‘Av‘.
: ...m 3.)
3.32.9. =
1.5... Bya‘we

$11....

$11

«mumwfiw
.

vuvvodw
a: as... .

r.
‘ , (:3.
5:22.}.
i .i?
.. :5:

§‘!.‘ It... I V I .
ulru1l ll»... 0 I 3 I1
v0. _ I.zh..,;.v.?.‘y 511;)!«31

in». 1|}

) nun) .7

 
 

3)..» Uh .. :I

. '1 L s. .
I at :uohhlv ring

:4 2.3:.
i: :5..-
a:

\
.hvflc‘lfifis \
I D it

11:.“ [lito’v

. .II; ‘ .1 .2

‘Dktin‘fhflk\

-a’ '0‘; lit, 75‘ ni‘n..1$§l
K . "(nol‘ ll. A

VII’I: .
"v‘i‘l’t-IA lib-r
11,:I~‘llv I.-.)’\Io i.

 

BIID|Iup

IlllfllllllTllllllllllll'llllllll

Z 1293 01563 4813

 

 

 

 

 

 

 

 

This is to certify that the

dissertation entitled

Modification of Mechanical Properties of Sapphire
Fibers by Energetic Ion Implantation

presented by

Heyun Yin

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Materials Science

 

 

    

 

Major professor

“MC/977

MSU is an Affirmative Action/Equal Opportunity Institution 0» 12771

MODIFICATION OF MECHANICAL PROPERTIES OF SAPPHIRE FIBERS BY
ENERGETIC ION IIVIPLANTATION

By
HEYUN YIN

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

DOCTOR OF PHILOSOPHY

Department of Materials Science and Mechanics

1997

ABSTRACT

MODIFICATION OF MECHANICAL PROPERTIES OF SAPPHIRE FIBERS BY
ENERGETIC ION IMPLANTATION
By

Heyun Yin

Effects of 175 keV ion implantation into single crystal sapphire fibers have been
studied, which include bend strength variation, modification of abrasion resistance,
indentation fracture toughness, and sub-surface microstructure alteration. Behavior of
implanted sapphire fibers in N iAl composite matrices was also investigated.

By means of three point bend tests, Weibull characteristic bend strength
measurements showed that implantation of Mg“ and Ar” has modest effects on the bend
strength of unabraded fibers for doses ranging from 1 x 10" to 2 x 1017 ions cm‘z.
However, the abrasion resistance of implanted sapphire fibers can be increased
significantly such that over 90 percent of bend strength can be retained after abrasion,
whereas the same abrasion process causes 50 percent strength loss for unimplanted fibers.
The measured surface compressive stress on implanted fibers using microindentation
technique was approximately 2 GPa. The existence of the compressive stress is found to
be closely related to the observed abrasion resistance increase. TEM observation indicates
that the sub-surface microstructure resulting from ion irradiation was heavily damaged but
still crystalline, and no amorphization was observed. The fabrication process for
production of N iAl/A1203 composites leads to severe strength deterioration in both
unimplanted and implanted fibers.

The relationships between bend strength, surface flaws, abrasion processes,

surface stress distributions, and indentation fracture mechanics are discussed.

 

 

 

 

 

 

Copyright by

Heyun an
1997

 

ACKNOWLEDGMENTS

This project was supported by the U. S. Department of Energy under grant #DE-
FGOZ85ER45205, by the Composite Materials and Structure Center, Michigan State
University, and by Ford Motor Company.

I would like to express my sincerest appreciation and gratitude to Professor D. S.
Grummon, my major advisor, for his encouragement, guidance and immeasurable support
throughout my study at Michigan State University.

I would like to extend my appreciation to the Ford Motor for provision of the
Varian 350-D ion implantation machine, to Dr. R. Noebe at NASA Lewis Research Center,
Cleveland, OH, for provision of MA] alloy, and to Dr. J. Lucas for his assistance with
nanO-indentation experiments at Sandia National Laboratory.

1 want to extend grateful acknowledgment to my advisory committee Dr. T. Bieler,
Dr. M. Crimp, Dr. D. Grummon, Dr. W. McHarris and Dr. H. Schock for their guidance.

Also, I want to give my thanks to my colleagues and friends for their great help.

A special tribute is owed to my wife, Rong, for being so understanding and

supportive throughout my study.

iv

TABLE OF CONTENTS

Table of Contents ........................................................................... v
List of Tables .................................................................................. vii
List of Figures ................................................................................. viii
I. Introduction ............................................................................. 1
II. Review of Literature ................................................................... 5
2.1 Physical and Mechanical Properties of (It-Alumina .......................... 5
2.1.1 Crystal Structure ........................................................ 6
2.1.2 Intrinsic Disorder Mechanisms and Diffusion ...................... 9
2.1.3 Mechanical Properties .................................................. 10
2.2 Continuous Sapphire Fiber Reinforced NiAl Composites .................. 14
2.3 Strength Degradation of Sapphire Fibers due to Surface Damage ......... 17
2.4 Ion Implantation Process ....................................................... 18
2.4.1 The Cascade Model .................................................... 19
2.4.2 Radiation Damage in Solids ........................................... 24
2.4.3 Radiation Effects ....................................................... 30
2.4.3.1 Rapid Quenching Effects ..................................... 31
2.4.3.2 Radiation Enhanced Diffusion ............................... 31
2.4.3.3 Precipitation Effects ........................................... 32
2.4.3.4 Residual Stress Induced by Ion Implantation .............. 33
2.4.3.4.] Ion Implantation-Induced Stress State ............. 34
2.4.3.4.2 Surface Stress Model ................................ 35
24.3.4.3 Experimental Determination of Surface Stress. . . . 38
2.4.3.5 Radiation Hardening and Softening Effects ................ 39
2.5 Modification Of Hardness, Toughness and Wear Resistance of Sapphire
by Ion Implantation ............................................................. 42
2.6 Post-Implantation Heat Treatment ............................................. 44
2.7 Summary of Literature Review ................................................ 45
III. Experimental Procedures .............................................................. 47
3.1 Materials .......................................................................... 47
3 .2 Ion Implantation Process ....................................................... 48
3.2.1 Ion Beam generation ................................................... 48
3.2.2 Target Set Up and Affecting Factors ................................. 54
3.3 Abrasion Procedure ............................................................. 59
3.4 Three Point Bend Tests ......................................................... 61
3.5 Indentation Fracture Toughness Tests ........................................ 64
3 .6 Post-implantation Annealing ................................................... 67
3 .7 Fabrication of NiAl/A1203 Composite diffusion Bonding .................. 67
3.8 Preparation of TEM Samples .................................................. 72
3 .9 Electron Channeling Pattern (ECP) and
Atomic Force Microscopy (AFM) ............................................ 73

3.10 Transport Calculation of The Range and Damage Profile .................. 74

IV. Experimental Results and Analysis .................................................. 78
4.1 Three Point Bend Strength without Abrasion ................................ 78
4.2 Effect of Abrasion on Bend Strength ......................................... 85
4.3 Surface Residual Stress Determined by Micro-Indentation Fracture ...... 93
4.4 Nam-Indentation Hardness .................................................... 100
4.5 Annealing Effects ................................................................ 102
4.6 Sapphire Fiber Strength Degradation During NiAl Fabrication ............ 110
4.7 Fiber Surface Morphology ..................................................... 117
4.8 Fracture Surface Morphology .................................................. 125
4.9 TEM Investigation of Microstructure ......................................... 131
4.9.1 TEMObservation ...................................................... 131
4.9.2 Channeling Effects ..................................................... 142
V. Discussion .............................................................................. 143
5.1 Chemical versus Ballistic Effects .............................................. 143
5 .2 Residual Surface Stress and Bend Strength .................................. 147
5.3 Residual Surface Stress and Fracture Behavior .............................. 158
VI. Summary of Results and Conclusions ............................................... 173
VII. Bibliography ............................................................................ 177

 

Table 2.1

Table 3.1
Table 3.2
Table 4.1
Table 4.2
Table 4.3
Table 4.4

Table 4.5
Table 5.1

List of Tables

Page
Comparison of Physical Properties of Several Materials Relevant to
High Temperature Interrnetallic Composites ............................. 15
Properties of Sapphire Fiber ............................................... 48
Median Ranks ............................................................... 63

Three Point Bend Strength of Sapphire Fiber with Different Doses. . 80

Effect of Abrasion on Bend Strength of Sapphire Fibers .............. 88
Micro-Indentation Crack Length at Different Dose ..................... 93
Characteristic Bend Strength of Sapphire Fibers at Different

Annealing Temperature ..................................................... 104
Chemical Composition of B—NiAl ......................................... 123
Comparison of Tensile Stress in Near-Surface Zone ................... 153

vii

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6

Figure 3.7

Figure 3.8
Figure 3.9

Figure 3.10
Figure 3.11

Figure 4.1

Figure 4.2

List of Figures

(a) Schematic of Stacking Sequence of Anion and Cation in A1203

Structure ..................................................................... 7
(b) Crystal Structure of A1203. ............................................ 8
Schematic of Cascade Trajectories in Planar BeO Caused by 5-keV

I127 Bombardment ........................................................... 21

The Number of Displaced Atoms in the Cascade as a Function of

PKA Energy according to the Model of Kinchin and Pease ........... 23
A Schematic Representation of The Variation in Size and Position

of Implantation Induced Amorphous ..................................... 27
Schematic Representation of The Principles of Implantation Induced
Stress Model ................................................................. 36
The Variation of 25g Knoop Microhardness with Dose for Ti
Implanted into Sapphire .................................................... 40
Annealing Process of an Amorphous Material .......................... 46
Schematic Description of Varian 350 - D Ion Implanter ................ 49
Schematic of Freeman Ion Source Structure ............................. 51
Mg and Ar Mass Spectrum Generated in 350D Implanter ............. 53
Schematic of Rotating Fixture for Ion Implantation .................... 55
Oscilloscope Display of Beam Current ................................... 56
Schematic of Abrasion Apparatus ......................................... 60
Schematic Representation of the Crack geometry around a Vickers
Indentation ................................................................... 65
Schematic of Cutting Tool for MA] Diffusion Bond Assembly ....... 69
Schematic of EDMed NiAl Diffusion Bond Plate ....................... 7O
Schematic Of N iAl/Sapphire Fiber Diffusion Bond Assembly ........ 71

The Distribution of Mg+ and Ar+ Implanted into Sapphire Fiber At
Different Energy, Calculated by TRIM (90.05) ......................... 70

Weibull Plot of Bend Strength for Mg” Implanted Sapphire without
Abrasion .....................................................................

Weibull Plot of Bend Strength for Ar+ Implanted Sapphire without

81

viii

 

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6
Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12
Figure 4.13
Figure 4.14

Figure 4.15

Figure 4.16

Figure 4.17

Figure 4.18

Figure 4.19

Figure 4.20

Abrasion ..................................................................... 82
Comparison of Bend Strength of Mg” and Ar+ Implanted Sapphire

Fibers without Abrasion ................................................... 83
Effect of Cleaning Methods and Testing Environments on Bend
Strength of Unimplanted Sapphire Fibers ............................... 84
SEM Observation on the Fracture Surface for Both Unimplanted

and Implanted Sapphire Fiber. ............................................ 86
Effect of Abrasion on Bend Strength of Sapphire Fibers .............. 89
Weibull Plot of Bend Strength for Mg+ Implanted Sapphire Fiber
Abraded 20 and 60 minutes ................................................ 90
Weibull Plot of Bend Strength for Ar+ Implanted Sapphire Fiber
Abraded 20 and 60 minutes ................................................ 91
ESEM Images on the Surface Topography of Sapphire Fibers for
Different Implantation and Abrasion Conditions ........................ 92
SEM Images of The Indentation Traces on Sapphire Fibers at

Different Doses (load 100 grams) ......................................... 95
Variation of Integrated Surface with Doses of Mg“ Implantation into
Sapphire Fibers ............................................................. 96
Variation of Indentation Fracture Toughness with Doses .............. 97
SEM Observation on the Trace of Lateral Crack Propagation ......... 99

Comparison of Nam-indentation Hardness and Bend Strength
of Sapphire Fibers .......................................................... 101

Effects of Annealing Temperature on the Bend Strength Of Implanted
Fibers ......................................................................... 105

Weibull Plot of Bend Strength for Ar“ Implanted Sapphire Fiber at
Annealing Temperature of 1200 °C for 2 hours ......................... 106

Weibull Plot of Bend Strength for Mg+ Implanted Sapphire Fiber at
Annealing Temperature of 1200 °C for 2 hours ......................... 107

Weibull Plot Of Comparison of Bend Strength for Mg+ Implanted
Sapphire Fiber Annealing 2 hours in Air Environment ................. 108

Comparison of Bend Strength Weibull Plot for Unimplanted Sapphire
Fibers and As-received Sapphire Fibers Etched From NiAl/A1203
Composite ................................................................... 1 12

Comparison of Bend Strength Weibull Plot for Implanted Sapphire
Fibers Etched from NiAl/Sapphire Fibers Composite .................. 113

Figure 4.21
Figure 4.22

Figure 4.23
Figure 4.24

Figure 4.25

Figure 4.26

Figure 4.27
Figure 4.28

Figure 4.29

Figure 4.30

Figure 4.31

Figure 4.32

Figure 4.33

Figure 4.34

Figure 4.35

Figure 4.36

Figure 4.37

Figure 4.38

Figure 4.39

Fragmentation of Sapphire Fibers Etched From NiAl/A1203

Composite ................................................................... 114
Effects Of Different Processes on the Bend Strength Weibull Plots
for Both Unimplanted and Mg Implanted Sapphire Fibers ............ 115

SEM Observation on a Clean As-received Sapphire Fiber Surface . .. 118

SEM Observation on the Surface of Sapphire Fiber Subjected to
Chemical Etch Solution .................................................... 119

SEM Observation on the Surface of Extracted Fibers from N iAl/Ale3
Composite ................................................................... 120

High Magnification SEM Observation on the Surface of Extracted
Fibers From NiAl/Ale3 Composite ...................................... 121
EDS Composition Analysis on the Residues in Extracted Fibers . . .. 122

SEM Observation of Ridge on the Surface of Etched Sapphire Fibers
from NiAl/Ale3 ............................................................ 124

SEM Observation on Fracture Surface Variation with Three Point
Bend Strength. .............................................................. 126

SEM Observation on Fracture Surface Feature of Sapphire Fiber
Which Are Subjected to Three Point Bend Strength .................... 127

SEM Observation on Fracture Surface Feature of Sapphire Fiber
Which Fractured During Consolidation Process of NiAl/A1203

Composite ................................................................... 129
SEM Observation on Internal Pores in Sapphire Fibers Which Fracture
During Consolidation Process of NiAl/AIZO3 Composite .............. 130
SEM Observation on Internal Pore of Sapphire Fiber Which Fractured
during Consolidation (Pore on Edge) .................................... 132
TEM Observation on 7 x 10“ Mg” cm‘2 Implanted Sapphire Fiber at
Room Temperature ......................................................... 135
TEM Observation on 2 x 10'7 Mg+ cm’2 Implanted Sapphire Fiber at
Room Temperature ......................................................... 136
High Magnification TEM Observation on 2 x 1017 Mg+ cm’2 Implanted
Zone .......................................................................... 137
TEM Observation on 1 x 1016 Ar“ cm'2 Implanted Sapphire Fiber at
Room Temperature ......................................................... 138
High Magnification TEM Observation on 1 x 10"5 Ar+ cm’2 Implanted
Zone .......................................................................... 139
Electron Channeling Pattern of Sapphire Fibers ........................ 141

Figure 5.1
Figure 5.2

Figure 5.3
Figure 5.4
Figure 5.5

Figure 5.6

Figure 5.7

Figure 5.8

Figure 5.9
Figure 5.10

Figure 5.11

The System of MgA1204-A1203 ............................................ 144

Schematic of V-notch Circular Surface Flaw on Sapphire Fiber and
Theoretical Calculation of Strength Changes with Crack length ..... 148

Schematic of Implanted Zone and Stress Distribution .................. 154
AFM Topography on Sapphire Fiber Surface ........................... 159
SEM Observation on the Interface between Sapphire Fiber and

Abrasive Particles ........................................................... 160
AFM Image on Surface Profile and Surface Roughness Analysis

on Abrasive Paper .......................................................... 161
Schematic of Radial and Lateral Cracks Evolution Under Point
Indentation ................................................................... 162
Curvilinear Coordinate System for Boussinesq Axially Symmetric

Point Load P ................................................................. 165
Principal Stress Contour at Head of Point Load ........................ 166
Stress Distribution Beneath Sharp Point Indentation .................. 167
Stress Distribution during Indentation Process .......................... 169

 

CHAPTER I

INTRODUCTION

Alpha-aluminum oxide (Oi-A1203), commonly called alumina in its polycrystalline
form, or sapphire in its single crystal form, is an extremely stable metal oxide with a
melting point of 2323 K. Desirable characteristics of single crystal sapphire fibers for
reinforcing composite materials include high room temperature strength and high modulus.
Sapphire fibers exhibit high strength and excellent creep resistance at temperatures as high
as 1500°C [T ressler, 1992] and are thus an excellent candidate for use in metallic-,
intermetallic-, or ceramic-matrix composites intended for use in the hot and corrosive
environments found in projected aerospace, energy conversion and advanced transportation
applications. The most comprehensive application of sapphire fibers, and where much
work has been undertaken, is the use of sapphire fibers as reinforcement for B—NiAl
intermetallic matrices [Darolia, 1991; Shah, 1993; Bowman, 1995].

The main deficiency of sapphire fiber is its high sensitivity to surface flaws, or any
surface damage caused by abrasion during fiber-fiber contact, hard particle impact, or
cleaning procedures. Surface damage causes a dramatic reduction in strength, and thus
may limit the potential of sapphire fibers as a reinforcement phase in advanced composite
materials. In previous studies, sapphire fibers have been found to lose 30 percent of their
ambient temperature tensile strength during handling and cleaning ['I‘rumbauer, 1992], and
33 to 50 percent after fabrication by the power cloth technique [Draper, 1992].

Realistic fabrication protocols for composite components also place stringent
demands on the bend strength of the fibers [Morscher, 1992], which is sensitive to pre-
existing or process-induced surface flaws. In addition to abrasion during handling and lay-
up, surface damage during processing also comes from surface chemical reactions with

matrices and impurities, chemical reactions with binders, residual stresses resulting from

fiber-matrix coefficient expansion mismatch, and hot processing stress states [Draper,
1994; Bowman, 1995]. It is therefore of interest to develop methods to provide sapphire
fibers with improved resistance to process-induced surface damage during high-
temperature consolidation and service.

The realization that the surface plays a critical role in the mechanical properties of
sapphire fibers has resulted in applications of several techniques by which sapphire fiber
surfaces are treated. These include surface sizing ['I‘rumbauer, 1992], surface glazing and
quenching [Kirchner et al., 1969], and impurity doping [Sayir, 1992].

Sizing involves organic coating on the fiber surface with materials such as E15-LV
hydroxypropyl methylcellulose (Dow Methocelm). However, impurities (such as NaCl,
Fe and sulfated ash) found in the sizing have the potential to degrade fiber strength during
composite processing [Trumbauer, 1992].

Glazing and quenching by packing sapphire rods in chromium or calcium
compounds, followed by firing at high temperature [Kirchner et al., 1969] introduce
compressive surface layers. The compressive stress results from a chemical reaction taking
place at the surface and yielding products with greater volume. For instance, the volume
increases 31% after the reaction: Ca0 + 2 A1203 => CaAl4O7. It is apparent that the
sapphire surface is damaged by such reactions.

Impurity doping has been employed to improve the mechanical properties of
alumina by a combination of solution and precipitate strengthening. Studies by Sayir et a1
[1992] showed that the bend strength of undoped laser heated float zone sapphire (LI-IFZ)
fibers degraded with an increase in temperature at low and intermediate levels
(25<T<900°C), while Mg“, Ti“+ doped and Mg2+ + Ti4+ co-dOped sapphire did not show
a strength degradation in the same temperature range. However, solubility of the aliovalent

species is limited by the necessity to form charge-compensating point defects in the bulk,

which may in turn adversely affect creep properties.

 

 

Ion implantation has been used to improve the surface properties of solid materials
without altering their bulk properties. During the ion implantation process there is a
physical modification of a shallow surface region, normally 0.3 to 0.4 pm thick. This area
exists either as a metastable and strained crystalline system, as an amorphous phase, or as a
surface layer in which fine precipitates exist as a result of implantation and/or subsequent
heat treatment of the implanted material. Further, these highly defected structures may
produce large residual stress states in the near—surface and sub-surface regions, as deep as
~ 0.4 um. As will be shown, these stresses have interesting effects on strength and
abrasion resistance of sapphire fibers. The possible effects of implantation-induced
stresses in modifying surface behavior are potentially important and can involve changes in
crack nucleation and crack growth behavior, as well as in plasticity mechanisms [Bumett,
1985; Hioki, 1986; Tressler, 1992]. The technique may thus be suited to the task of
producing a useful near-surface modification of sapphire fibers without adversely changing
their bulk properties.

In this study, Mg” and Ar‘ ion implantation was used to improve the surface
properties Of sapphire fibers. Modification of surface properties by ion implantation was
characterized by measurement of the retention of three-point bend strength of fibers
subjected to a controlled abrasion process, and by measurement of surface compressive
stress, which is believed to effectively retard propagation of surface flaws during abrasion.
Both implanted and unimplanted fibers were also consolidated in liNiAl matrices by a
standard hot pressing process. Sapphire fibers extracted from the consolidated composites
were then studied by measurement of residual bend strength, fiber fragmentation, and
observation of surface and fracture surface morphology. Weibull statistical methods were
used to establish the relationship between failure probability and characteristic strength. In
addition, a theoretical calculation of the distribution and cascade damage accumulation from

energetic ions implanted into sapphire was carried out using the well-known TRIMl ion

 

' TRIM stands for W [Ziegler, 1985], detailed information about TRIM is given
in Section 3.10, page 74.

 

transport code. The effects of compressive stress on the stress intensity factor at surface
crack tips were also evaluated. Transmission electron microscopy microstructural studies
of implanted sapphire fibers transverse sections were used to correlate the existence of a
buried compressive stress zone with fracture behavior.

The following chapter presents a review of recent work on the basic properties of
sapphire fibers, strength degradation of fibers caused by surface damage, and the use of
ion implantation for modification of surface properties. Chapter III describes the
experimental details and procedures. The results and discussion are presented in Chapter
IV. Further discussion on the relationship between fracture mechanics and the abrasion
process, and the relationship between fracture behavior and compressive stress induced by
implantation is given in Chapter V. Finally, a summary of results and conclusions from the

work are presented in Chapter VI.

CHAPTER II
REVIEW OF THE LITERATURE

As has been stated in Chapter I, sapphire fibers suffer from a pronounced
dependency of strength on surface damage, which may occur during handling and
composite processing. The issue of surface damage has attracted extensive research
[T rumbauer, 1992; Draper, 1992; Sorensen, 1993; Shah, 1993; Asthana, 1995; Tewari,
1993; Bowman, 1995]. These investigations attempted to uncover possible mechanisms
for surface damage and strength degradation, and proposed protective methods to shield
sapphire fibers from surface damage. In this chapter, the literature on properties and
behavior of single crystal sapphire fiber is reviewed, beginning with physical and
mechanical properties of sapphire, followed by a discussion of the properties of continuous
sapphire fiber reinforced NiAl intermetallic composites. Surface property modification by
ion irradiation, as well as effects of irradiation, is reviewed in a subsequent section. Basic

principles of ion implantation are also reviewed, and the principles of some measurements

of mechanical properties are presented.

2. 1 Physical and Mechanical Properties of a—Alumina
(It-Alumina has useful properties including thermal expansion well matched to metal

matrices, high heat capacity, good thermal shock resistance, desirable sintering behavior,
and excellent mechanical properties. Therefore, it has widespread practical applications for
structural components, insulating materials and other specialized systems in the Optical,
nuclear and electronic industries. In this section, a detailed review will be focused on its

crystalline structure, disorder mechanisms and mechanical properties.

2.1.1 Crystal Structure

(it-Alumina has a rhombohedral-hexagonal crystal structure with the space group
R3C, which can be described as having oxygen ions in a hexagonal closest packed
arrangement with the aluminum ions occupying two thirds of the octahedral interstitial
sites. As is conventional with rhombohedral solids, sapphire is usually described in terms
of a hexagonal unit cell with Miller-Bravis indices. The lattice parameters of the structure
are as follows: a0 = 5.12 A, or = 55°17', n = 2 (rhombohedral); or equivalently, ao = 4.75
A, co = 12.97 A, n = 6 (hexagonal). Figure 2.1 schematically shows a schematic of the
stacking sequence of anions and cations in the A1203 structure [Kronberg, 1957].

In the structure of (it-alumina, large oxygen ions are stacked in the sequence A-B-A-
B, thus forming a hexagonal closest packed array of anions. The cations are placed in the
octahedral sites of this basic array and form another type of close packed planes which are
inserted between the oxygen layers. To maintain charge neutrality, only two thirds of the
octahedral sites available are filled with cations. Since the vacant octahedral sites also form
a regular hexagonal array, three different types of cation layers can be detailed, depending
on the position of the vacant cation site within the layer, namely a, b, and c, stacking in the
sequence a-b-c-a-b—c. Therefore, the complete stacking sequence of anion and cation layer
is A-a—B-b-A-c-B-a-A-b-B-c-A, repeating only after the sixth oxygen layer, or after the
sequence a-b-c is repeated twice.

The structure of (it-alumina results in coordination of 6 and 4 for the cation and the
anion, respectively. The ionic radii for this coordination are 0.053 nm for Al3+ and 0.138
nm for 02‘ [DOrre and Hiibner, 1984].

In addition to Ot-A1203, there are many other forms of alumina. One of them is y-

A1203, which has a simple cubic crystalline structure.

[1100] [1‘2’10] [oirol

 

 

 

mic] ll 1001
O oxygen
. aluminum
0 empty octahedral site

Figure 2.1 (a) Schematic of stacking sequence of anion and cation
in A12 03 structure. The large open circles are the oxygen ions,
which form an hcp array; the small filled circles are the alunimum
cations, which stack in an ABC sequence; and the small Open
circles are normally empty octahedral interstices. A] , A2 , and A3
are the smallest lattice vectors lying in the basal plane. (After
Kronberg: 1957)

 

 

 

 

 

 

[1010] i ‘3‘
— . 3 s
[1120] Q ‘ . 110]
- .l [1100]
[0110] [12I0]
0 aluminum
. hole

Figure 2.1 (b) Distribution of aluminum ions and holes on the hexagonal
unitcell. The smallest rhombohedral cell describes the position of the

cations in Ot-A1203.

2. l .2 Intrinsic Disorder Mechanisms and Diffusion

Before reviewing the defects in A1203, let us introduce KrOger—Vink notation. In
this notation, positive and negative effective charges are marked by the superscript“ o ”
and “ ’ ”, respectively. For the indication of effective charge neutrality the symbol “ x ” is
used. The position of the ion is given by subscript which denotes the regular lattice sites of
the anion or cation by their respective symbols, or an interstitial Site by the symbol “ i ”.
Thus Al.” is an interstitial Al ion, 00" represents an oxygen ion in its regular lattice site.

DOrre and Hfibner [1984] summarized the possible disorder mechanisms in A1203,

namely:
Frenkel disorder of A1 AIM" 4: Al.“ + V A,”
Frenkel disorder of O 00" c: O," + V0”
Schottky disorder null 4:) 2V M... +3Vo“
Electronic disorder null 4: e' + h‘ .

Where e stands for an electron, h for a hole, and null means electronic neutrality.

These types of defects in the ionic Ot-alumina crystal are important because they are
related to intrinsic or extrinsic diffusion mechanisms in a-alumina, where diffusion
controls the properties of the material at elevated temperature.

Intrinsic diffusion means that the point defects through which the atomic mobility is
accomplished are created exclusively by temperature. Atomic disorder of the crystal is
caused only by thermal entropy. The equilibrium constants of defect formation and the
formation energies indicate that intrinsic defect concentrations are very small [Kingery,
1976]. Typical high-temperature intrinsic defect concentrations in a—alumina are in the
range 10'9 or lower, and the concentration of Schottky defects in Ot-alumina near the
melting point is of the order 10'”. That means that in tit-alumina, intrinsic defects should
not be important, even close to the melting point, because the defect number is smaller by
several orders of magnitude than the number of native defects formed to compensate for

aliovalent impurities [DOrre and Hiibner, 1984]. Many investigations have been done on

10

the relation of intrinsic diffusion and electrical conductivity, optical absorption, sintering
and creep [Yee and Kroger, 1973; Dutt and Kroger, 1975].

In contrast to sluggish intrinsic diffusion where intrinsic point defect concentrations
are very small in A1203, extrinsic diffusion arising from aliovalent impurities or dopants
strongly affect the diffusion—controlled properties in A1203 because they greatly increase the
defect concentration. For instance, Mg is thought to occur in A1203 as Mg A; . The
neutrality condition is given by [Mg A; ] = 3 [Ali'”] for Frenkel disorder of Al to be
dominant, and [Mg M'] = 2 [Vo”] for Schottky disorder to be dominant. Therefore, for even
a very small dopant content, in the ppm range, the concentration of native defects that
results from the requirement of charge balance is much greater than the number of defects
formed by the intrinsic disorder mechanism. If Frenkel disorder dominates, the coefficient
of Al self-diffusion can be written as:

DA, oc [Al,”']exp(-AH/RT)
where [Alim] = 1/3 [Mg “'1 which is the order of 10‘, greater than the intrinsic disorder
concentration of 109.

In summary, defect formation and mobility are important to Ot-A1203. Small

amounts of impurities may control the concentration of the diffusing defects and lead to

dissolution or precipitation process, which is directly related to the properties of Ot-A1203.

2.1.3 Mechanical Properties

(it-Alumina stands out as a useful engineering structural ceramic due to high
hardness, high wear resistance, low friction coefficient, high resistance to corrosion by
practically all chemical reagents, a very high resistance to hot corrosion in air, and high
thermal stability. Meanwhile, it has well known drawbacks, i.e., its low strength as
compared with the theoretical strength, the large statistical spread Of its strength values, its
great brittleness, its low toughness, and large susceptibility to thermal and mechanical
shock loading.

11

Single crystal Ot-alumina (or sapphire) has very high elastic properties. For
instance, Young’s modulus is 461 GPa parallel to the c—axis, and 335 GPa perpendicular to
the c-axis [W atchman, 1960]. Its tensile strength is 2.1 to 3.4 GPa at rOom temperature
(parallel to c-axis), and its flexural strength is 1035 MPa parallel to the c-axis and 760 MPa
perpendicular to the c-axis. Knoop hardness values of 1900 parallel to the c-axis, and
2200 perpendicular to the c-axis are reported by the manufacturer [Saphikon, 1992].

The fracture energy of single crystal (it-alumina also shows anisotropy. Wiederhem
[1969] found preferential cleavage on the {loiol and {T012}p1anes, where the fracture
energy values are 7.3 J/m2 and 6.0 J/m2 at room temperature, respectively. The fracture
toughness is 2.47 MPax/t—ri .

The crystallographic anisotropy of sapphire is also illustrated by its plastic
deformation at elevated temperature. There are three known slip systems in sapphire:
(0001)1/3<1 1§O> basal slip, {1I00}<1T01> prism slip, and {01 i1 }1/3<oi 12> (or
possibly { 0T 12} 1/3<01T1> or { 1T23}1/3<01'1'1>) pyramidal slip [LagerlOf, 1984].
For c-axis sapphire in tension, four different fracture mechanisms dominate at different
temperature ranges [Jones, 1990]: 25-9000C, 900-16000C and 1600-20000C. The tensile
strength of c-axis sapphire decreases from 3.8 GPa at room temperature to 1.4 GPa at
600°C in the lowest temperature region. It is believed that the mechanisms of brittle failure
and stress corrosion failure are responsible for the strength degradation [Wiederhom,
1978]. The second temperature region starts around 900°C and extends to the onset of
macroscopic plastic deformation at 1600°C. In this range, tensile strength rises back to
2 GPa at 900°C, then decreases with further temperature increase. At 2000°C, tensile
strength is only 0.5 GPa. Slow crack growth and plastic deformation are major
mechanisms for the tensile deformation in these two temperature regions.

Sapphire is brittle at low temperatures and therefore the strength of sapphire at room
temperature is highly flaw dependent. According to Griffth theory, a brittle material

contains small flaws which act as stress concentrators when an external stress is applied.

12

Due to the absence of plastic deformation, stresses cannot be relaxed by local plasticity, and
the local stress in the vicinity of the most severe flaw may reach the theoretical strength of
the material, causing failure. The relation between strength and crack length can be
described byO' = 1/ Y (27E / C)V2, where 0' is the strength, 7 is the fracture energy, E is
Young’s modulus, C is the flaw size, and Y is a geometrical factor.

DOrre and Hiibner [1984] summarized the types of flaws which have been
observed in single crystal tit-alumina. They are (1) intrinsic flaws, including pores,
impurity particles, dislocation-nucleated flaws, and twin-nucleated flaws; and (2) extrinsic
flaws, including edge flaws due to large pores, stepped flaws due to pore groups, and
regular machining-induced flaws.

Residual pores come from shrinkage voids inherent in the single crystal (it-alumina
fiber growth process where voids form by entrapment of liquids behind the solid surface,
or from incomplete sintering in polycrystalline alumina. Fracture-initiating pores are
usually located underneath the surface and must have a sufficient size to act as fracture
origins because they must compete with other Strength-controlling defects such as surface
defects induced by machining. Pore groups beneath the surface can act as fracture origins.
Upon application of a load, individual pores of the group can link together to produce a
large flaw [Evans and Toppin, 1972].

Dislocation-nucleated flaws are associated with dislocation slip systems. At high
temperature, basal slip systems, prism slip systems, or pyramid slip systems can be
activated during loading. The pile-up or interaction of generated dislocations can serve as a
stress concentrator, assisting crack extension. Pollack and Hurley [1973] observed the
strain rate dependence of the strength of sapphire whiskers and attributed the failure to the
Orowan mechanism Of crack extension.

Mechanical twinning can also cause failure of (It-alumina. Twins on both basal and

rhombohedral planes have been found. For instance, the basal twinning system, i.e.,

(0001)<1I00> has been described by Kronberg [1957], whereas the rhombohedral

 

 

13

twinning system is { 10T1}<10I2> [Heuer, 1966]. Preferential cracking along a twin-
crystal band was observed by Stoel and Conrad [1963]. In the bending fracture of single
crystal (it-alumina, Heuer [1966] found the presence of deformation twinning. As
compared to plastic deformation by dislocation glide, twinning is favored by low
temperature and high strain rate. Twinning is a very important feature in the mechanical
behavior of alumina because it occurs during most types of mechanical loading.

The possibility of plastic deformation during room temperature indentation and
abrasion of sapphire has been verified by several researchers [I-Iockey, 1971; Stijn, 1961;
Becher, 1975; Duwell, 1962]. The occurrence of plastic deformation in this normally
brittle material is considered to be a consequence of the nature and magnitude of the local
stresses developed under a pointed indenter or an irregularly shaped abrasive particle. It
has also been observed that high dislocation densities can be produced in the near-surface
region by mechanical polishing with a fine diamond abrasive compound, and that plastic
deformation by both slip and twinning occurs during microhardness indentation at room
temperature. Thus, deformation during abrasive surface finishing can influence the
subsequent mechanical behavior by introducing a variety of surface damage artifacts.

Plastic deformation properties also control the abrasive wear behavior of sapphire.
Abrasive wear is the result of a mechanical machining operation such as sawing, grinding,
lapping, and polishing. It may also occur during solid particle indentation. Abrasive
particles are small and irregular in shape so that a sharp comer of the particle is usually in
contact with the substrate. This small area produces a high localized stress concentration,
and the particles penetrate the surface. Moving the particle along the surface generates a
long plastic groove and a complex crack pattern. During the abrasion process, the material
removal involves plastic flow and cracking. The lateral cracking mechanism accounts for
the occurrence of plastic deformation and fracture [Lawn, 1975, 1980, 1984].

It can be concluded that mechanical properties of (rt-alumina vary with crystalline

orientation, surface flaws, temperatures and environments.

14

2.2 Continuous Sapphire Fiber Reinforced NiAl Composites

The intermetallic NiAl compound offers new opportunities for developing low-
density, high strength structural alloys which might be used at temperatures higher than
currently possible with conventional titanium and nickel-base alloys. B—phase NiAl (50
at% Ni, 50 at% Al) has four key advantages [Darolia, 1991]: its density of 5.95 g/cm3 is
approximately two-thirds the density of nickel-base superalloys; its thermal conductivity is
four to eight times that of nickel-base superalloys; it has excellent oxidation resistance; and
its simple ordered bcc (B2—CsCl) crystal structure makes plastic deformation potentially
easier compared to other intermetallic compounds.

However, the widespread use of MA] as a high temperature structural material has
been limited because of its poor room temperature ductility and toughness, and its poor
high temperature strength. The ductility is highly dependent on aluminum content, grain
size, impurity content and texture, and the ductility increases with temperature. It has been
found [Darolia, 1991] that the tensile ductility of polycrystalline NiAl at room temperature
is about 0 ~ 2%, but rises to about 6% for <110> single crystal NiAl containing 0.1% Hf.
For single crystal NiAl, the plastic deformation behavior is highly anisotropic.

The fracture toughness of binary NiAl is also highly anisotropic. For example, a
typical value of about 8 MPax/in at ambient temperature is obtained from a sample oriented
in the <100> direction while 4 MPax/Z was found for the <110> direction [Darolia,
1991]. Like ductility, the fracture toughness increases with temperature due to increased
plasticity at the tips of the growing cracks.

The yield strength of MN drops rapidly with increasing temperature. The tensile
strength of single crystal NiAl oriented in <100>, for example, decreases from 880 MPa at
room temperature to 200 MPa at 1600°C [Darolia, 1991]. It is obvious that the high
temperature strength of unalloyed NiAl requires improvement to be competitive with the
superalloys. There are six ways by which the strength of NiAl can be improved:

elimination of grain boundaries, solid-solution strengthening, metallic precipitate

15

strengthening, intermetallic-precipitate strengthening, dispersion strengthening, and
composite strengthening.

The use of sapphire fibers as a reinforcement in the MA] is well known to improve
its high-temperature strength [Shah, 1993; Bowman, 1995]. While a relatively weak fiber-
matrix bond is desirable in brittle matrix composites from the fracture toughness
perspective, a strong bond is a prerequisite for high-temperature strengthening, where the
fibers are the load-bearing constituents. In the case of MA], rather than improving the
room temperature bond, the use of sapphire fiber as a reinforcement material has been
recommended to improve its hi gh-temperature strength with a strong fiber-matrix bond
[Asthana, 1995].

The reason that sapphire fiber was chosen as one of the most compatible
reinforcement candidates in MA] composite is that it is chemically stable in almost all the
intermetallic compounds, with well matched coefficients of thermal expansion, and also
because of its commercial availability. For example, the coefficient of thermal expansion of
sapphire fiber is 9.1 x 106/C at 1400 °C, much closer to 16 x 106/C of MA] than either
SiC (5.5 x 106/C) or Si3N4 (3.7 x IO‘IC). Table 2.1 lists a comparison of the physical
properties of several reinforcement candidate materials [Shah, 1993].

Table 2.1. Comparison of Physical Properties of Several Reinforcement Candidates
Materials in MA] Interrnetallic Composites

 

Materials Density (g/cm3) TIll (0C) Modulus (GPa) CI'E (106/C) Temperature 0C

 

 

NiAl 5.96 1638 177 16.0 1000
A1203 4.0 2050 524 9. 1 1400
SiC 3 .2 2600 690 5.5 1500

Si3N4 3.18 1899 379 3 .7 1500

 

 

16

Bowman [1993, 1995] has shown that the fracture strength of NiAl/sapphire fiber
composite (30 volume percent) is 288 MPa at 300K, compared with 120 MPa for
monolithic N iAl at room temperature; the fracture strength of N iAl/sapphire fiber composite
still remains at 102 MPa at 1200 K, compared with 70 MPa for monolithic NiAl. Kumar
[1992] reported that the fracture toughness increases to 9 MPas/fi for N iAl containing 15
vol% sapphire whiskers from 6 MPax/E for monolithic NiAl, based on studies using
chevron-notched short rod specimens. Kumar further found that a volume ratio of sapphire
whiskers below 7.5% has no effect on fracture toughness.

Recently, Shah [1993] used DuPont’s FPO alunrina fibers (which are
polycrystalline) as reinforcement in MA] composite (approximately 35 volume percent).
The results showed that the yield strength of aligned FP” alumina fiber reinforcement NiAl
composite increased from 3.44 MPa for monolithic N iAl to 56.63 MPa at 1473 K. In
contrast to N iAl/aligned FP" alumina fiber composite, the chopped FPO alumina fiber
reinforcement NiAl behavior showed very poor consolidation, and the yield strength at
1473 K was only 3.90 MPa. Also, NiAl/aligned FPO alumina fiber composite has better
creep resistance than monolithic N iAl or NiAl/chopped FPO alumina fiber composite.
These results indicate that aligned fibers improved the high temperature strength and creep
resistance significantly, and that aligned fiber composites are more easily fabricated.

However, several issues concerning the use of sapphire fiber as reinforcement in
MA] are still under study. The fast concerns the interfacial reactions with the matrix.
Recent research [Bowman, 1994; Draper, 1990, 1994] has shown that interfacial reactions
degrade fiber strength, thus limiting the potential of the composite in load-bearing
applications. The properties of NiAl/sapphire fiber composites fall below expectation by the
rule-Of-mixtures. One current approach to this problem is to employ coatings which are
thermochemically stable to prevent reaction between fiber and the matrix; this preserves the
physical properties and control the debonding and sliding stress, ensuring composite

strength [Schalek, 1995]. The second issue concerns bend curvature that the fibers can

l7

withstand as a function of time and temperature. Fragmentation of extracted fibers from
NiAl/A1203 composites [Draper, 1992; Bowman, 1993] indicated that the fiber fracture
must occur either during hot pressing fabrication process of N iAl/A1203 composite or
during cooling from consolidation temperatures, in which thermal stress due to mismatched
CI‘E might cause damage to the fibers. This requires fibers to tolerate small bend radii so
that they will fracture at relatively high bend strain with respect to the room temperature
failure bend strain [Morscher, 1992].

The main drawback of sapphire fibers as a reinforcement is due to its inherently
brittle nature. Sapphire fibers thus exhibit very high sensitivity to surface flaws, which
limits the fiber strength. The following section will summarize current knowledge

regarding sapphire fiber strength degradation due to surface damage.

2.3 Strength Degradation of Sapphire Fibers due to Surface Damage

It has been well documented that sapphire fibers are highly susceptible to damage
by abrasion during fiber-fiber contact, hard particle impact, and cleaning procedures.
Draper [1992] Observed the presence of particles adhered to sapphire fibers etched from a
Fe-Al matrix. These fibers failed predominately at locations near the attached particles
during tensile tests. Trumbauer [1992, 1994] identified four types of flaw population
which significantly affect sapphire fiber strength: type A are the surface flaws attributed to
handling and abrasion damage; type B are volume or internal flaws attributed to shrinkage
voids which form during the manufacturing process; type C are localized fiber surface
reaction flaws introduced during the flame-cleaning procedure, and type D are self-abrasion
surface flaws introduced on unsized fibers. Furthermore, he found that flame—cleaning to
remove the sizing applied by the manufacturer of Saphikon fiber produced a significant loss
in strength, though it was apparent that other cleaning methods, such as the RCA1 process,

were much less destructive. He also found that self-abrasion treatments performed on

 

1 RCA process is based on a two-step oxidizing and complexing treatment with hydrogen peroxide solution
which act to remove organic surface films, alkali ions, and some other metallic ions.

18

unsized fibers produced a significant (30%) strength degradation, and further concluded
that procuring unsized fibers increased the probability that self-abrasion damage would
occur during shipping and handling. In addition to abrasion during handling and lay-up,
mechanisms of fiber damage during processing include chemical reaction with matrices and
impurities, chemical reaction with binders, and residual stresses resulting from fiber-matrix
coefficient expansion mismatch, as well as hot processing stress states. Trumbauer [1992]
reported that the tensile strength of sapphire fibers could easily be reduced to less than 69
MPa by implementation of a coating process followed by a simulated MMC fabrication
cycle, and Tressler [1992] further pointed out that the defects which caused failure of
sapphire fibers embedded in composites would most likely be defects generated during the
handling and manufacturing processes.

Bend tests on small diameter (40-85 tun) sapphire fibers [Morscher, 1992] revealed
localized kinking associated with failure sites. This kinking was not present in the large
diameter (150 um) fibers, presumably because the high volume allows more subgrain
formation and stress relaxation. However, high temperature three-point bend tests on
individual fibers have shown the fiber surface to be the origin of failure for sapphire fibers
of all sizes. Thus, surface flaws are probably the limiting factor on sapphire fiber bend
strength.

So far, it has been stated that sapphire fibers have important applications in high
temperature NiAl composites. Meanwhile, they also have a significant drawback, namely,
high sensitivity to surface damages which cause severe strength degradation. The
following section will document how ion implantation can overcome the disadvantage and

improve mechanical properties of sapphire.

2 .4 Ion Implantation Process
Ion implantation is a low-temperature vacuum surface treatment process involving

ion-energies typically in the range of 50 to 500 keV. Energetic ions penetrate the surface of

19

the target (host) material and come to rest in an approximate Gaussian distribution. During
the implantation process, the energetic ions collide with the substrate or target’s atoms,
transfer energy to the host atoms and then cause further cascades. The cascade effect
finally results in a great physical or chemical change near the substrate surface (usually 0.3
to 0.4 um deep) without altering the bulk material properties. In this section, ion
implantation processes and applications are reviewed, beginning with general description of
ion bombardment cascade model, followed by discussion of radiation damage and other
related effects. Finally, a summary of the literature on modification of mechanical

properties of sapphire by ion irradiation is presented.

2.4.1 The Cascade Model

When a bombarding particle strikes a solid target, the energy of the bombarding
particle will be transferred to a stationary lattice atom by both nuclear collisions and by
interaction with host atom electrons. Inelastic nuclear collisions result in the displacement
of host atoms from their structure-sites. These displaced atoms may then proceed to
displace other host atoms until the energies of both the incident ions and the recoiling host
atoms are insufficient to produce further displacements. For most materials the energy
needed to displace an atom from its structure is about 25 eV (which is so called
displacement threshold energy). Thus for ions with a few hundred kilo electron volts
energy, the number of host atom displacements per incident ion is expected to be large.

Electronic collision in which the incident ion interacts with electrons of host atoms
can result in the weakening of the host-atom-bond by ionization and by the formation of
charged defects such as color centers. Such processes are more efficient at higher ion
energies, whereas displacements prevail at lower energies, with the competition between
these processes producing the typically Gaussian damage and implant species concentration
profiles. Theoretical models have been developed for predicting both the energy

partitioning between the collisional and electronic energy loss processes, and the spatial

 

20

distribution of the resultant damage patterns. One of these models, known as TRIM-90
due to Ziegler [1985], has been used in this study. This Monte Carlo simulation calculates
the penetration of ions into solids, the final distribution of the ions, and the kinetic
phenomena associated with the ion’s energy loss, target damage, ionization and phonon
production.

The lattice atom first struck and displaced is called the primary knock-on atom or
PKA. Because a PKA possesses kinetic energy transferred by incident ion, it becomes an
energetic particle and is capable of creating additional lattice displacements. Subsequently
displaced lattice atoms are known as hi gher-order knock-on, or recoil atoms. An atom is
considered to have been displaced if it comes to rest sufficiently far from its original lattice
site so that it cannot return spontaneously. It must also be outside the recombination region
of other vacancies created in the process. The displaced atom ultimately appears in the
lattice as a substitutional or an interstitial atom. The ensemble of point defects created by a
Single primary knock-on atom is known as a ‘displacement cascade’. Figure 2.2 is the
schematic of a cascade process due to impact by a single ion.

The simplest theory of the displacement cascade is that of Kinchin and Pease
[1955]. Their analysis is based on the following definitions and assumptions:

1) Definition of parameters:

E is an energy possessed by the PKA, Ed is displacement threshold energy (25 eV
for most materials); EC is so—called cutoff energy (below which the moving atom cannot
transfer enough energy to an electron of the medium to remove the electron from whatever
bound state it may be in), and is related to interatomic potential; T is the transferred
energy; Pd is displacement probability; and 0(5) is the number of displaced atoms
produced by a single energetic incident ion.

2) The cascade is created by a sequence Of two-body elastic collisions between
atoms;

3) The displacement probability is given by

21

—o

r;
\

.l-r“
.
A

/
'-—-
-‘

       
 
  

/
/

,1:

Iodine path
/ A \ _ Beryllium path
\\\~A ——-- Oxygenpath
\’. 0 Oxygen absorption point
. Be absortion point

Replacement collisions

Figure 2.2 Schematic of cascade trajectories in planar BeO caused by S-KeV I127
bombardment (after Carter, 6., and Colligon, J. 8., Ion Bombardment of Solid, 1968).

22

0 (for T < Ed)
1 (for T > E)

(2.1)

4) The energy E; consumed in displacing an atom is neglected in the energy
balance of the binary collision that transfers kinetic energy to struck atoms;

5) For all energies less than EC, electronic stopping is neglected, and only atomic
collisions are considered; 9

6) The energy transfer cross section is given by a hard-sphere model;

7) The arrangement of the atoms in the solid is random, effects due to the crystal
structure are neglected.

Under these assumptions, the number of displaced atoms ‘0(E) produced by

incident ions can be deduced:

r0 (forO<E<Ed)

1 forE <E<2E

v(E)=l ( " d) (2.2)
E/2Ed (for2Ed<E<Ec)

[EC/215, (forE>Ec)

 

The following plot (Figure 2.3) illustrates the four regions predicted by the model, i.e.
when the energy of an incident ion is less than the displacement energy E; , the number of
displaced atoms is zero; when the energy is between Ed and 2 Ed , the number is 1; when
the energy is greater than 2 Ed , but lower than cut-off energy EC , the number increases
linearly with incident ion energy; Finally, if the energy exceeds the cut-Off energy EC , the
number of displaced atoms reaches maximum value of EC /2 Ed .

In summary Of the cascade process, primary atom makes multiple collisions, either
elastic or inelastic, with the lattice atoms before coming to a final resting position. Energy
loss is transferred to the lattice in form of electronic excitation when there are inelastic

contributions to the loss mechanism, or in kinetic energy of motion of the lattice atoms

23

 

Number of displaced Atoms (v)

 

 

 

 

PKA Energy (E)

Figure 2.3 The number of displaced atoms in the cascade as a function of PKA
energy according to the model of Kinchin and Pease (after Kinchin and Pease:
Rep. Prog. Phys, vol. 18, 1955).

24

when the collision is elastic. In former case, secondary electron emission and emission of
electromagnetic waves is observed. In the latter case, the struck atoms are displaced from
their equilibrium position in the lattice. If the received energy is sufficient and properly
directed, the struck atoms influence the surrounding atoms immediately. A vacancy is
created at the original atom position whilst the displaced atoms become foreign atoms

elsewhere in the lattice.

2.4.2 Radiation Damage in Solids

As ion bombardment proceeds, it is anticipated that radiation causes atomic
displacements and the production of defects in the target (or host material), which influence
the material’s properties. This is so-called radiation damage. Radiation damage is directly
associated with ion energy, incident ion spices, target material, and target temperature.

Ion energy determines the stopping power or specific energy loss, (dE/dx). Here,
E is the ion energy and x is the distance which is usually measured along the direction of
incidence of the ions. The total specific energy loss is taken to be the sum of three

separable components: nuclear, electronic and charge exchange

dE/dx = (dB/(1.x): + (dE/dx)n + (dE/dx)ch (2.3)

At lower energy, nuclear stopping dominates and is responsible for most of the angular
dispersion of an ion beam. At higher energy, electronic collisions are the more important
and in slowing down to rest from these energies the bulk of the particles energy is
dissipated in the form of electronic, rather than nuclear, motion. Provided a recoiling lattice
atom receives an energy in excess of the displacement energy, Ed, it can leave its lattice site
to become permanently displaced within the solid. In most cases of interest, lattice atoms
recoil from primary collisions with kinetic energies far in excess of the displacement

energy, and are capable of penetrating many atomic distances into the surrounding lattice.

25

The intensity of incident ions, irradiation time and irradiated area can be related

together by so-called ‘ion dose’, as defined:

D: L
Aqe

(2.4)
where t is the implant time in seconds, I is the integrated beam current in Amperes, A is the
effective scanned area in cm2, D is the dose level in ions/cmz, e is electron charge (1.602 x
10’19 coulombs) and q is the ion charge state.

The choice of ion spices affects the extent of radiation damage because of ion mass
and chemical nature. It is understandable that heavy ions can transfer more energy to lattice
atoms than lighter ones, and thus cause more collisions and more displaced lattice atoms.
Chemical interaction between incident ions and lattice atoms affect the production of defects
and microstructure. For instance, implantation of gaseous ions such as N or Ar into
sapphire causes blistering within the implanted region [I-Iioki: 1989; McHargue: 1987].

For different classes of materials, their chemical bonding might range from ionic to
covalent to metallic. The type Of bond affects production of defects and atomic
displacements during ion irradiation, and further influences the extent Of irradiation
damage. Generally, ion irradiation produces amorphous microstructures in ionic or
covalent bond-type materials more easily than in metallic materials. Naguib and Kelly
[1975] proposed a bond-type criterion to predict the amorphization for materials. They
suggested that materials with Pauling ionicities less than 0.6 are more easily made
amorphous by ion irradiation. At 300 K, the amorphization dose for Ot-SiC, for which the
Pauling ionicity is 0.12 (covalent), is 10'3 of that required for tit-A1203, whose Pauling
ionicity is 0.63 (ionic bonding) [McHargue et al, 1986].

In order to estimate the magnitude of radiation damage, two characteristics of the

damage must be specified, i.e., the nature and the number of defects. The following

26

paragraph will review the radiation damage induced during ion implantation process. First,
defects produced by ion implantation include:

(i) points defects (vacancies and interstitial);

(ii) impurity atoms (atomically dispersed transmutation products);

(iii) small vacancy clusters (depleted zone);

(iv) dislocation loops;

(v) dislocation lines; and

(vi) cavities.

The formation of voids, dislocation loops and networks, for example, originates
from point-defect aggregation. In their subsequent evolution, those extended defects serve
as additional sinks for continuously generated point defects. Pedraza and Mansur [1986]
pointed out that it is the generation of point defects and their evolution during irradiation
which perturbs long-range order in crystalline host target. Take a Frenkel pair for example,
which contains a vacancy and an interstitial. If an interstitial of the appropriate species is
generated at or it attains by diffusional migration such a site, it will have a higher tendency
of remaining in it if there is a nearby vacancy. The role of the vacancy in the trapping
process is that of allowing for partial volume relaxation. The formation of the complex
thus creates a center of short-range order. In this way, damage accumulation eventually
may lead to the host material becoming amorphous.

The relationship between microstructure and irradiation damage accumulation was
studied by Burnett and Page [1985]. According to their proposed model, three
microstructural regions may arise from room temperature ion-implantation into crystalline
materials (Figure 2.4). In region I, at low doses, a damaged but still crystalline solid
solution is formed. In region II, at intermediate doses, amorphous material is initially
formed at the peak of the displacement damage profile, resulting in a subsurface amorphous
layer which thickens with increasing doses. In region 111, at sufficiently high doses, a true

surface amorphous layer is formed which also thickens with increasing dose.

27

«$3

SEQ U E 33 honest .m .2 we 33% a.— cunt E K
35» 2255 .5 .m LSKS a??— 33908“ 083m a E .83 30:995. come—353.
a u .a2w8 ofizsmbo woman—«e a H .3505 28s 3290.5 @0265 ecu—3535
me 9028.“ Ea one“ 5 count? 05 we aces—H883. ouaoaom < ta am

A a- Eu mac: 8% «Bob

22 : S 2 o—

ecu—Euro

 

\

28

Microstructure alternation due to ion implantation can be characterized by
mechanical property changes. Through the implantation of 300 keV Ti+, Cr+, Y+ and Zr'"
into planar sapphire [Burnett and Page, 1985], it was found that while doses below that
which are required for the onset of amorphization, implantation resulted in surface
hardening; however, the Ti+ and Cr+ implanted specimens showed a less rapid decrease in
hardness with increasing dose compared to the Y"' and 21” implanted sapphire. Compared
to unimplanted sapphire, the results for the Ti'*' and 0‘" which created a region 11
microstructure showed little change in hardness.

Direct TEM observations on the implanted region of sapphire [McHargue: 1986,
1987, 1991] also revealed a high density of ‘black spots’, which are typical of point-defect
clusters that exist in a-A1203 implanted with 2x1016 Cr“/cm2 (280 keV) at room
temperature. For 300 keV Zr’ implanted sapphire at dose of 4x1016ions/cm2, a subsurface
amorphous layer was produced extending from 40 nm to 100 nm from the free surface,
whereas the material was still crystalline but damaged outside the implanted region.

Second, according to Kinchin-Pease theory, the number of atoms displaced is
related to the incident energy (see Equation 2.2); i.e., the number of atoms displaced is
proportional to the incident energy for the energy between E; and Be, Norgett et al [1975]
proposed a modified method to calculate the number of displacements. This new method is
based on the Kinchin-Pease model and the half-Nelson model, and is outlined by following
equation:

1) The number of Frenkel pairs Nd generated by a PKA of initial kinetic energy E
is given by:

N, = kE/ZEd (2.5)

where E is the energy available to generate atomic displacement by elastic collisions.
2) The displacement efficiency k is given the value 0.8, independent of PKA

energy, the target material, or its temperature.

29

3) The inelastic energy loss is calculated by the method of Lindhard, i.e.:

I? = E/[1+ k8(€)l

g(e) = 3.400881/6 + 0.4024415”4 + e

k = 0.13372,”5 (21 / .11.)"2 (2.6)
e=[A2E/(A, +A,)][a/z,z,e2]

a = (97:2 /128)U3a0[Z.2’3 + 233]”2
where do is the Bohr radius, e the electronic charge, Z, and Z, are the atomic numbers of
the projectile and target and A , and A2 are the mass numbers of the two atoms.

It should pointed out that at higher energies (a few keV), especially for low
temperature implantation into covalent materials (such as semiconductors or insulators), the
linear Kinchin-Pease theory cannot accurately predict the damage level. This inability has
been attributed to the effects of energy spikes. Basically, when the nuclear energy loss per
atomic plane is high, it is possible to conceive of a volume surrounding the ion track as
either (a) a thermal spike, in which the average energy supplied to lattice atoms
substantially exceeds the heat of melting, or (b) a displacement spike, in which an almost
continuous network of displaced atoms is created. Thermal spikes can give rise to excess
damage by a process in which the local hot spot surrounding the ion track extends out
considerably beyond the original collision cascade dimensions. This superheated region
may ultimately be quenched into an amorphous state in times of the order of 10‘12 s to give
considerably more damage than expected from collision theory. Alternatively,
displacement spikes can give rise to excess damage by spontaneous collapse to an
amorphous state, when the defect (or displacement) density attains a critical level.

Temperature affects the degree of damage and alters the nature of the observed
damage structure. Two types of annealing that can take place during implantation; namely,
(a) dynamic annealing of damage produced by single ions, and (b) thermal annealing of

damaged material due to a general rise in target temperature during implantation. In type (a)

30

annealing, where damage increases with dose rate, it is believed that cascades can overlap
before a single cascade damage has completely annealed. Type (b) annealing of the
damaged layer can result from very high dose and high dose-rate implantation conditions,
which produce a bulk temperature rise during implantation or it may occur if the target is
deliberately heated. In such cases, an increase in dose rate (beam current density) can
produce a reduction in apparent damage because of the high target temperature attained.
The latter (bulk heating) effect can have several consequences for the generation of a
disorder structure during the initial stages of implantation. It will be detailed in section

2 .4. 3 .7.

When an incident ion enters a crystalline target surface, two additional phenomena
which affect the extent of radiation damage may appear: focusing and channeling.
Focusing refers to the transfer of energy and atoms by nearly head-on collision along a row
of atoms. Channeling is the complementary process whereby atoms move over longer
distance than normal in the solid, along open directions in the crystal structure, being kept
in the channel by glancing collisions with the atomic wall. Focusing and channeling affect
both the number and configuration of displaced atoms in a cascade. First, atoms moving
along the crystallographic direction favorable to focusing or channeling lose energy only by
glancing collisions with atoms ringing the axis of motion. The energy transfer in those
collision is well below Ed . Second, the focused or channeled atoms are able to move
much larger distances than ordinary knock-ons before conring to rest. As such, displaced
atoms that have been created by focusing or channeling mechanisms contribute
disproportionately to radiation effects, such as diffusion—enhanced creep and void growth,

and can significantly extend the depth of the damage zone.

2.4.3 Radiation Effects
As has been stated, radiation damage originates from production of point defects.

Interaction between atomic mobility and aggregation of point defects and dissipation of

31

ions’ high energy along a cascade result in so-called radiation effects, including quenching
effects, diffusion effects, stress effects, precipitate effects, as reviewed in following

section.

2.4.3.1 Rapid Quenching Efiects

The energy deposited in a collision cascade is dissipated through atomic motion in
times of order of 10‘12 s, presumably by lattice thermal conductivity. Typically a cascade
volume can be expected to contain 2 1000 atoms [Appleton, 1984]. Some research
reported from theoretical calculations that Spike temperature in copper dropped from
2000°C to 500°C in 3 x 10'11 s [Dienes, 1957]. The high local "temperature" should cause
some homogenization around the thermal spike, but the ultimate material interactions
induced locally and at long range depend strongly on the particular material system. The
quenching rate of a spike cascade can be estimated to be about 1013 K/s. Thus, a variety of
new material properties can be expected from this extremely rapid quenching. The ultimate
result, however, will depend on the defect interaction and annealing properties of the solid,

as well as on the details of the cascade.

2.4.3.2 Radiation Enhanced Difi‘usion

In contrast to thermal diffusion, radiation enhanced diffusion is a diffusion effect
which is activated by bombardment with high energy particles. This effect can be produced
during self-implantation, or by irradiation with non-doping particles (or—particles, neutrons,
protons). Under energetic ion bombardment, diffusion can be greatly enhanced because
huge numbers of point defects, which mediate normal thermally activated diffusion, are
produced. Ion bombardment not only creates large numbers of defects, which can
accelerate substitutional, interstitial or vacancy diffusion mechanisms, but also establishes
solute atom and point defect concentration gradients, which can act as strong driving forces

in the mixing process. The effect of radiation-enhanced diffusion can be calculated

32

approximately by assuming that the diffusion coefficient is proportional to the vacancy

concentration [Tsuchimoto, 1970]:

2
3Nv(x, t) =Dv a_1;___j2x,t) NvixJ) +805) (27)

V

where Nv(x, t) is the vacancy concentration, D, is their diffusion coefficient, 1,, is the
recombination lifetime of vacancy, and g( x) is the generation rate. g(x) may be estimated
from the ion range, the straggling, the amount of energy released in the form of nuclear

recoils and the displaced energy, i.e.:

1%)
g, = "’m" (2.8)
454

where go is the maximum generation rate, j is the current density, Ed the vacancy-
formation energy, and (dE ldx)“,max the maximum energy deposited into atomic processes.
The vacancy concentration Nv(x) and diffusion coefficient could then be calculated as a
function of depth. Some researchers have confirmed the phenomena of radiation-enhanced
diffusion. According to Deamaley [1973, for heavy ions having energies in the vicinity of
50 keV during typical implantation conditions, the irradiation enhanced diffusion coefficient
has been estimated at 10'15 cm2 seC'l. Such a diffusion coefficient is typically equivalent to
a temperature of about 450°C in copper, 600°C in iron, 200°C in aluminum and about
1000°C in Silicon with respect to vacancy controlled diffusion.

2.4.3.3 Precipitation Effects
Generally, as ion dose is increased, more and more atoms are introduced into the
lattice and it is of considerable interest to determine whether or not the material so formed

remains a solid solution indefinitely. In conventional alloying methods, when the

33

concentration of impurity exceeds the maximum solubility limit in the target material and the
diffusion conditions are suitable, then precipitation of a second phase results. In a
conventional system, precipitation effects can only be observed in a limited number of
material systems because supersaturation occurs when there is a reduction in temperature of
a solid solution. In case of ion implantation, there is, in principle, no such limitation since
ions of any element can be injected into any solid irrespective of solubility considerations.

There are basically four processes that govern precipitation: (a) nucleation, which
means that the small region of the second phase must overcome interface energies in order
to form; (b) diffusion of atoms to the nuclei to supply additional material so that the second
phase can grow; (c) a reaction occurs where the atoms at the interface rearrange themselves
into the structure of the new phase; and (d) diffusion of excess atoms from the reaction
away from the interface.

At low temperature where thermally activated diffusion of the implanted atoms is
virtually absent, it might be concluded that the system so formed would in fact remain as a
supersaturated solid solution. But, on the other hand, if the implantation temperature was
sufficiently high that impurity diffusion became significant, agglomeration and precipitation
might be expected. Therefore, post-implantation annealing usually is undertaken when it is
desired to make precipitation take place.

Previous studies have shown the occurrence of epitaxial recrystallization of Cr, Mn,
Ni and Xe implanted into sapphire during post-implantation thermal annealing since there is
little or no solid solubility of such ion species in sapphire [Ohkubo, 1986]. For Mg-
implanted sapphire, Mg has been found to segregate to the (lOI 1) prismatic free surface
[Mukhopadhyay, 1988].

2.4.3.4 Residual Stress Induced by Ion Implantation
After ion implantation, the implanted solid surface is generally in a stressed state,

either compressive or tensile, depending upon the depth. The existence of stress on the

34

solid surface greatly influences measured surface properties. The following section will
deal with ion implantation induced stress states, proposed surface stress models, and

experimental measurement of surface stress.

2.4.3.4.1 Ion Implantation-Induced Stress States

Generation of defects by displacement processes results in a volume change within
the implanted layer. However, if this change is constrained by either underlying or
surrounding material, large stresses may be generated in this relatively thin implanted layer.
Both compressive and tensile stresses have been observed, depending upon the nature of
the host material and ion species [EerNisse, 1971; Burnett, 1985].

Sapphire is known to undergo volume expansion when exposed to ion
bombardment [Krefft, 197 8]. Because the defect density varies with depth as a result of
the ion stopping power of the target atoms, the generated stress is not uniform. Krefft,
however, assumed that the stress was distributed uniformly along the expansion direction,
and used the concept of integrated stress, S , to represent the stress generated by ion
implantation, which is the integral over the entire ion damage range. Their results have
shown that the integrated stress increased with doses up to 1x103 (N/m) for the sapphire
plate which was implanted with Ar‘" at 500 keV. They also have shown that light ion
implantation into sapphire which had been previously implanted with heavy ions may
relieve the compressive stress resulting from Ar+ implantation. Furthermore, they found
higher absolute stress values to result from implantation along [1120] and [01 T0] which
were caused by higher lattice expansion along [0001]. In other words, defect formation
and corresponding lattice expansion occurred more rapidly during irradiation perpendicular
to the c—axis.

Specht [1994] found that the density of A1203 in midrange decreases by 4% after
Cr+ implantation while the volume expansion in the midrange was only ~0.2%. He
attributed the density reduction of A1203 to high energy transfer collisions that knock Al

35

and O atoms deep into the crystal and give rise to excess vacancies and deep interstitial.
Therefore, it is expected that the stress state in the nridrange about 40 nm from free surface
is tensile.

Using both cantilever bending and indentation fracture techniques, Burnett and
Page [1985] found that the generation of near-surface compressive stresses in sapphire
implanted with Y"' and Ti+ initially increased with ion dose until a critical dose ( 8 x 1016
ions cm'z) was reached. Beyond this dose, stress relaxation was observed. The stress,
averaging over 4<XD>1, can reach 6.4 GPa for 5.6 x 1016 Ti+ cm'2 and 7.6 GPa for 1.7 x
1016 Y"' cm'2 at 300 keV.

Hioki et al [1986] studied the effects of implantation temperature on the
compressive stress. They showed that for implantation at 300 K, the integrated stress
increases monotonically with doses up to 1 x 1016 Ni+ (300 keV). At 100 K integrated
stress increases rapidly with an increase in dose and reaches a maximum value at just below
Da (the amorphization dose is 5 x 10" N i/cmz), followed by a marked decrease at higher
dose. They also determined that the average compressive stress reaches a value of 9 GPa at
100 K and 2 GPa at 300 K. Therefore, ion implantation at a lower temperature is much

more effective in generating a large surface compressive stress.

2.4.3.4.2 Surface Stress Model

As discussed in section 2.4.2, accumulation of irradiation damage leads to
amorphization of crystalline target material when implantation exceeds a certain dose.
Amorphization results in stress relief. After amorphization, the integrated stress is
considered to be the sum of those contributions from the portions of the damage and range
profiles lying in the still-crystalline material, together with that from the stress supported
within the amorphous layers. Burnett and Page [1985] proposed a surface stress model

 

1 <XD> is the standard deviation in the depth of the damage peak position in the EDEP-l computer code

which predicts the ion distribution. If the distribution is assumed to be Gaussian, then 96% of all damage
will lie in a layer 4<XD> thick.

36

which deals with the role of amorphization in stress relief based on the following
assumptions:

(i) the variation of surface stress (S ) with dose is linear prior to amorphization;

(ii) thermal effects, dose rates, electronic effects, etc., are negligible;

(iii) the amorphous material is mechanically homogeneous; and

(iv) after amorphization, the compressive stress in the surface is taken to consist of
two components: the stress supported within the still-crystalline but damaged material (Sc)
and the stress supported within the amorphous material (5,, ). The former stress, Sc,
includes two sources: Sm due to the presence of the implanted atoms and Scd due to the
formation of other defects, e. g., Frenkel pairs.

Figure 2.5 is a schematic representation of the principles of the implantation-
induced stress model. Initially, at doses below that for amorphization (region I), the
integrated stress S is the sum of the implanted depth of the expected stress profile arising
from implantation. Upon amorphization (region 11) the integrated stress is the sum of the
portion of the Gaussian implant profile remaining within the crystalline material, Sc , and
the stress level supported within the now amorphous material, Sa. At higher doses a
surface amorphous layer is formed (region III), and the stress contribution from the
crystalline material is further reduced.

Once the range and damage profiles have been obtained, the model is controlled by
four parameters, namely:

(i) a constant of proportionality, a , which may be determined experimentally;

(ii) the stress supported within the amorphous layer, 02,;

(iii) the partitioning of the stress contributions from the still-crystalline materials
into components due to the damage (Sad) and due to the implanted atoms (Sea), which may
be represented as B = Seal (Sm + Scd ); and

(iv) the critical energy density at which amorphization occurs, 95 cm, which may

also be determined experimentally.

37

 

Ammg rKH sack flee ..~..& $2.55 .SKS
E68 $2: ecosefiéoufieua8_ .8 33353 05 we aoafieomoauu ozmaonom Wm osmE

a + o 083v.
mm + omuw m mum we .53an
lll seep

lull Snow

ll. 3%

ssans

‘—

Kw:—

 

 

38

If B is small, i.e., the stress in the crystalline material is damage controlled, several
features of integrated stress might be obtained in terms of comparison ofO'a and am,
where am is the stress in the crystalline phase at the onset of amorphization (i. e., the
maximum stress level obtained):

(i) if the amorphous layer supports no stress, then the integrated stress falls off
dramatically after amorphization due to the initial rapid thickening of the amorphous layer,
though at higher doses the stress maintains a nearly constant value;

(ii) if a}, = am the integrated stress will continue to rise after amorphization, again
gradually leveling off at higher doses; and

(iii) at intermediate values of 02,, an initial decrease in integrated stress will be
observed; however, by the time region III is reached, S may start to increase slowly.

Burnett and Page obtained a good correlation of the stress model parameters with
experimental data for sapphire implanted with Tit and Y+ [Page, 1985]. Although this
model described the role of amorphization in stress relief, the accuracy of this model is
limited by the assumption of a sharp crystalline to amorphous transition, with an associated
abrupt change in mechanical properties. In any case, this model has enabled an estimate of
integrated stress values in terms of the stress relieving effects associated with the

production of amorphous materials.

2.4.3.4.3 Experimental Determination of Surface Stress

Cantilever Beam Bending This method was first introduced by EerNisses
[1971]. Based on the fact that volume expansion of targets will occur after ion
implantation, the volume changes of the implanted surface layer are monitored by
measuring the bending of the cantilever beam-shaped samples irradiated on one side. This
bending occurs as a result of stress induced by lattice damage which causes the lattice to
expand. In the direction normal to the implanted surface, the crystal lattice is free to
expand, but in the lateral direction it is restricted by the much thicker underlying

39

undamaged lattice; and this will result in a lateral stress T. The integral, S, of this stress
over the depth of the damaged layer can be expressed by:
E612

S = Era—v) (27)

where 6 is the deflection of the beam, 1' is the thickness, 1 is the length, and E andv are
Young’s modulus and Poisson’s ratio, respectively.

It should be pointed out that E and 0' must be chosen to correspond to the
crystallographic orientation of each particular sample orientation if the lateral stresses are to
be compared for elastically anisotropic materials. This technique is only applied to low dose
implantation where the elastic properties of the slightly damaged lattice in the bombardment
layer could be assumed to be the same as for the undamaged material.

Indentation Fracture Techniques The model, proposed by Lawn and Fuller
[1984], was developed for evaluating stresses in the surface of brittle materials from
changes in indentation-induced crack dimensions. The basis of the model is a stress
intensity formulation incorporating the solution for a penny-like crack system subjected to a
constant stress over a relatively thin surface layer [Lawn, 1975; 1980]. Since the model is
focused on the measurement of thin layer surface stresses, which corresponds reasonably
well to the implanted layer produced by ion implantation, it may be applied in the present
work. Full description on the technique will be addressed in section of experimental

procedures.

2.4.3.5 Radiation Hardening and Softening Efiects
As reviewed above, an implantation process is expected to create radiation damage
below the surface of a solid. As a large number of foreign atoms are forced into a host

lattice, they can exist either as solutes in a metastable and strained crystalline system, as an

40

amorphous phase, or as a surface layer in which fine precipitates exist as a result of
implantation induced heating or subsequent heat treatment of the implanted material.

Both hardening and softening by radiation in metallic and ceramic materials have
been observed [Burnett 1985; Hioki: 1986], leading to changes in nricrohardness, wear
and friction rates, and stress-strain behavior. Hardening induced by radiation is associated
with solid solution strengthening, precipitate hardening, defect hardening and the existence
of compressive stress in the surface; softening has often been correlated with the presence
of an amorphous layer. One hardening and softening observation by Burnett and Page
[1985] is shown in Figure 2.6. The result shows that Knoop microhardness of Ti-
implanted sapphire at 300 K increases from 3800 at dose of 2 x 1016 ions/cm2 to 4750 at a
dose of 9 x 1016 ions/cmz. Beyond this dose, Knoop nricrohardness decreases with dose
until maximum dose of 5 x 1017 ions/cmz.

Ion implantation can produce substitutional or interstitial solid solutions, depending
on the composition and the basic characteristics of the system and the respective
equilibrium relationships. The modification of mechanical properties can be produced at
the surface of the host material through one or a combination of the following: (a)
formation of substitutional solid solution plus residual strain effects; (b) formation of
interstitial solid solution plus residual strain; or (c) the production of high compressive
stress due to second phases or volume expansion, which is associated with production of
defects. Point defects and impurity atoms are believed to contribute negligibly to hardening
compared to the effect of the larger defect clusters.

Precipitate hardening is associated with the formation of fine precipitates which can
impede dislocation movement and thus produce a marked increase in mechanical strength.
There is usually a misfit between the precipitate particle and the matrix in which the particle
is lodged. If the precipitate volume is larger than the matrix it replaces, the particle acts as a

point center of compression and creates a stress field in the surrounding solid. On the other

Knoop Hardness

Figur
Ti irn‘

 

41

 

 

 

 

 

 

 

 

| I

| I

I I

I I

I I

I I

. H 1111

I I

I I

I I

I I

I I

I I

I I

I I

I I

i 1» :--

I I

I I

. \\

I

i 10‘
— I I

I I

I I

' 'l

I I

I
W l l l llllll l l '1 1
02x10'6 1x10” 5x10"

Dose (Tr cm'2 )

2.6 The variation of 25 g Knoop microhardness with dose for
planted into sapphire (After P. J. Burnett and T. F. Page, J. Mater.

 

Sci, 2

30 (1985)).

42

hand, if the precipitate occupies a smaller volume than the material that has been replaced,
there will be internal tensile stresses in the solid around the foreign particle.

The stress state induced by implantation within the implanted layer obviously
affects the surface properties. In most cases, compressive stress occurs due to volume
expansion. External applied load is needed to overcome the compressive stress. The
existence of compressive stress can effectively retard surface crack propagation, and thus
enhance surface properties such as hardness and wear resistance.

Surface softening has been clearly correlated with the formation of an amorphous
layer, and the degree of softening has been shown to be dependent upon the depth of the
amorphous layer [Page, 1985].

2.5 Modification of Hardness, Toughness and Wear Resistance of Sapphire by Ion

Implantation

The surface structural change leads to surface property change. McHargue [1982]
reported that a room-temperature Cr" and Zr+ implantedl single crystal A1203 lattice was
significantly damaged but remained crystalline. In this case, most of both Cr+ and Zr+
moved into a substitutional lattice sites, as indicated by Rutherford Back Scattering (RBS)
results, increasing hardness by solution strengthening. Regarding indentation hardness,
McHargue [1982] and O’Hem [1990] reported that the relative hardness of Cr'I' implanted
sapphire (150 keV and 180 keV at room temperature) is 1.31 on the c-axis, and 1.23 on the
a-axis. Furthermore, from the results of RBS analysis, McHargue pointed out that a major
portion of the hardness increase in the Cfi-irnplanted material is due to the nearly random
distribution of the ion species, perhaps interacting with the damaged Al sublattice
[McHargue, 1982].

Substrate temperature modifies the hardening effect of ion implantation by
influencing transient annealing and diffusion. Hioki et al [1986,1989] showed that at low

 

1 Ion implantation conditions are 10" to 1017 Cr” cm'2 at 280 to 300 keV and 2 x 10'6 Zr” cm'2 at 150 keV
at ambient temperature.

43

temperature (100 K), the relative hardness of the planar sapphire implanted with 300 keV
Ni‘I’ could be increased by a factor of 1.5 at dose of 2 x 1015 ions cm'z. However, above
this dose, the relative hardness decreased rapidly with increased doses, reaching 0.6 at
1x1017 ions cm‘z. For the same material implanted at 300 K and 523 K, the hardness
increases monotonically with the dose (up to 1 x 1017) [Hioki, 1986, 1989]. They
concluded that for 100 K implantation the increase in hardness before the dose of 2 x 1015
ions cm“2 was due to radiation-defect-hardening, and the following rapid decrease in
relative hardness apparently resulted from the formation of an amorphous layer. The
hardness increase at higher substrate temperature (300 K to 523 K) may involve
contribution from solid solution hardening or precipitate hardening because of the
increased concentration of implanted Ni‘I’ [Hioki, 1985]. Bull [1991] concluded that the
increased annealing expected in damaged but crystalline samples as the substrate
temperature is increased has little effect on the total radiation hardening. Some change in
hardness was attributed to thermally-induced migration of implanted ions into substitutional
lattice sites.

The fracture toughness, ch, evaluated by the Vickers indentation method, was
found to increase with increasing doses in sapphire samples which were implanted with
300 keV Ni'I' at 100 K, 300 K and 523 K, dose up to l x 1017 ions cm'2 [Hioki, 1986].
The relative fracture toughness of planar sapphire ranged from 1.12 to 1.18 for specimens
implanted either at 4x1015, 4 x 1016 Cr'I' cm'2 at 150 keV or 1x1017 0* cm'2 at 180 keV
at room temperature [O’Hem, 1990]. It is believed that compressive stress induced by ion
implantation or softening and increased plasticity due to amorphization may contribute to
the fracture toughness increase. Also, substrate temperature plays a very important role in
the increase of fracture toughness. The results from Hioki [1989] have shown that at a
given dose the implantation at 100 K is more effective in increasing KIC than implantation
at 300 K or 523 K. This is consistent with the observation that the compressive stress

induced by implantation at 100 K is 3 to 9 times larger than that produced at 300 K.

44

Ion implantation can substantially modify surface and near-surface tribological
properties (such as wear and friction coefficients) of sapphire plates. By performing
lubricated pin-on disc tests on sapphire plates which were implanted with Ti'I' (to dose of
3.2 x 1015 and 7.7 x 1016 ions cm'2) and Zr‘I' (to dose of 1.7x1016 ions cm'2) at 300 keV,
it was found that ion implantation resulted in an increase in the coefficient of friction above
that of the control by as much as a factor of three [Bumett, 1987]. Studies on the wear
properties of sapphire [Ramos, 1992] reported that, compared with unimplanted sapphire
plates, the wear scar of a disk which was implanted with 150 keV 1017 Ti'I' cm'2 was very
low (less than 20 nm compared to 130 pm of unimplanted sapphire). No transfer film was
formed in the disk wear track, and only a few wear particles were present on either side of

the track.

2.6 Post-Implantation Heat Treatment

In section 2.4, it was stated that ion implantation produces a variety of structures,
ranging from crystalline, to metastable solid solutions with a large concentrations of point
defects and dislocations, to amorphous phases. These modification of surface structure are
directly related to surface properties.

The response of materials to ion implantation damage varies markedly from one
material to another, and thus post-implantation annealing behavior of surface structure also
significantly differs. Generally, damage recovery during subsequent thermal annealing is
expected to occur by epitaxial recrystallization of the amorphous layer. The final state will
be dependent both on the annealing environment and on the extent of recrystallization.

The effects of post-implantation thermal annealing on structural and mechanical
properties of sapphire specimen implanted with a variety of ions have been studied by a
number of researchers [Naramoto, 1983; McHargue, 1987; McCallun, 1990; White, 1987;
Ohkubo, 1986; Potter, 1992]. White [1987] demonstrated an annealing process for an

amorphous material in which a-A1203 was implanted stoichiometrically with Al (4 x 1016/

45

cm2, 90 keV) and O ( 6 x 1016 / cmz, 55 keV ) at liquid nitrogen temperature. The
crystallization of the resultant amorphous A1203 during thermal annealing can be
schematically illustrated as shown in Figure 2.7. Two stages are involved in the
crystallization process. The first step in the annealing process is the conversion of the
amorphous film to the crystalline 'y-phase of A1203 . The second step is the conversion of
Y-A1203 into Ot-A1203. This transformation takes place at a well defined interface which
moves from the original amorphous/crystal interfacial region toward the free surface.
Rutherford Backscattering Spectroscopy (RBS) is often used to track the
crystallization of an amorphous material during thermal annealing. By means of RBS,
McHargue et al [1982] reported that upon annealing, the recovery of the Al sublattice in
A1203 implanted with Cr at high dose began at 800°C and the O sublattice at 1000°C, and
most of the Cr moved into substitutional lattice sites. In contrast to this, the results of RBS
showed that the A1203 implanted with Zr was more resistant to recovery and no tendency
for Zr to move into preferred sites was found. McHargue et al [1982] reported that the
relative hardness showed an increase of about 28% for A1203 implanted with Cr, and about

only 19% after the implanted A1203 was annealed at 1200°C.

2.7 Summary of The Literature Review

Summarizing this review on ion implantation into sapphire, it can be stated that ion
implantation changes implanted sapphire surface microstructure, and creates a compressive
stress region within the surface. Mechanical properties of implanted sapphire such as
flexure strength, hardness, wear resistance are greatly improved. Ion implantation
provides a practical and feasible means to overcome sapphire fibers’ drawback, i.e., their
high sensitivity to surface flaws which greatly degrade mechanical properties. As will be
seen, application of ion implantation to sapphire fibers has interesting implications for

performance of sapphire fibers during handling and fabrication processes.

46

 

 

 

 

amorphous stochoimetric
-A1203 / <
. A1293 s implanted at
liquid nitrogen
Anneal te ure

 

   

 

 
 

 

 

Figure 2.7 Annealing process of an amorphous layer in which Ot-A1203 was
implanted stoichiometrically with Al and O at liquid nitrogen temperature (after

C. J. White: Mat. Res. Sym, I987)

CHAPTER III

EXPERIMENTAL PROCEDURES

In this chapter, a detailed description of experimental procedures will be given,
beginning with the materials used in the work. In the section on the ion implantation
process, the basic design of the Varian 350D irnplanter, its operation principles, and
modifications made to the ion source and end station will be described. Temperature rise
induced by ion irradiation will be briefly discussed. Following this, three point bend tests,
abrasion protocols and nricroindentation test methods are described. Weibull statistical
analysis was employed to analyze the characteristic bend strength. The fabrication process
of NiAl/Ale3 sapphire fiber composites will be detailed in section 3.7. Section 3.8 gives
the full description how a transverse section TEM sample of a single crystal sapphire fiber
is prepared. At the end of this chapter, transport codes which theoretically describe the

distribution of energetic ions inside the target will be discussed.

3.1 Materials

Commercial single crystal Ot-A1203 fibers were obtained from Saphikon, Inc.,
Milford, NH. These fibers were grown by Edge-Defined, Film-fed Growth (EDFG)
technique. The typical properties of sapphire fibers are summarized in Table 3.1. The total
impurities in the pure A1203 were claimed to be less than 100 ppm. The orientation of the
fibers is parallel to c-axis, i.e., the [0001] direction. The average diameter is about 140
um. According to technical data from Saphikon, Inc., these fibers were not provided with
a sizing. The Young’s modulus was given as 414 GPa.

Before ion implantation the fibers were soaked in ice water for 1 hour and rinsed
with deionized water to remove possible contamination particles from the fiber surface, in

accordance with the procedure described by Trumbauer [1992]. The fibers were

47

48

subsequently allowed to dry for a minimum of 24 hours. The cleaned and dried fibers

were wrapped in acetone-washed aluminum foil and stored in a vacuum desiccator prior to

 

 

 

further use.
Table 3.1 Properties of Sapphire Fiber.
Tensile Strength Young’s Modulus Shear Modulus Poisson’s Ratio CTE
(GPa) (GPa) (GPa) (x 10-6 °C)
0.27 ~ 0.30
2.1 ~ 3.4 414 175 orientation 8.8 (c-axis)
(c-axis) dependent 7.9 (a-axis)

 

 

 

 

 

The NiAl matrix material used in this study was donated by NASA-Lewis Research
Center in Cleveland, OH. The binary NiAl powder was consolidated by hot isostatic
pressing. The cast code was P-2306.

3 .2 Ion Implantation Process

The ion implantation process used here involves the generation of an ion beam,
mass analysis, acceleration of ion beam to high energy, rastering the beam, ion beam
adjustment, target set up, and ion beam dose measurement and calculation. The following

sections will detail this process.

3.2.1 Ion Beam Generation

A Varian 350-D medium-current ion implantation machine (on consignment from
Ford Motor Co.) was used to carry out ion implantation. This implanter consists of the
following major subsystems: a Freeman ion source, an ion analyzer section, the
acceleration tube, a quadrupole lens, a beam mask, a target chamber, and the vacuum and
coolant systems. The implanter has an energy range from 30 to 200 keV. Figure 3.1 is a

schematic illustration of the 350D implanter.

 

49

assuage es comm 55> co eoeaeoaoe oaeaoEom 3 some

xo mm 8:33 .033 850m :0
n/ 0 358 cc. _

\ \ 8.6.8 358

\

  
      

20ch .8200

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

omcmnoxo
:oéoflam
Ewen :0.
I - 6 I l accuse
L coma-m5.
\ a r r /
.3893 women... mouse zoom 30..
/ .2553
2339‘ 008.9 :9:

o 3
8:20.33.

50

After gaseous or solid substances are ionized in the ion source, the resulting ion
beam is magnetically separated according to ion mass directly after extraction from the
source. The ions are then accelerated by an electrostatic field. Finally, the ion beam is
directed towards the sample. In order to obtain homogenous implantation, raster scanning
of the beam is generally used. The implanted dose is determined by time integration of the
beam current divided by the area of coverage (see Equation 2.3).

The Freeman ion source is comprised of an oven heater, an arc chamber, a tungsten
filament and an extraction slit. Figure 3.2 schematically shows the structure of the
Freeman source. This segment was pumped by a diffusion pump to a vacuum of 1 x 10'6
Torr. Pure magnesium powder was placed inside the oven heater. When the oven was
baked at 500°C (magnesium’s melting point is 651°C, the vapor pressure is around 10‘4
Torr at 500°C ) for a period of time, vaporized species drifted into the arc chamber along
with the Ar working gas. The generation of ions originates inside the arc chamber which is
a cylindrical chamber of molybdenum. A tungsten filament is mounted horizontally
through the arc chamber and slightly forward of the chamber axis, with its ends protruding
out of the top and bottom of the chamber. Ionization begins with the application of power
to the frlarnent, causing it to heat up. When the filament reaches a certain temperature, it
begins to emit electrons. Since the arc chamber is at a more positive potential than the
filament, the electrons are attracted to the chamber walls, and accelerate toward them. A
fraction of the electrons collide with the gas molecules inside the chamber, resulting in the
formation of one or more species of positively charged ion. Several different species of ion
are usually formed during the process.

Positively charged ions are extracted through a slit from the plasma by electrical
attraction to an electrode which is maintained at 30 kV negative potential with respect to the
arc chamber. The ion beam exits the source and enters the magnetic field of a 90-degree
double focusing electromagnet. The magnet acts as a mass analyzer which separates the

ions in the beam according to their charge and atomic mass. The ionized species travel

E
i3
>

GAS

51

0 R FRBON SOURCE

      
  
  

ll
\\\\
I!

U III I MA NET .- IROUGH
_ PORT

I
II ACCEIJDECEL (EXTRACTOR)
ELECTRODE

ROUND ELECTRODE

/
—* EXTRACTION ION

\‘ BEAM

\\\\

I
I

7/////////<

 

 

 

 

WZWfl/A

 

 

7////////A

Figure 3.2 Schematic of Freeman ion source structure.

52

through a homogenous magnetic field and are constrained to a circular trajectory which was

given by

R = (%‘-’-) (fln’iwz (3.1)

where M is the ion mass in atomic mass units, V is the accelerating potential in volts, n
the charge state of the ion, B the magnetic field in gauss. The particles of the same charge
state and energy, but of differing mass, originating from a common source would follow
paths of different radii, and thus can be separated. Through calibration, the desired ions (in
present work single charge 24Mg+ or 4°Ar+ was selected) were selected by determining
corresponding magnetic field strength which aligned this beam with an aperture slit. Figure
3.3 is a mass spectrum of Mg and Ar generated in 350D implanter.

Upon completion of pre-analysis, the selected ion beam was extracted from the
terminal and accelerated toward the target chamber. A high voltage acceleration tube,
connecting the high voltage terminal to the bearnline and the target chamber, is employed to
provide the final beam energy. In the current study, the ion beam was accelerated to an
energy of 175 keV in the acceleration tube.

The ion beam diverges as it gains energy through the acceleration tube. To contrOl
this, a quadrupole doublet lens is located just beyond the grounded end of the acceleration
tube. The lens uses high voltage to focus the beam onto a small, symmetrical spot on the
target.

After focusing, the beam is scanned vertically and horizontally by two pairs of
electrostatic deflection plates. This enables a relatively large area to be covered uniformly
with ions, using a small diameter, intense ion beam. The horizontal deflection plates and
vertical deflection plates are driven by a specially controlled triangular wave output.

Neutral particles formed in the accelerator tube, and in the bearnline beyond and in

the analyzer region, are removed from the ion beam by superimposing a constant d—c offset

 

 

 

53

 

 

 

 

 

 

 

 

26Mg+

 

25Mg+
24Mg+
Ar++ /
8. 8. . . 3
V In In In (D
Dial Setting

Ar+

 

 

7.53

 

 

Mg and Ar Spectrum

Energy: 175 KeV
Extraction: 30 KV, 5A
ARC: 60 V, 0.7A

Vaporizer Temp: 520°C
Source Pressure: 5x10'6’1‘orr

 

 

 

Figure 3.3 Mg and Ar mass spectrum generated in 350D implanter.

54

voltage to the horizontal plates, causing the beam to be deflected plus 7 degrees or minus 7
degrees off its central axis in the horizontal direction. The neutral particles in the beam are
unaffected by the d-c offset voltages and continue on the original path, where they are
trapped by a beam dump in the region beyond the scanner and centered between the
bearnline.

3 .2.2 Target Set up and Afiecting Factors

At the target end of the irrrplanter, there is a specially designed working chamber.
Inside the chamber, a specially designed rotating fixture (rotating speed 1 rpm) allows 60
fibers (150 mm length each)‘to be processed simultaneously with a nearly uniform
irradiation of 360 degrees of fiber circumference without breaking the vacuum. On the
back side of the fixture, there is a custom designed Faraday cup. This Faraday cup is
placed at the central axis of the chamber, and perpendicular to incident ion beam direction.
There is a 1 cm2 circular hole at the center of the panel. This Faraday cup is connected to
an Ampere meter so that it can monitor the ion beam current variation and also continuity of
irradiation process. A schematic view of the fixture is shown in Figure 3.4 (a).

Dose calculation is based on following formula (referring to Equation 2.3)
where D is the dose (ions/cmz), I is the beam current in Amperes, t is implantation time in
seconds, A is beam scanning area in cm’, q is the ion charge state, and e is a the electron
charge (1.602 x 10'19 coulombs).

In the measurement of ion beam current, beam current versus X or Y sweep was
displayed on an oscilloscope, where beam current can be adjusted to maximum value. A
typical display of beam current is illustrated in Figure 3.5. The four different regions are:

a. flat and equal baseline section indicating total overscan of beam on mask;

b. rise and fall as beam spot is swept over edge of mask;

Radiation Shielding

Aluminirk

Beam LA

 

its»???

 

55

F“) Faraday Cup

  
   

Central Hole with Area 1 cm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(b)

Figure 3.4 Schematic of rotating fixture for ion implantation on sapphire
fibers: (a) view of fixture set up; (b) estimation of dose error due to fibers

shadow effects.

56

 

Beam Current

 

Scan Voltage

Figure 3.5 Oscilloscope display of beam current vs. X or Y scan Voltage. Four

regions:

a. Flat and equal baseline section indicating correct total overscan of beam on mask;

b. Rise and fall as beam spot is swept over edge of mask;

c. Flat section as beam is swept over main Faraday indicating stable source condition
and constant Faraday reading;

(1. Central dip. This is a qualitative indication of the beam focus and alignment of the
system.

57

c. flat section as beam is swept over main Faraday cup indicating stable source
condition and constant Faraday reading;

(1. central, only observed on the flag. This is a qualitative indication of the beam
and alignment of the system.

The reading of beam current can obtained from the main Faraday cup. After
adjusting the beam current to maximum value and to full scanning area, i.e. 91 cm2, the
dose can be calculated by equation 2.3 for the exposured time in which fibers were
irradiated.

Two factors which cause ion dose error and beam heating on the sample Should be
considered. One was self-shadowing due to the 140 11 diameter on fibers at the back side
of the fixture. In the current study, a total of 60 fibers were mounted in the rotating fixture,
separated by 6 degrees from each other. As shown in Figure 3.4 (b), the fiber in the
position B is covered by the project ion of a fiber in position A, which is separated by 6
degrees from fiber B. The tangential speed of a fiber is

V = (UR (3.2)

where (1) is the rotating speed (1 rpm), R is the radius of the fixture (50 mm). Assume the

project of A moves over B (distance is 2r) at a uniform speed V, = (11R sin(¢/2), where
4) =6 degree, I is the radius of a fiber (0.07 mm). Since the fixture is dynamically rotating,

the relative speed is

V" = 2mRsin(¢/2) (3.3)

Within one revolution, the covered time of fiber at position B by project of fiber A can be

obtained by: -

 

 

, = 35 = ,r (3.4)
v. amino/2)
Similarly, the covered time by next fiber on the position of the fiber B is given by:
At, = r 4) (3.5)
arRsin(2 * —)

Only half of the 60 pieces of fiber cause the shadowing effects on the fiber B, therefore, the

total covered time within one revolution is given by:

 

 

T=i ff 1 + +OOO+-————-1 (3.6)
‘ kasin(£) sin 2 * (—) sinn * (fl)
2 2
The shadowing coefficient can be obtained by:
,, = _T_
(l/w)

3° 1 1 1
=2: + +ooo+——— (3.7)

‘ R sin(2) sin 2 * (fl) sinn * (fl)

2 2 2

The self-shadowing effect was thus estimated to be approximately 11 percent.

The other factor was the beam heating effect on the sample. Present implantation
was canied out without target cooling at a nominal beam current of 80 uA to cover 91 cm2
of scanning area, and some degree of heating effect was expected. The precise temperature

rise depends upon a number of factors. These include the dimensions of the specimen and

59

the ion beam, the emissivity of the surface and the losses by conduction to the target
holder. By assuming that the sample is in poor thermal contact with the target support (the
thermal conductivity of sapphire is 0.086 cal/sec - cm - °C, from Saphikon, Inc.) and that
conduction losses are negligible, the Stephan-Boltzman law is used to estimate the

temperature rise due to ion bombardment. By the Stephan-Boltzman law [Kern, 1950]:

E = ()'(T‘4 - T04) (3.8)

where E is the beam energy deposited on the sample, 0' is the Stephan-Boltzmann constant
(5.67x 10'12 watts cm'2 K"), T is the temperature of the bombarded specimen at
equilibrium and To is the temperature of the surrounding. Thus a beam density of 0.879
trA/cm2 at an ion energy of 175 keV corresponds to 0.154 watt of power for each square

centimeter of the target, and the temperature rise is estimated to be less than 180°C.

3 .3 Abrasion Procedure

Abrasion tests of the fibers were performed on a specially designed rolling-contact
abrasion apparatus. Figure 3.6 Shows the experimental arrangement. The sapphire fibers
were placed between the load plate and the slider, which was covered by 1000 grit abrasive
paper (average particle size is 5 pm). The fiber axis was kept perpendicular to the
direction of the slide. The static load plate weighed 30 grams. The mobile slide, covered
with abrasive paper, drove the fibers to roll back and forth between the load plate and the
sandpaper by the force of friction. A 1 rpm motor was connected to the mobile slide. The
speed of the mobile slide was thus given by V = 25.4 sin(t) (mm/sec). The abrasive paper
was changed for each abrasion run. The abrasion protocol was carried out at room
temperature without lubricant. Abrasion time ranged from 5 to 60. nrinutes. As will be
shown later, this arrangement produced abrasion damage equivalent to that produced in
tumble mill reported by Trumbauer [1992].

 

'veP-per Fiber Specimen

(b)EnlargedVicwofContactAreaBetweenFlberSpecimcn,StaticLoadPlateandAbnsivePaper

Figure 3.6 Schematic plane view of abrasion apparatus.

61

3 .4 Three Point Bend Tests

Three point bend tests were carried out at room temperature individually on a
screw—driven micro-tensile load frame. The span length was 3 mm. All samples were
tested at a constant crosshead speed of 0.15 mm/min at room temperature. The rupture

load was monitored by a computer system. The rupture strength was calculated by:

a =LI'L— (3.9)

7rR3

where F f is the rupture load (N), L is the span length (3 mm), and R is the specimen radius
[Callister, 1985].

The number of test samples was at least 20 pieces for each condition. The bend
strength distribution of each sample group was characterized using Weibull distribution
[Johnson, 1964; Holland, 1989], since single crystal sapphire exhibits brittle fracture at
low temperature. The Weibull distribution function makes it possible to estimate a
population of infinite size from a small amount of data [Weibull: 1951]. In the current
study, all bend strength is characterized by so-called Weibull characteristic strength, which
will now be described.

The Weibull distribution function can be expressed as:

F(x)=1-expl:-[x-x“] ] (3.10)
It)

 

where the parameters are:

F ( x ) statistical fraction of specimen that failed at given stress or lower

x stress

xu stress below which no specimens failed ( called location parameter)

x0 characteristic strength, stress at which 63.2 percent of specimen failed

62

m Weibull slope or modulus.

The three Weibull parameters, xu , x0, m, are assumed to be constant of the
material. When xu is assumed to be zero, the three-parameter Weibull distribution function
becomes known as a two-parameter Weibull distribution function. The Weibull slope or
modulus m provides an indication of scatter or variation in the strength distribution; a small
value of m indicates that a significant amount of scatter exists in the strength distribution.

Equation (3.4) can be rearranged to form an equation for a straight line as follows:

ln(ln(

 

= _ _ 3.11
1—F(x))) m(ln(x x“) lnxo) ( )

In this form a plot of the distribution function should be linear in a coordinate system where

the ordinate is ln(ln( 1
1 - F (x

 

))) and the abscissa is ln( x — xu ).

Johnson [1951, 1964] proposed the so-called median rank method to obtain the

values of F( x). In the method, the data set is arranged in order of increasing stress or

rupture strength. Each value then has an order number according to its position in the list.

F ( x ) is obtained by the following formula:

n—(l-ln2)-(21n2—l)(n_1)
F(x): N ”'1 (3.12)

 

 

where n is the position number in the ordered list and N is size of the data set. One of the

median rank examples is shown in Table 3.2.

63

Table 3.2 Median Ranks

 

 

 

Sample Size = N
List 1 2 3 4 5 6 7 8 9 10
1 0.500 .2929 .2063 .1591 .1294 .1091 .0943 .0830 .0741 .0670
2 .7071 .5000 .3864 .3147 .2655 .2295 .2021 .1806 .1632
3 .7937 .6136 .5000 .4218 .3648 .3213 .2871 .2594
4 .8409 .6853 .5782.7 .5000 .4404 .3935 .3557
5 .8706 345 .6352 .5596 .5000 .4519
6 .8909 .7705 .6787 .6065 .5481
7 .9057 .7979 .7129 .6443
8 .9170 .8194 .7406
9 .9259 .8368
10 .9330

 

 

The value of F ( x ) and x - xu can be plotted as lnln[%1_ F(x )1] as a function of the

log of strength. The plot will be linear for the correct value of xu , which is also called the

location parameter.

If the original plot of the data is a straight line, then xu is assumed to be zero (i.e.,

the minimum stress below which no specimen can fail is zero). If the original plot is

concave downward, then there is some finite stress below which no specimen will fail.

The true value of xu can be found by substituting assumed values into the expression
x - xu until the Weibull plot becomes linear. The bend strength data in this study have
been characterized by Weibull’s characteristic strength, 0'0, i.e., the strength for failure
probability of 0.632.

 

3 .5 Indentation Fracture Tests

Figure 3.7 is a schematic diagram showing mutually orthogonal radial crack
systems produced by Vickers indentation. Theoretically, the technique of indentation
fracture test, proposed by Lawn and Fuller [1984], is based on two assumptions: (1) the
depth of the layer (d ) is small compared to that of the crack (6' ); and (2) the stress is
uniform within this layer. Two main types of cracks exist: radial/median cracks on a
symmetry plane normal to the surface and containing the load axis; and lateral cracks, on
shallow subsurface planes approximately normal to the load axis. The radial crack traces
on the specimen surface should constitute a sensitive indicator of the residual stress level.

Two components of the stress intensity factors are considered. First, there is the

residual stress intensity factor which is produced after indentation. It can be expressed by :

_ 1P
Kr—c—3l-2' (3.13)

where P is the peak contact load, c is the characteristic crack size, and x is a dimensionless

factor which represents the intensity of the persistent field. Here, x = 0.016 g , where E is

Young’ modulus and H is the hardness [Hioki, 1986].

The second intensity factor arises from the surface stress. The stress, a", , is

assumed to be uniform over the depth (d) of implanted layer, and d<<c (thin layer

approximation). The stress system is considered as a surface line force, 03d. The

appropriate stress intensity factor is:

Ks=2 ‘I’O’, dl/Z (3.14)

where ‘I’is a dimensionless crack geometry term of value about unity.

65

 

 

 

Median Crac
Lateral Crac

 

 

 

 

2c2

 

201

 

 

 

Figure 3.7 Schematic representation of the crack geometry around a Vickers
indentation in a brittle material. Cracks have half penny-like geometry center
about central deformation zone (shaded region). Dashed lines show lateral
cracks and dis the thickness of the irradiated compressive stress layer.

66

Finally, provided that the surface stress layer does not alter the characteristic of the
elastic-plastic properties embedded in the 1 term, by the principle of superposition one can

obtain the stress intensity factor for the radial crack system:

1
K = g+2wosd§ (3.15)
c2

Considering the equilibrium fracture condition the crack extends when K reaches the level

of the critical stress intensity, Kc (constant of bulk material), the expression can be

rewritten:

l
= Kc — 2wosd§ (3.16)

finwl’é

Now, using the reference state of zero surface stress to eliminate x , inserting c = CO at 0', =

0, we finally have:

(3.17)

3
1_(._,).
or as = Kc——Cl— (3.18)

Therefore, the stress evaluation is reduced to a measurement of relative crack dimension in
the stressed and un-stressed states.
Vickers indentation tests were carried out on an interfacial debonding system

located in the Composite Materials and Structure Center at Michigan State University. This

67

system was controlled by a computer in which loading and unloading rates and position of
an indenter were precisely controlled. This system also provided an optical microscope in
order to observe the surface topography at a maximum magnification of 100X. The
Vickers indentation load was pre—set to be 100 grams throughout the tests. The indenter tip
had pyramid shape and a 4 um diagonal in length, and was made by Magwen Diamond
Pro., MA. The test method was based on ASTM E384-89 (Standard Test Method for
Microhardness of Materials). During the test, the sapphire fibers were cemented to a glass
slide with cyanoacrylate adhesive so that the fibers were stable. The indenter contacted the
sapphire fiber at a velocity of 20 pun/sec, and the time of the full load application was 15
seconds. After indentation, the leg length of the cross diagonal was measured by using
optical microscopy, and the measured cross diagonal 2c1 and 26‘; were recorded as
illustrated in Figure 3.7. For each implantation condition, at least 20 points of indentation
were made, and the average of the diagonal length was then determined. The morphology
of indentation traces was observed in SEM (Hitachi S-2500C) after light coating with Au.

3 .6 Post-Implantation Annealing

Implanted sapphire fibers were annealed in both vacuum and air environments.
Vacuum annealing was carried out in MT S 810 hydraulic servo system with a centorr
vacuum furnace. The vacuum was around 1 x 10'5 Torr. Annealing temperature ranged
from 1073 K to 1473 K for 1 hour. After annealing, the chamber was furnace cooled
down to room temperature. The air environment annealing was only carried out at 1273 K

for 1 hour. It was also furnace cooled down to room temperature.

3.7 Fabrication of MAI/111203 Composite by Difi'usion Bonding

The bulk matrix material of cast MA] was sectioned into approximately 2.5 x 0.5 x
0.3 cm plates by means of EDM. The plates were first mechanically polished on both sides
with SiC paper ranging from 240 to 600 grit. Further polishing was continued using

68

polishing cloth and alumina powder of size ranging from 600 grit, 5 um, 0.3 pm. When
one side of the plate was polished to a surface fmish of 0.5 pm, another side of the plate
was superglued on a steel block, and then polished down. The final thickness was around
0.5 mm. The whole block with both sides polished was ultrasonically cleaned in acetone
until the glue holding the plate is dissolved. As a result of the grinding, the bonding plates
had parallel sides that helped ensure a uniform load distribution during hot pressing.

The grooves to place individual fiber on the polished NiAl plate were made by
using a Material Science LTD, Model MK-II electro-discharge machine, with setting of R =
3, C = 3, and V = 3. A custom plunger tool, as shown in Figure 3.8, was used to machine
about 250 um diameter grooves with approximate 500 um spacing. Figure 3.9 shows an
EDMed bonding plate. After machining, the plates were degreased in a cleaning solution,
then ultrasonically cleaned in warm water for 10 minutes, dried with a hot air gun, then
ultrasonically cleaned again in hexane for another 15 minutes and finally naturally dried.

Sapphire fibers were placed into the groove with careful handing. After placement,
both ends of the plate were taped together using a small amount of double-sided tape to
ensure the that fibers remained in place. (The tape was finally burned off during vacuum
hot pressing). The assembly of NiAl/sapphire fibers, schematically shown in Figure 3.10,
is in a manner similar to the foil-fiber-foil technique [Mackey, 1991]. To prevent the MA]
plates from bonding to the TZM compression plates, boron nitride coated alumina plates
were placed between the plates and the assembly.

The diffusion bonding process was carried out at a pressure of 2 x 10'5 Torr in a
vacuum furnace of a MT S810 system. The heating rate was controlled at 36 K/min from
room temperature to the maximum holding temperature 1523 K, and then a constant stress
of 15 MPa was applied to the composite assembly for 3 hours with holding the temperature
at 1523 K. The cooling rate was 34 K/min from the holding temperature to 723 K and then

69

Connection bar
/ With EDM

 

 

 

/ Steel sheet ~0.2
mm thick

Figure 3.8 Schematic of cutting tool for MA] diffusion bond assembly plates.

7O

 

\\

0.2 ~0.25

0.35

 

 

 

unit: mm

Figure 3.9 Schematic of EDMed NiAl diffusion bond plate.

sapphire fiber

71

  
 
     

NiAl Plates

  

OQQOQOQOOOO’V
. 44.x.:.;.;.;.;.:.;.;.:
..........,...

666664646666
66

     

6
9.9
6

 

6
'6
o
6

  
  

EDMed grooves

Figure 3.10 Schematic of NiAl/sapphire fiber diffusion bond assembly.

72

the furnace power was turned off and furnace cooled. According to previous experience
[Schalek, 1995], assembly slipping could be minimized during the heating stage of the
process by applying a preload of 0.5 MPa to the assembly. Once the hold temperature was
attained and held for 5 minutes, the full stress was then slowly applied over a period of 5
minutes. When the furnace was cooling down, the stress was removed to minimize the
composite fracture during cooldown.

The N iAl/A1203 composite in the current study includes N iAl with control sapphire
fibers assembly, and MA] with fibers implanted with 1 x 10“5 Ar”, 4 x 10“5 Ar”, 2 x 10“5
Mg”, 5 111016 Mg”, 7 x1016 Mg” and 2 x 1016 Mg"/cm2 assemblies, respectively. The
fiber volume fraction for each assembly is approximately 7%.

The sapphire fibers were etched from NiAl/A1203 composite using an enchant
solution. The solution consisted of 50% H20, 33% HNO3, and 17% HCl (by volume).
The solution temperature was maintained at 63°C. After the NiAl/A1203 dissolved and the
solution cooled, it was filtered through Whitman filter paper, and then rinsed by distilled
water to remove the residual etchant. After the fibers on the filter paper dried, they were
carefully collected and stored.

3 .8 Preparation of TEM samples

Initially, the ultra-microtome technique was attempted for preparation of sapphire
transverse TEM samples. This attempt failed because sapphire fibers are very brittle. The
sectioned samples were not uniformly electron transparent, and serious damage was caused
by the diamond knife. Later, the preparation of transverse fiber TEM specimens was
successfully accomplished by a technique similar to that reported by Nutt [1985]. A single
layer of parallel sapphire fibers (50 pieces of sapphire fibers, 40 mm length each) was
placed between two foils of 99.9% pure aluminum. The assembly was then diffusion-
bonded in an argon atmosphere in an MTS 810 system by hot-pressing with 15 MPa stress

at 500°C for 45 minutes, producing a composite monotape. The samples of 3 mm disc

73

were obtained by an ultrasonic core drill on the 6 mm by 40 mm monotape. After this, the
disc was placed on an SPI Pyrex specimen mount using crystal-bond adhesive. One side
of the disc was polished on 600 grit sandpaper until two thirds of the, sample thickness was
removed. The stub was then soaked in acetone to remove the crystal-bond. Since the
sapphire fibers were very brittle, a molybdenum ring was attached to support the polished
side by using a high temperature epoxy compound from Gatan which can withstand high
temperatures arising from the ion mill. The sapphire fiber disc with Mo-ring was again
mounted on a Pyrex stub by crystal bonding. The sample was then polished to a final
thickness of approximately 60 um. Following this, the sample was dimpled on both sides
by a Gatan dimple grinder. One technique used in dimpling to control the dimpled depth
was to monitor the color change of sapphire fibers illuminated by a polarized transmitted
light. According to the color spectrum of sapphire with thickness, the corresponding
thickness could be controlled to 40 to 50 um. If a very perfect dimple on the sapphire fiber
disc was obtained, the time required for perforation during ion mill could be dramatically
reduced. Finally, the disc was ion thinned on both sides for perforation using the Ion
Technology argon-beam ion mill operating at 5 kV, 100 uA per gun at a 20° incidence angle
at the beginning, changed to 150 during final perforation. The TEM specimen was
observed in a Hitachi H-800 microscope in the Department of Materials Science and
Mechanics at Michigan State University, and in a Jeol 2000 FX at The Electron Microbeam
Analysis Laboratory at the University of Michigan in Ann Arbor.

3 .9 Electron Channeling Pattern (ECP) and Atomic Force Microscopy (AFM)

The surface microstructure of implanted sapphire fiber was also studied by ECP
(Electron Channeling Pattern) using CAMSCAN model 44FE. The principle of ECP is that
when the impinging electron beam from SEM interacts with the planes of the crystal being
illuminated, electron diffraction occurs; as this beam is swept across the crystal, the angle

of incidence changes, resulting in changing diffraction conditions. Thus, when the

74

imaging mode is either back-scattered electron or specimen current, the resulting diffraction
pattern may be observed. The electron channeling patterns were obtained by operating the
SEM in the channeling mode, i.e., basically by rocking the beam about a stationary point
within each grain of interest. For metal and ceramic crystals which are free of plastic
deformation, this process generates sharp, fine structured electron channeling patterns; the
presence of dislocation damage causes loss of higher order lines, broadening of other lines
and, for sufficiently large deformation, the loss of contrast altogether.

In the current study, ECP patterns were obtained for unimplanted and implanted
sapphire fibers. The ECP pattern change with ion dose was compared with the TEM
information on the same condition sapphire fibers.

A N anoScope HI Atomic Force microscope (Digital Instruments Inc.) was used to
extract topographical information from sapphire fibers and also to generate a quantitative
surface profile of the abrasive paper. Basically, the AFM has a very sharp tip which
protrudes from the underside of a small, flexible cantilever. A laser beam reflects off the
top side of the cantilever onto a split photodiode. Height variation on the sample surface
deflect the cantilever causing the position of the laser beam on the photodiode to change.
The differential voltage from the photodiode elements provides a sensitive measure of the
cantilever deflection (surface height of the sample).

By NanoScope III AFM, surface topography information on implanted and
unimplanted fibers was obtained by scanning areas ranging from 5 um by 5 um to 100 pm
by 100 um. Further, surface roughness was analyzed by sectioning the scanning area.

3.10 Transport Calculations of The Projected Ion Range and Damage Profile
TRIM-90 (TRansport of Ions in Matter) designed by Ziegler [1985] calculates the

penetration of ions into solid using a Monte-Carlo simulation. This program accepts a

description of a complex target made up of compound materials, with up to three layers

made up of different materials. It calculates both the final distribution of the ions and all

75

kinetic phenomena associated with the ion’s energy loss: target damage, ionization and
phonon production. The calculation of cascades, target displacement, replacement
collision, etc., is theoretically based on following assumptions:

(1) Incident atoms having atomic number Z 1 and energy E collides with an atom of
atomic number 22 within the target. After the collision , the incident ion has energy E, and
the struck atom has energy E2. The displacement energy E; for the target, E, is the binding
energy of a lattice atom to its site, the final energy of a moving atom, below which the atom
is considered to be stopped.

(2) A displacement occurs if E, >E, (the struck atom is given enough energy to
leave the site). A vacancy occurs if both E, >E, and E,>E, (both atoms have enough
energy to leave the site). Both atoms then become moving atoms in the cascade. The
energy E, of atom E, is reduced by E, before it has another collision. If E,< E ,, then the
struck atom does not have enough energy and it will vibrate back to its original site,
releasing E, as phonons.

(3) If E, > E, and E, > E, and Z, > 2,, then the incoming atom will remain at the
site and the collision is called a replacement collision, with El released as phonons. The
atom in the lattice site remains the same atom by exchange. If E, < E, and E, < E, and Z, >
2,, then Z, becomes a stopped interstitial atom.

(4) If E, < E, and E, < E,, the E, becomes an interstitial and E, + E, is released as
phonons. If the target has several different elements in it, and each has a different
displacement energy, then E, will change for each atom of the cascade hitting different
target atoms.

The results of TRIM show the number of displacement collisions which records
how many target atoms were set in motion in the cascade with energies above their
displacement energy; and the number of replacement collision which reduces the number of

vacancies, and the interstitial atoms.

76

TRIM also calculates sputtering, which is ballistic removal of near surface atoms
from the target. When a cascade gives a target atom an energy greater than the surface
binding energy of that target, the atom may be sputtered. To actually be sputtered, the
atom’s energy normal to the surface must be above the surface binding energy when it
crosses the plane of the surface. The sputtering of a surface is described by a “sputtering
yield”, which is defined as the mean number of sputtered target atoms per incident ion.

The parameters which are used in calculation of TRIM are:

Ions: Mg (mass = 23.99), Ar (mass = 39.95)
Energy: 175 keV and 120 keV, respectively

Ion Angle to Surface: 0 degree
Target: Al,O3, 40% of Al and 60% of 0

Bottom Depth: 1 ttm

Density: 3.98 g/cm"
Displacement Energy: 18 ev for Al Atom and 72 ev for O Atom in Al,O3
Total Ions calculated: 4000

Typical TRIM calculation results are shown in Figure 3.11.

77

 

0.0014

1

175 KeV Ar+

   

I

0.001 2

0.001

 

0.0008 175 KeV Mg+

0.0006

0.0004

Distribution of Mg (Atoms/Angstrom/ ion)

0.0002

 

 

 

0 50 100 150 200 250 300 350 400

Depth (nm)

Figure 3.11 The distribution of Mg+ and Ar+ implanted into sapphire fiber at
different energy, calculated by TRIM (90.05).

 

CHAPTER IV

EXPERIMENTAL RESULTS AND ANALYSIS

In the following section, experimental results in current work will be presented.
These results include bend strength of sapphire fibers after different implantation
conditions, effects of abrasion on bend strength of control and irradiated fibers, retained
bend strength after a process of composite fabrication, annealing effects on bend strength
of implanted fibers, estimation of surface stress by micro-indentation technique,
microstructure studies by TEM , surface topography after abrasion and fracture surface
observed using SEM and ESEM.

4.1 Three Point Bend Strength without Abrasion

Single crystal sapphire fibers were implanted at different doses of 2 x 1015, 4 x
1016, 5 x 1016 and 2 x 1017 Mg+ cm'2, and 1 x 1015, 2 x 1015 and 4 x1016 Ar"' cm'2,
(Note: Zero dose refers to unimplanted or control condition). The characteristic bend
strength is summarized in Table 4.1. The mean strength and standard deviation is also
given in Table 4.1. For Mg+ implantation, it can be seen that the characteristic bend
strength changed very slightly with dose, from 9.31 :1: 0.48 GPa for unimplanted fibers to
9.47 :l: 0.33 GPa for the fibers implanted with a dose of 4 x 1016 Mg+ cm'z. At the
maximum dose of 2 x 1017 Mg cm'z, bend strength decreased slightly to 8.21 :1: 0.32
GPa. The measurement of bend strength on unimplanted sapphire fiber is close to that
measured by other investigators on similar material [Sayir, 1992]. The Weibull modulus is
about 6.8, consistent with modulus values on Saphikon sapphire fibers reported in the
literature [Bowman, 1993; Davis, 1994; Sayir, 1992]. It is apparent that Mg+ implantation
has only a very modest effect on bend strength, producing a maximum increase of only 1.5

percent for irradiation of 4 x 1016 Mg cm'2, and degrading unabraded strength by

78

79

approximately 8.8 percent for a dose of 2 x 1017 Mg+ cm‘z. In the case of Ar
implantation, both measured characteristic strength and mean strength are slightly lower
than that of the control fiber. For instance, the characteristic bend strength at doses of 2 x
10“5 and 4 x 1016 Ar“/cm2 are 8.34 :1: 0.40 and 9.24 i 0.58 GPa, respectively.

Figures 4.1 and 4.2 present the Weibull plots of bend strength for different doses
of Mg and Ar+ implanted fibers. In these plots, two dashed lines represent 95 percent
confidence band for the bend strength of control fiber. It can be seen that most of the
strength data points fall within the confidence band for both Mg+ and Ar+ implantation.
Figure 4.3 compares the variation of characteristic strength with doses in both cases. It
shows that the bend strength of Mg and Art implanted fibers varies slightly from the
strength of control fiber. Therefore, it is obvious that without any surface damage by
handling or abrasion, ion implantation of Mg“ and Ar“ causes, at most, very modest
changes in the bend strength of sapphire fibers.

The scatter of characteristic bend strength data comes mainly from the brittleness of
sapphire fiber. Since sapphire fiber is sensitive to surface flaws, the depth of surface flaws
will affects the fracture behavior during three point bend tests. According to the principle
of the weakest link [Ochiai and Murakarni: 1981, 1988], the most severe flaw in the bend
gauge determines the fracture behavior.

It will be recalled that, in order to remove any contamination particles on the surface
of the sapphire fibers, these test fibers were cleaned by soaking in ice water for 1 hour,
followed by drying in air. This process might have had an effect on surface properties of
the fibers since sapphire fiber is very sensitive to surface condition. Figure 4.4 presents
the strength results for the fibers processed by different cleaning methods and tested in
different environments. It shows that the bend strength of ice-water cleaned fiber is
slightly higher than that of as-received fiber, while moisture environment has little effect on
the bend strength. For instance, the characteristic bend strength for ice-water cleaned fibers

is 9.3 GPa whereas the fibers were tested in dry air or 65% of humidity environments. For

80

 

 

 

 

Table 4.1 Three Point Bend Strength of Sapphire Fiber with Different Implant Doses
(175 keV)
Beam Estimated Characteristic Mean Number
Ion Current Temperature Dose Strength Strength Weibull of
Species Density ( oC) (ions/cmz) (GPa) (GPa) Modulus Samples
mA/cmz)
0 0 0 9.31 :t: 0.48 8.71 :1: 0.52 6.83 80
1.3 198 2 x 1016 8.87 :1: 0.52 8.28 :1: 0.59 6.57 35
2.2 252 4 x 10“5 9.47 :1: 0.33 9.00 :1: 0.39 9.43 60
Mg+ 1.0 171 5 x 10“5 9.26 :t 0.56 8.62 :1: 0.53 6.41 30
2.3 257 7 x 10‘6 8.91 :1: 0.44 8.48 :1: 0.20 9.26 25
1.2 188 2 x 10” 8.21 :1: 0.32 7.78 :1: 0.20 8.65 40
1.3 197 l x 10'6 9.15 :1: 0.33 8.31 :1: 0.42 4.96 25
Ar+ 1.1 179 2 x 10“5 8.34 :1: 0.40 7.34 :I: 0.50 3.46 25
0.88 160 4 x 10“5 9.24 :1: 0.58 8.55 :1: 0.43 6.28 25

 

 

 

 

 

 

 

 

 

81

 

 

 

 

 

 

 

 

 

 

1 . (F
‘I .......................................
.5, ...........................................
I,
0.1 II """""""""""""""""""""""""
g No Abrasion
8 a!
.n unimplanted
E 5 s 5!
“W: """"""" , “$1 """ 0 2,11016 Mg+crri2
5 ! i .I
a 0.01 E..........., ........... r...,..., ........ A 4x1016Mg+crri2
Z i I :1
- ------------ g ------------------ g ------------- x 5:11016 Mg+crri2
- I ' 16 -
a: 7x10 Mg+cm2
g g 0 2x1017Mg+cni2 I
0.001 t 1 n (all: 1 l I 11111]
1 10 100

Characteristic Bend Strength (GPa)

Figure 4.1 Weibull plot of bend strength for Mg+ implanted sapphire fiber
without abrasion. Two dashed lines represent 95 percent confidence band.

82

 

 

 

Failure Probability

 

 

 

 

1 a
; . ..( ............. ................. .............
0.1 s l; ............. ................. .............
: 2 ........ . ................. é ................. .............
No Abrasion
0.01 unimplanted
1:11016 Ar+cm'2.
+ -2
2:110l6 Ar cm
. . , X 4x1016 Ar+cm’2'
0.001 I 1 l l l 1 1 li 4L fl l l l lil

 

 

 

1 10 100
Characteristic Bend Strength (GPa)

Figure 4.2 Weibull plot of bend strength for Ar"' implanted sapphire fiber
without abrasion. Two dashed lines represent 95 percent band of unimplante
sapphire fibers.

83

 

12°00 "N z : 7 E 3 Y

1 0.00

 

 

 

 

 

 

 

 

 

 

E

g 8.00

s:

H

O

c:

2

‘7, 6.00

'O

C

d)

1::

£2

17. 4.00 1. .
E 0 Mg Ions
0

g I Arlons
5 2.00 , ,

 

 

 

 

 

0.00 _/\I 1 1 111111 1 1 111111 1 1 1 L1111
16 17 18
0 ”‘10 1X10 1x10
Dose (ions/cmz)

 

Figure 4.3 Bend strength variation for Mg and Ar+ implanted sapphire fibers
at different ion dose without abrasion. The error bar is one standard deviation.

84

 

 

IVE!

0.1

 

 

Failure Probability

E n D Ice-water clean, tested
i i 0 at 6596 of humidity

; 0 Ice-water clean, tested

A5 at dry air enrivonment

in A As-received, without
‘ clean, tested at 65%
of humidity

0.01 1 l I 1 I l I I I I I I l I l I l
l 10 100
Characteristic Bend Strength (GPa)

 

 

 

 

 

 

Figure 4.4 Effects of cleaning methods and testing environments on bend
strength of unimplanted sapphire fibers.

85

as-received fibers without cleaning, the bend strength is 8.28 GPa. It has been

documented that the strength degradation in the intermediate temperature region (between
200 and 600 0C) is due to stress corrosion, i.e., from chemical species in the environment
reacting at crack tips and lowering the bond strength, which in turn lowers the fracture
strength [Sayir, 1992; Shahinian, 1971; Heuer, 1966]. However, it is unlikely in this
research that stress corrosion cracks occur in sapphire fiber at room temperature.

Figure 4.5 shows the SEM observation on the fracture surface for both implanted
(dose of 5 x 10"5 Mg*/cm2) and unimplanted fibers. It illustrates fully brittle fracture
features and shows no substantial difference between the two conditions. The site of crack

initiation could not be determined.

4.2 Effect of Abrasion on Bend Strength

Abrasion causes markedly different bend strength degradation in control and
irradiated sapphire fibers. In the current work, fibers were implanted at doses of 2 x 1016,
4 x 1016, 5 x 1015, and 2 x 1017 Mg cm‘z, and were subjected to the controlled abrasion
process. The results show that there was a significant decrease in bend strength for
unimplanted fibers after abrasion, whereas the bend strength of implanted fibers was close
to original strength. Table 4.2 summarizes the bend strength change for different abrasion
conditions. Figure 4.6 represents the bend strength change with implantation doses of Mg“
at different abrasion times. It can be observed that abrasion causes strength deterioration,
but the extent of degradation depends on surface condition. For unimplanted fibers, the
bend strength retention is 51 percent of initial strength after abrasion of 10 minutes; the
value rises to 89 percent for fibers irradiated at a dose of 2 x 1016 Mg+ cm‘z, and reaches
93 percent for the highest dose of 2 x 1017 Mg cm'z. For two other intermediate doses,
strength retention is above 72 percent after 10 minute abrasion. Figure 4.7 and Figure 4.8
illustrate the Weibull plots of characteristic bend strength versus failure probability for
fibers implanted with Mg and Ar+ at abrasion times of 20 and 60 mins, respectively. In

 

(b) 5 x 1016 Mg + /cm2

Figure 4.5 SEM observation on the fracture surface for both unimplanted and
implanted sapphire fiber.

87

these figures, two dashed lines still represent the 95 percent confidence band of bend
strength for as-received fibers without abrasion, and are included for reference purposes.

It is clear that all lines of failure probability versus strength have been shifted to the left,
i.e., characteristic bend strength has been degraded due to abrasion. However, the extent
of degradation depends significantly upon irradiation dose. It is worth noticing that the
curve of failure probability vs. characteristic bend strength for as-received fibers shifted left
away from the lower band of the confidence interval, while the curves of implanted fibers,
either Mg+ or Ar+, still fell within the band of the confidence interval. Statistically, one can
say with confidence that abrasion causes significant strength degradation on as-received
fibers, but no significant degradation on implanted fibers.

The strength loss of unimplanted fibers is comparable to the losses reported by
Trumbauer [1992] in which a tumble-mill self-abrasion treatment performed on unsized
sapphire fibers resulted in 30 percent strength degradation. Therefore, the abrasion
conditions in the current work are somewhat more severe than in a tumble mill. Relating
these results to the sapphire fiber handing and shipping process as well as the hot pressing
process for sapphire fiber reinforcement composite material, it is obvious that surface
damage by abrasion processes will significantly reduce strength.

The fiber surface topography was examined by ESEM (Environment Scanning
Electron Microscope) after abrasion testing. Figure 4.9 shows ESEM rnicrography of
sapphire fibers. It is clearly evident that the surface condition of unimplanted fibers is
worse than that of implanted fibers after abrasion whereas the surface conditions of the
implanted fiber changes little after abrasion. Even though little direct evidence about the
depth of surface flaws after abrasion is available, this topographical information is still
consistent with the results of bend strength, in which bend strength retention of
unimplanted fibers is 48 to 62 percent while retention can reach 90 percent or above for
implanted fibers after abrasion.

88

Table 4.2 Effect of Abrasion on Bend Strength of Sapphire Fibers

 

 

 

 

Abrasion Rupture Strength Strength
Dose Time Weibull Characteristic Mean Retentionl

(ions/cm?) (min) Modulus (GPa) (GPa) (96)
0 6.83 9.31 8.71 100

5 4.51 5.70 5.20 61

O 10 5.16 4.73 4.35 51

20 3.52 5.78 5.20 62

40 5.04 5.12 4.70 55

60 4.29 4.51 4.08 48

0 6.57 8.87 8.28 95

5 7.09 8.22 7.71 88

2x1016Mg+ 10 6.88 8.33 7.81 89
20 6.78 8.40 7.87 90

40 4.41 7.73 7.04 83

60 7.60 7.79 7.34 84

0 9.43 9.47 9.00 102

4x1016Mg+ 10 4.71 7.08 7.47 76
20 4.19 7.48 6.79 80

0 6.41 9.26 8.62 99

5 4.95 7.42 6.83 80

leomMg+ 10 5.35 6.74 6.23 72
20 4.43 7.57 6.87 81

40 8.14 8.26 7.79 89

60 4.66 7.58 6.95 81

0 8.65 8.21 7.78 88

5 10.20 8.45 8.06 91

17 + 10 10.97 8.66 8.28 93
2x10 Mg 20 5.54 7.84 7.23 84
40 6.99 8.09 7.56 87

60 5.20 8.15 7.48 88

0 4.96 9.15 8.31 98

1x1016A1'+ 20 4.68 6.44 5.87 69
60 3.93 6.83 6.17 73

0 3.46 8.34 7.34 89

2x1015Ar” 20 3.92 7.05 6.36 76
60 6.71 7.14 6.66 77

0 6.28 9.24 8.55 99

4x1016Ar+ 20 7.28 7.71 7.23 83
60 5.82 7.54 6.88 81

 

 

 

 

 

 

 

1 The strength retention is the ratio of bend strength to that of as-received condition, i.e. to the

characteristic bend strength 9.31 GPa.

 

Bend Strength (GPa)

89

 

 

 

 

 

_x_ t = 20 min.

-"°--- t=60min.

o /\/ 1 1 1 1 1 1 1 f

N 7

035, 2x10‘6 1x10‘
Dose (Mg/cmZ)

 

 

 

 

 

 

Figure 4.6 Effect of abrasion on bend stength of sapphire fibers.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1
0.1
.51‘
3
«I
.n
2
a
2
g l ' l
g 0.01 : ------------- fi-~---,---~---, ---------------------------------- l -----------
I ........................... .I. ..................... . Legend
' t = 20 minutes D Unimplanted
' ' ‘ ' x 2111016 Mg+ cm'z
. 1 -
000 (1) A 4111016 Mg+cm2
c 5111016 Mg+cm’2
0 2x101‘7Mg4'cm'2
b 0.1 : .......................
é ....... 2...; .....................................
.0 :
2
n- ...t..... ............................................
2
2
E 0.01 E ' o ..............................................
1: I 'l
1.. ...........................................................................
t= 60 minutes
0.001 1 .11....1 1 .

1 10 100
Characteristic Bend Strength (GPa)

Figure 4.7 Weibull plot of bend strength for Mg+ implanted sapphire
fibers abraded 20 and 60 minutes, respectively. The two dashed lines
represent 95 percent confidence band of bend strength for unimplanted
sapphire fibers at room temperature without abrasion.

91

0.1
g
3
as
.n
2
n.
2
2
'5 0.01
IL
1 = 20 minutes Legend
El Unimplanted
4 2 16 + —2
0.001 (1) A 1x10 A’ m
0 2x1016 Ar+cm'2
o 4x1016Ar+cm-2
>,, 0.1
g
IE
«I
.n
2
n.
2
2
.2 0.01
t=60minutes
0.001

 

1 1 0 1 00
Characteristic Bend Strength (GPa)

Figure 4.8 Weibull plot of bend strength for Ar+implanted sapphire
fibers abraded 20 and 60 minutes, respectively. The two dashed lines
represent 95 percent confidence band of bend strength for unimplanted
sapphire fibers at room temperature without abrasion.

92

 

(a) Non-implantation without abrasion (b) Non-implantation with abrasion (60 min)

 

 

 

(c) Implantation without abrasion (d) Implantation with abrasion (60 min)
(2 x 1017 Mg'l'cm'z) '( 2 x 1017Mg +cm-2)

Figure 4.9 ESEM images on the surface topography of sapphire fibers for different
implantation and abrasion conditions. (a) unimplanted and unabraded; (b) unimplanted
and abraded 60 minutes; (c) implanted (2)110" Mg’cm" ) and unabraded; (d) implanwd
(21110" Mg’cm") and abraded 60 minutes

93

4.3 Surface Residual Stress Determined by Micro-Indentation Fracture

The Vickers indentation technique was used to estimate the residual stress in the
implanted zone according to the methods described in 3.5. Table 4.3 lists the
corresponding indentation crack length at different dose. Each average crack length was

measured on basis of at least 20 indentation cracks.

Table 4.3 Micro-indentation Crack Length at Different Dose

 

Dose unimplanted 1x10" Ar‘lcm2 21110“s Mg‘lcm2 5x10“ Mg‘lcm2 2711011 Mg’ch2

 

 

Crack Length (pm) 32.47zt3.1 23.59:l:3.3 22.52:l:3.8 22.03125 20.45i3.7

 

 

 

Figure 4.10 shows the SEM images of indentation crack traces for different
implantation conditions. It is obvious that the indentation crack length on the implanted
sample surface is shorter than the unimplanted fibers. According to Equation 3.18, the
corresponding surface stress can be calculated. Here, it was assumed that unimplanted (or
control) fiber was at a zero stress state. The integrated residual compressive stress in the
implanted region increases as a function of dose as shown in Figure 4.11. The residual
compressive stress was found to reach a maximum 2.8 GPa at a dose of 2 x 1017 Mg+
cm'2. This measured residual compressive stress is the same order as that measured by
cantilever beam bending by Burnett and Hioki [Bumett, 1985; Hioki, 1989].

This technique can also be used to evaluate the apparent fracture toughness of
implanted sapphire fibers. Using the equation of Lawn [1980] for calculating radial
cracking around a Vickers indentation, the apparent fracture toughness Kc may be written

as:

Kc = 0.0139 €3,- P cm (4.1)

94

where c is the radial crack length in indentation, H is the hardness of sapphire fiber (2000
Kg/mmz, from Saphikon, Inc.), E is Young’s modulus along c-axis of sapphire fiber (414
GPa), and P is load (100 g in the present study). The apparent fracture toughness was
found to increase with implantation dose as shown in Figure 4.12. Compared with
unimplanted fibers, the fracture toughness of sapphire fibers at a dose of 2 x 1017

Mg+ cm'2 increased to almost twice as that of unimplanted fibers. For other doses, the
apparent superficial fracture toughness increases over 50 percent.

Changes in the indentation fracture behavior by ion implantation will arise mainly
from (1) the generation of surface residual stress [Burnett and Page, 1984, 1985] and/or
(2) the increased surface plasticity accompanied by the implantation-induced surface
amorphization [Hioki, 1986]. The increase in measured Kc has been qualitatively
attributed to radial cracks shortened by the existence of surface compressive stresses
generated by ion implantation. It is well known that ion implantation has two main effects
on the indentation fracture behavior of brittle materials. First, a suppression of lateral crack
breakouts takes place, together with a suppression of subsurface lateral crack propagation
or nucleation. Second, a small but significant decrease in the extent of the radial crack

trace on the test surface yields an apparent increase in the KC value.

It should be pointed out that the fracture toughness KC change due to ion
implantation is a so called ‘apparent’ fracture toughness. Because of ion implantation, the
near surface microstructure has been changed, therefore, the resistance of surface to
fracture has also been changed, but not the material below the irradiated depth. Apparent
fracture toughness is different from bulk material’s fracture toughness.

Since micro-indentation tests were directly done on sapphire fiber surfaces, it is
necessary to make some comments on the technique. It has been shown that indentation
fracture technique can measure the compressive stress induced by ion implantation [Burnett
and Page, 1984, 1985; Hioki, 1986]. The indentation fracture around a Vickers

indentation trace in brittle single crystal materials, however, is strongly anisotropic, i.e., it

95

(a) control

 

(c) Dose = 5 x 1016 Mg cm— (11) Dose = 2 x 1017 Mg+ cm'2

Figure 4.10 SEM images of the indentation traces on sapphire fibers at different doses
£14m 1020 grams). (a) unimplanted; (b) 2x10" Mg’ cm"; (c) 5x10“Mg’cm"; (d) 2x10"
g‘ cm .

96

 

 

 

 

 

 

 

 

 

4 ’\/

A 3

U

n.

8

3

.13

V)

g 2

8

8

a.

E

O

U

'3

.3 1 1- ........................................................

'8

1: I Mg" ion

, : : O Ar+ion
0 /\/1 .‘1 1 1 1 1 1 1 11 1 1
No
16 17

Dose1x1o 1X10

Dose (Ions/cmZ)

Figure 4.11 Variation of intergrated surface stress with doses of Mg"' implantation
into sapphire fibers. The error band is one standard deviation. The stress is
compressive.

97

 

 

 

 

 

 

 

 

 

4 (V
E 3
a
n.
I
V
tn
m
‘c’
g 2
a
a
*9 r
2
a
'6
E 1 1- ..............................................
LL
I No Irradiation
O Mgion
§ A Arion
O /\/[ 1 1 1111111 1 1 111111
No 16 17
Dose1x10 1x10

Dose (ions/cmZ)

Figure 4.12 Variation of surface indentation fracture toughness (Kc) with
doses. The error band is one standard deviation.

98

changes with the orientation of the planes of easiest cleavage with respect to the tensile
components of the indentation stress field. Burnett and Page [1984] have summarized a
stereogram of single crystal sapphire, which shows the crystallography of possible
median/radial crack systems. For instance, for a (1012) test surface, only one of the

{21 10} cleavage planes forms, i.e., (1210) is perpendicular to the surface and produces a
[1011] crack trace. Of the {1012} cleavage forms, both (0112) and (1102) are very
nearly perpendicular to the test surface producing the near-orthogonal < 2201 > traces.
Therefore, the tested surface should be kept perfectly flat in order that the crystallography
could be determined and measurement of radial cracks could be accurate and reproducible.
In our current indentation test, the diagonal of pyramid Vickers indenter is 4 um.
Compared this diagonal with corresponding curve with radius of 70 um, the error is
0.05%. The contact surface, thus, is assumed to be flat. In addition, extreme care was
taken to align one of pyramidal edge with respect to [0001] direction. The tested surface
should thus be one of prism planes.

It is worth noting that lateral fracture becomes particularly important in wear
processes where lateral crack break-out combined radial cracks can lead to substantial
material removal. In general, lateral break-out in the current work was a rare occurrence,
even though some lateral break-out takes place as shown in Figure 4.13, both for
unimplanted and implanted sapphire. This observation is different from the case of silicon
[Burnett and Page, 1984], in which ion implantation considerably reduced the lateral break-
out and surface-lifting, due to sub-surface lateral cracks. However, Burnett and Page
[1984] observed the extent and number of the lateral cracks underneath the surface is
reduced after implantation by using reflected polarized light micrography. Again, stresses

induced by implantation are presumed to be responsible for the lateral crack retardation.

 

(b) 4 x 1016Ar+ cm‘2

Figure 4.13 SEM observation on the trace oflareral crack propagation.

100

4.4 Nana-Indentation Hardness

The variation of nano-indentation hardness on sapphire fibers with implantation
doses of 2 x 1015, 2 x 1017 Mg+ cm'2 is shown in Figure 4.14. Compared with
unimplanted fiber, nano-indentation hardness reaches a maximum value at the dose of
2 x 1016 Mg+ cm’z. However, the hardness at dose of 2 x 1017 Mg* cm'2 is approximately
equal to that at unimplanted state. In the preceding section, it has been presented that the
compressive stress induced by ion implantation monotonically increases with ion dose, in
other words, the measured compressive stress at a dose of 2 x10‘17 Mg+ cm’2 is 2.8 GPa,
larger than the dose of 2 x1016 Mg” cm’2 (2 GPa). A question arises to why nano-
indentation hardness reaches a maximum value in an intermediate compressive stress in
stead of at maximum compressive stress? For a better answer to the question, the
characteristic bend strength corresponding to each implantation condition are also presented
in Figure 4.14 for reference, which are the same data as in Figure 4.3. To some degree,
the trend in hardness changes is consistent with the observed bend strength changes. As it
is known, hardness measurement is associated with plastic deformation behavior at the
ahead of indenter tip. Hardness is usually measured on the indenter’s contact area left due
to plastic flow after removal of load. Therefore, the extent of elastic-plastic and elastic
recovery at the head of indenter influences the hardness measurement. In other words,
irradiated surface properties are closely related to the hardness measurement. According to
the model described in 2.3.1 [Bumett, 1985], Burnett and Page predicted that
amorphization of surface should occur at doses over 1 x 1017 ion cm'2. Before the
occurrence of amorphization, the irradiated surface is still crystalline but damaged. The
production of defects such as vacancy clusters, dislocation loop and existence of
compressive stress leads to surface hardening, as discussed in section 2.4.3.5. The
production of an amorphous surface layer causes a surface compressive stress due to the
volume change that accompanies the transformation. The deformation of amorphous

phases occurs by viscous flow rather than by dislocation slip or cleavage fracture.

Nana-indentation Hardness (GPa)

101

 

80 _ 7 : 1o

 

, :
O Nana-indentation
Hardness

       

 

 

O)
C

 

 

I Bend Strength J

.b
O
CO
(9d9) mfiuens puee

 

N
O

 

 

 

0 5 10 15 20

Dose (1016 Mg+ cm'2)

Figure 4.14 Nona-indentation hardness of sapphire fibers with ion implantation
dose. Corresponding bend strength included for comparison.

102

However, TEM observations, discussed later in section 4.9.1, particularly in Figure 4.37,
on the near-surface microstructure of sapphire fiber implanted with the highest dose of 2 x
1017 Mg+ cm‘2 does not show amorphization within the implanted region. Due to TEM
resolution, it does not exclude possible partial amorphization within the implanted zone, as
observed by McHargue [1987]. Another possibility for the hardness behavior might arise
from the fact that the penetration depth of nano-indentation exceeds the implanted zone
thickness. The penetration depth in the case of 2 x 1017 Mg“ cm’2 is about 1100 nm while
the whole implanted zone is 350 nm (as shown in Figure 4.37 later). The measured
surface stress is an integrated surface stress over the implanted zone. It is unknown how
the stress changes with ion distribution because the ion distribution inside the implanted
zone is supposed to be Gaussian distribution as verified by many investigators by using
Rutherford Backscattering techniques [Ohkubo, 1986; Hioki, 1985]). Beyond the
implanted zone, it is not clear how the existence of compressive stress affects the
indentation displacement.

In Figure 4.14, bend strength change with dose shows a similar trend as hardness
does. The relationship between bend strength and compressive stress will be explored

further in section 5.2.

4.5 Annealing Efi‘ects

It is known that ion implantation is a non-equilibrium process, which introduces
massive foreign atoms into host substrate, thus causing lattice or crystalline damage. It has
been shown that Mg+ and Ar+ implantation into sapphire fibers in the current study has only
modest effects on the their bend strength. But the retained strength of implanted fibers after
abrasion is higher than that of unimplanted fibers. These characteristics are attributed to the
introduction of compressive stress by ion implantation. Questions arise regarding the
effects on the bend strength of sapphire fibers after implanted fibers undertake post-
implantation treatment, because the microstructure created by the non-equilibrium

103

implantation process can be expected to change after heat treatment. In this section, the
effects of annealing on the bend strength of sapphire fibers will be presented and
discussed.

The fibers implanted with Mg+ and Ar“, together with unimplanted fibers, were
annealed in a vacuum environment for 2 hours. The annealing led to the disappearance of
the violet coloration produced by implantation, indicating that color centers had undergone
recombination reactions. Table 4.4 summarizes the bend strength for annealing
temperatures ranging from 800°C to 1400°C. In general, bend strength decreases with
increase of the annealing temperature as shown in Figure 4.15. For control fibers, the
bend strength decreased 23 percent from 9.31 GPa for the as-received condition to 7.15
GPa for fibers annealed at 1400 °C. Over the same temperature range, the strength
decrease was about 21 percent for 5 x 1016 Mg’t/cm2 implanted fibers, and 28 percent
decrease for 2 x 1017 Mg+l cm2 implanted fibers. Annealing of Ar+ implanted fibers
showed similar behavior. Figures 4.16 and 4.17 show Weibull plots for unimplanted and
implanted fibers after annealing 2 hours at 1200°C vacuum environment, respectively.
These figures show that all Weibull curves for both Mg and Ar+ have shifted to the low
bound of 95 percent confidence interval of as-received bend strength, even for high dose
implantation case (2x10l7 Mg+ ). Furthermore, comparing Figure 4.1 with Figure 4.16,
or Figure 4.2 with Figure 4.17, one can see that the only change between them is that the
Weibull curves of annealed fibers shifts to the left, i.e., bend strength is reduced.

In comparison, Figure 4.18 shows the Weibull plot of bend strength for Mg“
implanted fibers annealed in air. The bend strength of as—received fiber decreases about 10
percent from 9.31 GPa at room temperature to 8.34 GPa at 1000 °C. The decrease is about
36 percent for 2 x10'7 Mg“ cm‘2 implanted fibers (bend strength decreases from 8.31 GPa
at room temperature to 5.23 GPa after annealed in air environment). Annealing in air
environment is more detrimental to the retained strength than a vacuum environment, and

that more prior irradiation causes a greater reduction in strength.

104

Table 4.4 Characteristic Bend Strength of Sapphire Fibers at Different Annealing

 

 

 

 

Temperature
(GPa)
Temperature 25°C 800°C roooec 12000c 14000C
0 9.31 8.61 8.41 7.01 7.15
2x1016Mg+ 8.87 8.41 8.41 8.10 8.11
4x1016Mg+ 9.47 8.30 7.82 7.37 6.25
5x1016Mg+ 9.26 8.66 8.19 7.01 7.36
7x1016Mg+ 8.92 8.35 7.22 6.08 6.93
2x1017Mg+ 8.21 7.46 6.60 6.19 5.92
1x1016Ar+ 9.15 6.94
2x1016Ar+ 8.34 5.63
4x1016Ar+ 9.24 6.19

 

 

 

105

 

 

 

 

 

 

 

1O
\ \ \ a
8 i ............................... in“ ........ .3, .............................
........ \ \ . a 4?-
A \A\
G , \
e 6 A
s
u
D
C
.8
m 4 .. ...........................................................................................
'D
C
Q .
m 2
3 lunimplanted
2 b ----------------------------------------------------------- g ‘ 16 + _2
g 0 5x10 Mg cm
A 2x10”Mg*cm'2
0 . 1'
25 500 1000 1500

Annealing Temperature (C)

Figure 4.15 Effects of annealing temperatures on the bend strength of
inplanted fibers. Two dashed line is the curve fit which indicates that
three point bend strength of sapphire fibers decreases with increase of
annealing temperature.

106

 

 

IIII

.......................................................................................

0.1

1
I
I
I
I
I
I
§
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

O
I
I
I
I
I
I
I
I
I
I
O
I
I
I
I
I
I
I
I
I
I
.'
I
I
I
I
I
I
I
I
I
I
I

I

 

. 1200C, Vacuum l
,.........-f ......... 1:1 Unimplanted .

"""""" I """ ""7

* """"""""" ‘ """" 1 o 2x1016Mg+cm2

 

0.01 1

Failure Probability

III'

' A 7x1016Mg+ 611i2

 

 

3 5 g o 211017 Mg+ cm2
0.001 II I 1 'llllli j 1 l I .lIi
1 10 100
Characteristic Bend Strength (GPa)

 

 

 

 

Figure 4.16 Weibull plot of bend strength for Mg+ implanted sapphire fibers
at annealing temperature of 1 200 (C for 2 hours. The two dashed lines
represent 95 percent confidence band of bend strength for unimplanted
sapphire fibers at room temperature.

107

 

 

 

 

 

 

 

 

 

 

t
p-
r-
l-
P
0.1 :
Q E
3 C
{I
.D h
2
m . ......................... , ........ r ................ .
2 . , . I . 1200 0C,Vacuum
: E = E
I=li 0.01 :...........; ................ 4 ......... D Unimplanted
“- : i I 3 I i 16
: ............ ' ........... g .......... 0 1x10 Ar"cm‘2 .
h A 2x1016A1'+cm'2
; 5 5 O 4x1016Af+ 0111'2
0.001 I I I IIIIII I J ITIIII
1 10 100

Characteristic Bend Strength (GPa)

Figure 4.17 Weibull plot of bend strength for Ar+ implanted sapphire fibers
at annealing temperature of 1200 °C for 2 hours. The two dashed lines
represent 95 percent confidence band of bend strength for unimplnated
sapphire fibers at room temperature.

108

 

 

 

 

 

 

 

 

 

1
:3
-'§
n 0.1 : ........................................................................................
2 . A on
m h
8 ' : . :
.2 H : : : .
7. """"""""" 3 """"""" 3 """"" Annealed 1000°C,A1r‘
u. - i E 3
I A an D Unimplanted
i E i -2
.. ............. g ........... o 2x1016Mg+cm .
. . . 1
. . . A 2x10 7 Mg+cmi2
0.01 .‘I I I I I I ILL I I I I I I I I
1 10 100

Characteristic Bend Strength (GPa)

Figure 4.18 Weibull plot of comparison of bend strength for Mg+ implanted
sapphire fibers annealing at 1000 0C for 2 hours in air environment.

109

It is expected that thermal diffusion or segregation occurs for the impurity atoms
induced by ion implantation during annealing. The impurity atoms which come to rest
within the implanted region are mobile once the thermal driving force is available.

Studies of Mg implanted sapphire after thermal annealing revealed that Mg may
segregate to the [0001] basal plane, or diffuse deeply into bulk material, or volatilize from
the surface, depending strongly upon the annealing environment [Baik, 1985, 1986;
Mukhopadhyay, 1988; Jardine, 1987]. Annealing in air led to surface enrichment of Mg
[Baik, 1985, 1986]. The surface concentration of magnesium was found to decrease
monotonically with increasing temperature above 13000C. Below 1300°C, the surface
concentration of magnesium was found to be very non-uniform [Baik, 1985]. In addition,
SIMS analysis indicates that the concentration of Mg decreases dramatically with depth for
air annealed sapphire which was implanted with Mg [Jardine, 1987]. Annealing in air
probably allows oxygen to diffuse into the implanted sapphire to compensate charge
imbalances created by the diffusing-out Mg ions. By using low-energy electron
diffraction, Baik, er al [1985] showed the presence of two-dimensional ordered overlayer
structures in which Mg is present in form of small “islands”, suggesting that Mg ions form
a substitutional solution with host Al in the A1203 surface. In the current study, no attempts
were made to relate the formation of the substitutional solution with surface topography
change. However, based on the fact that bend strength of 2 x 1017 Mg+ implanted sapphire
fibers after annealing in air environment decreased to a larger extent than control fibers, it is
postulated that the formation of the substitutional solution might alter the surface
topography, and further deteriorate the surface uniformity.

Hioki et al [1985] verified that a spinel compound NiAIQO4 was formed on the Ni
implanted sapphire surface after the sample was annealed in air at 1000 °C. This thin film
spinel compound affects the flexural strength because thermal properties of the thin film are
different from the underlying bulk material. In this case, the thermal expansion coefficient

110

of NiA1204 is 6% smaller that that of or-A1203, and results in a residual surface compressive
stress while the specimen is cooled down to room temperature.

Previous research [Mukhopadhyayz 1988] on Mg“ implanted sapphirel by Auger
electron spectroscopy has shown that no trace of Mg on the surface was found after the
implanted samples were annealed under vacuum environment. Instead, Ca was found to
segregate strongly to the surface although the bulk concentration of Ca was known to be
very low. The absence of a magnesium signal was attributed to a rapid loss of surface Mg
through a vaporization process which increases with decreasing partial pressure of 02. The

decomposition reaction of MgO is:

MgO (solid) = Mg (gas) + 1/2 02 (gas) (4.2)

The vaporization flux of magnesium will be proportional to the equilibrium vapor pressure
of the Mg-containing species.

In summary, Mg segregation to the surface and formation of precipitates after
annealing in air, as well as Ca segregation to the surface during annealing in vacuum, and
degradation of bend strength is not fully understood at this time. It is reasonable to
postulate, based on previous research [Mukhopadhyayz 1988; Baik: 1985, 1986] and the
current bend strength measurements, that any reaction of Mg or Ca with host A1203 results

in change on the surface uniformity which in turn affects fracture behaviors.

4.6 Sapphire Fiber Strength Degradation During Composite Fabrication
The following section describes fiber strength degradation after etching from

N iAl/A1203 composite, and discusses the efficiency of ion implantation on protection of

sapphire fiber during the composite fabrication process. As described in section 3.7,

 

1 Total impurity in the sa hire was 40 ppm. The samples were implanted with 200 keV Mg” ions to a
dose of 1 x 10” to 2 x 10 7 ions/cmz. The samples were annealed at vaccum (1045 to 10'7 Pa), temperature
(1300°C) for 2 hours.

111

NiAl/A1203 composites were fabricated, and then sapphire fibers were extracted from the
composite after hot-press consolidation, and, finally, bend strength was obtained by three
point bend tests on the etched fibers.

The extracted fibers from the MA] composite with as-received fibers had serious
strength degradation, which decreased from 9.31 GPa to 3.36 GPa, as shown in Figure
4.19. In comparison, the bend strength of extracted fibers from the MA] composite with
implanted fibers also showed degradation to some degree, exhibiting residual strength
slightly higher than as-received fibers, as shown in Figures 4.20. The average
characteristic bend strength for extracted fibers from the NiAlfrmplanted fiber composite
(both Mg+ and Ar*) is as follows: for Mg implanted fiber, the retained strength increased
from 3.33 to 4.51 GPa; for Ar implanted fiber, the retained strength from 4.13 to 5.35
GPa. However, this retained strength is substantially lower than the strength before
fabrication (8.56 GPa for 2 x 10"5 Mg+ cm‘z, 8.34 GPa for 2 x10"5 Ar" cm‘z), and lower
than the strength after annealing (8.1 GPa for 2 x 10‘6 Mg+ cm'z, 5.63 GPa for 2 x 10'6 Ar“
cm’z, after vacuum annealing at 1200 0C).

The percentage of fiber breakage occurring during composite fabrication is shown
in Figure 4.21. The majority of fibers etched from a 25 mm long NiAl/A1203 plate
fabricated by foil/fiber/foil technique were less than 4 mm in length, accounting for over 80
percent of the extracted fibers. Unimplanted and implanted fibers both have the same
general trend of fragmentation.

The complicated factors associated with the fabrication process can also be
demonstrated by comparison of bend strength among control, abraded and extracted fibers.
Figure 4.22 compares the bend strength variation for unimplanted and Mg implanted fibers
in conditions of as-received, abraded 60 minutes and extracted from NiAl/A1203 composite,
respectively. For unimplanted fibers, bend strength decreases in order from 9.31 GPa at
as-received condition, to 4.51 GPa after abrasion for 60 minutes, and to 3.36 GPa after

extraction from the composite. Similarly, for Mg implanted fibers, the strength decreases

112

 

0.1 ~ ------- a -------- ----------- --------- é ----------- a ---------

 

 

Failure Probability

D Unimplanted

i § § 2 E s o Unimplanted.
i 5 § E i etched from
R """" """"""" """"" """""" 5 """ NiAl/unimplanted
3 : ‘ : sapphire fibers
composite

001 I I llzlllli I I I :11 II: I I IIII
. .

 

 

 

 

 

 

0.1 1 10 100
Characteristic Bend Strength (GPa)

Figure 4.19 Comparison of bend strength Weibull plots for unimplanted sapphire
fibers, which include unimplanted sapphire fibers as reference and etched
unimplanted fibers from NiAl/unimplanted sapphire fiber composite.

 

 

113

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 .
:3
IE 5
a I
n I
g 0.1 _ 5.,
g =
a 3 Legend
"L § En El unimplanted sapphire
_ 0 A0 5 ' fibers as a reference
0 unimplanted sapphire
. ; : in fibers etched from MA]
4 Ar implantation ‘ composrte
. . .1.... 1 . .1.... Aimp'anwdsapphi‘efibe’s
0-01 (1 ) i- f - - - etched from MA] cgmposite
' . (D=1)810‘6 Ar+ cm , or
g 0? 2x10] Mg+ c1112 ).
g8 o O implanted sapphire fibers
. 0 § etched NiAl composite
5,9; 0 § (D=4xl 31A? cm‘2 , or
Ao D 51110“5 Mg+ cm'z).
g A0 go E]
'8 9 E i D
.D A : : D
e O 1 _ : : I:
n- C E O 5
8 _ 9 s U
2 ' E
E . U
i . .
Mg implantation :
0.01 AL glIJJILL 41 L1_i__L1 1 1_11_11
0.1 1 1O 1 00

Characteristic Bend Strength (GPa)

Figure 4.20 Comparison of bend strength Weibull plots for
implanted sapphire fibers etched from N iAl/sapphire fibers
composite. Bend strength for both unimplanted sapphire fibers
and unimplanted sapphire fibers etched from NiAl/sapphire fibers
composite are also included.

114

 

 

E No Irradiation

6Mg

1

7x10

16Ar

E 2x10

   

§

 

 

 

 

22E we 33580..

Fiber Length (mm)

Figure 4.21 Fragmentation of sapphire fibers etched from NiAl/A1203 composite.

115

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1
0.1
E
3
a
n
2
°' 1
t : : I :1
‘2 E ..... ‘ ......... Z ............... 3 ...... U ..............................
1' I
L n I 4 I
r» ..... 4' U . lanted .. .......... . Legend
0.001 (1) . f ° °°""°'
E AAg l '3 Abraded 60 mins
- A i A etched from MA!
' ' ' ‘A 3 l t '
. .......... 1;! ..........................
A 3: .
Q 0.1 —. o ,
E 1- A DI! I!
a . ............................................. 9.] ...........................
E I ."§ i s
g . ...................................... mi .......... .................
.3. . . I : :
'5 s s 2 . s - s
e 0-01 2 2 a f 2! '
' 5x101‘Mg+cm2
F- ----- OE! implant“ ---------------------------------------------
0.001 ; L lilju] i l l I'LIII '1 l Lillll

 

 

 

0.1 1 10 100
Characteristic Bend Strength (GPa)

Figure 4.22 Effects of different processes on the bend strength Weibull
plots for both unimplanted and Mg implanted sapphire fibers. Two
dashed lines represent 95 percent confidence band of bend strength for
unimplanted sapphire fibers at room temperature. The top figure is for
unimplanted sapphire fibers, the bottom is for implanted sapphire
fibers.

 

116

from 9.26 GPa at as-implanted condition, to 7.58 GPa after abrasion for 60 minutes, and
to 4.51 GPa after extraction from the composite. This clearly shows that the fabrication
process leads to severe strength degradation on sapphire fibers. This process involves not
only possible abrasion and point load effects, but also fiber-matrix reactions and other
elevated temperature effects.

Three possible mechanisms can contribute to the strength loss. The first
mechanism should be related to fabrication temperature. In the preceding section,
annealing at fabrication temperature led to fiber strength degradation for both as-received
and implanted cases. The reason is that impurity segregation to the surface during
annealing could alter the fiber surface uniformity. During the fabrication process, the
NiAl/sapphire fiber composite was held at constant temperature of 1523 K for 3 hours.
Therefore, it is expected that impurity segregation to surface would take place. The second
mechanism could be related to the role of hot pressing. During the fabrication operation,
matrix could exert point loads on the sapphire fibers. The extracted fiber fragmentation
strongly suggests that during fabrication process, fibers are subjected to stress, though it is
also possible that a circumferential compressive stress around the fiber arises from the
contraction due to mismatch of thermal expansion coefficient between fiber and matrix
during fabrication cooling process. However, Draper et al [1994] reported that the strength
of the fibers etched from a powder cloth composite was nearly the same as the strength of
fibers sputter-coated with matrix and exposed to elevated temperature. They concluded that
the contribution from the hot press or HIP process might not be a major factor in the
degradation of A1203 fiber strength, and that the effect of the pressure from hot pressing
may cause a degradation in fiber strength if all contributions from matrix reaction were
eliminated. The third possible mechanism for strength loss is related to the fiber-matrix
reaction. Sapphire is thermodynamically stable in NiAl at 1523 K. However, chemical
reaction on sapphire during fabrication of NiAl/sapphire composite may take place. Further

SEM observations on the surface topography will be presented in the next section.

1 17
4.7 Fiber Surface Morphology

A high magnification SEM morphology of a washed as-received sapphire fiber is
shown in Figure 4.23, which demonstrates a very smooth character. Few flaws or pores
were observed. This is in contrast to another investigation [Davis, 1995], in which flaws
were observed in the as-received fibers that were believed to be associated with porosity
developing during fiber growth. The SEM morphology of an as-received sapphire fiber
subjected to the extraction chemical solution, shown in Figure 4.24, exhibits the same
surface features as shown in Figure 4.23. The absence of pitting or surface morphology
alternations indicate that the fiber extraction process has no effect on the fiber.

Figure 4.25 is a low magnification SEM morphology of sapphire fiber etched from
NiAl/A1203 composite. In Figure 4.25(a) there are impurity particles adherent on the

surface, ranging in size from approximately a few to 15 microns. These adherent particles

are believed to come from the matrix. In contrast, the etched surface of 7 x 1016 Mg‘l'lcm2
implanted sapphire fiber shows different sized pore features, as shown in Figure 4.25(b).
These pores evidently result from chemical reactions between fiber and matrix during
composite fabrication process. These pores usually correlate with failure origins and
account for the strength loss. Figure 4.26 is a high magnification SEM observation on
etched Ar“ implanted fiber surface along with extracted unimplanted fiber surface. One still
can see many residues attached on the surface.

The residues in Figure 4.25 were analyzed by EDS as shown in Figure 27. The
EDS results on the residues on control fibers show that a small amount of Ni exists as a
compound component in which Al and O are major compound elements. On the 7 x 10"
Mg+ cm‘2 implanted fibers surface, Mg was also found. These results are similar to that
reported by Draper [1994]. It is likely that the reaction with the matrix resulted in pores or
adherent particles on the fiber surface. The strength of sapphire fiber is dependent upon

surface roughness. If the surface roughening due to the pores or adherent particles was

118

 

Figure 4.23 SEM observation on a clean as-received sapphire fiber surface.

119

 

Figure 4.24 SEM observation on the surface of sapphire fiber subjecting
to chemical etch solution, which is composed of 50 % P120, 33 % HNO, and
17 % HCl (by volume).

n:

 

16 -
(b) 7 x 10 Mg+ cm2 implanted fiber

Figure 4.25 SEM observation on the surface of extracted fibers from
NiAl/Alz O, composite.

 

(a) control fiber

 

6 -
(b) 1 x 101 + cm implanted fiber

Figure 4.26 High magnification SEM observation on the surface of
extracted fibers from NiAl/Al 203, composite.

122

 

 

 

 

 

 

 

 

 

 

 

150-
1
I
O
3 l I I f ‘l°__lrt r I I I I I
20
WNW
(a)
I
1 .
JAM
‘l j r 1 I I 1 l
10 15 20
WNW

(b)

Figure 4. 27 EDS composition analysis on the residues shown m Figure 4 ..25
(a) extracted unimplanted fibers; (b) extracted 7x10 Mg cm ’implanted fibers.

123

more severe than the pre-existing flaws, fiber strength was reduced.

Upon heat treatment, the fiber surface morphology was also found to be strongly
influenced by diffusion effects. The surface morphology of an etched sapphire fiber
processed for 3 hours at 1523 K and with an applied stress of 5 MPa is shown in Figure
4.28. There are thermally grooved grain boundary imprints on the surface and small pores
were sometimes associated with these ridges. These ridges may arise from matrix sintering
to the fiber, and are diffusion-related phenomena which is driven by the reduction of grain
boundary and surface energies. It is believed that any polycrystalline material will
eventually result in ridges on the fibers’ surface given enough time at a high enough
temperature. Since the ridges extend to outward from the surface, the strength loss due to
ridges may not be as severe as with the other types of flaws.

In principle, MA] is thermodynamically compatible with A1203. However, from
the SEM observations on the fibers extracted from MA], as shown in Figure 4.26, it is
evident that an interfacial chemical reaction took place between A1203 and NiAl. Since the
fabrication process was carried out without a binder, the reaction must be associated with
both the surface chemistry of the fiber, and the matrix composition. Table 5.1 lists the
chemical composition of MN for different fabrication processes [Bowman, 1994]. (Note:
we have the same source of materials from the NASA Lewis Research Center in

Cleveland, OH). One possible reaction mechanism is that: A1203 reacts with carbon

Table 4.5 Chemical Composition of NiAl

 

 

 

Ni Al C O N
Starting Powder 50.3 49.7 0.01 1 0.046 0.006
Powder Cloth 50.2 49.8 0.101 0.21 1 0.005

 

 

Binderless 50.1 49.8 0.023 0.073 0.005

 

124

 

Figure4.28 SEMobservationofridgeonthesurfaceofetchedsapphirefiber
fromN'rAl/Alzq.

125

because A1203 and carbon can react to form a volatile aluminum suboxide at temperatures
as low as 1100°C (Note: carbon contamination may also come from the vacuum furnace).
The chemical reaction of interest and the standard free energies, AGO (J/mol), are as

follows [Davis, 1995]:

A1203 (S) + 9/2 C (S) = 1/2 A14C3 (S) + 3C0 (g)
AGO =1260904.5 - 575.5 T

where T is the temperature.

Bowman [1994] found that fibers extracted from powder-cloth processed material
in which organic binder (poly(methyl methacrylate) PMMA) was used often had C-rich
areas on the fiber surface. These C-rich areas contain residue left from incomplete
volatilization of the binder materials; these deposits may act as fracture initiation sites and

thus become detrimental to fiber strength.

4.8 Fracture Surface Morphology

In Figure 4.5, it has been shown that the fracture surface of the three point bend test
on sapphire fiber demonstrates brittle fracture features, for both unimplanted and implanted
cases. The crack initiation origin is almost invisible since the crack size is very small (as
discussed in chapter V). Figure 4.29 further compares fracture surface features from low
strength to high strength. It apparently shows brittle fracture features for both cases, i.e.,
no distinct difference on the fracture surface was observed.

However, for the sapphire fibers extracted from an NiAl/fiber composite, fracture
surface features can be divided into two categories. The fast group is the fracture surface
of extracted fibers after subjecting to three point bend test, as shown in Figure 4.30, where
brittle fracture features are characteristic of both extracted unimplanted and implanted

fibers. The other is the fracture surface which result from the fiber fragmentation during

 

unimplanted implanted

Frgure 4.29 SEM observation on fracture surface variation with three point bend strength.
Low, medium and high bend strength are 7.2, 8.5 and 9.5 GPa, respectively.

127

 

(b)

Figure 4.30 SEM observation on fracture surface feature of sapphire fibers which were
mbjecwdmdneepointbardtestafwrdreywemexuacwdfiunNrAVAlzgcmnposiw
material. (a) unimplanted sapphire fibers; (b) 5x10“ Mg’ cm”2 implanted sapphire fibers.

128

the fabrication process of NiAl/fibers composite, which is the so called “Naturally
Fragmented” surface. Two subcategories about this kind of fracture surface can be
classified. The first kind of feature is shown in Figure 4.31. The features are still in the
nature of brittle fractures, similar to the feature in Figure 4.5, but shows more coarse
characteristics for both implanted and unimplanted cases. This is probably related to the
fabrication process, in which the fibers were subjected to 1523 K at stress of 5 MPa for 3
hours. During the process, creep or slow crack propagation may have taken place.

The second is the feature originating from internal pores, as shown in Figure 4.32.
This fracture surface can be divided into four zones, i.e., ‘region 1’ is crack initiation zone
where the internal pore acts as a stress concentrator; ‘region 2’ is crack propagation zone,
where crack propagated in the radial direction and left behind a flat propagation path;
‘region 3’ is transition zone from crack propagation to brittleness fracture; Finally, ‘region
4’ is brittle fracture. The size of internal pores ranges from a few tenth micrometers to 1
micrometer. The internal pores are attributed to shrinkage voids inherent from the
Saphikon fiber growth process (EDFG), in which sapphire fiber is drawn from the melt,
and voids form by entrapment of liquid behind the solid surface. The void density in

Saphikon fiber ranges from 33 to 105 pores/mm3 while size ranges from less than 0.1
microns to 2 microns.

Two possible fracture mechanisms might initiate the fracture behavior for the
naturally fragmented fibers. In Figure 4.31, the surface flaw which is externally stressed is
the control step for fiber fragmentation, similar to the fracture behavior in three point
bending. Surface flaws may arise from original surface imperfection, or from abrasion
during consolidation assembly, or surface faceting due to creep at consolidation
temperature (because creep may be activated at this temperature). Once the crack is loaded,
and the fracture criteria is reached, fracture occurs quickly. Brittle fracture dominates the
process. This is why no crack initiation site was able to be detected from SEM

observation.

129

 

(b)

Figure 4. 31 SEM obsuvatiorr on fracture surface feature ofsapphire fibers which
durin consolidation process of N"rAl/Al,0, composite. (a) unimplanted sapphire
fibus; (b) 5x10 ‘ Mg cm" implanted sapphirefibers

130

 

Figure 4. 32 SEM obsuvation on internal pores in sapphire fibus which fractrned during
consolidation process of NiAl/Al,03 composite. (a) pore occurred in unimplanted sapphrre
fibers;(b)highmagnificationimageof(a); (c)poreoccurred.inMgimplantedsap phire
fibus; (d) high magnification rmage of (c). Region 1: pore site (crack uritiation site); region
2. crack propagation; region 3: fracture mode transition; region 4: brittle fracture.

131

In Figure 4.32, failure resulting from creep of internal pores is the controlling
mechanism. It is apparent that the crack due to internal pore has to propagate a certain
distance until the stress state of crack reaches the fast fracture condition. Figure 4.32
shows that the creep deformation is about a few micrometers. Beyond the creep zone,

brittle fracture features are distinct.

The control mechanism is dependent upon the competition between surface flaws
and internal pore creep. Figure 4.33 shows that an internal pore initiates slow crack
growth, and propagates to the surface edge. Failure resulting from internal pore creep

dominates the process.

4.9 TEM Investigation of M icrostructure
In this section, TEM observations on irradiated near-surface microstructure and

relevant channeling effects during ion irradiation will be reported and discussed.

4.9.1 T EM Observation and ECP Results

The microstructure in the implanted region depends on several factors, including
ion energy, ion dose, ion species and combinations of ion and target as well as implantation
temperature. For ions and targets of relatively high atomic number and for relatively low
ion beam energy, the energy lost by the ion is concentrated in a thermal spike which can
lead to amorphization along a single ion cascade by the melting of a small region followed
by rapid cooling. For relatively light atoms and high energies, amorphization follows an
accumulation of independent single atomic displacements rather than a collective, thermal
process. As proposed by Burnett and Page [1984], amorphization occurs after an ion dose
reaches a certain threshold for a given ion implantation temperature, energies and
combinations of ion and target. This is because at low temperature, recovery is suppressed
and defects accumulate as the dose is increased. If recovery is sufficiently inhibited, a

concentration of defects may be reached where the long-range order of the crystal lattice is

132

4.... ‘

2"")? ’5
- S»,
r ‘ ‘
.9; "I”!

r b- . _
49“}; ' .

b

’.-\I

 

Figure 4.33 SEM obsuvatiorr on internal pores in sapphire fibers which occurred near
edge. The fibus incurred during consolidation process of NiAl/Al,O3 cemposite.

133

destroyed and an amorphous state is produced. Before the amorphization dose, the
implanted region is still crystalline but damaged.

Figure 4.34 shows the cross section TEM results for a sapphire fiber implanted
with 175 keV Mg+ of 7 x 1016 ion/cm2 at room temperature. Because of limitations on the
resolution of the TEM, the diffraction pattern actually overlays the implanted region and the
substrate. From the present micrography it can be seen that the surface region is still
crystalline but there are many small networks or dark contrast features which are suspected
of being small dislocation loops [McHargue, 1991] in the implanted region or in the buried
layer of incident ion species.

There are four different regions in the implanted zone, i.e., free surface; near-
surface region which is about 40 nm beneath the free surface, and is free of visible loops or
networks but contains many small (~2 nm) features which exhibit contrast consistent with
that of voids; the buried layer which is 200 nm to 400 nm below the near-surface region;
and finally, the un-damaged fiber interior.

The characterization can be seen more clearly in Figure 4.35, which shows the
TEM cross section observation on the sapphire fiber implanted with 2 x 1017 Mg'1’lcm2 at
room temperature. The TEM photo was taken using a JOEL 2000FX in the Michigan Ion
Beam Laboratory in Ann Arbor. The insert diffraction pattern of the buried layer clearly
shows that this region at dose of 2 x 1017 Mg'l'lcm2 is still crystalline but a more heavy
density of dark contrast features are present. Figure 4.36 is a high magnification TEM
observation of this buried layer. In this case, the width of near-surface is about 50 nm and
the buried layer is about 350 nm.

Although Ar has very different chemical properties than Mg, comparison of the
behavior of Ar and Mg implanted fiber shows little difference between two cases, from
bend strength measurement to abrasion resistance, to TEM microstructure observation.
Therefore it is suspected that the ballistic collision dominates the effects of implantation

rather than the chemical effect for these two particular cases. Figure 4.37 is a TEM

134

observation on sapphire fiber implanted with l x 1016 Art/cm2 at room temperature.
Similar features as Figure 4.34 and 4.35 were observed. Figure 4.38 is a high
magnification TEM observation on the implanted region. There is a high density of dark
contrast features within the buried layer and no bubbles or macro-scale voids were visible.
There is no evidence for the formation of amorphous zones in the buried layer region. In
Figure 4.37, the near-surface is about 40 nm wide, and the buried layer is about 180 nm.
Microstructural change in the subsurface induced by ion implantation is a
complicated phenomenon and depends upon many factors. McHargue [1989, 1991],
Burnett and Page [1984] and Hioki [1986] have observed microstructure in sapphire
implanted with different ions at a given implantation condition such as ion species, ion
energy, substrate temperature, ion dose, and temperature. For instance, McHargue [1991]
reported that the TEM microstructure of (x-Ale3 implanted with 4 x 10"5 Zr” cm'2 (280 keV
at room temperature), combined with RBS spectra, has following features: the region from
the free surface to 40 nm is still crystalline but damaged; region between 40 to 100 nm is
amorphous; the region below 100 nm is also crystalline, but there is a high density of
defects. However, TEM investigation by Specht [1994] on sapphire substrate implanted
with 160 keV at 4 x 1016 Cr+ cm2 and 1 x 1017 Cr" cm2 at room temperature showed no
evidence for the amorphous formation in the implanted region. The TEM features are
similar to the current observations, i.e., near-surface is about 60 nm and the buried layer is
about 170 nm. A mass of dark contrast features (which the author considered as an array
of dislocation loops and networks) is present within the buried layer. Based on TEM
observation, Specht suggested that sapphire samples implanted at room temperature exhibit
a high amorphization threshold because a local equilibrium is established where vacancies
and interstitial recombine at the same rate that are produced. For amorphization at such a
high dose (2x10l7 Cr’ cm'2 ) in single crystal A1203, Specht proposed that accumulation of

defects may serve as a mechanism of amorphization. These defects come from high-

135

IonDistributionbyTRIM

 

 

 

Free Surface

Figure 4. 34 TEM observation on 7x1016 Mg+ cm'2 implanted sapphire fibus (insert' rs the
diffraction pattern of the implanted zone). Near-srn'face region is defines as the region between
free sin-face and the left side tail of implanted rons distribution; Buried layer rs defined as the

ion distribution region.

136

 

350

Figure 4.35 TEM observation on 2 x 10" Mg‘ cm'2 implanted sapphire fibers at room
tempuature. Thediffractionpatremistakenfromtheinsideofimplanwdmne.

137

 

Figure 4.36 High nmgnification TEM observation on 2 x 10" Mg‘ cm’2 implanted zone.

138

 

I
13°

Figure 4.37 TEM obsuvation on 1 x 10" Ar‘ cm‘2 implanted sapphire fibers at room
tempuature. The diffraction pattern is taken from the inside ofimplanted zone.

139

 

Figure 4.38 High rmgrrification TEM observation on 1 x 10" Ar‘ cm'2 implanted zone.

140

energy-transfer collisions which knock Al and O deeper into the crystal and leave a
shallow vacancy and a deep interstitial.

Mechanical properties are closely related to subsurface microstructure. In the case
of ion implantation, both the injection of the implanted ions and the creation of point defects
cause a volume increase in the implanted region. Since the material is free to expand only
in one direction (normal to the free surface) the constraints of the substrate to hold the
lateral dimension constant produce a biaxial compressive stress in the implanted region. If
the implantation produces residual compressive stress large enough, it will affect the
mechanical properties by increasing the applied stress necessary to place the stressed
surface into tension, thus reducing the probability of propagating a pre-existing flaw, or by
affecting the crack tip stress fields.

Though the biaxial residual compressive stress is produced on the ion end of the
range region, the stress distribution in the midrange is totally different. Specht [1994]
found that the density1 of A1203 in midrange decreases by 4% after Cr+ implantation while
the volume expansion in the nridrange was only ~0.2%. He attributed the density reduction
of A1203 to high energy transfer collisions that knock Al and O atoms deep into the crystal
and give rise to excess vacancies and deep interstitials. Therefore, it is expected that the
stress state in the rrridrange about 40 nm from free surface is tensile. It seems controversial
because compressive stress has been measured on the implanted surface by indentation
technique. Keep in mind that the penetration depth during micro indentation is greater than
the thickness of the rrridrange (40 run), while in the nano-indentation measurement the
average penetration depth is about 900 nm. This issue will be treated in further discussion.

Figure 4.39 shows an electronic channeling pattern comparison of unimplanted and
implanted fibers, which illustrates the surface microstructure is still in crystalline. This
observation was consistent with TEM observations, which confirms that there is no large

scale amorphization formation on the implanted sapphire fiber surface.

 

1 The density profile was measured by x-ray reflectivity. Detail can be referred to R. A. Cowley and T. W.
Ryan, J. Phys., D 20 (1987) 61.

141

n j 1; n 11 f, [I 11 p r ——

00000 300mm ——

 

(c) Dose . 4x10“ Mg‘ cm" ((1) Dose = 2x10" Mg‘ cm"

Frgure 4.39 Electron channeling pattern for control and implanted sapphire fibus. The
fiber axes is on the image plane.

142

4.9.2 Channeling Effects

It is interesting to note that the measured implantation depth (or range) is about 0.4
pm ( in case of 7 1th16 and 2 x 1017 Mg’ cm'z, 175 keV), which is about 33 % greater
than the calculated depth (about 0.3 um) by TRIM. For the case of Ar” implantation, the
measured depth is about 0.22 um while calculated depth is 0.14 um. The discrepancy
between the observed and calculated damage range may be attributed to a channeling effect
(see section 2.4). Channeling is a process whereby atoms move over a distance in the solid
along an open direction in the crystal structure [Carter: 1968]. In the current experiment,
ion implantation was carried out by implanting ions perpendicular to the c-axis, or, in other
words, the ion beam trajectory was parallel to the basal plane. In the hexagonal crystalline
structures of A1203, the c/a ratio is 2.73. The most open structure is in a direction
perpendicular to the c-axis, or parallel to the basal plane. Ions moving along this
crystallographic direction favorable to channeling can lose energy mainly by glancing
collisions with atoms ringing the axis of motion. Therefore, channeled ions are able to
move much larger distances than original knock-ons before conring to rest.

The reason that the channeling travel distance will be large is that since the distance
of approach of the primary knock-ons and the lattice atoms are always large (= 1/x/2
atomic spacing) only small energy transfer will occur, compared to a random sequence of
collisions where, in a head on collision, all primary energy may be dissipated. On the other
hand, channeled ions can materially reduce the anticipated defect production rate,
particularly at high ion energies. These production of defects is dependent upon the
interatonric potential. TRIM (version 90.05) doesn’t take into account the probability of
channeling. Therefore, the calculation results from TRIM can only represent a lower

bound for penetration depth.

CHAPTER V

DISCUSSION

The results in the preceding section have shown that Mg’ and Ar” implantation have
little effect on the bend strength of unabraded sapphire fibers, and that there is an apparent
compressive stress zone with a magnitude of 2 GPa in the implanted zone. Further TEM
observations revealed that the implanted zone consists of different regions. The most
striking results, the observed changes in bend strength retention during abrasion process,
have been postulated to be associated with the implantation induced compressive stress.
Based on these experimental results, this section will further explore this hypothesis, but
first, the general nature of the accumulated lattice damage structures will be discussed.

5.1 Chemical versus Ballistic Eflects

It may be recalled in section 2.1 that sapphire belongs to the space group R3C. The
02' anions are hexagonal closed-packed, and the Al” cations occupy two-thirds of the
available octahedral sites. The displacement energy was found to be 18 eV for aluminum
and 72 eV for oxygen. It is possible to substitute isovalent or aliovalent impurities for Al”,
but such impurities cannot be added without creating charge-compensating defects. The
vacant octahedral sites are structural vacancies, and are ordered to balance the electrostatic
force in the crystal [McHargue: 1991].

Since Mg+ and Ar+ have different chemical properties, the defects produced by each
species may be supposed to be different. In the case of Mg, for example, a spinel phase,
MgAl,O4 , may form when the MgO content exceeds the solubility. The spinel phase is
normally found in form of precipitates. Figure 5.1 shows the pseudo-binary phase
diagram of MgAle4 - A1203 system. As seen from the phase diagram, Mg solubility in

A1203 is very limited. For instance, while the solubility reaches a maximum of 1% at the

143

144

 

  

 

 

 

 

 

2150
2100
8 I
.E 2050 b
2 l'
a i-
9 L
3 2000
E D
Q l-
l- r
L spinelss
1950 l-
. r
" r
' I
1% I I I L l I j I l I I I 11 I L L L 1
MgAl204 60 70 80 90 Al203

Mol % A1203

Figure 5.1 The system MgAle4 -Ale3. (After Dorre and Hubner: Alumina,
Springer— Verlag Berlin, Heidelberg 1984.

145

eutectic at 1975 °C, the solubility is 1400 ppm at 1830 °C, and only 300 ppm at 1630 °C
[Dorrez 1984]. At room temperature, the solubility of Mg is thus effectively negligible.
Mg+ can be incorporated into sapphire as MgO in several different ways [Lagerlofz

1983] as follows:
2Mg0 4: 2MgM' + 20; + v0“ (5.1)
3Mg0 a 2MgM' + 300K + Mgi” (5.2)
3Mg0 4» NM" a» 3MgM' + 300" + Alf“ (5.3)
The Schottky reaction is
null <=> 2VAL'" + 3V0” (5.4)

The anion Frenkel reaction is
00" a Oi" + V0” (5.5)
and the cation Frenkel reaction is

A1,," a» Alf” + VM'" (5.6)

Upon the implantation of Mg+ into sapphire, a large quantity of defects are
produced due to energetic collisions between ions and the host crystal (non—equilibrium
defects). For instance, the color of the fibers was observed to change from the original
transparent to violet coloration in case of Mg”, and black in case of Ar+ implantation. These
colorations come from oxygen vacancies containing trapped electrons (i.e., the F center (an
oxygen vacancy containing one electron) and the F center (an oxygen vacancy containing
two electrons) [McHargue, 1991].

These defects exist along with the formation of displacement spikes created by
collisions between energetic ions and host crystal atoms. The displacement spike consists
of multiple vacancies surrounded by interstitial atoms. It is the displacement spike that

creates a damaged region where a structure differs from the original lattice. When the

146

displacement spike grows or collapses, defect clusters, or dislocation loops, or dislocation
tangles and networks are created. McHargue [1991] reported that for (X-A1203 implanted
with 2 x 10“5 Cr/cm2 (280 keV) at room temperature contained a high density of dark
contrast features identified as defect clusters, and further confirmed that these defect
clusters were stoichiometric, interstitial dislocation loops. In the current TEM observation,
similar dark contrast features were found within the implanted region for both Mg+ and Ar+
implantation. Further confirmation on the nature of these features is needed in future
studies.

Ar+ implanted into sapphire can exist in form of interstitials, defect clusters, or in
form of gas within voids created by irradiation damage, depending upon implantation
conditions. Gas in the latter form can develop into a bubble or blister. Hioki [1985]
reported that blister formation in A1203 occurs only at high doses. For instance, the
threshold dose of blister formation for 400 keV nitrogen implanted in A1203 at 300 K was
about 2 x 1017 N cm’z. McHargue [1987] also reported that 30% of the sapphire (0001)
surface was covered by blisters after implantation of 1 x 1017 Ar“ cm’2 (230 keV) at room
temperature; in contrast, only 1% coverage consisting of smaller bubbles for a dose of
1 x 10"5 Ar“ cm‘z. Obviously, blistering is associated with the processes of inert gas
evolution and accumulation, instead of chemical reactions with host crystal.

Hioki er al [1985] also compared the mass effects of ion implantation on the flexural
strength of planar sapphire. They found that heavier ions were more effective at low doses
in strengthening. Their results showed that at the dose of 5 x 10" ions cm‘z, 800 keV Ar+
implantation at room temperature can increase the flexural strength by 60%, but only 20%
for Ni They concluded that radiation damage due to ballistic collision entirely accounts for
the observed strengthening effects.

In the current work, it was observed that after annealing, colorations due to ion
implantation disappeared for both Mg‘ and Ar+ implantation. This is because diffusion or

aggregation of implanted ions occurs during annealing, thus causing annihilation of color

147

centers. Based on measurements of bend strength from Mg+ and Ar+ implanted sapphire
fibers, and effects of abrasion and annealing on the retained bend strength, it was found
that there was little difference between the effects of ion species on fiber properties, though
Mg“ and Ar” have very different chemical properties. Thus, ballistic interactions apparently
dominate the effects of the implantation process, primarily, as will be shown, through

creation of a special residual stress distribution within the implantation zone.

5.2 Residual Surface Stress and Bend Strength

Comparing the results of the bend strength for both unimplanted and implanted
sapphire fiber, as shown in Figure 4.1 and Figure 4.2, ion implantation has little effect on
bend strength behavior, even with the presumed existence of ~2 GPa surface compressive
stress due to ion implantation. According to simple superposition principle, if there is a
certain magnitude of compressive stress which exists in front of a pre-existing surface
flaw, an increase in magnitude of external applied stress should be required to initiate
tensile fracture the sample during a three-point bend test. The explanation of the question
as to why the existence of compressive stress does not affect unabraded fiber bend strength
can be rationalized by considering the nature of pre-existing flaws and the interaction of
surface compressive stress with external stress states. The external stress states in three-
point bend loading are different from the local stress states arising during the abrasion

process.

Three point bend strength of unimplanted fibers

As shown in Figure 5.2(a), assume a single circular V-notch surface flaw with a
length c existing as an initial flaw on a sapphire fiber. The diameter of the fiber is 2b.
After a bending moment, M , is applied to the fiber, the outermost layer of the fiber endures

either maximum tensile stress or maximum compressive stress. Here, only tensile state is

148

M Impl ted Zone

 

A\\\\\\\_,\\\'

.'.". s u A

....
.....

’ :1: i ". ,"
C .5? : ~.
1 :17 913 Z see enlarged view

 

 

 

 

 

 

\\\\\<\_‘\\\\\l\2~ ~\\\\i\‘\\\\\

 

 

e- <—
M

M
(a) (b)

Undamaged terior Sub-Near- urface Veryxeapsurface

 

 

 

Figure 5.2 (a) Circular V-notch crack on unimplanted fiber surface; (b) Circular V-notch
crack on the implanted fiber surface; (c) Theoretical calculation of bend strength changes
with crack length.

3-pts Bend Strength (GPa)

14

12

10

 

149

 

I I I I I I I I I I I I I I I I I I I L I

In Ln In In
C. P. N “2
O O O 0

Crack Length (um)
(C)

(Figure 5.2 continue)

 

0.45

150

considered. The stress intensity factor, K I , in the front of a crack of a length c , can be

written as ['l‘ada, 1973]:

K, = $75170», c ) (5.7)

where P is the applied load, L is the fiber length between load points, and F ( b, c) is a

geometrical factor, which can be described as:

 

 

 

 

 

 

 

 

 

5
( H -c THY 5(b—c)3 35( -c)4 (-c)5}
F(b,c)=- 1+ +— +— +— 0537
-c 2 b 8 b 16 b 128
(5.8)
i.e.,
3 2 3 4 5
(b H -c TH) 5(1)-..) 35( ) (—c)}
K1=—oo\/;c_ l+— +— +— +— — 0537
b—c b 8 b 16 b 128
(5.9)

where do is the bend strength. Note that in the present case, since c << b, equation (5.9)

can be reduced to

K, 21.124200x/E (5.10)

When K, reaches the critical stress intensity factor ch (K10 =25 MPant for

single crystal A1203 [Tressler, 1990]), the fiber will fail instantaneously. Thus taking the
bend strength 0'0: 9.3 GPa, the corresponding critical crack length by equation (5.10) is

18 nm. it should be pointed out that 18 nm crack length is difficult to detect with our
available equipment. The relation between bend strength 0'0 and critical crack length c can

be plotted as shown in Figure 5.2(c).

151

Ochiai and Osamura [1988] calculated the average crack length in terms of an
energy release rate. According to them, the flaw size Cm can be calculated by:

a, = (1/1.12)[E,Gc‘/(nCm)]1/2 (5.11)

or
E ,0;
m = —————2- (5.12)
1.25no,
where of is strength, E f is Young’s modulus of the fiber, Go. is critical elastic strain
energy release rate of the fiber, given by 06": KCZ/Ef (Kc is the critical stress intensity

factor for mode I). Taking of =9.3 GPa for unabraded fiber, E f=414 GPa, Kc =2.5

MPax/E , the corresponding flaw size is again 18 nm. The calculation results are
consistent for both approaches.

This analysis basically states that the maximum crack length in sapphire fiber is 18
nm to be consistent with the observed 9.3 GPa bend strength in three point bend loading.

Three point bend strength of implanted fibers

Before discussing the bend strength of implanted fibers, it is necessary to consider
sputtering effects, since the relevant surface crack length is only 18 nm long. To assess the
effects of sputtering on the surface removal of sapphire fiber, TRIM codes were used to
theoretically calculate the sputter yield for the given ion implantation conditions.
Displacement energies for A1203 were used (18 eV for Al and 72 eV for 0). During the
simulation, a total of 10,000 ions were counted. The calculated sputter yield was 0.1429
atoms/ion for Al and 0.2333 atoms/ion for 0. Therefore, about 7 nm of the surface was
removed during the implantation process, for an implantation dose of 5 x 1016 ion/cm2
(assuming implantation was canied out on a planar substrate, and the density of A1203 is

3.97 g/cm3).

152

Since the measured bend strength of implanted fibers in the current work showed
slight change compared with unimplanted fibers, and since compressive stress of 2 GPa
was measured in implanted fibers, two questions arise: First, considering the 7 nm surface
removal by sputtering, the length of pre-existing flaws may have been reduced to as little as
11 nm, consequently, by equation (5.10), the bend strength should be increased to 11.96
GPa. However, the highest bend strength was found to be only 9.5 GPa at the dose of 4
x10" Mg“ cm‘z. At some doses, bend strength actually showed slight decreasel. Thus, the
first question is: why is there no increase in bend strength even if surface crack length
becomes shorter? Secondly, if the presumed 2 GPa compressive stress is superimposed on
the bending stress, the bend strength of implanted fibers should increase to a value of 11.3
GPa even in the absence of sputter losses. Why does the compressive stress not contribute
to bend strength? Both of the questions are more easily approached by consideration of
details of the sub-surface stress distribution in irradiated sapphire fibers.

The penetration depth in compressive stress measurement by indention technique is
about 2 pm, which actually penetrate through the implanted zone. The compressive stress
is treated as if it were distributed over an equivalent depth. Such consideration does not
represent the actual induced stress distribution. In Chapter 2, it has been stated that during
the implantation process, high energy transfer collisions knock host atoms deep into the
host, leaving a shallow vacancy region. Most implanted ions come to rest in projectile
range region which is deeper inside the host. The host volume expands in the project range
since the introduction of mass of foreign atoms, but it is constrained by the underlying
substrate. Therefore, the stress state in the shallow vacancy region is different from the
projectile range, and may in fact be tensile [Specht, 1994].

In Figure 4.34, TEM observations on implanted fibers revealed that there are four
different regions in implanted zone, i.e., the free surface, the very-near-surface zone

(VNS) between free surface and the buried layer or ‘sub-near-surface’ zone, the sub-near-

 

1 For example, the bend strength was 8.87 GPa at dose of 2 x 10" Mg+ cm'z, 8.91 GPa at dose of 7 x 10“
Mg” cm”, and 8.21 GPa at dose of 2 x 10” Mg” cm'z.

153

surface zone (SNS), and the unaffected interior of the fiber. The depth of VNS is about
40 nm deep (after a 7 nm sputter loss), and the SNS is about 200 to 400 nm deep for the
implantation conditions examined. A schematic diagram in Figure 5 .3 was constructed to
demonstrate the relationships between relevant regions and surface stress distribution.
Results from TRIM calculations are also included in the figure for implantation projectile
range reference.

The compressive stress in sub-near-surface zone (SNS) should be balanced by
VNS zone and fiber interior. Therefore, VNS zone and fiber interior are both in a tensile
state. It is reasonable to assume the resulting tensile stress is unifomrly distributed in these
two regions. Since sapphire is an ionic crystalline it is difficult to cause lattice distortion on
the neighbor atoms, and since the indentation penetration is 2 pm, it is reasonable to
assume that the compressive stress is balanced over the indentation penetration region. The
magnitude of tensile stress in very-near—surface region at each implantation dose and
consequently the measured bend strength, measured compressive stress, sputtering yield,

theoretical bend strength calculation are listed together in Table 5.1.

 

 

 

Table 5.1 Comparison of Magnitude of Tensile Stress in Very-Near-Surface (VNS),
Measured Bend Strength, Measured Compressive Stress at Each
Implantation Dose.
Dose Sputtered Estimated Predicted Estimated Measured Estimated Estimated
(ions/cm’) Thickness Crack Bend Compressiv Bend Tensile Bend
(nm) Length Strength e Stress Strength Stress Strength
(nm) (GPa) (GPa) (GPa) (GPa) (GPa)
mote 1) (Note 2) ( Note 3) (Note 4)
0 0 18 9.31 0 9.31 0 9.31
1x10" Ar‘ 1.4 16.6 9.7 1.8 9.15 0.036 9.66
2x10" Mg’ 2.8 15.2 10.17 2 8.87 0.04 10.13
5x10" Mg’ 7.0 11 11.96 2.2 9.26 0.044 11.91
2x10" Mg‘ 28 0 23 (Note 5) 2.6 8.21 0.052 22.95

 

 

 

 

 

 

 

 

 

Microstructure Regions

Surface Crack
Before Implantation

Surface Crack
After Implantation
(Dashed-line
represents the
original surface;
Shadow for stress
state) ‘
TRIM 1
Calculation :

l

1
Surface Stress I.
Distribution 1 g i '

154

   

‘ Fiber C—Axis f.

m; Ill/"A4 ‘

 

 

 

 

Substrate

Figure 5.3 Schematic of implanted zone, surface crack geometry before and after
ion implantation, and surface stress distribution. There four region in an implanted
zone: free surface, near-surface (40 nm), buried layer (220 nm, depending on ion

implantation conditions), and substrate.

155

Note: 1. Sputtered thickness is based on TRIM sputter yield calculation, i.e., sputtered
thickness is proportional to ion dose.
2. Predicted bend strength is calculated by equation (5.10).
3. Estimated tensile stress is calculated based on the thickness of VNS and un-
damage fiber interior.
4. Estimated bend strength is the difference between the predicted bend strength and
estimated tensile stress.

5. Theoretical strength (G / 2n), G = 144 GPa.

Comparison from this table, it can be seen that ion implantation has two effects on
the bend strength, i.e., tensile stress in near-surface causes strength to decrease, and the
compressive stress enhances the bend strength. The effects of compressive stress on the
bend strength are discussed as follows.

Assume a same circular V-notch surface flaw pre-existing on the sapphire fiber.
The compressive stress region induced by ion implantation is depicted as shadow area
underneath the fiber surface, as shown in Figure 5.2(b).

Consider a crack retarded by the compressive stress region. The additional stress
intensity factor Ka generated by the compressive stress in the front of a crack can be

calculated by following relationships according to Lawn and Fuller [1984]:

K“ = 2W03d% (5.13)

where d is the thickness of the implanted region, 1;! is a crack geometry term (4/1t2), and o;
is the surface stress. Taking d =0.4 um (taken from the TEM observation), the additional

stress intensity factor from the compressive region is:

K“ = - 1.03 Minn/Z (5.14)

Therefore, the new fracture criteria is:
chl = KIC — K

= 2.5 + 1.03

156

= 3.53 MPax/m (5.15)

In other words, if the stress intensity factor in front of a crack, K, , is greater than K ,C’ ,

fracture occurs; otherwise, the crack is arrested in the implanted zone.

Now, to resume the argument on the bend strength of implanted fibers. It was
shown that the maximum crack length is 18 nm, which has been reduced to 11 nm
considering sputtering effects for a dose of 5 x 10'“, to maintain 9.3 GPa bend strength.
During three point bend test, the pre-existing crack with a presumed length of 11 nm
propagates from initial position under constant loading as an external bending moment is
applied to the fiber. The tensile stress in the VNS -Zone (0.044 GPa) may cause the onset
of crack propagation at a low stress. But when the crack reaches the compressive SNS-
Zone at a depth of 40 nm, the stress intensity factor K I may be greater than K ,C’ , the crack
will not be arrested. This is because the crack is now 40 nm long. Specifically, at this

point, the stress intensity factor can be calculated using equation (5.10) by taking c = 40

nm, and 0'0 = 9.3 GPa, one can obtain:

K]: 3.71 MPax/m (5.16)

Comparing K I and K ,C’ , it turns out that:

K, >K,C’, (5.17)

Thus, the propagating crack continuously propagates through the implanted region, despite
the existence of the compressive stress zone.

Based on the above analysis, it is important to realize that in the current three point
bend test of 140 um fiber, the pre-existing crack length is very small (about 11 nm) to

157

maintain the measured bend strength (9.3 GPa). With increase in pre-existing crack length,
the bend strength decreases rapidly. Thus, even though there is a compressive stress
region which is beyond the scale of the pre-existing crack, once the crack fails
spontaneously upon application of external load, it becomes unstable. The existence of the
compressive stress region couldn’t retard the crack propagation, even though the magnitude
of the compressive stress is very large. Unfortunately, the compression stress induced by
ion implantation in our work is only about 2 GPa. This, therefore, is why there was no
significant change in the bend strength of implanted fibers when compared with that of the
unimplanted ones.

Now, to resume discussions on the sputtering effects on fiber surface morphology.
Kawalski [1982, 1986, 1987, 1990] has developed a theoretical model to predict the
surface morphology during ion bombardment. In the model, changes in the real surface
profile caused by ion erosion depend on the angle of ion-beam incidence and the angle
between beam direction and normal of the surface. Experimental results on 99.5% A1203
show that with a large ion incidence angle the A1203 surface becomes smooth or decreases
in roughness, while the mean roughness of an ion-sputtered surface at a right angle is much
greater than the unsputtered one. In our implantation process, the sample rotates around
the central axis of a fixture. The angle of beam incidence with the normal of the surface
changes with time. Thus there is not a clear relationship between ion sputtering and surface
roughness for the present experiments. AFM (Atomic Force Microsc0py) was used to
obtain the sapphire fiber topography after implantation, as shown in Figure 5.4. The mean
roughness difference between sputtered fiber and unsputtered fiber is less than 3 nm.
Furthermore, whether the roughened surface consists of sharp surface flaws is doubtful.
This result implies that the effect of ion sputtering on the fiber surface morphology is minor
at this stage.

In summary, the tensile stress in near-surface region causes bend strength

degradation slightly, compensating for crack shortening by sputter loss. The existence of a

158

buried compressive stress zone establishes a new fracture criterion of 3.53 MPax/Z.
When a pre-existing surface crack ( with length 11 nm) propagates to the implanted zone
(40 nm beneath the fiee surface), the corresponding stress intensity factor is 3.71

MPa x/r—n , which is greater than the new fracture criterion. The implanted zone cannot
arrest crack propagation in three-point bend loading. Therefore, no significant increase in

bend strength was observed in current work.

5.3 Residual Surface Stress and Abrasion Behavior

In section 4.2, it has been well documented that abrasion leads to a significant
strength degradation for unimplanted fibers, whereas the strength retention for implanted
fibers is considerably higher after abrasion. This is intuitively puzzling in view of the
failure of ion implantation to modify bend strength directly. It is reasonable to assume that
abrasion causes surface damage to a certain extent. In the abrasion process, sapphire fibers
are forced into contact with hard, sharp and irregularly shaped particles. Some of them
may be considerably larger than the average particle size. Figure 5 .5 illustrates the SEM
observation on the contact interface between fiber and abrasive particles. The surface
roughness profile is shown by AFM image section analysis in Figure 5.6. The average
particle size is 5 um in diameter. Figure 5.6 (c) schematically illustrate the contact position
between sapphire fiber and SiC particle during abrasion process. Therefore, the abrasion
process used in this study may be compared directly to the micro-indentation process where
sharp contact takes place on the sample surface.

Figure 5.7 depicts one complete loading and unloading cycle during indentation
[Lawn, 1975]:

(a) Initial loading The sharp indenter induces a zone of irreversible deformation
about the contact point. The size of this zone increases with load;

(b) Critical zone formation At some critical indenter load, a crack suddenly

initiates below the contact point, where the stress concentration is greatest;

159

x lttm/div
z 100 nm/div

   

(a) (b)

. 0.5 urn/div
0.5 urn/div x .
z 30 nm/div z 40 nm/dlv

 

 

Figure 5.4 AFM topography on sapphire fiber surface (a) camel fiber without abrasion;
(b) control fiber with abrasion; c) implanted fiber (5x10‘5Mg+crri2) without abrasion; (d)
implanted fiber (5x10‘6 Mg+cm' with abrasion.

 

Figure 5.5 SEM observation on the interface bewteen sapphire fiber and abrasive
particles.

 

 

 

 

 

(a)

Section Analysis: L 3.477 um
RMS 671.48 nm

 

/_
/
Fi.segure56AFM onanaliymiis SC bas eppapegls surfac eprofile. (a)hnSeurface
p fl Twopai eofarr ouhng
al ltionanalysis s,(cc)Shem:11tlic fharpp acbew afiber andethe
bas epaper.

162

1“ #—
+ i-
T

(b) (e)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 5.7 Schematic of radial and lateral cracks evolution under point indentation.
Radial crack forms during loading (+) half cycle, lateral crack during unloading (-) half
cycle. Fracture initiates from deformation zone (dark region). (a) initial loading, (b)
critical zone formation, (c) stable crack growth, ((1) initial unloading, (e) residual stress
cracking, (f) complete unloading. (after B. R. Lawn and M. V. Swain, J. Mater. Sci, 10,
1975).

163

(c) Stable crack growth Increasing the load causes further, stable extension of
the median crack;

(d) Initial unloading The median crack begins to close (not heal);

(6) Residual stress cracking Relaxation of deformed material within the contact
zone just prior to removal of the indenter superimposes intense residual tensile stress upon
the applied load; and

(f) Complete unloading The lateral crack continues to extend and may cause
chipping.

The stress distribution beneath the point-load indentation is known as Boussinesq
problem as depicted in Figure 5.8. Timoshenko [1951] has given the elastic solutions to

the problem in polar coordinating system as follows:

P 1 _ _
0'" = E{(l—2V{7-%(r2 +22) V2]-3rzz(r2 +22) 5/2}

000 = L(1_2V){_ri2+;§2_(r2 +Z2)-l/2 +z(r2 +ZZ)-3/2}

2a

31’ 3 2 2 —s/2
0' =-—z r +z 5.18
‘1 2n ( ) ( )
on __. __,22(r2 + 22)”
o,9=o&=0

The principal normal stresses can be transformed by following relationship:

164
tan2a = Zon/(ou —o,,)

0'll = 0,, sin2 a + all cos2 a — 20',z sinacosa
(5.19)
022 = 099

03, = 0,, cos2 a + Ca sin2 a + 20‘,z sin acosa

The stress contours at the head of the indenter tip can be plotted by using
Mathematica, as shown in Figure 5.9. Both 0’ll and 0'33 are contained within planes of

symmetry through the normal load axis, with 0'“ everywhere tensile, 0'33 everywhere
compressive. an is perpendicular to the plane which contains load axis and a” and is
tensile with a conical 4’ < 51.8° below the indenter and compressive outside this region. In
these stress contours, point load, P, is 30 mg. The stress values are labeled on each
contour.

If there is a 2 GPa compressive stress zone in the front of the indentation tip, the
resulting stress contours are changed. Generally, original tensile portions contract into a
lower stress value, while compressive contours expand to a high compressive stress value.
This is important because existence of compressive stress substantially affects indentation
crack evolution as discussed as follows.

In order to better understand compressive stress effects and indentation crack
evolution, a schematic is constructed in Figure 5.10 to show the stress distribution in
implanted fibers, along with TEM microstructure observation and indentation stress
contours which was imposed by the 2 GPa compressive stress. From Figure 5.10 (c), it
can be seen that the stress changes from tensile to compressive. It is the compressive stress
state that inhibits the median crack evolution. Since the tensile stress in the near-surface is
small compared with the tensile stress on caused by indentation, it thus is negligible.

During the indentation process, a penny-like median crack is formed upon the

165

 

 

Figure 5.8 Polar coordinate system for Boussinesq axially symmetric
point load P. 0'99 is perpendicular to the plane.

166

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C
3 o
*3
Q
. N... ..
P E
H
a V
1 ,2. N x
,V‘
o 022
O
H
x
*<'
‘N
.0 g 033
1",
.N x
,V

 

 

 

 

y (100 nm)

Figure 5.9 The stress contours in the front of point load indentation during sharp
point contact. on and 033 are contained within planes of symmetry through the

normal load axis, with on everywhere tensile, 0’33 compressive. 622 is

perpendicular to the plane containing load axis and an and is musile with (Ml-80
below the indenter and compressive outside this region.

167

Fiber C-Axis

TEM
(Ar+, 175 keV
1x1016 ions cm'2)

(a)

 

 

Regions
Surface Stress
Distribution 5_I .
(b)
a ' 3 b
as «r >. 8
‘5 ‘5 3 2
§ 5 5 m
u- z m
(C) rv///\J|

 

Figure 5.10 Schematic of the surface stress distribution in the implanted zone, and
stress contour beneath sharp point contact during implanted fiber abrasion process.
(a) TEM observation on 175 keV, 1x10l6 Ar+cm‘2; (b) Distribution of
compressive stress induced by implantation; (c) Stress contour on the fiber surface
produced by an abrasive particle (after superimposed by compressive stress).

168

application of point load P. Figure 5.11 shows median crack formation during an
indentation process. According to Marshall [1979], the corresponding stress intensity

factors are:

 

K, = 1;” (arising from elastic driving force) (5.20)
c
LP . . .

K, = T (arisrng from resrdual stress) (5.21)
c

K s = as ( we )1/ 2 (arising from surface stress) (5.22)

where xe, x,, w are dimensionless constants related to crack geometry, and w: 4/1r2,
x,+ xe= 0.0139E/H, xe > 3 x,, E is Young’s modulus, H is the hardness, E = 414 GPa,
and H = 20 GPa for sapphire [l-Iioki: 1986]; P is the point load (30 mg), c is the median
crack length and Us is the surface stress (2 GPa).

By the principle of superposition, the stress intensity factor is the summation of the

three components, i.e.,

K, +K, +K, = KC (5.23)

Considering the loading and unloading relation separately gives:

 

£52”. + 7:37; ”Sm,“ )1/2 = Kc PT (5.24)
t
3; + 262% + “we )1/2 = KC Pl (5.25)

where P‘ is the maximum load, P Tand P istand for loading and unloading cycles,
respectively. When the unloading half cycle begins, the effective load in residual stress

 

169

 

 

17+Pr

 

 

 

 

 

 

 

 

 

(b)

Figure 5.11 Stress distribution during indentation process. P is an
applied external load, Fe is the elastic component of the field, and Pr is
the residual stress. (a) during loading; (b) unloading.

 

170

zone reaches a maximum. Therefore, P"I was substituted for P in the second term of
equation (5.25). The process of median crack evolution can be described as follows: the
median crack reaches its maximum length, C‘, at the end of the loading half cycle, and then
begins to contract when the unloading half cycle begins, and finally reaching an equilibrium
length, C" , when the point load P has been completely withdrawn. Thus, the median
crack contraction arises from elastic recovery.

For unimplanted fibers (stress-free surface), taking 0'3 = 0, during the loading
cycle, AP > 0, and Ac > 0, the maximum crack length, C‘, at maximum loading
P = P'is:

C‘ = [(x. + raw/19]”3

(5.26)
Substitute x8: 0.23, x,= 0.057, P': 30 mg, and Kc = 2.5 MPax/m into equation
(5.26), one obtains:

C ‘= 106 nm

in!”

Similarly, during the unloading process, the residual portion 3/2
c

 

becomes constant, and
differentiating equation (5.25) relative to c by taking as = 0, the final equilibrium crack
length is:

Cv = [ 1.1" /Kc]2/3 (5.27)

Substitute xc= 0.23, P’: 30 mg, and Kc = 2.5 MPaxhn into equation (5.27), one

obtains:

C" = 91 nm.

 

171

By equation (5.10), the bend strength corresponding to this crack length is:

oo = 4.15 GPa

This calculated bend strength is close to measured bend strength for abraded unimplanted
sapphire fibers.

Likewise, for implanted sapphire fibers (stressed surface), the maximum crack
length during the loading cycle can be obtained by:

First, differentiating equation (5.24):

do
5; = 20¢, + x, )/[31t,.1/2 + 4o,(my)1/2c] (5.28)

Then integrating in the interval of load [0, P' ]:
P. _ C. 3 0.5 0.5
10%, + 1,)dP — I0 (2 ch — 205(71111) c )dc (5.29)

Substituting x,= 0.23, x,= 0.057, P': 30 mg, as = -2 GPa and Kc = 2.5 MPax/Frn into

equation (5.29), the maximum crack length during loading is:
C ' = 90 nm.

Similarly, in order to obtain the final equilibrium crack length, first differentiating equation
(5.25), and then integrating at the interval [P' , 0], gives:

0 CV 3 .
1,..x.dP= C] (5196” -20.(n'vr)°’c)dc (5.30)

 

172

The final crack length is
C" = 32 nm.

By equation (5. 10), the bend strength is:

c0 = 7.01 GPa

This calculated strength is also reasonably close to the measured bend strength for
implanted fibers after abrasion.

The indentation fracture mechanics analysis on median crack evolution has good
agreement with measured bend strength retention for abraded sapphire fibers. Without
compressive stress, indentation can generate a median crack with a length three times as
long as that with compressive stress. The existence of compressive stress can change
stress distribution in the front of indenter tip, and effectively inhibit median crack growth.
For the fiber abrasion process, this effect in turn allows retention of high bend strength for
implanted fibers after abrasion.

In summary, for unimplanted fibers, abrasion produces deeper crack and thus
causes fiber bend strength degradation. For implanted fibers, the irradiation-induced
compressive stress of 2 GPa did not enhance fiber bend strength, even though a high
fracture resistance is established in the implanted region. This is because the beginning of
the compressive zone is buried deep enough not to affect pre-existing crack tips extension.
By the time pre-existing cracks have grown into the compressive zone, their additional
length brings the stress intensity level above the critical value even when the excess
compressive stress is considered. In the case of implanted fibers subjected to the abrasion
process, however, a compressive stress can effectively retard the median crack evolution,

and thus retain a higher bend strength.

 

CHAPTER VI

SUMMARY OF RESULTS AND CONCLUSIONS

In the current work, effects of 175 keV Mg” and Ar" implantation into single
crystal sapphire fibers were studied. TRIM calculations showed that the ion range of 175
keV Mg“ implanted into sapphire is 190 nm with straggling 52 nm. The sputter yield
during ion implantation is 0.14 atoms/ion for Al, and 0.23 atoms/ion for O; For 175 keV
Ar“ implantation, the ion range is 178 nm with straggling 42 nm.

TRIM calculations correlated with TEM observations if focusing and channeling
effects are considered. TEM observations on implanted fibers with doses of 1 x 1016 Ar+a
7 x1016 Mg+, and 2 x 1017 Mg'*'/cm2 showed that there were four distinct regions in the
implanted zone, i.e., free surface; a near-surface zone which was 40 ~ 50 nm underneath
the free surface; a buried layer beginning at a depth of ~ 40 nm which was 200 nm to 400
nm thick and contained a high density of dark contrast features; and below this, the un-
damaged fiber interior. Diffraction patterns in these areas showed that the microstructures
in these regions were crystalline after implantation.

By three-point bend tests, a characteristic bend strength of 9.31 GPa was measured
for as-received fibers. Mg+ and Al’" implantation into the fibers changed the characteristic
bend strength only slightly. The Weibull distribution of the bend strength of all implanted
fibers fall within the 95 percent confidence interval of the bend strength of unimplanted
fibers.

The ability of ion implantation to enhance wear resistance was apparent. The bend
strength of unimplanted fibers decreased about 49 percent after subjected to an abrasion
process, whereas Mg+ and Ar+ implanted sapphire fibers retained up to 93 percent of their

original strength. In terms of Weibull statistical analysis, abrasion caused significant

173

174

strength degradation on the unimplanted fibers but no significant degradation on implanted
fibers.

The observed wear resistance enhancement was attributed to generation of a

subsurface compressive stress region induced by ion implantation. By micro indentation

technique, the magnitude of compressive stress was estimated to be about 2 GPa. The

superficial fracture toughness was also found to increase monotonically with ion dose.

Based on these results, following conclusions are drawn:

. The characteristic bend strength of Saphikon single crystal sapphire fiber is 9.31:1: 0.48
GPa.

175 keV Mg+ implantation into sapphire fibers has little effect on the bend strength for
unabraded fibers. A slight enhancement in bend strength, to a maximum of 9.47 i
0.33 GPa, resulted from a dose of 4 x1016 ions cm‘z.

. Similarly, 175 keV Ar+ implantation also shows little effect on the bend strength for
unabraded fibers. Measured bend strength varies from 8.34 :1: 0.4 GPa to 9.24 :1: 0.58
GPa.

. Abrasion for 60 minutes causes 52 percent bend strength degradation in unimplanted
fibers.

. Mg+ implantation substantially enhances the fiber abrasion resistance, resulting in over
90 percent of bend strength after abrasion.

. Ar+ implantation also allows over 90 percent of bend strength retention after abrasion.

. The magnitude of the compressive stress induced by implantation is estimated to be at
least 2 GPa, as measured by Vickers micro-indentation techniques.

. The superficial fracture toughness of implanted fibers monotonically increases with ion
dose.

. Mg+ implantation causes slight increase in nano-indentation hardness in intermediate

implantation doses, and a decrease at the dose of 2 x 10'7 Mg+ cm‘z.

10.

11.

12.

13

14.

15.

175

Residual bend strength of extracted unimplanted fibers from fiber-reinforced NiAl
composite is sharply reduced, to 3.36 GPa. Residual bend strength of extracted Mg”
implanted fibers is reduced to 4.51 GPa. Similarly, residual bend strength of extracted
Ar+ implanted fibers is reduced to 4.13 GPa.

Annealing fibers at 1400 0C for 2 hours produces 23 percent bend strength decrease for
unimplanted fibers, and 21 to 28 percent for both Mg“ and Ar+ implantation.

TEM observations on implanted fibers show that there are four regions, i.e., the free
surface, the near-surface region (40 nm) beneath free surface where there are few
implanted ions, and the buried layer (200 nm to 400 nm width) where there is a high

density of dark contrast features, and the un-damaged fiber interior.

. Diffraction analysis in TEM observations shows that implanted region was still

crystalline.

Fracture mechanics calculations show that a compressive stress establishes a higher
fracture criterion of 3.53 MPa 47a in the implanted zone. However, since the zone is
buried ~ 40 nm below the surface, the compressive zone can not arrest pre-existing
crack propagation during three-point bend tests.

Indentation fracture mechanics calculations suggest that indentation events during
abrasion produces deeper cracks in unimplanted fibers, whereas the compressive stress

in implanted fibers can effectively resist crack extension during abrasion.

Results from the current work have shown that ion implantation is a useful

technique to enhance surface abrasion resistance, and it indicates that ion implantation can

be a powerful method to overcome the high surface sensitivities of sapphire fibers, which

usually causes substantial strength degradation.

In order to continue the progress toward a better understanding of the efficiency of

ion implantation in modification of surface properties, the following recommendations are

made for future work:

 

1.

176

Implantation of different ions such as Zr’, Cr’ should be carried out to consider the
chemical effects in irradiation damage.

High dose and low temperature implantation resulting in surface amorphization, which
in turn affects surface properties.

TEM observation on characterization of defect clusters, possible diffusion and
precipitate effects, and relationships between defect production and implantation

conditions.

 

CHAPTER VII

BIBLIOGRAPHY

Appleton, B. R., "Ion Beam and Laser Mixing Fundamental and Application," Ion
,.ed J. S. Williams and J. M. Poate, Academic
Press, 1984.

Asthana, R., Tewari, S. N., and Bowman, R. R., “Influence of Fabrication Technique on
the Fiber Pushout Behavior in a Sapphire-Reinforced NiAl Matrix Composite,”
Met. Mat. Trans. A Vol. 26A, 209-223 (1995).

ASTM Designation: E 382-89, “Standard Test Method for Microhardness of Materials,”
ASTM, 1989.

Becher, P.F., “Abrasive Surface Deformation of Sapphire,” J. Amer. Ceram. Soc., Vol.
59, No. 3-4, 143-145 (1976).

Becher, P.F., “Fracture-Strength Anisotropy of Sapphire,” J. Amer. Ceram. Soc., Vol.59,
No 1-2, 59-61 (1976).

Bowman, R. R., Misra, A. K., and Arnold, S. M., “Processing and Mechanical Properties
of A1203 Fiber-Reinforced NiAl Composites,” Met. Mat. Trans. A, Vol. 26A, 615-
628 (1995).

Bowman, R. R., Misra, A. K., and Locci, I. E., “Unresolved Technical Barriers to The
Development of A Structurally Viable A1203/NiAl Composite,” HITEMP Review
1992: Advanced High Temperature Engine Materials Technology Program, NASA,
HITEMP, 57-1 to 57-12 (1993).

Brisset, F., and Vassel, A., “Compatibility of SiC and A1203 Continuous Fibers with
Titanium Aluminum Matrices,” W, edited by R. Doralia, J. J.
Lewandowski, C. T. Liu, P. L. Martin, D. B. Miracle and M. V. Nathal, The
minerals, Metals & Materials Society, 739-747 (1993).

Bull, 8.1., and Page, T.F., “Thermal Effects on the Microstructure and Mechanical
Properties of Ion Implanted Ceramics,” J. Mater. Sci., Vol. 26, 3086-3106 (1991).

Bull, S. J. and page, T. F., “The Effects of Ion Implantation on the Friction Behavior of
Sapphire,” J. Mater. Sci., Vol. 30, 5356-70 (1995).

Burnett, P. J ., and Page, T. F. , “Changing the Surface Mechanical Properties of Silicon
and (it-A1203 by Ion Implantation,” J. Mater. Sci., Vol. 19 3524-45 (1984).

Burnett, P. J., and Page, T. F. ,“Changing The Surface Plasticity and Hardness of

Sapphire by Ion Implantation,” WWW, edited by R. E.
Tressler and R. C. Bradt, 669- 680 (1984).

177

 

178

Burnett, P.J. and Page, T.F., “An Investigation of Ion Implantation-Induced near-Surface
Stresses and Their Effects in Sapphire and Glass,” J. Mater. Sci., Vol. 20, 4624-
4646 (1985).

Burnett, PL, and Page, T.F., “The Friction and Hardness of Ion-Implanted Sapphire,”
Wear, Vol.114, 85-96 (1987).

Callister, W.D., “MW,” Wiley, New York, 1985.

Carter, G., and Colligon, J .S., W, American Elsevier
Publishing Company, Inc., New York (1968).

Chasorr, E., Mayer, T. M. and Howard, A. J, "Kinetics of Surface Roughening and

Smoothing During Ion Sputtering," MW, eds. S. M.
Yalisove, C. V. Thompson and D. J. Eaglesham, Materials Research Society,

Pittsburgh, PA 1994.

Darolia, R., “NiAl Alloys for High-Temperature Structural Applications,” JOM, March,
44-49 (1991).

Darolia, R., “NiAl for Turbine Airfoil Applications,” Wes, edited by R.
Doralia, J. J. Lewandowski, C. T. Liu, P. L. Martin, D. B. Miracle and M. V.
Nathal, The minerals, Metals & Materials Society, 495-504 (1993).

Davis, J. B., Yang, J., and Evans, A. G., “Effects of Composite Processing on the
Strength of Sapphire Fiber-Reinforced Composites,” Acta Metal]. Mater., Vol. 43,
No. 1, 259 - 269 (1995).

Deamaley, G. Freeman, J. H., Nelson, R. S., and Stephan, J., "Ion Implantation,"
W, eds. S. Amelinckx and R. Gevers, Vol. 8, North-
Holland Publishing Company, 1973.

Dienes, CL and Vineyard, G.H., WW, Wiley, New York, 1957.

Dorre, E., and Hiibner, H. Alumina, Spinger-Verlag Berlin, Herdelberg, 1984.

Draper, S.L., Eldridge, J .I., and Locci, I.E., “A1203 Fiber Strength Degradation in
MMC’s and IMC’s,” HITEMP Review 1992: Advanced High Temperature Engine
Materials Technology Program, NASA, HITEMP, 17-1 to 1713 (1992).

Draper, S. L., and Locci, I. E., “A1203 Fiber Strength Degradation in Metal and
Interrnetallic Matrix Composites,” J. Mater. Res., Vol. 9, No. 6, 1397 - 1441
( 1994).

Duwell, E.J., “Friction and Wear of Single-Crystal Sapphire Sliding on Steel,” J. Appl.
Phys., Vol.35, No.9, 2691-2697 (1962).

Dutt, B. V., and Kroger, F. A., “High Temperature Defect Structure of Cobalt-doped or-
Alunrina,” J Amer. Ceram. Soc., Vol. 58, 420-427 (1975).

EerNisse, E.P., "Sensitive Technique for Studying Ion-Implantation Damage," Appl.
Phys. Let., Vol. 18, 581-583 (1971).

179

Green, D.J., “Compressive Surface Strengthening of Brittle Materials," J. Mater. Sci.,
Vol. 19, 2165-2171 (1984).

Heuer, A. H., “Deformation Twinning in Corundum,” Phil. Mag., Vol. 13, 379-393
(1966).

Hioki, T., Itoh, A., Noda, S., Doi, H., Kawamoto, J., and Kamigaito, 0., “Effect of Ion
Implantation on Fracture Stress of A1203,” Nucl. Inst. Meth. Phy. Res., B7/8,
521-525 (1985).

Hioki, T., Itoh, A., Noda, S., Doi, H., Kawamoto, J., and Kamigaito, 0.,
“Strengthening of A1203 by Ion Implantation,” J. Mater. Sci. 1et., Vol. 3, 1099 -
1101 (1984).

Hioki, T., Itoh, A., Noda, S., Doi, H., Kawamoto, J ., and Kamigaito, 0., “Modification
of he Mechanical Properties of Ceramics by Ion Implantation,” Nucl. Inst. Meth.
Phy. Res., BB9, 657-664 (1989).

Hioki, T., Itoh, A., Ohkubo, M., Noda, S., Doi, H., Kawamoto, J., and Kamigaito, 0.,
“Mechanical Property Changes in Sapphire by Nickel Ion Implantation and Their
Dependence on Implantation Temperature,” J. Mater. Sci., Vol.21, 1321-1328
( 1986).

Hockey, B.J., “Plastic Deformation of Aluminum Oxide by Indentation and Abrasion,” J.
Amer. Ceram. Soc., Vol. 54, No. 5, 223-231 (1971).

Holland, F.A., and Zaretaky, E.V., “Investigation of Weibull Statistics in Fracture
Analysis of Cast Aluminum,” NASA TM-102000, 1989.

Jardine, A. P., Mukhopadhyay, S. M., and Blakely, J. M., “Mg Implantation and
Characterization of Sapphire Surface,” Mat. Res. Soc. Symp., Vol. 74, Materials
Research Society, 365 - 373 (1987).

J askowiak, M.H., and Setlock, J .A., “Mechanical Behavior of Sapphire Reinforced
Alumina Matrix Composites,” HITEMP Review 1993: Advanced High

Temperature Engine Materials Technology Program, NASA, HITEMP, 70-1 to
70-11 (1993).

Jaskowiak, M.H., Eldridge, J .I., Hurst, J .B., and Setlock, J .A., “Interfacial Coating for
Sapphire/A1203,” HITEMP Review 1991: Advanced High Temperature Engine
Materials Technology Program, NASA, HITEMP, 84-1 to 84-11 (1991).

Jayne, D. T., and Matthiesen, M. M., “Removing Hydroxypropyl Methylcellulose Sizing
from Sapphire Fibers by Ultraviolet/Ozone Cleaning,” J. Am. Ceram. Soc., Vol.
78, 2861-64 (1995).

Johnson, L. G., “The Median Ranks of Samples Values in Their Population with an
Application to Certain Fatigue Studies,” Indust. Math., 1951.

Johnson, L.G., ‘ _ - --_ -_ '
Co., New York,1964.

 

'. 61-,” Elsevier Publication

180

Jones, L.E., Newcomb, S.A., and Tressler, R.E., “Creep and Rupture Behavior of Single
Crystal Sapphire Filaments,” HITEMP Review 1990: Advanced High Temperature
Engine Materials Technology Program, NASA, HITEMP, 47-1 to 47-16 (1990).

Kern, D.Q, "Wafer," McGraw-Hill Book Company, Inc., 1950.

Kinchin, G.H, and Pease, R.S., Rep. Progr. Phys., Vol. 18, l (1955).

 

Kingery,W.D., Bowen, H. K., and Uhlmann, D. R., Wes, 2nd ed.,
New York; John Wiley and Sons, 1976.

Kirchner, H. P., Gruver, R. M. and Walker, R. E., “Strengthening Sapphire by
Compressive Surface Layers,” J. Appl. Phys., Vol. 40, N0. 9, 3445 - 3452
(1969).

Kowalski, Z.W., "Ion-Bombardment Modification of the Surface Morphology of Solids,"
J. Mater. Sci., Vol. 25, 2875-2879 (1990)

Krefft, GB, and EerNisse, E.P., “Volume Expansion and Annealing Compaction of Ion-

Bombarded Single-crystal and Polycrystalline (it-A1203,” J. Appl. Phys., Vol.45,
No.5, 2725-2730 (1978).

Kronberg, M.L., “Plastic Deformation of Single Crystals of Sapphire: Basal Slip and
Twinning,” Acta Metall., Vol.5, No.9, 507-524 (1957).

Kumar, K. S., Mannan, S. K., and Viswanadharn, R. K., “Fracture Toughness of NiAl
and NiAl-Based Composites,” Acta. Metal]. Mater., Vol. 40, 1201-22 (1992).

Lagerlof, K.P.D., Mitchell, TE, and Heuer, A.H., “Deformation and Diffusion in

Sapphire (OI-A1203)? Waterman eds. R.E. Tressler and
RC. Bradt, Plenum Press, New York, 199-212 (1984).

Lagerlof, K.P.D., Pletka, B. J ., Mitchell, T. E.., and Heuer, A. H., “Deformation and
Diffusion in Sapphire (or-A1203),” Radiation Effects, Vol. 74, 87 - 107 (1983).

Lawn, B.R., and Fuller, B.R., “Measurement of Thin-Layer Surface Stress by Indentation
Fracture,” J. Mater. Sci., Vol.19, 4061-4067 (1984).

Lawn, B.R., and Swain, M.V., “Microfracture Beneath Point Indentation in Brittle
Solids,” J. Mater. Sci., Vol.10, 113-123 (1975).

Lawn, B.R., Evans, A.G., and Marshall, D.B., “Elastic/Plastic Indentation Damage in
Ceramics: The Median/Radial Crack System,” J. Amer. Ceram. Soc., Vol.63, No.
9-10, 574-581 (1980).

MacKay, R. A., Brindley, P. K., and Froes, F. H., “Continuous Fiber-Reinforced
Titanium Aluminide Composites,” JOM, May, 23-29 (1991).

Marshall, D.B., Lawn, B.R., and Evans, A.G.,"Elastic/Plastic Indentation Damage in
Ceramics: The Lateral Crack System," J. Amer. Ceram. Soc., Vol. 65, 561-566
( 1982).

 

181

Marshall, D. B., and Lawn, B. R., “Residual Stress Effects in Sharp Contact Cracking:
Part 1 Indentation Fracture Mechanics,” J. Mater. Sci., Vol. 14, 2001 - 2012
(1979).

Marshall, D. B., and Lawn, B. R.,"Residual Stress Effects in Sharp Contact Cracking:
Part 2 Strength Degradation, " J. Mater. Sci., Vol. 14, 2001-2012 (1979).

Martz, S. W., and Tressler, R. E.,"The Threshold Stress Intensity for Slow Crack Growth
in C-Axis Sapphire Fibers,”HITEMP Review 1992: Advanced High Temperature
Engine Materials Technology Program, NASA, HITEMP, 81-1 to 81-11 (1992).

McCallum, J. C., White, C. W., Sklad, P. S., and McHargue, C. J ., “Annealing
Environment effects in Solid-Phase Epitaxial Regrowth of Fe-Irnplantation A1203,”
Nucl. Inst. Meth. Phys. Res., B46, 137 - 143 (1990).

McHargue, C. J ., “The effects of Ion Implantation on the Structure of Ceramics,” JOM
No.7, 40 - 44 (1991).

McHargue, C. J., Farlow, G. C., Begun, G. M., Williams, J. M., White, C. W .,
Appleton, B. R., Sklad, P. S., and Angelini, P., “Damage Accumulation in
Ceramics During Ion Implantation,” Nucl. Inst. meth. Phys. Res., B16, 212 - 220
(1986).

McHargue, C. J ., Farlow, G. G, Lewis, M. B., and Williams, J. M., ‘Implantation of
Gases into Sapphire,” Nucl. Inst. Meth. Phys., B19/20, 809 - 812 (1987).

McHargue, C. J ., Farlow, G. C., Sklad, P. S., and White, C. W., “Iron Ion Implantation
Effects in Sapphire,” Nucl. Inst. Meth. Phys. Res., B19/20, 813 - 821 (1987).

McHargue, C. J. Naramoto, H., Appleton, B. R., White, C. W., and Williams, J. M.,
“Structure and PropertiesU of Single Crystal A1203 Implanted with Chromium and
Zirconium’ u; -_. : -._--.-u,eds .S..T Picraux
and W. J. Choyke, Elsevier Science Publishing Company, Inc., New Y,ork 147-
153 (1982).

 

McHargue, C..,J Yust, C. S., Angelini, P, Sklad, P..S, and Lewis, M.B., W
W, eds. C. Brooks and F. Almond, Adam Hilger, Bristol, 803-
812(1986).

Mchargue, C.J., and Yust, C.S., “Lattice Modification in Ion-Implanted Ceramics,” J.
Amer. Ceram. Soc., Vol. 67, No. 2, 117-123 (1984).

Misra, A. K., Bowman, R.R., Locci, IE, and Matthiesen, M. M., “Interface Mo
Coaitngs For A1203NiAl Composite,” HITEMP Review 1992: Advanced High
Temperature Engine Materials Technology Program, NASA, HITEMP, 58-1 to 58-
13, (1993).

Morscher, G.N., Sayir, H., and Unal, 0., “The Bend Properties of Sapphire Fibers at
Room and High Temperatures,” HITEMP Review 1992: Advanced High
Temperature Engine Materials Technology Program, NASA, HITEMP, 18-1 to 18-
14, (1992).

182

Mukhopadhyay, S.M., Jardine, A.P., Blakely, J .M. and Baik, S., “Segregation of
Magnesium and Calcium to the (1010) Prismatic Sutface of Magnesium-Implanted
Sapphire”, J Amer. Ceram., Vol. 71, 358-62 (1988).

Naguib, H. M., and Kelly, R., Radiation Effcts, Vol. 25, 1 (1975).

Norgett, M. J ., Robinson, M. T., and Torrens, I. M., “A Proposed Method of Calculating
Displacement Dose Rates,” Nucl. Eng. Desi., Vol. 33, 50 - 54 (1975.

Nutt, S.R., and Wawner, F.E., “Silicon Carbide Filaments: Microstructure,” J. Mater.
Sci., Vol. 20, 1953-1960(1985).

Ochiai, S., and Murakarni, Y, “Theoretical Prediction of Tensile Strength of Fibers as a
Function of Thickness of Brittle Zones on Fiber Surfaces,” Met. Trans. A, Vol.
12A, No. 7, 1155 - 1161 (1981).

Ochiai, S., and Osamura, K., “A Computer Simulation on Tensile Strength of Surface-
Damages Fibers,” Met. Trans. A, Vol. 19A, No.6, 1499 -1506 (1988).

O’Hem, M.E., McHargue, C.J., White, C.W., and Farlow, G.C., “The Effect of
Chromium Implantation on the Hardness, Elastic Modulus and Residual Stress
of A1203,” Nucl. Inst. Meth. Phys. Res., B46, 171-175 (1990).

O’Hern, M. E., Romana, L. J., McHargue, C. J., McCallum, J. C., and White, C. W.,
“The Effect of Crystl Orientation on Damage Accumulation in Chromium-Implanted
A1203,” ASTM STP 1125, 740 - 748 (1992).

Ohkubo, M., Hioki, T., Suzuki, N ., Ishiguro, T., and Kawamoto, J ., “Structure and
Magnetic Properties of Fe-Irnplanted Sapphire,” Nucl. Inst. Meth. Phys. Res.,
B39, 675 - 679 (1989).

Ohkubo, M., Hioki, T., and Kawamoto, J., “Recrystalline-Driven Migration and
Resultant-Oriented Precipitation”, J. Appl. Phys., Vol.60, N015, 1325-1335
( 1986).

Pedraza, D. F., and Mansur, L. K., “The Effect of Point Defects on the Amorphization of
Metallic Alloys During Ion Implantation,” Nucl. Inst. Meth. Phy., B 16, 203 - 211
( 1986).

Pollack, J. T., and Hurley, G. F., “Dependence of Room Temperature Fracture Strength
on Strain-rate in Sapphire,” J. Mater. Sci., Vol. 18, 1595-1602 (1973).

Ramos, S.M.M., Canut, B., Gearn L., Thevenard, P., Bauer, M., Maheo, Y., Kapsa,
Ph., and Loubet, J .L., “Structure, Mechanical and Tribological Properties of
Titanium Implanted Alumina,” J. Mater. Res., Vol. 7, No. 1, 173-184 (1992).

Sayir, H., DeGuire, M.R., Sayir, A., and Lagerlof, K.P.D., “Bend Strength of Undoped
and Impurity Doped Laser Heated Float Zone Sapphire Fibers,” HITEMP Review
1992: Advanced High Temperature Engine Materials Technology Program,
NASA, HITEMP, 20-1 to 20-17 (1992).

 

Schalek, R., Ph.D. dissertation, Michgan State University, 1995.

183

Shah, D. M., and Anton, D. L., “Alunrinua Fiber Reinforced Intermetallic Matrix
Composites,” Mum, edited by R. Doralia, J. J. Lewandowski,
C. T. Liu, P. L. Martin, D. B. Miracle and M. V. Nathal, The minerals, Metals &
Materials Society, 755-764 (1993).

Sorensen, J ., “Continuous Fiber Reinforced Interrnetallic Composites: An Overview,”
WW, edited by R. Doralia, J. J. Lewandowski, C. T. Liu, P. L.
Martin, D. B. Miracle and M. V. Nathal, The minerals, Metals & Materials Society,
717-726 (1993).

Specht, E. D., Walko, D. A., and Zinkle, S. J ., “Amorphization of A1203 by Ion Induced
Density Reduction,” Nucl. Inst. Meth Phys. Res., B84, 323-330 (1994).

Steijn, R.P., “On the Wear of Sapphire,” J. Appl. Phys., Vol. 32, No. 10, 1951-1958
(1961).

Stotel, F., and Conrad, H., “Fracture and Twinning in Sapphire (a— A1203 Crystals,”
Trans. AIME, Vol. 227, 1053-1060 (1963).

Stough, M.A., Hellmann, J .R., and Angelone, M.S., “Zirconia Coated Sapphire Fiber
Surface Degradation,” HITEMP Review 1993: Advanced High Temperature Engine
Materials Technology Program, NASA, HITEMP, 80-1 to 80-10 (1993).

Tada, H., Paris, RC. and Irwin. G. R., W. Del
Research Corporation, Hellertown, Pennsylvania, 1973.

Tewari, S. N., Asthana, R., and Noebe, R. D., “Interfacial Shear Strength of Cast and
Directionally Solidfied NiAl-Sapphire Fiber Composites,” Met. Trans. A, Vol.
24A, 2119-25 (1993).

Tewari, S. N., Asthana, R., Tiwari, R., and Bowman, R. R., “Zone Directional
Solidification of Sapphire NiAl Composite,” HITEMP Review 1992: Advanced
High Temperature Engine Materials Technology Program, NASA, HITEMP, 60—1
to 60-11(1993).

Tewari, S. N., Asthana, R., Tiwari, R., Bowman, R. R., and Smith, J ., “Influence of
Interfacial Reactions on the Fiber-Matrix Interafcial Shear Strength in Sapphire-
Reinforced NiAl(Yb) Composites,” Met. Mat. Trans. A, Vol. 26A, 477-491
( 1995).

Tewari, S. N., Asthana, R., Tiwari, R., Bowman, R. R., and Smith, J ., “Influence of
Interfacial Reactions on the Fiber-Matrix Interfacial Shear Strength in Sapphire
Fiber-Reinforced NiAl(Yb) Composites,” Met. Trans. A, Vol. 26A, 477-491
(1995).

Timoshenko, S., and Goodie, J. N., W, McGraw-Hill Book Company,
Inc., 1951.

Traumbauer, B.R., Hellman, I.R., Shelleman, DC, and Koss, D.A., “Effect of Cleaning
and Abrasion-induced Damage on the Weibull Strength Distribution of Sapphire
Fiber,” J. Am. Ceram. Soc., Vol. 77, 2017-2 (1994).

184

Traumbauer, B.R., Hellman, I.R., Shelleman, DC, 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, HITEMP, 16-1 to 16—15 (1992).

Tresseler, RE, and Newcomb, S.A., “Factors Influencing the High Temperature Strength
of Sapphire Fibers,” HITEMP Review 1992: Advanced High Temperature Engine
Materials Technology Program, NASA, HITEMP, 19-1 to 19-16 (1992).

Tressler, R.E., Langesiepen, R.A., and Bradt, R.C., “Surface-Finish Effects on Strength-
vs-Grain-Site Relations in Polycrystalline A1203,” J. Amer. Ceram. Soc., Vol. 57,
227-228 (1974).

Tsuchimoto, T., and Tokuyama, T., Radiation Effects, Vol. 6, 121 (1970).

Watchman Jr., J .B., Tefft, W.E., Lam Jr., D.G., “Elastic constants of synthetic single
crystal corundum at room temperature,” J. Res. Nat. Bur.Stds., Vol. 64A, 213-
228 (1960).

Wiederhom, S. M., “Fracture of Sapphire,” J. Amer. Ceram. Soc., Vol. 52, 485-491
(1969).

Wiederhom, S. M., Hockey, B. J ., and Robert, D. B, “Effect of Temperature on the
Fracture of Sapphire,” Philog. Mag., Vol. 28, 783 - 796 (1978).

Wiederhom, S. M., “Brittle Fracture and Toughening Mechanisms in Ceramics,” Ann.
Rev. Mater. Sci., Vol. 14, 373 - 403 (1984).

Yee, J ., Kroger, F. A., “Measurements of Electromotive Force in A1203 - Pitfalls and
Results,” J. Amer. Ceram. Soc., Vol. 56, 189-191 (1973).

Ziegler, J .F., Biersack, J.F., and Littmark, U., W,
Pergamon Press, New York (1985).