hardened during the first
cycle, such that the work hardening rate increased for the second and third cycles. Since
the uncoated fiber composite probably has poor fiber-matrix bonding, the change in slope
probably corresponds to inadequate loading of the fiber by the matrix. In this case, the
slope would no longer be an indication of the interfacial work hardening rate. The exact
cause of the change in slope is impossible to determine, but in all likelihood is a
combination of several factors including, fiber-matrix decohesion, matrix work

hardening, and interfacial damage accumulation.

204
The instantaneous CTE slopes of the nickel- and niobium-coated fiber composites

showed a consistent value of -0.06 x 10'6 K4. These results suggest that the interfacial
mechanical properties are relatively insensitive to the number of thermal cycle (at least
for the number tested) and that the interfacial bond strength persists. Furthermore, since
the fiber-matrix adhesion in these composites appears good, the steepness of the slope
implies that the interfacial work hardening rate is low. The low work hardening rate is
consistent with the high temperature behavior of NiAl and other bcc materials [Bowman,
1992]. The addition of nickel and niobium to the fibers, therefore, does not appear to
alter the work hardening rate of the matrix near the interface. Since the attainment of
perfect fiber-matrix bonding in the uncoated fiber composite is doubtful, a direct
comparison among the slopes of the coated and uncoated composites is unrealizable.

As a final note on the instantaneous CT E curves, the "double dip" behavior seen
in most of the instantaneous CTEs is believed to be real and not an experimental artifact
since it was reproducible in the same and different types of composites. This type of
behavior could represent the dynamic behavior occuring between the work hardening and
relaxation processes (such as strain hardening followed by dynamic recrystallization).

4.9.4 Thermal Expansion Behavior of NiAl-( Uncoated- and Coated-Saphikon)
Composites at Temperatures Near the Knee Temperature The previous
composite thermal strain versus temperature plots exhibited a hysteresis strain between
the heating and cooling half-cycle curves and, except for the titanium-coated fiber
composite results, a knee in the heating half-cycle plots. If the knee temperature
corresponds to the exhaustion of residual stresses and the onset of compressive plastic
matrix deformation, then heating the composite to a maximum knee near (or below) the
knee temperature and cooling should not produce a hysteresis strain (or net residual
strain) between the heating and cooling half-cycles.

To test this assertion, the composite thermal expansion behaviors were

investigated to maximum temperatures near the previously observed knee temperatures.

205
Since the Ti-coated fibers showed no heating half-cycle knee, a value of ~850 K, which

appeared to correspond to an elastic thermal response, was chosen as the maximum
temperature. Figure 97 shows plots of the axial strain versus temperature for 7-vol.%
uncoated and coated fiber NiAl composites. In all cases, a small amount of residual
strain was observed, but the amount of hysteresis strain was negligible. These results
imply that the application of nickel, niobium, or titanium coatings to the fibers do not
significantly alter the elastic expansion/compression behavior of the composites. These
results also provide additional evidence that mechanical clamping stresses dominate the
low-temperature load transfer process.

4.9.5 Discussion of The Efiects of the Heating Rate on the Thermal Expansion
Behavior of Uncoated, Niobium- and Nickel-Coated Saphikon Composites

Another important experimental parameter that potentially alters the composite
strain response is the heating or cooling rate. Though the general form of the plots should
be independent of the heating rate, the heating rate represents a kinetic component to the
strain response and affects the temperature at which the knee occurs [Garmong, 1974].
The plots in Figure 98 show the influence of the heating rate (2.54 K—min.'1, 3.75 K-min:
1, and 4.54 K-min.'1) on the composite strain response of uncoated- and coated-fiber
composites, while Table 17 summarizes the thermal expansion data. (The cooling data
was acquired when the furnace was turned off. This experimental arrangement resulted
in an average cooling rate of 3.35 K-min.'1. Also, the heating rate determines the
maximum attainable temperature since the furnace is power limited.)

Comparisons among theoretical predictions indicate disagreement of the ef f ect(s)
of heating rate on the knee temperature [Garmong, 1974; Dutta, 1993]. For instance,
Garmong predicts (and measures) that the position of the knee in the heating half-cycle
shifts to higher temperatures with slower heating rates, while Dutta et al. predict (and
measure) a shift to lower temperatures. This disagreement illustrates that the model and

the composite system determine the results of the study, and that the shift in knee

206

 

 

 

 

 

y- -I
b a
D II
P 1
1.6 - ..
P 1
D 1
H > 1
u-
g b 1
d 1.2 h -
m p 1
1 ’ 1
-
" ‘1
2 . .
"’ 0.8 - -I
E , .
U
I. ' q
8 > .
h 1
0- ‘b d
r -I
P 1
> +
p
o A L 4 l A A A I A A A l A A A l A A A l 4
300 500 700 900 1100 [”0
Temperature (K)
(a)
2 rfi" Y T Y j V 1 V V Y T V Y Y I V Y Y T V V V
p 4
D 1
p 1
L 4
1. 5r- ~
6 L a
-
E s 1
5 . .
P 4
3
r ' ~ ~
on ’ +
e - «
3 r i
81 ~ .
o: 5'- —r
t 4
1
D 1
b 1
A A l L L L 1 #4 A l L I A l A L4 1 A A A

 

 

 

0
300 500 700 900 l 100 1300 1500
Temperature (K)

(b)

Figure 97 The strain of (a) uncoated, (b) Ti, (c) Ni, and (d) Nb-coated fiber composites.

207

 

1500

 

 

 

 

 

 

 

1411fi<+11fiJTfi1—111*m 11111dii1Jijd11J4+1111
All 1 A
v A v A
v A v A
m I 1m
vl .l
l
T A1 Y A
Y A V A
1 AM Am
I
v AlK v A (
( . A e
V L
e\) r
r v A
v A uC m
r w
I A
r A w v H m
r A m . A m”
lm—l T 1m
r 7
7 r A
v A
1 A T A
W A
fi A
T I“ T I
1 A. r A
Y A
1
0 ' L0
Prbrhinbb—>+¥L_>->L o L bbkpthhthLkLPhh my
2 I. 0
I. 0

£95 3.2 38...;
285 3.2. 33.3

(d)

Figure 97 (con't).

208
temperature can not a priori be determined. In general, however, faster heating rates

allow less time per temperature increment for occurrence of thermally activated
processes, such as diffusion and creep . This indicates that increasing the heating rate
would shift the knee to higher temperatures.

Similar to the expression for the interfacial shear strength given by equation 423,
the rate dependent material properties of the interface should also influence the interfacial
shear stress. Once plastic flow begins, the shear stress, I, at the fiber-matrix interface is
described by 1,

r . m x m

t=K,(fl em) (E) , 4.30
where K, is a measure of the intrinsic matrix shear stress (which depends implicitly on
the amount of strain provided that work hardening occurs), 6' is a proportionality constant
that is a function of the fiber volume fraction, E is the matrix tensile strain rate, m is the
strain rate sensitivity of the matrix, x is a distance along the fiber, and df is the fiber
diameter [Kelly, 1972]. Equation 430 indicates that the shear stress increases when the
matrix strain rate sensitivity increases. Furthermore, this expression shows that when m
is small (near zero), the shear stress is constant along the fiber length and equal to the
matrix yield strength. Similarly, the tensile stress carried by the fiber, or, is a function of

the matrix strain rate sensitivity and is given by

o.--(3Wé-u>'[(..+,)”-(.—t)"‘].

where the symbols carry the same meaning as above. As might be expected, the fiber

 

tensile stress builds up most rapidly in strain-rate sensitive matrices (when m is large the

 

1This description oftherate dqrendentproperties compliments thediscussion ofthe work hardening rate in
section 4.9.33.

 

 

209
load transfer is more effective). Likewise, the fiber is loaded to higher stress levels with

increases in matrix strain rate.

For a given material, the strain rate sensitivity depends on a number of parameters
including grain size, composition, and temperature. Generally, increasing the strain rate
increases the yield stress, the flow stress, and the work hardening rate. In addition,
increasing the temperature increases the strain rate sensitivity of the material. The

average strain rates of the composites can be estimated from
Q. = a 0 Heating Rate . 432

Using this expression and the average CTEs listed in Table 17 for each heating rate, the
composite strain rates range from 2 x 10'3 3'1 to 4 x 10'3 8'1. Thus, though the heating
rates are quite different, the effects on the composite strain rate are relatively small.

Two cycles of the thermal response of a 7-vol.% NiAl-(Al203) fiber composite
are shown in Figure 98a and resemble the plots shown in Figure 83. Though no apparent
knee is present in the first cycle, the second cycle shows a knee at ~1230 K. Both cycles
show a small hysteresis strain. The thermal expansion curves for a 7-vol.% NiAl-(Nb-
A1203) composite are shown in Figure 98b. The plots are similar to those reported in
Figure 88, however, these plots show a consistent trend of decreasing hysteresis and
compressive residual strain with increasing cycle. These resuslts suggest the occurence
of plastic Shakedown. No trend(s) are discernible for the 7-vol.% NiAl-(Ni-A1203) fiber
composite plots shown in figure 98c.

The histogram in Fr gure 99 shows that for the uncoated and niobium-coated fiber
composites, the knee temperature increases slightly as the heating rate increases. In
performing the heating rate experiments, the same composite was used for each heating
rate. Therefore, the composites not only experience different heating rates, but are also

thermally cycled. The shifts in the knee temperatures are therefore a combination of

210

Table 18 Thermal expansion parameters for composites tested at different heating rates.

 

211

 

2 IUT'IYI'IIVIIII‘IUU'IIII

 

 

 

i :
£105- "‘
3 Cycle #2 ‘
5
i " ‘
.5 : Cycle #1
3 .

IE.'so.s- -
in h
o ’1 A A I A A A l A AA I A A A l A A A l A A A

 

300 500 700 900 1100 1300 1500
Temperature (K)

(a)

 

2ti'rlTllIIllTUIlIIU'I

 

 
 

Cycle #3 .- .....

  
  
 

Cycle #1

Percent Axial Strain
fl
I I I I l l I I T l I I I I

I l I 4 I l l l l

1 l

l

h

o AAlAAAlAAAlAAAlAAAIAAA

 

 

 

300 500 700 900 1100 1300 1500
Temperature (K)

(b)
Figure 98 The strain for (a) uncoated, (b) Nb-, and (c) Ni-coated fiber composites.

Percent Axial Strall

G
u

212

 

2.5“

fl
0
u
‘ I V

fl
V V I V

I

 

-o.5'

A A 1 L L A

 

l A L J l A A A A A A A l

 

 

300

Temperature (K)

(C)
Figure 98 (cond't).

700 900 l 100 1300

1500

213
thermal cycling and changing heating rates. Comparing the second and third cycles of

the niobium-coated fiber composite with the first cycle shows that the knee becomes less
prominent. In addition, the average CT E above the knee temperature increases. In
general, the knee temperature is relatively insensitive to the heating rates used.

The primary effect of a change in the heating rate is the amount of creep

deformation (and the consequent stress relief) that can occur.

4.10 Axial Matrix Stress Analysis

Matrix and fiber residual stresses arise from differences between the fiber and
matrix thermal expansion coefficients. Since these stresses affect the load carrying
capacity of the composite, it is important to measure them. These stresses are limited by
matrix flow to values corresponding to the von Mises stress condition. The application of
nickel, titanium, and niobium fiber coatings may alter theses stresses. This section of the
thesis describes a qualitative analysis procedure for calculating the temperature
dependence of the matrix stress using thermal expansion curves [Masutti, 1990]. In this
analysis, which assumes no sliding at the fiber interface (perfect bonding) and no yielding
of the fibers, the strain of the composite is assumed to be equal to the strain of the fibers,

8f . The fiber expansion is expressed as a sum of the thermal and mechanical strains,

1
e, = afAT +E—[qf - 2V fof], 433
r

where of, E4, and Vfare the thermal expansion coefficient, Young's modulus, and
Poisson's ratio of the fibers, respectively. of and of are the fiber axial and radial
stresses, respectively, and are assumed uniform throughout the fibers.

In the absence of an external axial load, ROM analysis gives

214

afivf = (03)“, 4.34

where (a: > is the average of the axial stresses in the matrix and Vf and Vm are the

volume fraction of the fibers and matrix. The distribution of the stress components in the
vicinity of the fiber-matrix interface during yielding of the matrix has been calculated by
Yano et al. [Yano, 1988]. These calculations indicate that (i) of has the same sign as all
and is on the order of 0.1 a} and (ii) although large gradients of the stress components
011“, of, and over are observed in the vicinity of the interface, the distribution of the
equivalent yield stress is nearly uniform and not very different from the absolute value of

(a: ). Thus in this calculation, the radial stress may be neglected. Combining equations

4—33 and 4-34, the average axial matrix stress can be calculated from

(of): %L5,(a,Ar- 5,), 435

where Ef = 414 GPa, cf: (5.96 + 0.00391) x 1061(‘1.

After meeting the requirements of the assumptions, the most critical parameter to
be determined is the initial (room temperature) matrix stress. In general, the initial
thermal stress in the composite is deduced by heating the composite to a high temperature
(~ 0.8 Tm), where the matrix stress becomes small and the absolute values of the flow
stresses are equivalent in tension and compression. Extrapolating the fiber thermal
expansion from the maximum temperature to room temperature allows the fiber thermal
strain, cf, to be determined. This strain is then used in equation 4-35 to determine the
initial matrix stress. In the present cases (see Figures 83, 86, and 88), the maximum
temperature of 1450 K corresponds to approximately 70% of the melting point of NiAl,
which may not be high enough to completely minimize interfacial stresses. Also, the

large difference in thermal response at the highest temperatures (large slope differences in

215

 

 

 

 

Figure 100 The knee temperature for uncoated, Ni-, and Nb-coated fiber composites.

216
the heating and cooling parts of the curves) for the nickel- and niobium—coated fiber

composites makes an absolute determination of the initial matrix stress difficult to
determine. Furthermore, in the uncoated fiber composite case, the condition of perfect
fiber-matrix bonding is probably not realized.

These deficiencies make quantitative calculations impractical; however, the
calculations can be performed to establish reasonable bounds on the stress to make
relative comparisons among the composites. To make qualitative comparisons, the
additional requirement of zero matrix stress at the highest temperature was imposed.

The results of these calculations are shown in Figure 100 to 102 for uncoated,
nickel-, and niobium-coated fiber composites, respectively. Two important points are
notable concerning the matrix stress of the uncoated fiber composite shown in Figure
100. First, the initial and final matrix stresses are approximately equal and appear
reasonable when compared to the room temperature matrix yield stress. Secondly, the
matrix stress never becomes compressive. The last observation is an artifact of the
additional imposed constraint of zero matrix stress at the highest temperature.

In contrast to these results, the nickel- and niobium-coated fiber composites show
a substantially smaller initial matrix stress and sizable compressive stresses. (Note that
the poor qualities of the cooling half -cycles result from noisy data and do not necessarily
reflect physical phenomena.) Also, because of the net residual strains, the matrix stress at
the end of a cycle is always greater than the initial stress of the next cycle. Furthermore,
the initial stress of the second cycle does not equal the matrix stress of the end of the first
cycle. This trend is also seen in the data reported by Masutti et al. and is a result of the
difficulty of determining the initial matrix stress [Masutti, 1990]. The large differences
seen between the initial and final matrix stresses are evidence of accumulated plastic
strain.

Figures 101 and 102 indicate that the maximum compressive stress for the nickel-

and niobium-coated fiber composites is approximately 40 MPa, which are approximately

217
50% of the NiAl yield stress. (Note that this represents a maximum compressive stress).

The creep behavior in NiAl follows that for metals and alloys and has been briefly
summarized in section 2.3.1.2 and Table 3. Furthermore, at 1175 K applied stresses of 30
MPa lead to strain rates among 10'5 and 104 S'1 [Noebe, 1993]. Comparing the
calculated stress of 40 MPa with the creep strain rates of 10'5 at 30 MPa, suggest that, at
1200-1450 K a significant amount of composite creep strain can occur during thermal
expansion of the nickel- and niobium-coated fiber composites.

These plots also provide a qualitative indication of the temperature range for the
transition from tensile to compressive matrix stress (~600 - 800 K). This temperature
range is well below the measured knee temperatures, but is suggestive of the behavior
shown in Figure 302. The last major distinction among the stress temperature curves is
the exhibition of compressive stresses during the cooling half-cycle for the niobium-
coated fiber composite. This behavior is a consequence of the mild concavity in the
cooling half-cycles of the thermal strain versus temperature plots seen in Figure 296. In
contrast, cooling half -cycles of the uncoated and nickel-coated fiber composites contract
in a linear manner. The concavity represents the occurrence of creep processes and
implies that the niobium coating produces more creep strain than either the uncoated fiber

or nickel -coated fiber.

4.11 Summary of the Results and Discussions

The results in this thesis serve as a step toward understanding the effects of
employing titanium, nickel, and niobium fiber coatings to alter the fiber-matrix interfacial
behavior in NiAl-A1203 composites. As described in section 1.3 and schematically

illustrated in Figure 23 the experimental approach consisted of investigating the as-

218

 

150rIIIUUTIITT‘ITTFIIIT'j'I

.—
G
9

 

Axial Matrix Stress (MPa)
m
e

 

AngLAAAIJALJAAALlAAAJAAL

 

 

-50
300 500 700 900 1100 1300 1500

Temperature (K)

Figure 100 Matrix stress for the a 7-vol.% NiAl-Saphikon composite.

 

"VIY'T'VT"VYVIV'UIYYY‘r‘VV
1

1500

s‘\._

D : 0‘
.

l’ V r
Cycle 02
/’
A i A A
1300

' Y ‘7 V
I

1

1100

V 1 V Y
\
r A .
see
Temperature (K)

I r V
l L A I

‘7“

V’-
A
A A

v I v
A I A
500

Z .

.3:
P .0,
AAAIALA AAAAAAAAAAAIAAAAAAA§
O O Q G.
o 8 3 3 N 3 '3
~

('0!) runs lure" may

V

 

 

 

 

 

VIVYYT‘Y‘V‘Y'VIYVY’TYVV§
A 1—
I.
~ , 4
A - .
1 i
i- —1
8
A 9 1
De
*0 1
A is
— -
~A
A 1":
V
b 1
A 1:
.. cg.
. 4 E.
A 1 a
A 1 0
8h
_ F
A 1
b 4
_. a§
* i
h
A
LAll

 

 

 

A A A L A .L Li A A. A A A A A §
0 6

g 3 3 3 «1 ° N

d e

(MIN) “”19 X113." I'm

219

(b)

 

 

 

 

vvv'vvv'vvvtvrvarvrf1Tfivu‘vvv
1
i- 1

A
l- : -1
.. a .
A 6‘ 1
> 4
n -1
A 1
A 1
* 1
A- 11
h 1
- 1
P 4
r- d
> +
A / 4
A 1
>- -1
A 1
A 4
A 1
AAAIA AlAAAlAAAlA AlAAAlAAAIAAA
2 8 8 3 2 ' a a 2
— 0 D

('0‘) “”8 "m" I‘m
1%?
v

 

1300 1500

1100

Temperature (K)

71”

(C)

Figure 101 Matrix stress for a Ni-coated fiber composite.

220

so

3: select-lea.
8m— 83 2: a 39

2:.

.28868 .3: “6286.92 a .5. $26 an2 NO— 253m

con

 

1 i I d ‘ ‘ 1 d ‘ ‘ r11“ 1‘ 4 1

«.3vo

 

’ P by P .P b b _ D D I b D

- ‘

1 4 d ‘ ‘ 1

D D b

no.

 

 

 

(MIN) ””38 WNW m"

A3

as £813.18...

 

Gem— 3— 0° - u so 86 8» 0 o n
1 4 it - 4 d 1 d 11 ‘i 4‘ dr [1 ‘ 4 q 4 ‘ I A. 1 e v C
V *‘b A
Y \ A! A
a. or
1 \ i/
t
T . / AA '0
v e19 // A
. \ A.
.\ .I.
v i l/[f
\
I no
. /
V
I
V
V
Y
I
V
v a. 3.6
Y
P ’ 1P — ’ h b h! ’ F ’ h b F D

 

 

 

 

('rl") ”“38 W“ m

3

0: actual;

 

8: can eo—— 8a 2:. 8m can
1 ‘ ‘ [q ‘ 1 1 4 ‘ d ‘ d 4 ‘ J .- 4 ‘ 11 1- id} ‘ J 0'.
A
. A
I “‘ 4
l .\ 1 a.
K

I \n A

\ A
v I.
1 *0k\“.. If! car L

x
.l ‘ / I. .
V
T
I
Y
Y
.I
V
. : 3.6
Y
b ’ b h F b L b 1P ’ F _ I D I h 5 b F b y r 1?

 

 

 

 

('JH) cents an.“ I'IIV

221
consolidated fiber properties, the thermal cycling behavior of the composites, and the

thermal expansion behavior of the composites.

The conclusions of these studies are sumarized in Table 19. These results indicate
that the fiber coatings (nickel, niobium, and titanium) produce considerable changes in
the fiber and composite properties. In particular, the as-composited fiber- fragment
aspect ratio distribution of the niobium-coated fibers and the uncoated fibers showed
approximately equal numbers of small, intermediate, and large ratios. In contrast, the
titanium- and nickel-coated fiber-fragment aspect ratios were skewed to the small ratios.

Dilatometry has been shown to be very effective in providing baseline
quantitative and qualitative data for understanding the thermal response of NiAl-
Saphikon composites. The instantaneous CTE of the composites were found to be
sensitive to plastic deformation processes and fiber-matrix interfacial bonding.
Dilatometry can, therefore, be used to indicate if the fiber-matrix interfacial bonds are
generally strong or weak. The fiber coatings affected the thermal expansion behavior of
the compoistes quite differently. All of the coatings increased the fiber-matrix adhesion;
however, the different hi gh-temperature deformation processes in the nickel-, niobium-,

and titanium-coated composites were clearly different.

222

 

 

 

 

 

 

 

 

 

 

 

89885 38280 | | 5.8.5 8% 02 m3 EESEH
"”8355 8988:— mmoj m3 §~20 9.5 8> 0:58 8:522
3322: vacate—5 $3 $3 .BwEO oZ mo> m8; 3x22
808.: adds

86$ 9am 238888.
528531 Amwev choEaI 8:2 33 Aflmddv 3.00.»; 35$

gram—2 25:25 x53 05 o>on< mmocnwsom skonoxanm mcaozonm 28m aooam<

Loni 38:25 .52»:an who .m>< oomtzm ABE came—m 3:55. Sci owflo>< mcuwoo poem

 

 

3:89:00 .BE c.3883 :< 9 03220”:
358.5 BaanoU

 

 

.mQEmcoue—oa acoqoaéaosbm 2: .6 $885 2 Eng.

 

CHAPTER FIVE
CONCLUSIONS

1.) Scanning electron microscopy and atomic force microscopy investigations of the as-
composited fiber surface morphology indicates that the FF and Saphikon fiber smface
roughness increased according to the fiber coating material as: uncoated, nickel coating,

niobium coating, and titanium coating.

2.) Hi gh-temperature (between 373 K and 1423 K) cyclic oxidation studies of uncoated
and coated fiber composites with ends exposed to the environment indicate severe
interfacial oxide scale formation. In contrast, composites with fiber ends embedded in

the matrix showed no presence of oxide scale formation.

3.) Two consolidation induced fiber fragment distributions were found: (1) a distribution
consisting of uncoated and niobium-coated (representing weak fiber-coating, or fiber-
matrix chemical reactions) fiber fragments having approximately equal numbers of short,
intermediate, and long fragments, and (2) a distribution represented by the nickel and

titanium coated (strong chemical reaction) fiber fragments consisting of short fragments.

4.) Room-temperature fractographic investigations indicate a ductile niobium interphase
region in and a thermally cycled 7-vol.% NiAl-(Nb-Saphikon) composite. In contrast, a
brittle interphase region was found in a 7-vol.% NiAl-(Ti-Saphikon) thermally cycled

four times.

223

224
5.) The nickel- and niobium-coated fiber composites's average heating half -cycle CTE,

measured above the knee temperature, were substantially less than the uncoated fiber
composite. These results indicate that the hi gh-temperature behavior of the nickel- and
niobium-fiber composites were fiber dominated and, hence, good fiber-matrix adhesion

was achieved.

6.) Strain ratcheting (composite shrinks upon each thermal cycle) occurred in the nickel-

and niobium-coated fiber composites.

7.) The thermal strain response of 5- and 7—vol.% titanium-coated fiber composites
lacked a knee and showed considerable Shakedown upon thermal cycling.

9.) Of the three fiber coating materials investigated (titanium, nickel, and niobium),
niobium emerges as the most favorable since the coating-matrix chemical reactions did
not increase the amount of consolidation induced fiber damage, enhanced hi gh-
temperature adhesion was achieved, and some evidence of low-temperature interfacial

ductility was observed.

BIBLIOGRAPHY

BIBLIOGRAPHY

Alam, MK. and Jain, SC, “The CVD Coating of fibers for Composite Materials,” .1.
Metals, 42, 56-58 (1990).

Adarnson, A.W., Physical Chemisfl of Surfaces, Interscience, New York (1960).

Amato, RA. and Yang, J.-M., "Status and Development of Nickel Aluminide (NiAl)
Composites," Mat. Res. Symp. Proc., 350, 203-212 (1994).

Amateau, M.F., ”Progress in the Development of Graphite-Al Composites using Liquid
Infiltration Technology," J. Comp. Mat, 10, 289-296 (1976).

Anton, DJ... and Shah, D.M., “Mechanical Evaluation of FF Alumina Reinforced NiAl
Composites”, Mat. Res. Soc. Symp., 273, 157-164 (1992).

Arnold, S.M., Arya V.K., and Melis, M.E., “Reduction of Thermal Residual Stresses in
Advanced Metallic Composites by Application of a Compensating/Compliant Layer
Concept,” HITEMP Review 1990: Advanced High Temperature Engine Materials
Technology Program, NASA, CP-lOOSl, 28-1 to 28-11 (1990).

Arnold, S.M., Arya V.K., and Melis, M.E., “Reduction of Thermal Residual Stresses in
Advanced Metallic Composites Based Upon a Compensating/Compliant layer Concept,”
J. Comp. Materials, 26(9), 1287 (1992).

Arsenault, R.J. and Taya, M., “Thermal Residual Stress in Metal Matrix Composite,”
Acta Metal]. Vol. 35, No. 3, 651-659 (1987).

Asthana, R., Tewari, S.N., and Bowman, R.R., ”Influence of the Fabrication Technique
on the Fiber Pushout Behavior in a Sapphire-Reinforced NiAl Matrix Composite,"
Metal]. and Materials Trans, 26A, Jan. 209-223 (1995).

Baker, I, Nagal, P., Liu, F., and Munroe, P.R., "The Effect of Grain Size on the Yield
Strength of FeAl and NiAl,” Acta Met. Mat, 39(7), 1637 (1991).

225

226

Barrett, C. and Massalski, T.B., Structure of Metals, 3rd ed., Chap 10, 1980, New York,
Pergarnon Press.

Barin, I. and Knacke, O., Thennochemical Promrties of Inorganic Substances, Springer,
New York (1993).

Bhatt, RT. and Grimes, HJ., Mechanical Behavior of Metal-Matrix Composites, edited
by J.E. Hack and M.F. Amateaux, TMS of AIME, 1983, pp. 51-64.

Bourke, M.A.M, Goldstone, J.A., Stout, M.G., and A. Needleman, in Fundamentals of
Metal Matrix Comflites, Eds. Suresh, S., Mortensen, A., and A. Needleman,
Butterworth-Heinmann, Boston, MA, 61-80 (1993).

Bowman, RR. and Noebe, R.D., ”Processing and Mechanical Evaluation of Continuous
Fiber Reinforced NiAl Composites,” I-IITEMP Review 1990. Advanced High
Temperature Engine Materials Technology Program, NASA, CP 10051, 40-1 to 40-13,
(1990) .

Bowman, R.R., personal communication (1994).

Bowman, RR, “Influence of Interfacial Characteristics on the Mechanical Properties of
Continuous Fiber Reinforced NiAl Composites,” Mat. Res. Soc. Symp., 273, 145-155
(1992).

Bowman, R.R., Noebe, R.D., Raj, S.V., and Locci, LE, ”Correlation of Deformation
Mechanisms with the Tensile and Compressive Behavior of NiAl and NiAl(Zr)
Intennetallic Alloys,” Metall. Trans, 23A, 1493-1508 (1992).

Bowman, R.R., Misra, A.K., and Arnold, S.M., "Processing and Mechanical Properties of
A1203 Fiber—Reinforced NiAl Composites," Met. and Mat. Trans, 26A, 615-628.

Broutrnan, L. J. and Krock, R.H., Comgite Materials Volume 1: Intergces in Metal
Matrix Comflites, Academic Press, New York, NY, 1974.

Broutrnann, L. and Krock, R.: Modern Comflite Materials, Addison-Wesley Publishing
Co., 1967.

Chawla, K.K., Comflite Materials, eds, B. llschner and NJ. Grant, Springer-Verlag,
New York (1987).

227

Clark, R.W. and Whittenberger, J .D., “Thermal Expansion of Binary CoAl, FeAl, and
NiAl Alloys,” in Thermal Exgision 8, ed., T.A. Hahn, New York Plenum Press, 189-
196 (1984).

Clyne, T.W. and Withers, P.J., An Introduction to Metal Matrix Comgites, Cambridge
Univ. Press, New York, NY (1993).

Clyne, T.W., “A Compressibility-based Derivation of Simple Expressions for the
Transverse Poisson’s Ratio and Shear Modulus of an Aligned Long Fiber Composite,” J.
Mat. Sci. Letts, B9, 336-339 (1990).

Copper, M.J., “The Electron Distribution in the Phases of CoAl and NiAl," Philos. Mag,
8, 811—821(1963).

Cotton, J.D., Noebe, RD, and Kaufman, M.J. , "Ternary Alloying Effects in

Polycrystalline fl-NiA], Structural Intennetallics, Eds. R. Darolia, J .J. Lewandowski, C.T.
Liu, PJ. Martin, D.B. Miracle, and M.V Natha], TMS, 513-522, 1993

Cox, I-IL, "l‘he Elasticity and Strength of Paper and other Fibrous Materials,” Brit. J.
Appl. Physics, 3, 73-79 (1952).

Crispin, RM. and Nicholas, M.G., "I‘he Wetting and Bonding Behavior of Some Nickel
Alloy-Alumina Systems,” J. Mater. Sci., 11, 17-21 (1976)

Curtin, W.A., “Exact Theory of the Fiber Fragmentation in a Single-Filament
Composite,” J. Mat Sci., 26, 5239-5253 (1991).

Delannay, F. Froyen, L., and Deruyterre, ”Wetting of Solids by Liquid Metals in Relation
to Squeeze Casting of MMCs, in Cast Reinforced MetaiComiosites, eds. S.G. Fishman
and AK. Dhingra, ASM, 81-84 (1988).

Darolia, R., “NiAl Alloys for High-Temperature Structural Applications,” J. of Metals,
43, 44-49 (1991).

Darolia, R., "NiAl for Turbine Airfoil Applications,” Structural Intennetaiics, Eds. R.
Darolia, JJ. Lewandoski, C.T. Lui, P.L. Martin, D.B. Miracle, and M.V. Natha], TMS,
495—504 (1993).

DeSilva, A.R.T. and Chadwick, G.A., "Thermal Stresses in Fiber Reinforced
Composites,” J. Mech. Phys. Solids, 17, 387-403 (1969).

228

Doychak, J. and Ruhle, M., ‘TEIvl Studies of Oxidized NiAl and Ni3Al Cross Sections,”
Oxidation of Metals, Vol. 31, Nos. 5/6, 431-452 (1989).

Doychak, J., Nesbitt, J. A., Noebe, RD, and Bowman, R.R., “Oxidation of A1203
Continuous Fiber-Reinforced/NiAl Composites,” Oxidation of Metals, 38, (1/2), 45-71
(1992).

Doychak, J ., Smialek, J .K., and Mitchell, T.B., "Transient Oxidation of Single-Crystal B-
NiAl," Met. Trans, 20A, 499 (1989).

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

Dutta, 1., Mitra, 8., and Wiest, A.D., ”Some Effects of Thermal Residual Stresses on the
Strain Response of Graphite-Aluminum Composites During Thermal Cycling," in
Residual Stresses rn Composites: Measurement Modeling, and Effects on Thermo-
Mechanical Behavior eds. EV. Barrera and I. Dutta, TMS 273-292 (1993).

Dwight, A.E., Intennetallic Communds, ed., J .H. Westbrook, Chap. 10, Huntington, NY,
R.E. Krieger Pub (1977).

Elssner, G., Suga, T., and Turwitt, M., "Fracture of Ceramic-to-Meta] Interfaces,” J. Phys.
Colloq. (France), 46 (C-4), 597 (1985).

Everett, R.K., “Deposition Technologies for MMC Fabrication,” Metal Matrix
Composites: Processing and Interfaces, eds, R.K. Everett and R.J. Arsenault, Acadenric
Press, Boston, MA, 103-109 (1991).

Garmong, G., “Elastic-Plastic Analysis of Deformation Induced by Thermal Stress in
Eutectic Composites: 1. Theory,” Metallurgical Transactions, Vol. 5, 2183-2190 ( 1974).

Garmong, G., “Elastic-Plastic Analysis of Deformation Induced by Thermal Stress in
Eutectic Composites: 1]. Thermal Expansion,” Metallurgical Transactions, Vol. 5, 2191 -
2197 (1974).

Garmong, G., “Elastic-Plastic Analysis of Deformation Induced by Thermal Stress in
Eutectic Composites: III. Therrnal-Cycling Damage,” Metallurgical Transactions, Vol. 5,
2199-2205 (1974).

Gebhardt, E. and Rothenbacher, R.Z., ”Untersuchgen im System Niob-Sauerstoff," Z.
Metallkd., 54, 623 (1963).

229

Glushko, V.I., Kovalenko, V.P., Mileiko, S.T., and T vardovsky, V.V., “Evaluation of the
Fiber Strength Characteristics on the Basis of the fiber Fragmentation Test,” J. Mat. Sci.,
28, 6307-6312 (1993).

Grimes, H.H., Lad, R. A., and Maisel, J .E, ”Thermal Degradation of the Tensile Strength
of Unidirectional Boron/Aluminum Composites," Metal. Trans, Vol. 8A, 1999-2005

(1977).

Groves, G.W. and Kelly, A., ”Independent Slip Systems in Crystals,” Philos. Mag, 8,
877-887 (1963).

Handwerker, C.A., Cahn, J.W., and Manning, J.R., "Thermodynamics and Kinetics of
Reactions at Interfaces in Composites," Materials Science and Eng., A126, 173-189
(1990).

Hartley, M.V., “The Effect of Isothermal Exposure on the Mechanical Properties of a
Continuous Fiber Reinforced Titanium Matrix Composite,” Mechanical and Physical

Behavior of Metallic and Ceramic Comflites, 9th Riso Int Symp., eds, 8.]. Anderson,
H. Lilholt and CB. Pedersen, Riso Nat. Lab., Denmark, 383-390 (1988).

Henrich, V.E., Surfaces and Interfaces of Ceramic Materials Ed. L.-C. Dufour, Kluwer
Academic, Norwell, MA (1989).

 

Hellman, J.R., Koss, D.A., Moose, C.A., Petrich, RR, and MN. Kallas, ”Interfacial
Shear Studies in Sapphire-Fiber Reinforced Niobium and Nickel Aluminide
Composites,“ in HITEMP Review, NASA CP-10051, 41-1 to 41-11 (1990).

Hindam, HM. and Smeltzer, W.W., ”Growth and Microstructure of a-A1203 on B-NiAl,"
J. Electrochem. Soc. 127, 1639 (1980).

Holliday, L. and Robinson, J., "Review: The Thermal Expansion of Composites Based
on Polymers,” J. Mater. Sci., 8, 301 (1973).

Howe, J .M., “Bonding, Structure, and Properties of Metal/Ceramic Interfaces: Part 1
Chemical Bonding, Chemical Reaction, and Interfacial Structure,” Int. Mat. Rev., Vol.
38, No. 5, 233-256 (1993).

Howe, J .M., “Bonding, Structure, and Properties of Metal/Ceramic Interfaces: Part 2
Interface Fracture Behavior and Property Measurement,” Int. Mat. Rev., Vol. 38, No. 5,
233—256 (1993).

230

I garashi, M. and Senba, H., “High Temperature Deformation and Oxidation Resistance of
NiAl Based Intennetallic Compounds,” Structural IntennetaMs, Eds, R. Darolia, J .J .
Lewandowski, C.T. Liu, P.L. Martin, D.B. Miracle and M.V. Nathal, TMS (1993)

lwasa, M. and Bradt, R.C., Structures and Promrties of MgO and A1303, Ceramics, Ed.
W.D. Kingery, Am. Ceram. Soc., Columbus, OH, 767 (1984).

James, M.R., and 0. Buck, in CRC Critical Reviews in Solid State and Materials Science,
61, 105-109 (1980).

Jansson, S. and Leckie, F.A., "Reduction of Therrna] Stresses in Continuous Fiber
Reinforced Metal Matrix Comosites With Interface Layers," J. Comp. Materials, 26(10),
1474 (1992).

Jedlinski, J. and Mrowec, S. , ‘The Influence of Implanted Yttrium on the Oxidation
Behavior of B-NiAl,” Mat. Sci. and Eng, 87, 281-287 (1987).

Jero, P. D., Parthasarathy, T. A. and Kerans, R.J., “Interface Evaluation in Ceramic
Composites," NATO AGARD Conference, Autalya, Turkey, 21-22 April 1993.

Katzman, H.A., “fiber Coatings for the Fabrication of Graphite Reinforced Mg
Composites,” J. Mat. Sci., 22, 144-148 (1987).

Kelly, A. ”Reinforcement of Structural Materials by Long Fibers," Met. Trans. 3, 2313
(1972).

Kelly, A. and Tyson, W.R., “I‘ensile Properties of fiber-Reinforced Metals:
Copper/Tungsten and Copper/Molybdenum,” J. Mech. Phys. Solids, Vol. 13, 329-350

(1965).

Khadkikar, P.S., Michal, G.M., and K. Vedula, “Preferred Orientation in Extruded Nickel
and Iron Aluminides,” Metall. Trans, 21A, 279-288 (1990).

Kieschke, R.R., Somehk, RE, and Clyne, T.W., ”Sputter Deposited Barrier Coatings on
SiC Monofilaments for use in Reactive Metallic Matrices - Part 1: Optimization of
Barrier Structure,” Acta Met. et Mat, 39,427-436 (1991).

Kingery, W.K., Bowen, H. K, and Ulhman, R., Introduction to Ceramics Wiley, New
Yorik, 210 (1976).

 

 

Kinloch, A.J., Gledhill, R.A., and Dukes, W.A., in Mhesion Sfcience a_n_d Technology,
ed. L.H. Lee, Plenum Press, 597 (1975).

231

Klein, M.J., Reid, ML, and Metcalfe, A.G., "Compatibility Studies for Viable Titanium
Matrix Composites,” AFML-TR-69-242, (1969).

Klomp, J.T., in “Ceramic Microstructures ‘86, Role of Interfaces,” eds. J.A. Pask and
A.G. Evans, New York, Plenum Press, 307, (1987).

Kyono, T., Hall, l.W., and Taya, M., Compgites ‘86: Recent Advances in Jam and the
United States edited by K. Kawata et al. Japan Society for Composite Materials, 1986,
pp. 553-561.

 

Larsson, L.O.K. “Thermal Stresses in Metal Matrix Composites,” ICCM/2, Metallurgical
Society of AIME, April 16, 1978, Toronto, Canada, 805-821.

Lenev, LM. and Novokhatskii, I.A., "Thermodynamic Characteristics of NiA1204,"
Russ. J. Inorgan. Chem., 10, 1307 (1965).

Ludwik, P., Elemente der Technologischen Mechanik, Springer, Berlin, p. 92 (1909).

Mackay, R.A., Brindley, P.K, and Froes, F.H., "Continuous Fiber-Reinforced Titanium
Aluminide Composites," JOM, 43, 23-31 (1991).

Manor, A. and Clough, R.B., “In-situ Determination of fiber Strength and Segment
Length in Composites by Means of Acoustic Emission”, Composite Science and
Technology, 45, 73-81 (1992).

Masutti, D., Lentz, JP, and Delannay, F., “Measurement of Internal Stresses and of the
Temperature Dependence of the Matrix Yield Stress in Metal Matrix Composites from
Thermal Expansion Curves,” J. Mater. Sci. Let, 9, 340-342 (1990).

McCullough, C. and Kieschke, R.R., ”Fiber Coating Performance in TiAl," Mat. Res.
Soc. Symp. Proc., 350, 119-124 (1994).

McDonald, J .E. and Eberhart, J.G., ”Adhesion in Aluminum Oxide-Metal Systems,"
Trans. AIME, 233, 512 (1965).

Metcalfe, A.G., Comflite Materials. Vol. 4 Metallic Matrix Composites, ed. K.G.
Kreider, New York, Academic Press, 269 (1974).

Metcalfe, A.G. and Klein, M.J., Comflite Materials, Vol. 1 InterfaLces in Comites ,
eds. Broutrnann and Krock, Academic Press, 125-168 (1972).

232

Min, B.K., in Advances in Commsite Materials and Structures, eds. S.S. Wang and Y.
Rajapakse, ASME, New York, in press.

Miracle, D.B., smith, RR, and Graves, J.A., ”A Review of the Status and Developmental
Issues for Continuously-Reinforced Ti-Aluminide Composites for Structural
Applications,” Mat. Res. Soc. Symp. Proc., 350, 133-142 (1994).

Miracle, D.B., “The Physical and Mechanical Properties of NiAl,” Acta Metal]. Mater.,
Vol 41, No. 3, 649-684 (1993).

Misra, A.K., Crandal], KS, and Bowman, R.R., “Modification of Interface in
A1203/NiAl Composite,” HITEMP Review 1992: Advanced High Temperature Engine
Materials Technology Program, NASA, CP 10104, 39-1 to 39-15 (1992).

Misra, A.K., “Thermodynamic Analysis of Chemical Interactions Between
Reinforcement Materials and Matrices in Aluminide-Based High Temperature
Intennetallic Matrix Composites”, Interfaces in Metal-Ceramic Comflites , eds. R.Y.
Lin, R.J. Arsenault, G.P. Martins and S.G. Fishman, TMS, 85-102 (1989).

Misra, A.K., Bowman, R.R., Locci, LE, and Matthiesen, M.M., "Interface Mo Coatings
for A1203/NiAl Composite,” HITEMP Review 1993: Advanced High Temperature
Engine Materials Technology Program, NASA, CP 19117, 58-1 to 58-13 (1993).

Moose, C.A., Koss, D.A., and Hellman, J.R., “Interfacial Shear Behavior of Sapphire-
Reinforced NiAl Composites," Mat. Res. Soc. Symp., 194, 293-299 (1990).

More], M. and Saravanos, D.A. “Computational Procedure to Tailor Compliant Layer
Characteristics and Processing History for Minimal Residual Stresses,” HITEMP Review
1990: Advanced High Temperature Engine Materials Technology Program, NASA, CP
10051, HITEMP, 44-1 to 44—9, (1990).

Morozuni, S., Kikuchi, M.and Nishino, T., “Bonding Mechanism between Alumina and
Niobium”, J. Mater. Sci. 16, 2137-44 (1981).

Murr, L.E., Interfacial Phenomena in Metals and Alloys, Addison-Wesley, Reading, MA,
(1975).

Meyers, M.A. and Chawla, K.K., Mechanical Metallurgy: Principles and Applications,
Prentice-Hal], New Jersey (1984).

Nagpal, P. and Baker, 1., "The 131th of Grain Size on the Room-Temperature Ductility
of NiAl," Scr. Met. et Mat, 24, 2381 (1990).

233

Netravali, A.N., Topoleski, L.T.T., Sachse, W.H., and Phoenix, S.L. , “An Acoustic
Emission Technique for Measuring fiber Fragment Length Distributions in the Single-
Fiber-Composite Test,” Comp. Sci. and Tech., 35, 13-29 (1989).

Nicholas, M.G., Surfaces and Interfaces of Cera_mic Materials, ed. L-C. Dufour, Kluwer
Academic, Norwell, MA (1989).

Noebe, R.D., Bowman, R.R., and Eldridge, J .I., “Initial Evaluation of Continuous fiber
Reinforced NiAl Composites,” Mat. Res. Soc. Symp., 194, 323-331 (1990).

Noebe, R.D., Bowman, R.R., and Nathal, M.V., “Physical and Mechanical Properties of
the B2 Compound NiAl,” Int. Mat. Rev., Vol. 38, No. 4, 193-232 (1993).

Ochiai, S., Okuda, S., Shimonmura, K., and Murakami, Y., “Effects of Interfacial
Conditions on the Shear Strength of the Interface and Critical Aspect Ratio in Tungsten-
Copper Composites,” Trans. JIM, Vol. 17, 649-654 (1976).

Ochiai, S. and Osarnura, K , “Multiple fracture of a Fiber in a Single Tungsten fiber-
Copper Matrix Composite,” Z Metallkde, 77, 255-259 (1986).

Ohuchi, ES. and Kohyama, M., “Electronic Structure and Chemical Reactions at Meta]-
Alumina and Metal-Aluminum Nitride Interfaces,” J. Am. Cer. Soc., 74 (6), 1163-87
(1991).

Okarnoto, H., ”Ni-Al,” J. Phase Equilibria, 14(2), 257 (1993).

Paley, M., and Aboudi, J ., ”Micromechanical Analysis of Composites by the Generalized
Cell Model," Mechanics of Materials, 14, 127-133 (1992).

Polvani, R.S., Tzeng, W.-S., and Strutt, P.R., ”High Temperature Creep in a Semi-
Coherent NiAl-NiyAlTi Alloy," Met. Trans, 7A, Jan. 33-40 (1976).

Park, Y.H. and Marcus, HL, Mechanical Behavior of Metal-Matrix Composites, edited
by J .E Hack and M.F. Amateaux, TMS of AIME, 1983. Pp. 65-75.

Parthasarathy, T.A., Marshal], DB, and Kerans, R.J., "Analysis of the Effect of
Interfacial Roughness on fiber Debondiing and Sliding in Brittle Matrix Composites,”
Acta Metal]. Mater., 42, No. 11, 3773-3784 (1994).

Pascoe, RT. and C.W.A. Newey, “The Mechanical Behavior of the Intermediate Phase of
NiAl”, Met. Sci. J., 2, 138-143 (1968).

234

Pauling, L. “The Nature of the Chemical Bond”, 2nd ed., New York, Cornell University
Press, 69 (1948).

Pettit, F.S., "Oxidation Mechanisms for Nickel-Aluminum Alloys at Temperatures
Between 900° and 1300' C," Trans. Met. Soc. AIME 239, 1296 (1967).

Pindera, M-J., Freed, A.D., and Arnold, S.M., "Effects of Fiber and Interfacial Layer
Morphologies on the Thermoplastic Response of Metal Composites," Int. J. Solids and
Structures, 30 (9), 1213-1238 (1993).

Pierik, S.M., Baker, DR. and Kaufman, M.J. , “Preliminary Interfacial Study on NiAl
Matrix Composites”, Fund_amenta_l Relationships Between Microstructure and
Mechanical Promrties of Metal-Matrix Comflites, Eds. P.K. Liaw and MN. Gungor,
TMS, 341-352 (1990).

Pierson, HO. and Randich, E, ”Titanium Diboride Coatings and Their Interaction with
Substrates”, Thin Solid Films, 54, pp. 119-128. (1978).

Polvani, R.S., Tzeng, W.S., and Strutt, P.R, ”High Temperature Creep in a Semi-
Coherent NiAl-NizAlTi Alloy," Metall, Trans. 7A, 33-40 (1976).

Pysher, DJ. and Tressler, R.E, “Creep Rupture Studies of Two Alumina-Based Ceramic
Fibers,” J. Mater. Sci., 27, 423-428 (1992).

Reddy, S.N.S, Betrabet, H.S., Purushotharnan, S.P., and Narayan, C., “Interface Mixing
Between Metals and Ceramics: Classification, Thennochemistry, and Processing,”
Ceram. Eng. Sci. Proc. 10[11-12], 1696-1707 (1989).

Reed-Hill, R.E., Physical Metallurgy Principles, 2nd ed., Litton Educational Publishing,
Inc. (1973).

Rusovic, N. and H. Warlimont, “Young's Modulus of fiz-NIA] Alloys,” Phys. Status
Solidi (a), 53, 283-288 (1979).

Schapery, R.A., ”Thermal Expansion Coefficients of Composite Materials Based on
Energy Principles,” J. of Composite Materials, Vol. 2, no. 3, 380-401 (1968).

Schulson, EM. and Barker, D.R., "A Brittle to Ductile Transition in NiAl of a Critical
Grain Size,” Scripta Metallurgica, 17, 519-522 (1983).

235

Shahinian, P., “High-Temperature Strength of Sapphire Filament,” J. Am. Cer. Soc., Vol.
54, No. 1, 67-68 (1971).

Shankar, S. and Seigle, L.L., “Interdiffusion and Intrinsic Diffusion in the NiAl ( 6) Phase
of the Al-Ni System,” Metall. Trans, 9A, 1978 1467-1476 (1978).

Shida, Y., Stott, F.H., Bastow, B.D., Whittle, DP, and Wood, G.C., “ Development of
Preferential Intergranular Oxides in Nickel-Aluminum Alloys at High Temperatures,”
Oxidation of Metals, Vol. 18, Nos. 3/4, 93-113 (1982).

Speyer, R.F., Thermal Analysis of Materials, Marcel Dekker, New York, NY. (1994).

Stacy, M.H., “Development in Continuous Alumina-Based fibers,” Br. Ceram. Trans. J .,
87, 168-172 (1988).

Stolof f , N.S., "Ordered Alloys-Physical Metallurgy and Structural Applications,"
International Metals Reviews, 29, No. 3, 123-135 (1984).

Stott, RH. and Ashby, DJ., “The Oxidation Characteristics of Electrodeposited Nickel
Composites Containing Silicon Carbide Particles at High Temperature,” Corros. Sci., 18,
183-198 (1978).

Strutt, PR and Kear, B.H., “Creep in Ordered Nickel Base Alloys,” High Temgrature
Ordered Intennetallic Alloys, eds. C.C. Koch, C.T. Liu and NS. Stoloff, Mat. Res. Sym.
Proc. 39, 288-292 (1985).

Taya, M. and Arsenault, R.J., Metal Matrix Comflites: Therrnomechanical Behavior,
Pergarnon Press, New York, NY. (1989).

Taya, M. and Mori, T., in Proc. IUTAM Symp.: Thermo—mechanical Couplings in Solids,
eds. H.D. Bui and Q.S. Nguyen, Elsevier, Amsterdam (1987).

Traumbauer, ER., Hellman, J .R., Shelleman, BL, and Koss, D.A., ”Effect of Fiber
Surface Cleaning Procedures on Tensile Strength of Sapphire Fibers," HITEMP Review
1992: Advanced High Temperature Engine Materials Technology Program, NASA, CP
10104, HITEMP, 16-1 to 16-9, (1992)

Tressler, RE. and Barber, D.J. , “Yielding and Flow of c-Axis Sapphire Filaments,” J.
Am. Cer. Soc., Vol. 57, No. 1, 13-19 (1974).

Tressler, RE, Moore, T.L., and Crane, R.L., "Reactivity and Interface Characteristics of
Titanium-Alumina Composites,” J. Mater. Sci., 8, 151 (1973).

236

Theocaris, P. S. and Demakos, GB, “The Influence of the Elastic Properties and
Dimensions of Coatings on the Stiffness of Encapsulated Fiber Composites,” Composites
Science and Technology, 45, 293-305 (1992).

Touloukiasn, Y.S., Kirby, R.K., Taylor, RE, and Desai, P.D., “Thermal Expansion -
Metallic Elements and Alloys,” IFI/Plenum, New York, 17a (1975).

Turwitt, M. Elssner, G., and Petzow, G., "Manufacturing and Mechanical Properties of
Interfaces Between Sapphire and Niobium," J. Phys. Colloq. (France), 46 (C-4), 123

(1985).

Tyson, W.R., ”Discussion of "Elastic-Plastic Analysis of Deformation Induced by
Thermal Stress in Eutectic Composites," Metallurgical Transactions A, Vol. 6A, 1674-
1677 (1975).

Urquhart, EE, Arnold, S.M., Pindera, M-J., and Aikin, B. J. M., "Simulation of
Experimentally Observed Thermal Expansion Behavior of FeCrAlY Based Composites,"
HITEMP Review 1993: Advanced High Temperature Engine Materials Technology
Program, NASA, CP10051, 54—1 to 54-11 (1993).

Vincent, H., Vincent, C., Ponthenier, J.L., Mourichoux, H., and Bouix, J ., "Elaboration en
Continu d'un Depot Mince de Carbure Ref ractaire en Surface des fibres de Carbone:
Caracterisation de la Fibre C/SiC", in Proc. 3rd European Conf. on Comp. Mat.
(ECCM3), eds. A.R. Bunsell, P. Lamicq, and A. Massiah, Elsevier, 257-264 (1989).

Wang, H.F., Gerberich, W.W., and Skowronek, C.J., “Fracture Meclmnics fo Ti/A1203
Interfaces,” Acta Metal]. Mater., Vol. 41, No. 8, 2425-2432 (1993).

Wakashima, K,Otsuka, M., and Umekawa, S., “Thermal Expansions of Heterogeneous
Solids Containing Aligned Ellipsoidal Inclusions,” J. Compos. Mater., 8, 391- 404
(1974).

Walston, W.S., field, R.D., Dobbs, J.R., Lahrman, D.F., and Darolia, R., "Microstructure
and High Temeprature Strength of NiAl Alloys,” Structural Intennetallics Eds. R.
Darolia, J.J. Lewandoski, C.T. Lui, P.L. Martin, D.B. Miracle, and M.V. Nathal, TMS,
523-532 (1993).

 

Ward-Close, CM. and Partridge, P.G., “A fibre Coating Process for Advanced Metal -
Matrix Composites,” J. Mat. Sci. 25, 4315-4323 (1990).

Warren, R. and Andersson, C.-H., "Silicon Carbide fibers and Their Potential for Use in
Composite Materials: Part II," Composites, 15, 101(1984).

237

Warwick, G.M., Kieschke, RR, and Clyne, T.W. "Measurement and Control of the
Stress State in Sputtered Diffusion Barrier Coatings on Monofilaments,” in Interfacial
Phenomenp in Compsoite Materials F.R. Jones (ed.) Butterwortlr, 267-275 (1989).

 

Warwick, CM. and Clyne, T.W. , “The Micromechanisms of Strain Ratchetting During
Thermal Cycling of Fibrous MMCs,” Fundamental Relationships Between

Microstructure and Mechanical Promrties of Metal-Matrix Commsites, eds, P.K. Liaw
and MN. Gungor, TMS, 209—223 (1990).

 

Warwick, CM. and Clyne, T.W., “Development of Composite Coaxial Cylinder Stress
Analysis Model and its Application to SiC Monofilament Systems,” J. Mat. Sci., 26,
3817-3827 (1991).

Wasynczuk, J.A. and Ruhle, M. in “Ceramic Microstructures ‘86, Role of Interfaces,”
eds. J.A. Pask and A.G. Evans, New York, Plenum Press, (1987).

Waterbury, M., Matikas, T., Karpur, P., and Krishnamurthy, K., “In-situ Ultrasonic
Imaging and Acoustic Emission Monitoring of Single fiber Fragmentation in Metal
Matrix Composites,” TMS Fall Meeting, Pittsburgh, PA, (1993).

Weber, C.H., Yang, J.Y., Lofvander, J.P.A., Levi, CG, and Evans, A.G., “The Creep

and Fracture Resistance of y-TiAl Reinforced with A1203 Fibers,” Acta Metal]. Mater.
Vol. 41, No. 9, 2681-2690 (1993).

Weddle, J.K., "Continuous Ceramic fibers," J. Text. Inst, 81, No. 4, 333-359 (1990).

Wei, W., “The Effect of Long Term Exposure on the Interfacial Chemistry and
Mechanical Properties of Metal Matrix Composites,” F‘undameng Relationships
Between Microstructure and Mechanical Promrties Of Metal Matrix Composites, eds,
M.N. Gungor and PK. Liaw, TMS, 353-370 (1990).

Wiederom, S.M., Hockey, B.J., and Roberts, D.E, “Effect of Temperature on the
Fracture of Sapphire,” Philos. Mag., 28, 783-796 (1978).

Whittenberger, J.D., Viswanadham, R.K, Mannan, SK. and Kumar, K.S., “1200 to 1400
K Slow Strain Rate Compressive Properties of NiAl-Ni2AlTi Base Materials”, High

Temgrature Ordered Intennetallic Alloys 11], CT. Liu, A.I. Taub, N.S. Stoloff and CC.
Koch, eds, Vol. 133, (Materials Research Society, Pittsburgh, PA, 621-626 (1989).

Whittenberger, J .D., Westfall, L.J. and Nathal, M.V., “Compressive Strength of a B2
Matrix NiAl-Nb Intermetallic at 1200 and 1300 K,” Scr. Metal]., 23, 2127-2130 (1989).

238

Whittenberger, J.D., “The Influence of Grain Size and Composition on 1000 to 1400 K
Slow Plastic Flow Properties of NiAl,” J. Mater. Sci., 23, 235 (1988).

Whittenberger, J .D., Gaydosh, B.J., and Kumat, K.S., "1300 K Compressive Properties of
Several Dispersion Strengthened NiAl Materials," J. Materials Sci., 25, 2771-2776
(1990).

Whittle, DP. and J. Stringer, “Improvements in High Temperature Oxidation Resistance
by Additions of Reactive Elements or Oxide Dispersions,” Phi]. Trans. R. Soc. Lond.
(GB), A295 (1413), 309 (1980).

Whitney, J.M. and LT. Drzal, “Axisymmertic Stress Distribution Around an Isolated
Fiber Fragment,” in Toughened Commsites, ASTM Special Technical Publication, 179-
196 (1985).

Wright, M.A., ”The Effect of Thermal Cycling on the Mechanical Properties of Various
Aluminum Alloys Reinforced with Unidirectional Boron Fibers,” Meta]. Trans, Vol. 6A,
129- 134 (1975).

Xia, 2., “Role of Coatings in Axial Tensile Strength of Long Fiber-Reinforced Metal-
Matrix Composites,” Acta Metal]. Mater. Vol 41, No. 7, 2097-2104 (1993).

Yano, T.; Yanagisawa, O., and Ohmuri, M., 'Defromation of a Fiber-reinforced
Composite Induced by Temperature Change and External Loading,” J. Jpn. Inst. Metals,
52, 229-235 (1988).

Yoda, S., Kurihara, N., Wakashima, K., and Umekawa, S., ”Thermal Cycling-Induced
Deformation of Fibrous Composites with particular reference to the Tungsten-Copper
System,” Metal Trans, Vol. 9A, 1229-1236 (1978).

Yoda, S., Takahashi, R., Wakashima, K., and Umekawa, S., ”Fiber/Matrix Interface
Porosity Formation in Tungsten fiber/Copper Composites on Therrna] Cycling," Meta]
Trans, Vol. 10A, 1796-1798 (1979).

Zener, C., "Theory of Growth of Spherical Precipitates from Solid Solutions," J. Appl.
Phys. 20, 950 (1949).

Zhang, H., Anderson, P.M., and Daehn, G.S., “Analysis of Thermally Induced Stress and
Strain in Continuous fiber-Reinforced Composites,” Met. Trans. A, Volume 25A, 1-11
(1994).

HICHIG QRIES

"11111111111111”!