. , .

 

LI E; RARY This is to certify tnat the
Michigan State dissertation entitled
University

 

 

 

PROCESSING AND CHARACATERIZATION OF PHASE
BOUNDARIES IN CERAMIC AND METALLIC MATERIALS

presented by

Liang Zeng

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Materials Science &

Engineering

 

 

 

 

Major Profe or’s Signature
June 6, 2003

Date

MSU is an Affirmative Action/Equal Opportunity Institution

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c-JCIFICJDateDuepes‘sz

PROCESSING AND CHARACTERIZATION OF PHASE
BOUNDARIES IN CERAMIC AND METALLIC MATERIALS
By

Liang Zeng

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Chemical Engineering and Materials Science

2003

ABSTRACT
Processing and Characterization of Phase boundaries in Ceramic and Metallic
Materials
By
Liang Zeng

The goal of this dissertation work was to explore and describe advanced
characterization of novel materials processing. These characterizations were carried out
using scanning and transmission electron microscopy (SEM and TEM), and X-ray
diffraction techniques. The materials studied included ceramics and metallic materials.

The first part of this dissertation focuses on the processing, and the resulting interfacial
microstructure of ceramics joined using spin-on interlayers. SEM, TEM, and indentation
tests were used to investigate the interfacial microstructural and mechanical property
evolution of polycrystalline zirconia bonded to glass ceramic MaCorm, and polycrystalline
alumina to single crystal alumina. Interlayer assisted specimens were joined using a thin
amorphous silica interlayer. This interlayer was produced by spin coating an organic based
silica bond material precursor and curing at 200°C, followed by joining in a microwave
cavity or conventional electric furnace. Experimental results indicate that in the joining of
the zirconia and MaCorTM no significant interfacial microstructural and mechanical property
differences developed between materials joined either with or without interlayers, due to the
glassy nature of MaCorTM. The bond interface was non—planar, as a result of the strong
wetting of MaCorTM and silica and dissolution of the zirconia. However, without the aid of
a silica interlayer, sapphire and 98% polycrystalline alumina failed to join under the

experimental conditions under this study. A variety of interfacial morphologies have been

observed, including amorphous regions, fine crystalline alumina, and intimate contact
between the sapphire and polycrystalline alumina. In addition, the evolution of the joining
process from the initial sputter-cure to the final joining state and joining mechanisms were
characterized.

The second part of this dissertation focused on the effects of working and heat
treatment on microstructure, texture, phase boundary movement, and mechanical
property evolution in Ti-6Al-4V wire. The as—received wire consisted of equilibrium or
and metastable B phases and had a moderately strong fiber texture with prism plane
normals aligned with the wire axis. The wire was worked by extrusion, solution heat-
treatrnent and water quenching, and aging. The extrusion process strengthened the as-
received texture. After solutionization and quenching, microstructural observations
showed the presence of many needlelike martensitic platelets in the prior B phase regions.
Texture analysis revealed that a secondary fiber with basal plane normals aligned with the
wire axis emerged at the expense of the initial texture, indicating that highly preferred
phase boundary motion (variant selection) occurred during the B—mt transformation. The
strength of the variant selection consistently increased with solutionization temperature
and time. In addition, the effects of dislocation type and density on variant selections
were further investigated. This implies that strategic prior deformation and heat
treatment can be exploited to design the resulting texture and microstructure and

consequently optimize the properties of titanium products.

 

To my wife and my parents!

iv

ACKNOWLEDGMENTS

I would like to thank my advisers, Professor Martin A. Crimp and Professor
Thomas R. Bieler for providing me the opportunity to perform this research and to
continue my graduate studies. Their continuous guidance, advice, and support were
invaluable to the completion of this work. Without their support and patience, this work
would not have been possible.

My thanks are extended to my guidance committee members Dr. Case, and Dr.
Velbel for being my committee members. In particular, special thanks should go to
Professor Case for his instruction, guidance and encouragement to complete this research.

The achievements of this work are also based on the help from the senior colleagues
in our research group. I would like to appreciate the help of Hong Geng, John Heckman,
Benjamin Simkin, Adwait Telang, and Jong-Gi Lee.

Many thanks go to Mr. Kenneth Singleton for the precious time he committed to
proofreading of my dissertation.

I gratefully thank parents who gave me their endless love, care and support. Finally,
I wish to thank my wife, Yanmei Li for her love and understanding. Especially, without
my wife’s patience and encouragement, I could not have completed this work.

Finally, I sincerely appreciate the support provided by the Composite Materials and
Structures Center at Michigan State University and Fairchild Fasteners, City of Industry,
CA, and Teaching assistant support from the Department which supplemented my

research studies.

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

PART I:

XII

INTERFACIAL MICROSTRUCTURE OF CERAMICS JOINED USING SPIN-

ON INTERLAYERS
Chapter I Introduction
1.1 Cerarrrics and Ceramic Processing
1.2 Joining ceramics
1.2.1 Joining without the use of interlayers
1.2.1.1 Diffusion Bonding
1.2.2 Joining with interlayers
1.2.2.1 Brazing
1.2.2.2 Transient liquid phase bonding
1.2.2.3 Joining with spin-on interlayers
1.2.3 Microwave heating and joining
1.2.3.1 Microwave processing and microwave effect
1.2.3.2 Interaction between microwaves and materials
1.2.3.3 Microwave joining
1.3 Indentation tests
1.4 Characterization techniques
1.5 Research strategy

Chapter II Experimental procedures
2.1 Goals and objectives of the proposed research
2.2 Materials and spin-on interlayers used in the current research
2.2.1 Joining materials
2.2.2 spin-on interlayers
2.3 Joining
2.3.1 Joining procedures
2.3.2 Materials joining systems
2.3.2.1 joining of crystalline materials with non-crystalline materials

2.3.2.2 Microwave and conventional joining of zirconia and MaCorTM

2.3.2.3 Joining of crystalline ceramics
2.3.2.4 Joining of polycrystalline alumina and sapphire
2.4 Characterizations
2.4.1 Optical microscopy and SEM
2.4.1.1 Sample preparations
2.4.1.2 Characterization techniques
2.4.2 TEM
2.4.2.1 TEM sample preparation
2.4.2.2 TEM Characterization

vi

WQOMMWWNH

2.5 Mechanical property of the interface 38

Chapter HI Experimental results 45
3.1 Introduction 45
3.2 As-received materials 45

3.2.1 zirconia and MaCorTM 45

3.2.2 Polycrystalline alumina 46

3.2.2.1 Interlayer thickness 46

3.3 Joining of crystalline ceramics with non-crystalline ceramics 47

3.3.1 Microwave joining of zirconia/MaCorTM 47
3.3.1.1 The interfacial microstructure of MaCorTM/zirconia joined at 1020°C

for 20 minutes using a spin-on interlayer 47

3.3.1.2 Indentation tests 49

3.3.2 Conventional joining of zirconia/MaCorTM 50

3.3.2.1 Effect of heating mode on the joining of zirconia and MaCorTM 51
3.3.2.2 The interfacial microstructure of MaCorTM/zirconia joined at 1070°C

for 60 minutes using a spin-on interlayer 52
3.3.2.2.1 Indentation tests 53

3.3.2.3 The interfacial microstructure of MaCorTM/zirconia joined at 1070°C
for 120 rrrinutes using a spin-on interlayer 53

3.3.2.4 The interfacial rrricrostructure of MaCorTM/zirconia joined at 1070°C
for 60 rrrinutes without the use of a spin-on interlayer 54
3.3.2.4.1 Indentation tests 55
3.3.2.5 Joining of MaCorTM/MaCorTM and zirconia/zirconia 55
3.4 Joining of crystalline ceramics with crystalline ceramics 56
3.4.1 Effects of heating mode on the joining of alumina/alumina 57

3.4.2 The interfacial microstructure of polycrystalline alumina/alumina joined at
1340°C for 20 minutes 57

3.4.3 The interfacial microstructure of polycrystalline zirconia/alumina joined at
1500°C for 20 nrinutes 58
3.4.4 Joining of sapphire and polycrystalline alurrrina 58
3.4.4.1 Joining of sapphire and 98% polycrystalline alumina 59
3.4.4.2 Joining of sapphire and 96% polycrystalline alumina 61
3.4.4.3 Indentation tests 63
Chapter IV Discussion and conclusions 97
4.1 Spin-on interlayer joining mechanisms 97
4.1.1 Surface and interface diffusions 97
4.1.2 Interfacial eutectic reactions 99
4.1.2.1 Silica and MaCorTM 100
4.1.2.2 Silica and alumina 100
4.1.2.2.1 Effect of impurities 101
4.1.3 Wetting and dissolutions 101
4.1.4 Interface formation 103
4.1.5 Phase development at the join interface 104

4.1.5.1 Joining of zirconia/MaCorTM 105

vii

4.1.5.2 Joining of sapphire and polycrystalline alumina 106

4.2 Microwave vs. conventional joining 106
4.2.1 Joining zirconia and MaCorTM 107
4.2.2 Joining of 98% polycrystalline alumina and alumina 107
4.2.3 Microwave joining 108

4.3 Joining without the use of interlayers 110
4.3.1 Joining of sapphire and polycrystalline alumina 111
4.3.2 Joining of zirconia and MaCor 112

4.4 Effect of silica spin-on interlayers 113
4.4.1 Joining involving glassy ceramics 113

4.4.2 Joining involving crystalline ceramics with little or no amorphous phase 113

Chapter V Conclusions and Future Studies 120
5.1 Summary and conclusions 120
5.2 Future studies ' 122
References 124
PART II

EFFECTS OF WORKING AND HEAT TREATMENT ON
MICROSTRUCTURAL AND CRYSTALLOGRAPHIC TEXTURE EVOLUTION
IN TI-6AL-4V WIRE

Chapter I Introduction 131
1.1 Introduction 131
1.2 Microstructure evolution in Ti-6Al-4V 132

1.2.1 Phase transformations and heat treatments 132
1.2.2 Crystallography of major phases in Ti-6Al-4V 136
1.2.3 Solution quenching and aging (SQA) 136
1.2.4 Deformation and processing 139
1.2.4.1 Deformation mode of or phase 139

1.2.4.2 Deformation mode of [3 phase 142

1.3 Texture evolution in Ti-6Al-4V 143
1.3.1 Deformation and recrystallization texture 144
1.3.2 Transformation texture and variant selections 145

1.4 Research strategy 151

Chapter II Experimental procedures 164
2.1 Research goals 164
2.2 Materials and fabrication 165

2.2.1 Materials and processing for the Crystallographic Texture & Microstructure
'(CTM) project 165

2.2.2 Materials and processing for the SQA project 166
2.2.3 Materials and fabrications for the Variant Selection (VS) project 167

2.3 Mechanical tests 169

2.3.1 Compression tests 169

viii

2.3.2 Micro-hardness tests 169

2.4 Optical microscopy and scanning electron microscopy (SEM) 170
2.5 TEM sample preparation and analysis 171
2.6 X—ray techniques 171
2.6.1. X-ray diffraction techniques 172
2.6.2. Texture 173
Chapter 111 Effects of working and heat treatment on microstructural evolution and
crystallographic texture of a and (1” phases in Ti-6Al-4V 182

3.1 Experiment results 182
3.1.1 As—received wire (B, M, and E) 183
3.1.2 Extruded specimens (Bex, Mex, and Eex) 184
3.1.3 Quenching (M-st20 & Mex-st20 and M-st60 & Mex-st60) 185
3.1.4 Ageing (M-st20-a & Mex-st20-a and M-st60-a & Mex-st60-a) 186
3.1.5 Crystal lattice measurements 188

3.2 Discussion 189
3.2.1 Phase transformations in Ti-6Al-4V 189
3.2.2 90° texture orientation rotation 191
3.2.2.1 Variant selections 191

3.2.2.2 Stress 192

3.2.2.3 Texture of high temperature B phase 194
3.2.2.4 Variant selections and activeness of slip system 194

3.3 Conclusions 196
Chapter IV Effects of solution temperature and time on crystallographic texture
and mechanical properties in Ti-6Al-4V wire 213

4.1 Introduction 214
4.2 Experiment results 215
4.2.1 Texture of as-received and extruded specimens 215
4.2.2 Texture of solution treated, quenched, and aged (SQA) specimens 216
4.2.3 Microstructure of SQA specimens 217
4.2.4 X-ray diffraction of serially sectioned SQA specimens 218
4.2.5 Mechanical properties of SQA specimens 219
4.3 Discussion 220
4.3.1 Variant selections and mechanical properties 221
4.3.2 Grain growth behaviors 222
4.3.3 Dislocation and ore-96 transformation 224
4.3.4 Model for the B—aa transformation 226

4.4 Conclusion 228
Chapter V Variant selections during aHB transformations 241
5.1 Introduction 242
5.2 Experiment results 243
5.2.1 Texture of as-received and annealed specimens 243
5.2.2 Texture of solutionized specimens after deformation 244

5.3 Discussion 244

ix

 

 

5.3.1 Effects of Dislocations
5.3.2 Deformation modes and dislocations
5.3.3 a—>B transformation and dislocation-interface interactions
5.3.4 Slip activity and dislocation reaction mechanism
5.4 Conclusion

References

245
246
246
249
255

264

 

LIST OF TABLES

PART I:
INTERFACIAL MICROSTRUCTURE OF CERANIICS J OINED USING SPIN-
ON INTERLAYERS

Table 1.1 shows the comparison of microwave pmf and other thermochemical driving
forces for ionic transport in NaCl. (p. 27)

Table 2.1 the experimental conditions and results for the joining of zirconia/MaCorTM in
the current research. (p. 40)

Table 2.2 outlines the experimental conditions for the joining of 3 mol% yttria-
zirconia/polycrystalline alumina, and 98% polycrystalline alumina/98%
polycrystalline alumina. (p. 41)

Table 2.3 outlines the experimental conditions and results for the joining of 98% poly-
crystalline alumina/sapphire via conventional heating. (p. 41)

Table 3.1 The experimental condition and results for ceramics/ceramics joining via
conventional heating in the present study. Please note, all experiments were
done twice to verify joining results. (p. 65)

Table 3.2 outlines the experimental conditions and results for the joining of 3 mol%
yttria-zirconia/polycrystalline alumina, and 98% polycrystalline alumina/98%
polycrystalline alumina. Please note, all experiments were done twice to
verify joining results. (p. 66)

Table 3.3 The experimental conditions and results for the joining of sapphire and
polycrystalline alumina via conventional heating in the current research.
Please note, all experiments were done twice to verify joining results. (p. 66)

Table 4.1 The dielectric properties of various cerarrrics used in the joining of zirconia
and MaCorTM using a silica interlayer. (p. 119)

Table 4.2 The dielectric properties of polycrystalline alumina used in the joining of
alumina and alumina using a silica interlayer at 2.45 GHz. (p. 119)

PART II

EFFECTS OF WORKING AND HEAT TREATMENT ON
MICROSTRUCTURAL AND CRYSTALLOGRAPHIC TEXTURE EVOLUTION
IN TI-6AL-4V

Table 1.1 Main phase transformations in Ti-6Al-4V. (p. 154)

xi

Table 1.2 The (a) main crystallography relations of or, B, a’, or”, and (I) phase

transformations in titanium alloys, the (b) lattice principle strains for B—m’
and B—mt” transformations. (p. 155)

Table 1.3 The (a) relationship between microstructure characteristics, mechanical

Table 1.4

Table 2.1

Table 2.2

Table 3.1

properties, and processing conditions [Boswell 1995, Boyer-B, C, E 1994],
and the (b) diffusion coefficients in Titanium alloys. (p. 156)

Slip and twin modes in or phase Titanium. (p. 157)

Chemical compositions of as-received Ti-6Al-4V alloys in weight percentage.
(p. 176)

Detailed nomenclature used in the Variant Selection (VS) project to evaluate

the influence of different deformation mode on variant selections of (at—)B—mt
transformation cycle. (p. 176)

Specimens examined using X-ray diffractometer, texture measurements, and
TEM technologies. Note, M stands for middle of the coil, ex for extruded, sq
for solutionization and quenching, a for aging. (p. 197)

Table 3.2 lattice parameters of Ti-6Al-4V specimens at various experimental processing

conditions. (p. 208)

Table 4.1 Lattice strain or linear dimensional changes in B—mt transformation. (p. 229)

xii

PART I:

LIST OF FIGURES

INTERFACIAL MICROSTRUCTURE OF CERAMICS JOINED USING SPIN-
ON INTERLAYERS

Figure 1.1

Figure 1.2

Schematic representation of direct ceramic-ceramic joining process without
the use of interlayers at elevated joining temperature and pressure [Redrawn
from Esposito 1998]. At the first stage of joining process (a), the interface
between the two polished ceramic parts can be seen as a macroscopic
inhomogeneity since no intergranular amorphous phase is present on the
surface of the grains at the interface. Driven by thermodynarrric capillary
force (b), the intergranular amorphous phase flows to the interface. As the
process continues (c), liquid phase flow, grain boundary sliding and grain
rotation can be activated to minimize the macroscopic inhomogeneity
between joint and base ceramics. Finally ((1), grain solution and
precipitation, grain growth and grain bridging occur, leading to
indistinguishable interface from the base ceramics. (p. 21)

Schematic representation of the sequence for diffusion bonding at a proper
joining temperature and pressure [after http://www.twi.co.uk/j32k]. Initially
(a) two ceramic surfaces are placed in contact, (b) at least one of the
surfaces begins to deform under high temperature and stress, (c)
deformation continues, leading to void shrinkage, and (d) the formation of
interface or joint. (p. 22)

Figure 1.3 shows the phase development for Ale3-SiC composite brazed by Ag—Cu-Ti

Figure 1.4

alloy from (a) initial phase distribution to (b) final experimentally
determined complicated phase distribution at the joint after brazing [Redraw
from Zurbuchen 1999]. (p. 23)

shows the sequential stages of the TLPB process for the joining of B-
SiAlON/B-SiAlON ceramics using B-SiAlON glass-forming composite
adhesives (mixing of Si3N4, Y203, A1203, SiOz powders) [Redraw from
Walls 1992, Hanson 1998]. Prior to the joining, adhesive powders are
sprayed onto one of the base materials. Around 1400°C, oxide components
react to form a Y-Si-Al-O liquid phase, leading to densification and
sintering of interlayer. At approximately 1600°C, the Si3N4 quickly
dissolves into the liquid, and alters the composition to B-SiAlON. At the
same time, the newly formed B-SiAlON grains grow and form an
interlocking network across the joint, leading to an indistinct interface
between the base B-SiAlON and the newly formed B-SiAlON materials. The
final microstructure composed of the base B-SiAlON and the newly formed
fi-SiAlON materials with an almost indistinct interface. (p. 24)

xiii

Figure 1.5

Figure 1.6

Figure 1.7

Figure 2.1

Figure 2.2

Figure 2.2

Figure 2.4

Figure 3.1

Figure 3.2

Figure 3.3

Heating patterns in (a) conventional and (b) microwave heating furnaces
[Redraw from Sutton 1992]. (p. 25)

shows the activation energy for (a) the microwave and conventional sintered
A1203 doped with MgO [Redraw from Janney 1988] and (b) the microwave
and conventional grain growth in A1203 doped with MgO [Redraw from
Janney 1991]. (p. 26)

Interactions of microwaves with different materials [Redraw from Sutton
1992]. (p. 27)

Schematic representation of spin-on interlayer joining method used in the
current study. (p. 42)

Representation of the ceramic joining set-up used in the current study. The
joined specimens are loaded using a small (20~85g) polycrystalline alumina
deadweight, to prevent slippage of the specimens. (p. 42)

Schematic representation of the cross-sectional TEM sample preparation
procedure used in the current study [after Geng 2003]. (p. 43)

shows (a) the indentation spacing and the (b) scheme of the test to check the
strength of the interface. (p. 44)

An TEM images of the (a) as-received glass-ceramics MaCorTM and (b)
sintered 3 mol% yttria-zirconia specimens. (p. 67)

A cross-sectional TEM image of the sintered, finely polished and cleaned 3
mol% yttria-zirconia specimens prior to coating showing the surface grains
morphology. (p. 68)

TEM images of the as-received rrricrostructures of 98% (a) and 96% (b)
polycrystalline alumina. The commercial 96% polycrystalline alumina (b)
consists of platelike or elongated OL-A1203 grains, with approximately 20%
intergranular grain boundary phase. The diffraction pattern of the
intergranular grain boundary phase reveals an amorphous structure,
consistent with that expected for the alurrrinosilicate glass [Brydson 1998].
However the 98% alumina (a) contains equiaxed (Jr-A1203 grains with little
grain boundary phase. (p. 69)

Figure 3.4 SEM image of silica interlayer showing the interlayer has a thickness in the

range of 250~320 nm thick after spinning at 3000 rpm and curing at 200°C
for 20 minutes. (p. 70)

Figure 3.5 An SEM micrograph of the microwave joined zirconia/MaCorTM showing the

bond interface. At the gross scale the interface is straight, locally the

xiv

zirconia contains grain boundary grooves. No interfacial phase(s) were
observed. (p. 70)

Figure 3.6 Low magnification TEM image of the interface between the 3 mol% yttria-
zirconia and the MaCorTM joined at 1020°C for 20 minutes using microwave
heating. Differential ion milling in the specimens results in the MaCorTM
being thinned back with respect to the zirconia, resulting in occasional
perforation of the MaCorTM specimen near the interface. (p. 71)

Figure 3.7 TEM image of the MaCorTM near the interface. Inserts show (a) a diffraction
pattern from a single mica plate and (b) a diffraction pattern from matrix
areas between the mica plates. (p. 71)

Figure 3.8 TEM image of the interface between the 3 mol% yttria zirconia and the
MaCorTM showing that the zirconia has been strongly wetted and dissolved
along grains boundaries, resulting in a non-planer interface. (p. 72)

Figure 3.9 A TEM image of the rrricrostructure of the zirconia in a joined 3 mol% yttria-
zirconia/MaCorTM specimen at a location at about 10 um away from the
interface. The microstructure of the zirconia at this location is similar to

that of the zirconia rrricrostructure near the interface (Figure 3.6). (p. 72)

Figure 3.10 A TEM image of the MaCorTM rrrierostructure in a joined 3 mol% yttria-
zirconia/MaCorTM specimen at a location at about 100 um away from the
interface. The microstructure at this location is similar to the microstructure
of the MaCorTM near the interface (Figure 3.5). Due to a range of mica
platelet orientations, the apparent dimensions differ from platelet to platelet.

(p. 73)

Figure 3.11 A Vickers indentation testing shows that the radial crack near the microwave

joined zirconia/MaCorTM interface with (a) and without (b) observable
porosity in the MaCorTM has different length on the vertical and horizontal
directions, and (c) propagates across the interface from zirconia into the

MaCor"M without deflection. (p. 74)

Figure 3.12 An SEM micrograph of the zirconia/MaCorTM joined at 1070°C for 60
rrrinutes using a silica spin-on interlayer showing the bond interface. (p. 75)

Figure 3.13 A TEM image the zirconia/MaCorTM joined at 1070°C for 60 rrrinutes using
a silica spin-on interlayer. Diffraction pattern from one of the rounded

plates near the interface revealed the mica structure (a), oriented to [542].
Diffraction pattern from the matrix of MaCorTM revealed an amorphous
structure (b). (p. 75)

XV

Figure 3.14 The interfacial rrricrostructure of the zirconia/MaCorTM joined at 1070°C for
60 minutes using a silica spin-on interla er showing intimate bonding
between zirconia and the matrix of MaCor . (p. 76)

Figure 3.15 A TEM image of the bulk MaCorTM microstructure. The center of the image
is about 5 pm away from the zirconia/MaCorTM interface. The specimens
were joined at 1070°C for 60 minutes using a spin-on interlayer. (p. 76)

Figure 3.16 A TEM image of the bulk zirconia microstructure. The center of the image
is about 5 pm away from the zirconia/MaCorTM interface. The diffraction
pattern is consistent with a two phases partially stabilized zirconia structure.

(p. 77)

Figure 3.17 A Vickers indentation testing shows that the radial crack near the joined
zirconia/MaCorTM interface propagates across the interface from zirconia
into the MaCorTM without deflection. (p. 77)

Figure 3.18 TEM brightfield micrograph of the zirconia/MaCorTM interface joined at
1070°C for 120 minutes usin a silica spin-on interlayer shows mica plates
incorporated into the MaCor Izirconia interface. (p. 78)

Figure 3.19 The interfacial microstructure of MaCorTM and zirconia joined at 1070°C for
120 minutes using a silica spin-on interlayer. Insert showing that selected
area diffraction pattern of the distorted plates continue to display the mica
structure. (p. 78)

Figure 3.20 An SEM micrograph of the zirconia/MaCorTM joined at 1070°C for 60
rrrinutes without the use of a silica interlayer showing the bond interface. (p.
79)

Figure 3.21 A TEM image of the zirconia and MaCorTM joined at 1070°C for 60 minutes
without the use of a silica interlayer. No interfacial phase(s) were observed
at the interface. (p. 79)

Figure 3.22 A TEM micrograph of MaCorTM (unjoined control) that was heated at
1070°C for 60 minutes. Insert shows the diffraction pattern from a single
rrrica platelet ([241] orientation). MaCorTM heated at 1070°C for 120
minutes showed similar microstructure. (p. 80)

Figure 3.23 An SEM image of the Vickers indentation crack on the zirconia/MaCorTM
joined at 1070°C for 60 rrrinutes without the use of a silica spin-on interlayer
in the zirconia side showing the radial crack propagates across the interface
without deflection. (p. 80)

xvi

Figure 3.24 The interfacial microstructure of MaCorTM and MaCorTM joined at 1070°C
for 10 minutes using a silica interlayer showing an amorphous interfacial
region, as indicated by the inset diffraction pattern. (p. 81)

Figure 3.25 TEM image of the MaCorTM/MaCorTM interface joined at 1070°C for 15
minutes using a silica interlayer showing that mica platelets obliquely cross
the whole joint region, making the original interface imperceptible. (p. 83)

Figure 3.26 The interfacial nricrostructure of the microwave joined 98% polycrystalline
alumina/98% polycrystalline alumina using a silica interlayer showing that
the equiaxed alumina grains cross the whole joint region, making the
interface very difficult to distinguish from the base material. (p. 84)

Figure 3.27 Low magnification TEM image of the interface between the 3 mol% yttria-

zirconia and the polycrystalline alumina joined at 1500°C for 20 rrrinutes
using microwave heating. (p. 85)

Figure 3.28 A TEM image of the interface between the 3 mol% yttria-zirconia and the
polycrystalline alumina showing that no observable reaction layers or
interfacial phases in the join region. (p. 85)

Figure 3.29 Optical microscopy images of sapphire/98% polycrystalline alumina joined
at (a) 1400°C for 180 rrrinutes, (b) 1450°C for 180 minutes, and (c, d)
1475°C for 180 rrrinutes using silica interlayer that illustrate two different
interfacial microstructures on the same joined bulk specimen. At some
locations, the reaction layer is approximately 3 pm thick (c) while at other
location reaction layer has a thickness only around 1 pm ((1). (p. 87)

Figure 3.30 TEM images of sapphire/98% polycrystalline alumina joined at 1475°C for
180 rrrinutes revealed two distinct interfacial microstructures at the joint. At
some locations (a), the interfacial phase has a thickness around 0.5~1 um,
while at other locations, a fine distribution of alumina crystalline phase
occurred at the joint (b) with indexed diffraction pattern (c). (p. 89)

Figure 3.31 (a) Optical microscopy and (b) TEM images of the sapphire/98%

polycrystalline alumina joined at 1475°C for 180+300 minutes using a silica
spin-on interlayer. (p. 90)

Figure 3.32 SEM images of the sapphirel96% polycrystalline alumina joined at 1475°C
for 180 minutes (a) with and (b) without the use of a silica spin-on
interlayer. (p. 91)

Figure 3.33 A TEM image of sapphire/96% polycrystalline alumina joined at 1475°C for
180 minutes using a silica interlayer. Insert shows the diffraction pattern
from the amorphous interlayer. (p. 92)

xvii

Figure 3.34 (a) Secondary and (b) backseattered electron SEM images of the
sapphire/96% polycrystalline alumina joined at 1475°C for 180+300
minutes using a silica spin-on interlayer. The bright contrast in the
polycrystalline alumina in the BSEM image (b) corresponds to grain
boundary phases, the gray contrast corresponds to or-A1203 phase, and the
black contrast corresponds to porosity. (p. 93)

Figure 3.35 A TEM image of the sapphire/96% polycrystalline alumina joined at 1475°C
for 180+300 rrrinutes using a silica spin-on interlayer. (p. 94)

Figure 3.36 An optical microscopy (a) and SEM (b) images of the Vickers indentation
crack on the sapphire/polycrystalline alumina joined at 1450°C for 180
minutes using a silica spin-on interlayer in the alumina side neat the joint
showing the radial crack propagates across the interface from polycrystalline
alumina into the sapphire without deflection. (p. 96)

Figure 3.37 An optical microscopy image of the Vickers indentation crack on the

sapphire/polycrystalline alurrrina joined at 1450°C for 180+300 minutes
using a silica spin-on interlayer in the alumina side neat the joint. (p. 95)

Figure 4.1 The schematic representation of the three mass transport mechanisms for
interface formation due to (a) surface and volume diffusion from surface
source to neck, (b) diffusion along the bond interface, and (c) bulk
deformation mechanism after [Derby 1982, 1984]. The arrows denote the
mass diffusion direction, while the dotted and bold lines represent the
interfacial void geometry before and after diffusions, respectively. (p. 116)

Figure 4.2 Equilibrium phase diagram of SiOz-KZO-Aleg [redrawn from Schairper
1954]. (p. 117)

Figure 4.3 The expected thermal groove profile shapes due to (a) Interfacial reactions
with groove depth d =1.13m(At)°'5 , (b) surface diffusion with depth

d =0.973m(B_,t)°'25 and width s =4.6(B,t)°'25, and (c) volume diffusion
with depth d =1.01m(19,,t)°-33 and width 5' = 5(B.t)°'” [After Mullins 1957,

2
1958, 1960]. Constants A=L—C°Z‘iQ—, L=—J—,Bs =m, and
kT c—co kT
B, = illigi’g , where m is the slope of the solid/liquid interface at the

groove root; Co is the initial concentration with a planar interface; C is the
concentration; 82 is the molecular volume and weight; M is the molecular
weight; x is the solubility of solid in the liquid; Dv and Ds are the volume
and surface diffusion coefficients, respectively; 8 is interfacial width; J is

xviii

PART II
EFFECTS

the flux, 7,1 is the solid/liquid interfacial energy, and ygb is the grain
boundary energy. (p. 118)

OF WORKING AND HEAT TREATNIENT ON

NHCROSTRUCTURAL AND CRYSTALLOGRAPHIC TEXTURE EVOLUTION
IN TI-6AL-4V

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4

HymLS

Figure 1.6

(a) Beta isomorphous phase diagram for Ti-6Al-4V showing three regions
that undergo different transformation behaviors [After Koul 1970] and (b) a
schematic diagram showing different microstructures under thermal
treatment with three different cooling rates [Maykuth 1971, Boyer-B 1994].

(p.158)

The lattice correspondence for B, or’ and or” transformations [From Davis
1979]. (p.159)

Primary B phase volume with different oxygen content as a function of
temperature [After Kahveci 1986]. (p. 159)

Major (a) slip and (b) twinning planes in or phase Titanium alloys. (p. 160)

(0002) pole figures of Ti-6Al-4V starting from initially randomly orientated
specimen rolled at 800°C followed by water quenching (a) and after further
annealing at 800°C for 1 hour (b) results in minor texture variations [After
Peters 1980]. Two major (002) pole intensity maxim are observed. One
major component, transverse texture, belongs to basal plane normals in the
TD aligned perpendicular to the SN and planes parallel to the RD. The
other one splits RD texture belongs to basal planes in the RD align about
20° with the SN. The schematic representation of cold rolling deformation (c)
and recrystallization texture (d) in pure Titanium sheet [After Keeler 1956],
results in slightly different crystal rotations of the basal planes, indicated by y
and 8. (p. 161)

Schematically represent the a—eB—m transformation texture cycle from (a)
initial or single variant with crystal orientation and the (002) pole figure to
the possible (b) (c) (d) variants of transformation B texture with the
corresponding with crystal orientation between or and B and the (110) pole
figures. (e) (f) (g) (h) (i) (j) are the possible six pole figures of room
temperature or phase variants from single variant of high temperature B
phase (c) after the B—>0t transformation. Due to crystal symmetry, only
three variant of high temperature B phase and six variant of room
temperature or phase are shown. (p. 162)

xix

Figure 1.7

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

(a) Relationship between the slip systems in {110}<111> and {112}<110>
and the preferred or variants and the simulated (b) and experimental (c) or

phase texture after or—>B—90t the transformation cycle with 75% reduction
deformed specimen [After Gey 1997]. (p. 163)

Schematic representation of typical industrial manufacturing flow chart. (p.
177)

Schematic representations of the specimen preparation procedures and
nomenclatures, from as-received wire, extrusion, to solution and aging
thermal treatment. (p. 178)

Schematic representations of the sample orientation for serial sectioning X-
ray diffraction techniques used in the current study. (p. 179)

Schematic representation of the specimen preparation procedures, from as-
received wire, to solution treated conditions. See table 2.2 for detailed
nomenclatures. The arrow mark referred to specimens’ orientations for
texture measurements. (p. 179)

Schematic representation of the specimen preparation procedures, from as-
received wire, deformed state, to solution treated conditions. All the texture
measurements were taken on the exact same surface plane after different
processing and deformation conditions, from as received, deformed, to heat-
treatment to monitor the surface texture evolutions. See table 2.2 for
detailed nomenclatures. The arrow mark referred to specimens’ orientations
for texture measurements. (p. 180)

Representation of the experimental set up used for normal compression test
(a) and related specimen orientation (b). The two cylinders inside the wire
represent the actual orientation corresponding to the original as-received
wire and how they were sectioned using EDM. The cylinder with
longitudinal direction (labeled as L) means the as-received wire axis is
parallel to the cylinder axis, while the cylinder in transverse direction
(labeled as T) means the as-received wire axis is perpendicular to the
cylinder axis. The compression axis was always parallel to the specimens’
axes. (p. 181)

Schematic representations of the sample orientation for (3) X-ray diffraction
techniques and (b) texture measurements used in the current study. (p. 181)

Crystal phases and their coherency relationship to each other. (p. 198)
shows the primary or phase pole figures of (a) the as-received middle of wire

M with wire axis out of page and (b) the extruded Mex with wire axis out of
page sectioned from middle of as-received wire, and (c) the corresponding

XX

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

crystal orientations with specimens. The crystal prism plane normal

direction, (1010)a, is aligned with the wire axis, and the (0001),I plane
normal direction is in the radial direction in the as-recived and extruded
specimens. (p. 200)

The overall texture variations from as-received wire through extrusion,
solutionization and quenching, and aging. (p. 201)

X-ray diffraction patterns of the wire M cut from middle of as-received coil
showing primary or and B phases, and as-quenched M—sq20 specimens

showing the peaks broadening due to the formation of or’ martensites. (p.
201)

shows the TEM image of as-received middle of wire M and the related
indexed diffraction pattern from primary B phase. (p. 202)

TEM rrricrographs of (a) the primary B phase in the extruded Mex specimens
showing high dislocation densities and (b) the primary or grains in another
thin foil. Inserts show the corresponding diffraction pattern from (a)
primary B and (b) 0!. phases. (p. 203)

shows the or phase pole figures of (a) the as—quenched middle of wire Mex-
sq60, (b) the aged Mex-sq60-a with wire axis out of page sectioned from
middle of as-received wire, and (c) the corresponding crystal orientations
with specimens. The pole figures indicated that a new fiber texture with the
crystal basal plane (0002),ll normal direction aligned with the wire axis and

the (1010)o prism plane normal direction perpendicular to the wire axis
after solutionization and quenching. This introduction of a new orientation
weakened the prior fiber texture having (0002),ll normal direction aligned
along radial directions. (p. 204)

Figure 3.8 shows (a) the TEM image of as—quenched M-sq60 sectioned from middle of

Figure 3.9

wire (b) the related diffraction pattern from primary B phase showing a

[111] B zone pattern and three or <1120> patterns superimposed, and (c) is
the schematic of (b). (p. 205)

X-ray diffraction patterns of the as-quenched M-sq60, M-sq60-a and M-
sq60-a (year later) specimens, showing the formation of orthorhombic 0t”
martensites and B phases due to artificial aging, and disappear of or”
martensites and B phases due to natural aging effect. Note, the M-sq60-a

and M-sq60-a (re-measured after a year later) are the specimen measured at
different time periods. (p. 206)

xxi

Figure 3.10 shows the TEM image of (a) the aged M-sq60-a specimens, and (b) the aged

M-sq60-a specimens one year later. Inserts show the corresponding
diffraction pattern from the corresponding indexed diffraction pattern from

primary B phase. (p. 207)

Figure 3.11 The lattice correspondence for B, and 01’ (or) transformations. (p. 209)

Figure 3.12 Schematic representation of internal stress build up between different

Figure 3.13

Figure 4.1

Figure 4.2

orientations of or grains, as well as with B grains upon cooling prior to
transformation to maintain compatibility due to large anisotropic thermal
expansion difference between the a and c-axis. Initially, (a) assuming these
two hexagonal or grains are well annealed at high temperature, and have a
90° mis-orientation. The (b) large anisotropic thermal contraction along c-
axes leads to the development of in plane or radial stress between or grains
upon cooling. In addition, if a B grain is located between these two or
grains, the in plane tensile stress could develop in the B grain due to cooling.
Note: the a-axes of the two or grains drawn are out of the page or radial
directions are in plane. (p. 210)

Schematic represent the possible (a) high temperature B phase texture, and
(b) the angles of (110)p with wire axis based on the Burgers relationship
[Burgers 1934] from a—>B, (c) high temperature B phase texture could lead
to the formation of or phase texture with the c-axis normal along the wire
axis upon quenching, and (d) the corresponding crystal orientations with
specimens. (p. 212)

The primary or phase (1010) and (0002) pole figures and schematic crystal
orientations of (a) as-received wire, (b) extruded specimens, and (b)
extruded specimens after solutionization at 925°C for 20 minutes and aging
(SQA) with wire axis out of the page. (p. 230)

The overall relationships (a) between major texture intensities and the
experiment solutionization conditions, and (b) the effects of solutionization
temperature and (c) time on the major texture intensities using statistical
analyzes via averaging the corresponding texture intensities of the same
temperature of different times and same times of different temperatures,
respectively. (p. 232)

Figure 4.3 TEM image of specimen after solution treated at 925°C for 2.4 ks and aging

 

at 500°C for 28.8 ks showing typical duplex microstructure, consisting of
equiaxed primary or and B grains. Diffraction patterns taken from one of the
transformed B grain regions reveals the Burgers orientation relationship of

<111>n||<11i0>a and (110)B||(0001)., between or and B phases, consistent
with other studies in bcc<-—>hcp transformation systems. (p. 232)

xxii

 

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 5.1

The overall relationships (3) between the grain sizes, prior B phase ratio, and
the experiment solutionization conditions, (b) the effects of solutionization
temperature on the grain sizes and prior B phase ratio using statistical
analyzes via averaging the corresponding grain sizes and prior B phase ratio
of the same temperature of different times, and (c) the average diffusion

distances of Ti, Al, V in or and B phases under the experimental
solutionization conditions. (p. 234)

shows the gradual sectioning of X-ray diffraction studies of specimens. The
(a) zigzag relationship between d-spacing of d(10I0)(l and dtooom. and (b) the
related mean, and standard deviation (sigma) of the duoiom and d(0001)a under
different solutionization conditions were identified. (p. 235)

The serial sectioning X-ray diffraction studies show, (a) the volume of hep
unit cell vs. processing conditions, (b) the variations of d spacings
dwoona/dnoima and duping/damp”, (c) the related mean and sigma of the
d(0001)a/d(1010)a and duoim/duoima with solutionization temperatures and
times. For comparison, the large variations of d spacings for as-received

and extruded wire suggested that macro-residual stress might cause the
lattice strain variations. (p. 237)

The overall (a) relationship between solutionization conditions and Young’s
Modulus and yield stress in the radial and wire directions, the (b) effects of

solutionization temperature and (c) time on the “1001131110511 properties using
statistical analysis. (p. 239)

Pole figures and schematic crystal orientations representations of the Ti-6Al-
4V plates showing that one single or phase variant can generate three or four
or. phase variants after ST or or<—>B transformation cycles in the a-I-B phase
field, including two perpendicular variants both have 90° nus-orientations
from prior orientations, and one variant having 30° nus-orientations from
prior orientation in the RD plane [After Moustahfid 1997]. (p. 240)

(a) reveals the overall relationships between major texture intensities and the

experiment conditions, (b) the experimental (1010).,, (0002),” and (110)B
pole figures with wire axis out of the page for specimen S40Q after
solutionization without prior annealing, and (c) the schematic 0t crystal
orientations. The levels of 90° nus-orientations from prior texture were
consistently lower in the prior annealed and followed by solutionization and
quenching specimens than that of as received one without annealing,
indicating the effects of defects on texture development. (p. 256)

xxiii

Figure 5.2 reveals the overall relationships between major texture intensities and the

Figure 5.3

Figure 5.4

Figure 5.5

Figure 5.6

experiment processing conditions after solutionization for 40 rrrinutes at

920°C. Different deformation, followed by the same heat treatment resulted
in different texture development. (p. 257)

showing (a) the (1010):, (0002)(,, and (110)3 pole figures of as-received
wire with wire axis out of the page, and the (b) schematic or crystal
orientations. The or phase consisted of one fiber texture with (0002)“ plane
normal along radial directions, The B phase consisted of a fiber texture with
the axisymmetric (110)B crystal orientations about the wire axis, and a few

of rotated or tilted (110)]; fiber texture aligned approximately 30° from the
wire axis. (p. 258)

showing the (1010),, (0002),, and (110)B pole figures with wire axis out of
the page of heat-treated specimens (a) L heat and (b) T heat after
solutionization for 40 minutes at 920°C, followed by quenching in sealed
quartz tube, and (c) the schematic (1 crystal orientations for both cases.
Note, the crystal orientation of (0001)“ and (110)p plane normal along wire
axis of T heat (deformed along transverse direction and heat treated) is
13.42 and 9.21 X random, respectively, which is 5.3 and 1.7 times higher
than the corresponding textures of specimens L heat deformed along

longitude direction and heat treated under the exact same conditions. (p.
260)

showing dislocations mutations based on the geometric and Burgers
relationship between or and B phases. Compression along the wire axis is
roughly out of the page. Compression in radial directions imposes
compressive stresses in directions roughly in the plane of the page. Neither

would activate much prism slip on (1010),, planes. The prism a

dislocations transform into [11 l](112)3 dislocations, but they cannot mutate
onto (110)B planes after or—>B transformation. However, either basal or

prismatic a screw dislocations could cross slip onto the two inclined
(101)B, (I01)B planes after (it—)B transformation. The c +a dislocations

could be directly transferred onto B phase (101)g, (101)5 planes, but would
have a large irrational Burgers vector. (p. 261)

shows the spatial construction of <100>, <110>, and <111> directions and
{100}, {110}, and {112} planes in bcc metal, and (b) planer representation for
projections of the ABCDX pyramids [After Zhi 1992]. In the ABCDX
pyramids, sides of hexahedron represents {100} planes, sides of the dashed
line pyramids represents {110} planes, and sides of the solids line pyramids
represents {112} planes. (p. 262)

xxiv

Figure 5.7 Schematic descriptions of effect of differential thermal expansion and
contraction of primary or grains (shaded boxes) on state of shear in B grains
during (a) heating and (b) cooling process. (p. 263)

XXV

Chapter I

Introduction

In recent years there has been rapid progress toward the development of new and
reliable ceramic joining techniques. Significant research in this area is being done on
ceramic joining because of the commercial importance of joining ceramics. Reliable
ceramic joining techniques would allow components to be joined from simpler parts that
have been pre-tested for flaws, thus maintaining high mechanical properties and
dimensional control, and lowering overall cost. The mechanical properties of joined
ceramics depend on the composition and structure of the interface, the joining technology
used, and reaction products that may be produced. Therefore, it is critical to have a
fundamental understanding of ceramic interfaces and interfacial reactions to further the
development of reliable ceramic joining methods. The first part of this project is an
attempt to investigate the interfacial rrricrostructure of ceramics joined using spin-on
interlayers, which has significant advantages over other joining approaches. Analysis
was carried out using optical, and scanning and transmission electron microscopy. The
mechanical properties of the interface were characterized using Vickers indentation
techniques to access the thermal residual stresses and interfacial strength. These
techniques were used to further characterize the relation between interfacial
microstructure and process variables such as time, temperature, and mechanical

properties of the interfaces.

1.1 Ceramics and Ceramic Processing

Ceramics range from silicates and oxides such as zirconia, titania and alumina, to
nonoxides such as nitrides, carbides, and borides of the transition elements. Ceramics are
attractive for numerous reasons because of their properties. They may be electrical
insulators, semiconductors, conductors, superconductors, heat conductors, or thermal
insulators. Ceramics have large potential for high temperature applications such as heat
engines or heat exchangers because they remain rigid at temperatures intolerable by
metals and superalloys [Santellal 1997, Schwartz 1990, Moorhead 1990, Woods 1985].

In industry, most ceramic fabrication processes begin with fine powders. Pressing
and slip casting are two major techniques used to form component shapes from ceramic
powders [Rieherson 1992]. Pressing is performed by placing the powders into a die and
applying pressure to obtain a green compact. The main limitation to this approach is that
the possible green compact shapes, such as rings or cylinders, are relatively simple
[Rieherson 1992]. Slip casting is performed by pouring particles that are suspended in
water into a porous mold, followed by drying and sintering. Although slip casting offers
more dimensional flexibility, it results in a relatively weak ceramic compact [Rieherson
1992]. In particular, the introduction of precise channels or holes within the slip cast parts is
problematic. Machining of cerarrrics is slow and expensive, making it impractical to

machine ceramic parts with complicated shapes from a ceramic billet.

One alternative approach to fabricate ceramics of complex geometry is to join
simpler subcomponents of densified ceramics to form the final component. This

approach has motivated a sustained level of research and development in ceramic joining

over the past decades. There are at least three reasons for joining ceramics: (1) to
assemble complex, multicomponent structures from simpler components of the same
materials, (2) to join dissimilar materials so as to produce a functional gradient in
material properties, and (3) to aid quality control by allowing defective sub-components
to be identified and removed early in manufacturing process [Johnson 1985, Hauth 1989,
Zdaniewski 1987, Shalz 1983]. Although improvement has been achieved in the
fabrication of ceramic materials that are reliably strong and tough, the successful
utilization of ceramics in many applications will depend on the ability to assemble or join
simple components into structures that will function effectively [Shalz 1993, Locatelli

1997].

1.2 Joining ceramics

In recent years there has been rapid progress toward the development of new and
reliable ceramic joining techniques. This work falls into a number of broad areas that
include: (1) fundamental studies of interfaces, (2) joining without the use of interlayers,
(3) joining using unconventional heating methods, such as microwave heating, (4)
diffusion bonding [Urena 1992, Cross 1993, Yano 1998], (5) brazing [Sandhage 1996,
Weldon, 1997], (6) the use of transient metal layers for joining [Shalz 1993, Gale 1999],
and (7) joining with spin-on interlayers [Seiber 1997, Lee 1997, Case, 1997, 1998]. This
list is not exhaustive and there are other ways to classify ceramic joining. However, the
objective of this chapter is to review ceramic joining and focus on the recent literature and
important findings related to the present research.

1.2.1 Joining without the use of interlayers

 

For direct ceramic-ceramic joining without the use of interlayer, adhesion and
bonding generally takes place through an intergranular amorphous phase present in the
base materials migrating to the joint [Binner 1998]; such intergranular phases can become
viscous at elevated temperatures. The driving force for this adhesion comes from the
tendency of the system to restore the equilibrium when a higher energy interface is
introduced [Esposito 1998]. The interface between two polished ceramic parts put in
contact can be viewed as a macroscopic inhomogeneity since no intergranular amorphous
phase is present on the surface of the grains at the interface. Adhesion is therefore driven
by the mass transport mechanisms that try to restore the equilibrium and minimize the
macroscopic inhomogeneity between interfacial region and base ceramics. The possible
mechanisms for bonding include liquid phase flow, grain boundary sliding and grain
rotation, grain solution and precipitation, grain bridging and grain growth [Nakamura
1997]. The principle step of direct ceramic-ceramic joining processes without the use of
interlayers is shown schematically in figure 1.1.

The presence of an intergranular phase appears to have two consequences: (1) it
provides an adhesive medium to hold the two joining pieces together under proper
temperature and pressure since the adhesion and chemical compatibility between
intergranular glassy phases and base ceramics is generally good, (2) it can allow the
grains near or at the interface in base materials to rotate and rearrange themselves at high
temperatures, which can lead to the strengthening of the joint [Walls 1992, Binner 1998].
This grain rearrangement is strongly dependent on the softening temperature of the glassy
phases, which in turn largely depends on the level of impurities, joining temperature and

pressure, and sintering aids already present in the ceramics. For a number of ceramics,

impurities and sintering aids are added to enhance the sintering densification rate. The
presence of these impurities and sintering aids can lead to significant intergranular
amorphous phases, which in turn degrade the high temperature properties of the ceramics.
Also, the need for additional fixtures to apply external loads to initiate grain rotation
across the joint and cause bulk deformation [Binner 1998] complicates the furnace
system and may be problematic in applying a large load to multiple parts simultaneously.
1.2.1.1 Diffusion Bonding

Diffusion bonding joins materials at high temperatures (from 0.6Tm to 0.8Tm,
where Tm is the melting point of the material) and pressures. Ideally, diffusion bonding
produces local deformation and/or melting at the joint, promoting surface contact, and
subsequently allowing atorrric or ionic migration across the joined interface to produce a
bond (Figure 1.2) [Santellal 1992, Sandhage 1996]. However, advanced structural
ceramics like alumina, silicon carbide, and zirconia do not deform easily except at very
high temperatures. In addition, they are difficult to densify due to the low diffusivities of
the chemical species of the base ceramics [Bates 1990]. Therefore, diffusion bonding can
be difficult and impractical, especially without interlayers [Santellal 1992]. For example,
Elssner et a1. [1981] found that 18 um coarse-grained alumina components could not be
bonded to each other via heating at 1700°C and 9.0 MPa applied pressure, due to the high
creep resistance of the alumina. However, under identical joining temperature and
applied stress conditions, diffusion bonding of an 18 um coarse-grained alumina to a 1
pm fine-grained alumina was achieved because the fine-grained alumina creeps enough

to facilitate the joining. Consequently, diffusion bonding has to occur within the range of

stresses and temperatures that induce creep in materials, leading to dimensional changes
and instability [Bates 1990, Elssner 1981].
1.2.2 Joining with interlayers
1.2.2.1 Brazing

Brazing with glasses, glass ceramics, or metals has been utilized by various
researchers for ceramic-ceramic joining. For example, SiC can be brazed using a calcia-
alumina glass ceramic [Ferraris 1998], a SiOz-CaO-Alzog glass [Ahn 1998], an Ag-Cu-Ti
alloy [Boadi 1997], and zirconia by CaO-MgO-SiOz-A1203 glass [Ashizuka 1996].

There are two main advantages of glass brazing for ceramic joining [1]: (1) the
adhesion and chemical compatibility between glass and base ceramics is good, (2) the
physical properties of glass such as viscosity and melting characteristics can be controlled
over wide ranges. However, glass bonds for high temperature applications are limited by
the relatively low glass transformation point, low fracture toughness, and susceptibility to
stress corrosion [Zdaniewski 1987, Sandhage 1996]. To increase the glass interlayer
fracture toughness and stress corrosion resistance, Zdaniewski et al. [1987, 1975] used
glass adhesives that can be crystallized under controlled joining conditions. Formation of
interlocking grains or needles across the joint is expected to result in joint toughening,
called crystallization toughening [Zdaniewski 1987, 1975]. However, the major
difficulty encountered in crystallization toughening of ceramic adhesives is the persistent
porosity in the joints, probably due to the evolution of gases absorbed in the ceramic
[Zdaniewski 1987, 1975].

Metal brazing generally relies on selective oxidation to achieve bonding between

the metal and base ceramics [Sandhage 1996]. However, the selection of available active

braze alloys is limited [Santella 1992]. Many common metal braze alloys are either
ductile, such as aluminum [Suganuma 1987] and nickel [Vegter 1998], or contain indium
or tin to allow melting at relatively low temperatures, such as Ag-Cu based alloys [Boadi
1997]. However, metal brazing often results in complicated interfacial reaction phases
and thick bond layers, with dimensions several microns thick or more [Locatelli 1997,
Ferraris 1998] that can result in large thermal residual stress at the joint. This stress is
caused by thermal expansion differences between brazing metal and base ceramic that
sometimes results in cracks or failure [Kovalev, Colombo 1998, Singh 1999]. As an
example of the complicated reaction phases generated due to metal brazing, Zurbuchen et
al. [1999] identified a series of complicated reaction layers and interfacial phases
between the parent ceramics, an alumina-silicon carbide whisker (Al203-Sij)
composite, and the brazing alloy, Ag—Cu-Ti, using TEM and EPMA (Figure 1.3). The
local interfacial phase distribution sequence from the SiC is
SiC/TiCI'I‘issi3/I’i3(Cu,Al,Si)30/Ag-Cu. The local interfacial phase distribution sequence
at the A1203 side is Al203/y-T10/Ti3 (Cu,Al,Si)30/Ag-Cu [Zurbuchen 1999]. Figure 1.3
shows the phase development for A1203-SiC composite brazed by Ag-Cu-Ti alloy.

1.2.2.2 Transient liquid phase bonding

In metal systems, the use of a transient liquid phase (TLP) for bonding superalloys
using alloys or layers of alloys is well known [Duvall 1974]. TLP has the ability to
produce a bond at a lower temperature than that at which it will be ultimately used. The
technology is currently being adapted to ceramic-ceramic joining, either using thin glass
layers such as oxynitrides for joining SiAION [Walls 1992] or alloy layers such as

CulPt/Cu and Cu/Ni/Cu for joining alurrrina [Shalz 1993]. In theory, it should be possible

to disperse the interlayer over time [Gale 1999]. This transient liquid phase will
eventually disappear and result in a microstructurally homogeneous joint that is similar to
the parent ceramics [Gale 1999]. However, in practice this has rarely been demonstrated,
and would take very long times [Gale 1999], probably even days or weeks especially for
bond layers several microns or more thick. In addition, the metal alloys or glassy layers
that have the desired phase behavior, wetting behavior, and physical properties for
joining are limited [Loehman 1999]. For example, Loehman [1981] investigated the
joining of Si3N4 using oxynitride glasses whose compositions match the intergranular
amorphous phases of the base Si3N4. During the joining, the oxynitride glassy interlayer
must redistribute itself and penetrate into the adjoining base Si3N4 ceramics [Loehman
1981, Shalz 1993]. Thus, this joining approach requires the development of different

interlayers to join different materials.

A few researchers have investigated interfacial microstructural evolutions of the
TLP bonded ceramics using transmission electron microscopy (TEM) and other
analytical techniques [Walls 1992, Hanson 1998]. In B-SiAlON/B-SiAlON ceramics
joined using B-SiAlON glass-forming composite adhesives (mixture of Si3N4, Y203,
A1203, Si02 powders), the rrricrostructure of the joint consisted of a primary joint zone
and a secondary joint zone on both sides of the primary joint zone (Figure 1.4) [Walls
1992, Hanson 1998]. At the joining temperature of 1400°C and above and 2.0 MPa
pressure, the oxide components react to form a Y-Si-Al-O liquid phase, leading to
densification and sintering of the interlayer. The Si3N4 then dissolves into the liquid,
altering the composition to B-SiAlON. At the same time, the newly formed B-SiAlON

grains grow and form an interlocking network across the joint. TEM analysis showed

that the joint thickness was approximately 1~2 pm, with an indistinct interface between
the base B-SiAlON and the newly formed B-SiAlON materials. The secondary joint
zone, determined by elemental analysis of the concentration across the joint into the
surrounding base material using electron micro-probe analysis (EPMA), had a thickness
of around 250 pm from the primary joint zone. The authors suggested that a gradual
change of the ceramic composition within the secondary joint zone is an ideal way of
relieving residual stresses at the joint that is introduced during the joining [Walls 1992,
Hanson 1998].
1.2.2.3 Joining with spin-on interlayers

Joining with spin-on interlayers, which uses high rpm spinning to produce very thin
(typically around 200 nm thick) interlayers, has been developed at Michigan State
University (MSU) [Case 1997, 1998]. The spin-on coating typically consists of an
organic based ceramic precursor such as Silicafilmm, Blackglasm, or sodium silicate,
deposited as a liquid onto the surface of the ceramics to be joined. The ceramics to be
joined are spun on a substrate spinner and pyrolyzed at low temperature to form a
ceramic layer. The thickness of the ceramic precursor can be controlled by adjusting the
spin rate, with films approximately 200~300 nm thick are produced at a spin rate of 3000
rpm for 20 seconds. This joining approach, which has many advantages over other
joining methods, produces joined interfaces with interlocking microstructures that have
little or no observable bond layer(s) or second phases [Zeng 2001] and have relatively
high interfacial strength. Other characteristics of joining with spin-on interlayers can be

listed as [Case 1997, 1998]:

1. This ceramic joining technique uses either zero loading or a minimal (20~85
gram) external loading that leads to zero or minimum dimensional changes and
eliminates the need for a fixture to apply large external loads to the specimens and
during joining.

2. It can accommodate complex exterior geometries and internal channels or holes.
The spin-on interlayers can be applied to only one of the surfaces to be joined.
Therefore, holes or channels can remain at precise dimensions, which would be
impossible for brazing or transient liquid phase bonding since interlayers would
melt, flow and obliterate the holes or channels.

3. Both conventional and microwave heating can be used to join the spin-on
interlayer ceramics.

4. However, spin-on interlayer joining requires that the surfaces be flat and polished
and may require the development of the different spin-on interlayers to join

different materials.

To date, a variety of ceramics have been successfully joined by the spin-on
interlayer techniques, including similar and dissimilar materials such as alumina/alumina,
MaCorTM/MaCorTM, SiC/SiC, alumina/zirconia, and zirconia/MaCorTM [Lee 1997, Case
1998]. Indentation tests showed that these bonds have relatively high interfacial
toughness with Vickers indentation induced crack near the interface propagated across
the interface without deflection [Lee 1997, Case 1998].

1.2.3 Microwave heating and joining

10

 

 

One limitation of conventional ceramic—ceramic joining is that the entire assembly
needs to be heated to the desired joining temperature. This requirement could lead to
many practical difficulties. Therefore, ceramic-ceramic joining by selective heating
using unconventional heating methods, such as microwave heating, has been the subject
of considerable attention.

1.2.3.1 Microwave processing and microwave effect

Microwave heating is fundamentally different from conventional heating (Figure
1.5). Electrical furnaces used for conventional heating of materials are composed of
heating elements, such as SiC, and insulation. For microwave heating, the cavity used to
heat the materials is composed of a metal shell and rrricrowave port through which
electromagnetic waves are allowed into the cavity from the microwave power supply
[Sutton 1992]. In conventional processing, heat is generated from an external source
which transfers thermal energy to the surface, and further into the center of the material
by conduction. However, in microwave heating, heat is generated within the material
through the interaction between the electromagnetic field and induced molecular motions.
When compared to conventionally heated materials, materials will be heated more
uniformly, and typically more rapidly, when heated using microwaves, within the limits
set by microwave absorption [Loehman 1983].

Due to the inherent heating characteristics associated with the microwave, many
attractive advantages have been experimentally demonstrated. In addition to reduced
processing times and energy savings, a number of researchers [Palaith 1989, Fanslow,
Janney 1991] have reported that microwave processing produces chemical and physical

reactions that are different from the reactions that would have occurred if the only effect

11

of the microwaves were to increase the temperature. For example, increased diffusion
coefficients [Janney 1991], enhanced chemical and physical reactions [Fanslow 1991],
improved microstructures and mechanical properties [Patterson 1992], smaller grain sizes
[Tian 1988], and lowered sintering activation energies [J anney 1988] have been observed
in microwave heating. These enhancements are known as the “microwave effect”.
Janney et al. [1988] reported that A1203 doped with 0.1 wt% Mg0 sintered under
vacuum using 28 GHz microwaves densified much faster than by conventional sintering.
The activation energy observed for conventional sintering was 575 KJ/mol. However,
the activation energy for microwave sintering, 160 KJ/mol, was only one-third of that for
conventional sintering [Janney 1988], suggesting the sintering process and mechanisms
were quite different for microwave and conventional sintering. In addition, J anney et al.
[1988] also compared the kinetics of grain growth in both microwave and conventional
annealed alumina samples. The rate of grain growth is greatly accelerated in microwave
annealing. The activation energy for microwave grain growth was 480 KJ/mol, which
was around 20% lower than 590 KJ/mol for conventional grain growth (Figure 1.6)
[Janney 1988, 1991], suggesting the grain growth process may be more similar for both
microwave and conventional annealing. To account for the observed athermal
microwave effects, J anney et al. [1991] have speculated that these enhancements occur as
a result of the electromagnetic field “coupling to lattice defects and other bulk crystalline
effects”. In addition, Lee and Lin [1998] measured and compared the dielectric loss
factor (tan 5) of sintered BaO-La203-4.7T102 ceramics with and without glasses addition.
The tan 8 of the sintered ceramics with an addition of glass is significantly higher than

the theoretically calculated values of ceramics and glass based on the sum rule weighted

12

by constituent volume fractions, due to the heterogeneous interface between glasses and
ceramics. This difference between the measured and calculated tan 6 indicates that large
dielectric losses occur at interfaces between materials with large differences in dielectric
loss factors, leading to a large increase in tan 8 at the interface [Lee and Lin 1998].
Recently, Rybakov et al. [1994] proposed a model in which a microwave field
induces oscillatory fluxes of ionic point defects near surfaces, generating a net current of
charged defects. Furthermore, Booske and their co-workers [Booske 1997, 1998,
Freeman 1995, 1998] illustrated theoretically and numerically, and confirmed
experimentally, that microwave fields induce an additional driving force (ponderomotive
force or pmf) that enhances the mass transport rate and ionic diffusion, not seen with
conventional heating. The pmf is an electromagnetic force that tends to move mass and

can be expressed as the following:

 

pmf = <59 q" (57' + EPI> [Booske 1997, 1998]

Where ( > denotes time averaging over one cycle,

q denotes electric charge of the species n,
n

Ci denotes the initial, equilibrium concentration value of the species n,

C": denotes the concentration perturbation from the initial value due to
microwave excitation for species n,

E" denotes the microwave field or potential,

E" denotes the microwave potential perturbation from the initial value due to

microwave excitation.

l3

The numerical model, based on continuum equations, predicted that the microwave
fields do not enhance the ionic mobility, but rather are responsible for a new driving force
for ionic diffusion [Booske 1997, 1998, Freeman 1995, 1998]. During powder sintering,
this new additional driving force pmf could drive defects to the crystal boundary and
enhance the long-range mass transport into the interior of the crystal. In addition, the
model predicted that the pmf should predominate near material surfaces or any abrupt
discontinuities, such as interfaces. Table 1.1 compares the microwave pmf with other
calculated thermochemical driving forces for ionic transport in NaCl. The authors
[Booske 1997, 1998, Freeman 1995, 1998] state that the value of the pmf is higher than
the solid-state diffusion driving force for materials joining between two flat NaCl/NaCl
surfaces. This can explain the rate enhancement during microwave joining observed by
many authors.

1.2.3.2 Interaction between microwaves and materials

Depending on the material type, microwave energy can be transmitted, absorbed, or
reflected. Metals are opaque to microwaves and thus are very difficult to heat by
microwave energy (Figure 1.7). Low loss ceramics such as A1203, Mg0, 8102, and most
glasses are transparent to microwaves at ambient temperatures [Sutton 1992]. However,
for these low loss ceramics, there is a critical temperature above which the ceramics
begin to absorb and couple more efficiently with microwave radiation. In addition,
adding conductive or magnetic phases or other additives in the form of fibers or particles
to the low loss or microwave transparent ceramics can greatly enhance their microwave
energy absorption. The added conductive and magnetic phases can absorb microwave

energy more rapidly than the matrix and thus can be selectively heated [Sutton 1992]. On

14

the other hand, ceramics exhibiting high dielectric losses such as Cu0, N10, and Mn02
can absorb microwave energy efficiently at room temperatures and heat up very rapidly
[Sutton 1992, Tinga 1988].

When microwaves penetrate and propagate through a material, internal electric
fields will be generated. These internal electric fields induce complicated motions of
charges such as electrons, ions, and dipoles. ”The resistance to the induced motions due
to inertial, elastic, and frictional forces causes losses and attenuates the electric fields”,
resulting in volumetric heating [Sutton 1992]. Generally, the dielectric loss factor (tan 8)

is used to describe these losses as the following [Sutton 1992]:

tanszI—f-
8r

Where 8:3 denotes effective relative dielectric loss factor

3; denotes relative dielectric loss factor

Therefore, when a material exhibits a high dielectric loss factor in the microwave
frequency range, it is expected that the dielectric material will absorb microwaves easily
and with very efficient energy conversion, resulting in rapid heating.

1.2.3.3 Microwave joining

Since the absorption of microwave energy varies with the composition and structure
of different phases, selective heating is certainly possible and presents opportunities for
commercial applications where partial or local heating is highly desirable. One of the
greatest characteristics of microwave heating that can be used for ceramic joining is the
possibility of selectively heating the interface. A large tan 5 dielectric interlayer can be

heated preferentially and join the base ceramics without heating the bulk materials,

15

minimizing the volume of thermally stressed ceramics [Loehman 1993, 1999]. Another
potential advantage of selective heating is the possibility of protecting certain heat
sensitive components by shielding them with a piece of metal during joining since metals
are opaque to microwaves [Loehman 1993].

Other advantages of microwave joining include (1) enhanced diffusion and lower
joining temperatures and (2) lower processing times and energy saving costs. For
example, Binner et al. [1998] investigated the microwave and conventional joining of
different grades of polycrystalline alumina. They found that relatively low purity
alumina, such as 85% and 94%, can be readily joined. However, higher purity 99.8%
alurrrina was “insufficient to lossy to couple” with microwaves, and thus could not be
heated to sufficient high temperatures due to the negligible amount of grain boundary
phases present, making joining difficult. Despite the sirrrilarity in bonding mechanism
based on viscous flow of the grain boundary phase, microwave joining can be a factor of
at least four times faster than conventional diffusion joining, resulting in significantly less
deformation associated with joining for the 85% alumina. Similar results were found by
Fukushima et al. [1988] when joining different grades of alumina. Although 92 to 96%
purity alumina were directly joined via microwave heating, higher purity 99% alumina
failed to join together without the use of a layer of lower purity alumina.

Although nricrowave joining of ceramics exhibits great potential for commercial
applications, many important process variables associated with microwave heating cannot
be adequately controlled, such as heating and cooling rates, accurate temperature control,

and temperature uniformity in the ceramics. As pointed out by Loehman [1993], “there

16

has not been enough experiences with microwave joining of ceramics to provide the
needed data on process control and reproducibility”.
1.3 Indentation tests

Vickers indentation tests, which use a pyramidal diamond indenter, are extensively
used to study the toughness, fracture and deformation properties of brittle polycrystalline
ceramics [Marshall 1977]. Cracks that initiate from the central deformation zone and
develop as “half-pennies” along indentation diagonals can provide deep insight into the
fundamental nature of damage modes in brittle solids and establish a basis for models of
wear and erosion resistance [Marshall 1979]. Furthermore, understanding the behavior of
indentation cracks near surfaces and interfaces is beneficial to the understanding of
failure mode of bonded dissimilar materials and the further development of joining
approaches [Lardner 1990, Lee 1998].

To evaluate the interfacial strength of bonded materials in comparison to the
strength of the bulk ceramics, various researchers have utilized Vickers indentation
testing on joined cerarrrics [Ye 1994, Nakamura—A 1997 , Rodriguez 1998]. For example,

by placing a series of Vickers indentations across the joint interface and comparing the

crack lengths of the far and near field interface indentations, the residual stress 0" can be

determined from the following relationship [Lee 1996, Ye 1994]:

1.5

4Cr —o.5 g —1

yr c,

 

0',=K.

Where Kc denotes critical stress intensity factor,
C' denotes indentation crack length near the interface,

C denotes indentation crack length far away the interface.

17

In addition, Vickers indentation cracks placed near the joint can also be used to
estimate the interfacial fracture toughness of the joint [Lee 1996]. In the absence of an
elastic mismatch across the joined materials, a crack can penetrate across the interface
without deflection when the interfacial fracture energy reaches approximately 60% of the
bulk fracture energy [Lee 1996]. For indentation tests on bonded dissimilar materials,
distortion of radial cracks may result from elastic mismatch. Lardner et al. [1990]
investigated the mechanisms of indentation crack propagation in cracks placed in glass
near a joined glass/epoxy interface. Initially, the crack propagated toward the interface
due to the modulus of glass being much higher than that of the epoxy. However, as the
crack approached the interface, the crack reoriented in a direction away from or parallel
to the interface due to the changing stress fields near the interface [Lardner 1990].

1.4 Characterization techniques

Development of ceramic joining methods requires a fundamental understanding of
ceramic interfaces and reactions. The ultimate goal of ceramic joining is to develop
joints with high interfacial strength and desirable interfacial microstructures between
similar or dis-similar materials. Therefore, knowledge of the relationships between
interfacial reactions and adhesion, the composition and structure of the interface, the
joining mechanism, and the nature of any reaction products is crucial for the development
of joining technology. Unfortunately, instead of using transmission electron microscopy
(TEM), most researchers have only use optical microscopy (0M) and scanning electron
microscopy (SEM) to characterize the interfaces. The reason may be related to the
difficulty of TEM sample preparation (described in Chapter 11). Although 0M and SEM

are useful for noting the presence of large-scale interfacial phases or interlayers, the

18

resolution of these techniques is still somewhat limited. Fine scale interlayers or defects
that can be revealed using TEM may not be visible using 0M or SEM. Therefore, more
detailed characterization and structural analysis of interfaces should be carried out using
TEM in conjunction with 0M and SEM. More importantly, interface examinations will
help in understanding the fundamental process in ceramic joining.

1.5 Research strategy

It is known that the mechanical properties of joined ceramics depend on the
composition and structure of the interface, the joining technology used, and reaction
products that may be produced [Loehman 1993, 1999]. Therefore, the initial objective of
this research was to characterize the interfacial nricrostructures of ceramics joined using
spin-on interlayers.

The interfacial microstructure of microwave joined polycrystalline partially
stabilized zirconia and glass-ceramics MaCorTM using spin-on interlayers were first
studied. In microwave joined zirconia/MaCorTM specimens, no silica interlayer,
interfacial phase(s) or other reaction products were observed. Presumably, the interlayer
diffused into the surrounding parent ceramics, leaving a perfectly bonded interface.

Because the glass-ceramic composite MaCorTM contains an amorphous matrix
phase, it is not clear if the spin-on interlayer is in fact necessary to successfully bond
zirconia and MaCorTM specimens. Thus, the initial objective of this study was modified
to more comprehensively and intensively assess the nature of nricrostructure development
in the bond region of partially stabilized zirconia and MaCorTM joined via conventional
versus microwave heating with or without a spin-on interlayer in relation to a number of

other processing variables. However, as the research progressed, it became clear that no

19

significant differences in the interfacial rrricrostructure developed between conventional
versus microwave joined zirconia and MaCorTM, due to the glassy nature of the
MaCorTM. Thus, the final objective of this research was to focus on the interfacial
microstructures of joined crystalline ceramics. Because of the difficulties in identifying
interfaces in joined polycrystalline ceranrics/polycrystalline ceramics, and because of the
broad applications of single and polycrystalline alumina in industry, the interfacial
microstructures of polycrystalline alumina and single crystal sapphire joined using spin-
on interlayers have been investigated. Through a step-by-step process, better
understanding of the nature of the joint associated with joining mechanisms and the

joining technology has now been successfully achieved.

20

 

 

Intergranular phase

    

 

 

Figure 1.1 Schematic representation of the direct ceramic-ceramic joining process
without the use of interlayers carried out at elevated joining temperature and pressure
[Redrawn from Esposito 1998]. At the first stage of the joining process (a), the interface
between the two polished ceramic parts can be seen as a macroscopic inhomogeneity
since no intergranular amorphous phase is present on the surface of the grains at the
interface. Driven by thermodynamic capillary force (b), the intergranular amorphous
phase flows to the interface at elevated temperatures. As the process continues (c), liquid
phase flow, grain boundary sliding, and grain rotation can be activated to minimize the
macroscopic inhomogeneity between joint and base ceramics. Finally ((1), grain growth
and grain bridging occur, leading to an indistinguishable interface from the base
ceramics.

21

 

(b)

   

(c) ((0

Figure 1.2 Schematic representation of the sequence for diffusion bonding at a
proper joining temperature and pressure [after http://www.twi.co.uk/j32k]. Initially (a)
two ceramic surfaces are placed in contact, (b) at least one of the surfaces begins to
deform under high temperature and stress, (c) deformation continues, leading to void
shrinkage, and (d) the formation of interface or joint.

22

are A120.

..,,L m. ._ .,,..,f __.-.1
. . _ , _, .

“ Elmer

Again

~e-flfis<ie-.AI.-,e‘

SiC AI203 _

A1203

 

(a) (b)

Figure 1.3 shows the phase development of a A1203-SiC composite brazed by Ag-
Cu-Ti alloy from (a) the initial phase distribution to (b) the final experimentally
determined complicated phase distribution at the joint after brazing [Redrawn from
Zurbuchen 1999].

23

         
    
      
  
 

  

Starting microstructure T=1400°C T=1 00°C

  

.tIIIIIllll.lli

’llll " ‘

I Ill
e IIII
_

‘

  

,n .‘i‘h‘lw

   

l'llllllll."'
ill.”

   

“‘

   

 

10 um

Figure 1.4 shows the sequential stages of the transient liquid phase bonding process
for the joining of B-SiAlON/B-SiAlON ceramics using B-SiAlON glass-fomring
composite adhesives (mixture of Si3N4, Y203, A1203, SiO2 powders) [Redrawn from
Walls 1992, Hanson 1998]. Prior to the joining, adhesive powders are sprayed onto one
of the base materials. Around 1400°C, oxide components react to form a Y-Si-Al-O
liquid phase, leading to densification and sintering of the interlayer. At approximately
1600°C, the Si3N4 dissolves quickly into the liquid. At the same time, the newly formed
B-SiAlON grains grow and form an interlocking network across the joint, leading to an
indistinct interface between the base B-SiAlON and the newly formed B-SiAlON
materials. The final microstructure is composed of the base B—SiAlON and the newly
formed B-SiAlON grains with an almost indistinct interface.

24

heating element microwave port

I i

 

 

 

 

 

 

sample

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Inst'tlation Furnace IMetal shell Cavity

(a) (b)

Figure 1.5 Heating patterns in (a) conventional and (b) microwave heating furnaces
[Redrawn from Sutton 1992].

25

 

  
    
 
  

 

 

 

3" fi 1 r r ' l

2 Microwave
3; 160 KJ/mol
8
is" 1’
a
'25; 0
on
'2 1

- Conventional

’ 575 KJ/mol
-2 n J . - r . r -
0.5 0.6 0.7 0.8 0.9
1000/1“ (°K")
(a)
I I

 

   
 

Microwave
480 KJ/mol

  
 
  

log (Sintering rate)
is

 

 

 

Conventional
-21 _ 590 KJ/mol
0.45 05 0.55 0.6 0.65

1000/1‘ (°K")

(b)

Figure 1.6 shows the activation energy for (a) sintering process in microwave and
conventional sintered A1203 doped with Mg0 [Redrawn from Janney 1988] and (b) grain
growth in microwave and conventional heated A1203 doped with Mg0 [Redrawn from

Janney 1991].

26

 

Table 1.1 Comparison of microwave pmf and other thermochemical driving forces for
ionic transport in NaCl [Booske 1997, 1998, Freeman 1995, 1998].

 

 

 

 

 

 

 

 

 

 

Transport driving force Typical magnitude (N)
1 Sintering 1pm powder—initial stage ~10'16
2 Sinterin g 1 pm powder—intermediate stage ~10"8
3 Joining two flat faces <10'18
4 Chemical interdiffusion (depending on .16
concentration gradient ~10 to °°
5 Tracer isotope diffusion <<10'fir
6 Microwave pmf ~10'19 to 10'15

 

 

Materials Penetration

transparent Total
W at...

None

\ . . opaque

.M (Fe) (Reflected)
'bsorbel' Partial to Total
(Cu0)

absorber Partial to Total

0 fl 0 0",‘0‘ (mixed)
M W0 (a) matrixm-ansparent

‘ fl ° ~ 0 ‘\‘ (b) fiber/particle/additives: absorber

O

 

 

 

 

 

 

 

 

 

 

Figure 1.7 Interactions of microwaves with different materials [Redrawn from
Sutton 1992].

27

Chapter 11

Experimental procedures

2.1 Goals and objectives of the proposed research

This research program focused on the interfacial microstructures of ceramics joined
using spin-on interlayers. The joined ceramic samples were characterized using different
techniques including optical microscopy (0M), scanning electron microscopy (SEM),
and transmission electron microscopy (TEM). The primary goal of the interfacial
assessment was to gain a complete understanding of the nature of the joint. The
mechanical properties of the interface were characterized using Vickers indentation
techniques to access the thermal residual stresses and interfacial strength. The
microstructure and mechanical characterization resulted in information about the
structure of the interfaces, and the fundamental ceramic joining processes. Furthermore,
the degree of bonding was assessed and related to mechanical properties of the joined
ceramics.
2.2 Materials and spin-on interlayers used in the current research

2.2.1 Joining materials

Three mol.% yttria-zirconia, glass-ceranrics MaCorTM, polycrystalline alumina, and
single crystalline alurrrina (sapphire) were joined in this research. .

The polycrystalline partially stabilized zirconia (PSZ) specimens used in this
research were sintered from powder compacts which were pressed in a uniaxial die at
approximately 32 MPa using commercial powders (TZ-3Y, Tosoh Corporation, Japan).

The powder compacts were sintered at 1375°C for one hour in a 2.45 GHz single-mode

28

microwave cavity (Model CMPR-250, Wavemat Inc., Plymouth, MI) [Case 1997, 1998].
During sintering, the temperature was controlled by adjusting an input power with
automatic sliding short and probe. After sintering, the zirconia specimens were
approximately 17 mm in diameter and 2 mm thick.

MaCorTM (Corning Code 9658, Corning, NY) is a fluorophlogopite mica reinforced
glass ceramics in the system of Si20-Al203-Mg0-K20-F, composed of approximately 55
volume percent of a randomly oriented fluorophlogopite mica platelet and 45 percent of
borosilicate glass matrix [Corning]. MaCorTM is widely used as an electrical or thermal
insulator for structural components in semi-conductor industry.

Two different commercial grades of polycrystalline alumina were used in this
research; 96% and 98%, which are often used as insulators for precision resistors, wear
resistant components for pump seals, and lids for hybrid microwave applications.
Commercial 96% and 98% grades of polycrystalline alumina were purchased from the

Coors Corporation.

Sapphire wafers with (1120) surface normals (3:0.3°), with flatness better than 25

micrometer (um) and a metallic impurity level less than 100 ppm, were purchased from
the Union Carbide Corporation.
2.2.2 spin-on interlayers
The spin-on interlayer used as the bonding agent was produced from SilicafilmTM
(Emulsitone Co., Whippany, NJ). SilicafilmTM is an organically based silica precursor
that was applied to the polished surface to yield a pure amorphous Si02 film
[Emulsitone]. This thin, amorphous silica film coating was the interlayer that aided in

joining the specimens. The SilicafilmTM was applied as a liquid and spun on a substrate

29

spinner to produce a thin film. The film was then cured or heated in air to provide an
amorphous silica layer with the high purity required for integrated circuit applications.
SilicafilmTM has an impurity level of less than one part per million of metallic ions
[Emulsitone].
2.3 Joining
2.3.1 Joining procedures

The technique used to join densified ceramics using spin-on interlayers was
developed here at Michigan State University (MSU) by previous researchers [Case 1997 ,
1998]. Figure 2.1 shows an overview of the typical spin-on interlayer ceramic joining
procedure used in the current study. More detailed system specific procedures follow the

common outline. The general joining procedure was as follows:

1. Mount the specimens to be joined to a flat aluminum disk with thermoplastic
using a hot plate.

2. Polish the specimens using an automated polishing system (LECO Corporation,
St. Joseph, MI) and a series of an abrasive diamond paste with grits ranging
from 35, 25, 17, 15, 10, 6, to 1 umfor 2~20 rrrinutes.

3. Demount the specimens from the aluminum disk by heating on the hot plate.

4. Clean all specimens in a 50:50 rrrixture of acetone and alcohol for 10~20
minutes in an ultrasonic bath.

5. Adhere the ceramic components to a substrate spinner (Headway Research, Inc.

Garland, Texas) using a double-sided tape.

30

6. Apply four to five drops of SilicafilmTM (Emulsitone Co., Whippany, NJ) to the
polished and cleaned surfaces of the ceramic components using a pipette.

7. Turn on the substrate spinner and spin at a rate of 3000 rpm for 20 seconds.

8. Separate the coated ceramic components from the double-sided tape using a
single sided razor blade.

9. Cure the liquid SilicafilmTM in a conventional resistance furnace for 20 rrrinutes at
200°C to form a thin amorphous silica layer (approximately 200~300 nm). This
thin amorphous silica coating was the interlayer that aids in joining the
specimens.

10. Put the specimens to be joined on an alumina setter and place the coated
surfaces of the specimens in contact.

11. Apply a small (20~85g) polycrystalline alumina deadweight (Figure 2.2) to the
specimens.

12. Heat the specimens to be joined in a single-mode cylindrical 2.45 GHz
microwave cavity or conventional resistance furnace.

13. Cool the joined specimens in a fumace-cooling mode.

The microwave heating was carried out using a cylindrical zirconia casket
refractory and the TM.“ microwave cavity (Model CMPR-250, Wavemat Inc., Plymouth,
MI) mode with a stepped input power sequence typically starting from 100 Watts and
increasing to approximately 1200 Watts in steps of 100 Watts every 3 minutes. During
the microwave joining, the joining temperatures were measured using an optical

pyrometer system (Accufiber Optical Fiber Thermometer, Model 10, Luxtron Co.,

31

Beaverton, OR), which is capable of measuring temperatures ranging from 500~1900°C
with an accuracy of i2°C [Case 1997, 1998].

The conventional joining was performed using a conventional electrical resistance
furnace (Carbolite, Aston Lane, Sheffield, England) with SiC heating elements. The
temperature was controlled to i2°C. The heating and cooling rates were approximately
10°C per minute. For all microwave and conventional joining, the joining time refers to
the time at the desired temperature.

2.3.2 Materials joining systems

Three different joint systems were produced via microwave heating: 3 mol.% yttria-
zirconia/MaCorTM, polycrystalline alumina/polycrystalline alumina, and polycrystalline
alumina/3 mol.% yttria-zirconia. Conventional furnace heating was used to produce
MaCorm/MaCorm, polycrystalline alumina/single crystalline alumina (sapphire), and 3
mol.% yttn'a—zirconia/MaCorTM joins. Table 2.1—2.3 list all of the different joined
specimens produced in this study.

2.3.2.1 joining of crystalline materials with non-crystalline materials
2.3.2.1.1 Microwave and conventional joining of zirconia and MaCorTM

In this study, 3 mol% yttria-zirconia specimens were joined to MaCorTM using
microwave and conventional heating to assess the differences between microwave joining
and conventional joining by comparing the times and temperatures necessary to achieve
successful joining using these two approaches. 10-mm X 10-mm X 1.8-mm specimens
were cut from the as-received MaCorTM billets using a low speed diamond saw. After
polishing, coating and curing, 3 mol% yttria-zirconia specimens were joined to MaCorTM

using microwave heating at 1020°C for 20 minutes. During the joining process, a 20-

32

 

gram polycrystalline alumina deadweight was applied to the specimens, providing a very
small normal load of 2000 Pa.

Table 2.1 outlines various experimental conditions and combinations of joined
specimens via conventional heating. In addition to the joining of MaCorTM/zirconia with
spin-on interlayers, two polished but uncoated MaCorTM/zirconia specimen pairs were
joined via conventional heating to compare the results with spin—on interlayer joining and
to determine the effect of the silica spin-on interlayer.

2.3.2.2 Joining of crystalline ceramics

In addition to the joining of crystalline materials with non-crystalline materials,
joining of crystalline ceramics was performed via microwave and conventional heating.
Table 2.2 outlines the experimental conditions for the joining of polycrystalline
alumina/3 mol% yttria-zirconia and 98% polycrystalline alumina/polycrystalline alumina.
Zirconia/alumina, microwave and conventional joining 98% polycrystalline
alumina/alumina specimens were joined using microwave heating at 1500°C for 20
minutes, and 1340°C for 20 minutes, 1340°C for 180 minutes, respectively. During the
joining process, a 20-gram polycrystalline alumina deadweight was applied to the
specimens, providing a very small normal load of 2000 Pa.

2.3.2.3 Joining of polycrystalline alumina and sapphire

Sapphire wafers with (1120) surface normals (i0.3°), and commercial 96% and

98% polycrystalline alumina were used in this study. Ten-mm diameter and 2~3 mm
thick specimens were cut from 96% and 98% polycrystalline alumina rods, respectively.
Initially, the polycrystalline alumina specimens were finely polished, and cleaned in a

50:50 mixture of acetone and alcohol for 10~20 minutes in an ultrasonic bath.

33

Fifty mm diameter and 0.45 mm thick of sapphire disks were sectioned into
rectangular 10mm X 10mm squares. The as-received sapphires were only cleaned in an
ultrasonic bath. Polishing was not required due to the high flatness of the sapphire. After
spin coating and curing, the coated surfaces were put in contact and joined by heating in a
conventional electric furnace. During the joining process, an 85 gram polycrystalline
alumina deadweight was applied to the specimen, providing a very small normal load of
8000 Pa. Table 2.3 outlines the experimental conditions for the joining of 96%, 98%
polycrystalline alumina with sapphire using silica interlayers at temperatures from 1400°
C, 1450°C and 1475°C for 3 hours via conventional heating.

2.4 Characterization
2.4.1 Optical microscopy and SEM

2.4.1.1 Sample preparation

Specimens were prepared for optical and SEM examination by cross-sectioning
with a low speed diamond saw. A portion of each sectioned specimen was mounted in
diallyl phthalate powder (LECO Corporation, MI). Using an automatic polisher, the
mounted specimens were polished using a series of diamond abrasive grit ranging from
25, 10, 3, to 1 microns. The final cleanings were done in a 50:50 mixture of acetone and
alcohol for 10~20 minutes in an ultrasonic bath. Thus, the specimens were ready for
optical microscopy examinations. In addition, prior to SEM examination, the polished
and cleaned specimens were carbon coated using a carbon evaporator (Denton Vacuum
DV-502) to avoid charging effects.

2.4.1.2 Characterization techniques

34

Optical microscopy was used to initially assess the integrity and thoroughness of
the bonds. Optical microscopy was performed using a Nikon optical microscope (Model
EPIPHOT 200, Japan). More detailed characterization of the interfaces was carried out
using SEM. The SEM observations were carried out using a Hitachi S-2500C, a
CamScan 44 FE SEM, or a JEOL JSM—35CF operated at an accelerating voltages of
15~25 kV and working distance of 15 m using secondary and backseattered electron
imaging techniques.

2.4.2 TEM
2.4.2.1 TEM sample preparation

To carry out TEM analysis, cross-sectional TEM thin foils were prepared by
sectioning, mounting the sections to 3 mm molybdenum rings, polishing, dimpling and
ion milling until perforation. Figure 2.3 shows an overview of the cross-sectional TEM
sample preparation procedure. The detailed sample preparation sequence used to form a

cross-sectional TEM thin foil is as follow:

1. Section the joined specimens into a 1 mm thickness slab using a low speed
diamond saw.

2. Continually section the slab into a 2.5x2.5 mm squares using a low speed
diamond saw.

3. Clean the joined specimen squares in a 50:50 mixture of acetone and alcohol.

4. Mount the 2.5x2.5 mm joined specimens square onto a 19 mm diameter

sapphire flat with CrystalbondTM mounting wax on a hot plate.

35

5. Mechanically attach the sapphire flat with the mounted joined specimens square
into a sapphire flat holder that fits into the polishing block.

6. Polish the joined specimen squares with a series of diamond lapping films,
ranging from 35, 25, 10, 3, 1, to 0.5 pm.

7. Separate the joined specimen squares from the sapphire flat using acetone,
followed by a rinse with ethanol.

8. Measure the joined specimen square thicknesses using a micrometer.

9. Glue the polished side of the joined specimen square to a 3 mm diameter
molybdenum ring using GatanTM G—l epoxy and hardener, followed by curing
on the hot plate for 10 nrinutes at 130°C.

10. Mount the joined specimen squares with molybdenum ring onto the sapphire
flat with Crystalbondm mounting wax on the hot plate.

11. Re-measure the total thickness of the sample assembly to determine the amount
of materials needed to be removed by polishing.

12. Mechanically re-attach the sapphire flat into the sapphire flat holder that fits
into the polishing block.

13. Polish the joined specimen squares to a thickness of approximately 70~90 pm
with a series of diamond lapping films, ranging from 35, 25, to 10 um.

14. Mount the joined sample assembly on a VCRTM Model D500i dimpler.

15. Dimple the joined specimen squares to a thickness of approximately 15~20 um
using a 3 pm diamond coated dimpling wheel with 1 pm diamond slurry.

16. Dismount the dimpled joined specimens square from the sapphire flat with

acetone, followed by a rinse with ethanol.

36

17. Mount the joined sample assembly on a Gatan DuoPostTM support

18. Ion mill using a Gatan PIPS at an angle of i3~i6° with 5 Kev Ar+ beam until
perforation.

19. Carbon coat the thin foil using a carbon evaporator (Denton Vacuum DV-502)

to avoid charging under the electron beam.

The original pieces and interfaces for joining of like ceramics such as 98%
polycrystalline alumina/98% polycrystalline alumina, may be indistinguishable optically,
leaving the original specimens are almost perfectly joined together. Thus, the standard
cross-sectional TEM sample preparation procedures (described above) had to be changed
to examine the interfacial microstructures. To determine the interfaces for dimpling, after
polishing and gluing, the specimens were ion milled at a high angle of 10° for a short
time. After ion milling, the interfaces were distinguishable with an observable thin
interface line. This preferential thinning was probably due to the difference in chemistry
and strength between the interfacial region and base ceramics. Following the short time
ion rrrillings, the specimens were then dimpled to 15~20 um and further ion milled to
electron transparency.

2.4.2.2 TEM Characterization

Detailed characterization of the interfaces was carried out using TEM. TEM was
used to investigate the effect of interlayers on the joining process and check for the
presence of interfacial phases or interlayer at the interface. In addition, the overall
interface microstructural features, such as interface roughness, wetting behavior, grain

sizes, and the possible base rrricrostructural changes due to joining process, were

37

characterized using TEM to facilitate the understanding of the fundamental materials
joining process. In this study, TEM examinations were performed using a Hitachi H-800
electron microscope operated at 200 kV. Both bright and dark field imaging were carried
out in conjunction with selected area diffraction. A double tilt holder allowed the
interface planes to be oriented parallel with the electron beam. The EMS software
package was used to assist indexing the experimental diffraction patterns [EMS].

2.5 Indentation tests

Vickers indentation tests were done on the joined ceramic 3 mol% yttria-
zirconia/MaCorTM and polycrystalline alumina/sapphire to assess residual stresses and
interfacial fracture toughness of the joint at the gross scale. The joined ceramics were
cross-sectioned with a minimum thickness of 2.5 mm, mounted, and finely polished using
a series of diamond abrasive grits, ranging from 35, 25, 10, 3, to 1 pm. A series of
Vickers indentations were placed across the joint interface with a minimum spacing of 2.5
times the Vickers indentation crack length to avoid the effect of strain field around the
prior indent (Figure 2.4). By comparing the crack lengths of the far and near field interface
indentations, the residual stress and interfacial toughness were determined [Lee 1996, Ye
1994].

Vickers indentation tests were done at a load of 5 Kg for 10 seconds with load
speed of 60 uni/sec. The indentation tests were done on the 3-mol% yttria-zirconia side
for joined zinconialMaCorTM specimens because indentation cracks in MaCorTM would be
repeatedly deflected, blunted, and arrested due to the randomly orientated
fluorophlogopite mica crystals in glass matrix [Grossman 1972, 1978]. In addition, the

indentation tests were done on the polycrystalline alumina side for joined polycrystalline

38

alumina/sapphire specimens to protect the expensive diamond indenter. The indentation
crack lengths were measured using both optical microscopy and SEM. Prior to SEM
observations, the indented specimens were carbon coated using carbon evaporator

(Denton Vacuum DV-502) to avoid charging under the electron beam.

39

Table 2.1 the experimental conditions and results for the joining of
zirconia/MaCorTM in the current research. Please note, all experiments were done twice
to verify joining results.

 

 

 

 

 

 

 

 

 

 

“3222:” be WEEKS? (°C) 10.2%,... Heating mode Coating
MaCorTM/zirconia 1020 20 Microwave SilicafilrnTM
MaCorTM/zirconia 1020 20 Conventional SilicafilrnTM
MaCorTM/zirconia 1020 40 Conventional SilicafilmTM
MaCorTM/zirconia 1020 60 Conventional SilicafilmTM
MaCorTM/zirconia 1070 20 Conventional SilicafilmTM
MaCorm/zirconia 1070 40 Conventional Silicafilm1M
MaCorTM/zirconia 1070 60 Conventional SilicafilmTM
MaCorm/zirconia 1070 60 Conventional No coating
MaCorTM/zirconia 1070 120 Conventional SilicafilmTM

 

 

 

 

 

 

 

40

Table 2.2 outlines the experimental conditions for the joining of 3 mol% yttria-
zirconia/polycrystalline alumina, and 98% polycrystalline alumina/98% polycrystalline
alumina. Please note, all experiments were done twice to verify joining results.

 

 

 

 

 

s . t be.. d w 11mm“? Joining C t. Heating
pecrmens o jorne [(333 ure time (min) oalng mode
Polycrystalline
alumina/polycrystalline 1340 20 SilicafilmTM Microwave
alumina
Polycrystalline
alumina/polycrystalline 1340 180 SilicartlrnTM Conventional
alumina
3 mol% yttria-
zirconia/polycrystalline 1500 20 SilicafilmTM Microwave
alumina

 

 

 

 

 

Table 2.3 outlines the experimental conditions and results for the joining of 98%
polycrystalline alumina/alumina, 98% polycrystalline alumina/sapphire, and 96%
polycrystalline alumina/sapphire via conventional heating. Please note, all experiments
were done twice to verify joining results.

 

 

 

 

 

 

 

 

 

 

 

 

 

. . . Joining Joining .
Specrmens to be joined temperature (0 C) time (min) Coating
Sapphire/98% polycrystalline alumina 1400 180 SilicafilmTM
Sapphire/98% polycrystalline alumina 1450 180 Silicafilm‘M
Sapphire/98% polycrystalline alumina 1475 180 SilicafilmTM
Sapphire/98% polycrystalline alurrrina 1475 180 No coating
Sapphire/98% polycrystalline alurrrina 1475 180+300 SilicafilmTM
Sapphire/96% polycrystalline alumina 1475 180 SilicafilmTM
Sapphire/96% polycrystalline alumina 1475 180 No coating
Sapphire/96% polycgstalline alumina 1475 180+300 SilicafilmTM

 

41

 

 

Polished specimen Spin-on material Hi gh-speed Ispinner Curing

Alumina dead weight \

Specimens \5

  
 

 

Setter

Joining specimen

Figure 2.1 Schematic representation of spin-on interlayer joining method used in
the current study.

   
 
   

Dead Weight

Silica film

eeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeee
.....
.....
.....
IIIII
.....

.....
.....
.....
.......
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
oooooooooooooooooooooooooooooooooooooooooooo
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee

Figure 2.2 Representation of the ceramic joining set-up used in the current study. The
joined specimens are loaded using a small (20~85g) polycrystalline alumina deadweight, to
prevent slippage of the specimens.

42

pom-e lea-.l

cross-sections

 

        

2 «a Glue on
Meeting

 

Ion milling 0&6".

Figure 2.3. Schematic representation of the cross-sectional TEM sample preparation
procedure used in the current study [after Geng 2003].

43

| 20 I l d22.5(2c)

 

 

 

(a)

a s i- .f

B

 

 

 

 

Interface

 

 

 

(b)

Figure 2.4 shows (a) the indentation spacing and the (b) scheme of the test to check
the strength of the interface.

Chapter III

Experimental Results

3.1 Introduction

This section of the thesis describes the microwave and conventional spin-on
interlayer joining of zirconia and MaCorTM, the microwave joining of polycrystalline
alumina with polycrystalline alumina, and polycrystalline alumina with polycrystalline
partially stabilized zirconia, and the conventional joining of polycrystalline and single
crystal alumina. The interfacial rrricrostructures of these joined samples have been
examined using optical microscopy (OM), scanning electron microscopy (SEM), and
transmission electron microscopy (TEM). The interfacial toughness evaluation of these
joined samples have been examined using indentation tests. These comprehensive
interfacial microstructural and mechanical properties studies of the joined specimens
resulted in a better understanding of the relationship between interfacial microstructures
and mechanical properties associated with the joining process.
3.2 As-received materials

Prior to joining, the rrricrostructures of the sintered 3 mol% yttria-zirconia, as-
received glass ceramics MaCorTM, and 96% and 98% polycrystalline alumina specimens
were evaluated using TEM. In addition, the silica spin-on interlayer thickness was
measured using SEM.

3.2.1 MaCorTM and 3-mol% yttria-zirconia
Figure 3.1 shows the microstructures of (a) MaCorTM and (b) zirconia. The

commercial glass ceramic MaCorTM consisted of approximately 55% unifomrly and

45

randomly distributed fluorophlogopite mica (KMg3AlSi3OroF2) crystals in glass matrix.
This highly interlocked microstructure is the key to MaCorm’s machinability, which
results from the fractures propagating by broaching along cleavage planes of the mica or
the mica-glass interfaces [Grossman 1978]. Then rrricro-cracks are repeatedly deflected,
blunted, and arrested, resulting in excellent machinability. Diffraction patterns (Figure
3.1(a)) of the matrix phase reveal an amorphous structure, consistent with that expected
for the borosilicate glass [Coming].

The sintered zirconia consisted of monoclinic/tetragonal plates in equiaxial cubic
grains with grain sizes approximately 0.5 urn (Figure 3.1(b)). The cross-sectional surface
grain morphology of the sintered and finely polished zirconia prior to coating was
evaluated using TEM. After polishing and cleaning, the zirconia shows an almost
perfectly flat surface without any wetting or surface grain boundary grooves (Figure 3.2).

3.2.2 Polycrystalline alumina

Figure 3.3 shows the rrricrostructures of as-received 98% (a) and 96% (b)
polycrystalline alumina. The commercial 96% polycrystalline alumina consisted of
elongated or-Al203 grains, with approximately 20% intergranular grain boundary phase,
due to the presence of impurities CaO, BaO, and Si02. Diffraction patterns (Figure
3.3(b) inset) of the intergranular grain boundary phase reveal an amorphous structure,
consistent with that expected for the alunrinosilicate glass [Brydson 1998]. In contrast,

the 98% alumina contained equiaxial or-Al203 grains with little grain boundary phase.

Diffraction patterns from these alumina grains revealed the or-Al203 structure, with the

particular grain in figure 3.3(a) oriented to [031].

3.2.3 As-spun and cured interlayer thickness

46

The silica spin-on interlayer thickness is an important factor in joining that can
affect the final interfacial microstructure and strength. Excess interlayer may result in a
thick bond region and modify the local microstructure. To determine the as-spun and
cured silica interlayer thickness, one coated sapphire sample was sectioned in half after
curing and examined using SEM. Figure 3.4 shows that the interlayer has a thickness in
the range of 250~320 nm in the current research, slightly thicker than the 200 nm of
manufacturer’s specification [Emulsitone].

3.3 Joining of crystalline ceramics with non-crystalline ceramics
3.3.1 Microwave joining of Zirconia/MaCorTM

This section describes the interfacial rrricrostructure of polycrystalline 3 mol%
yttria-zirconia/commercial glass ceramics MaCorTM joined by spin-on layer bonding.
After polishing, coating, and curing, 3 mol% yttria-zirconia specimens were joined to
MaCorTM using microwave heating at 1020°C for 20 minutes.

3.3.1.1 The interfacial microstructure of MaCorTM/zirconia joined at 1020°C

for 20 minutes using a spin-on interlayer

Cross-sectional SEM examination of the microwave joined zirconia/MaCorTM
specimens revealed no observable reaction layers or interfacial phases in the join region.
Figure 3.5 shows an example of a cross-sectional SEM image taken from one of the
zirconia/MaCorTM specimens joined at 1020°C for 20 minutes. In these images, the
interface is wavy and an interlocking microstructure results from gain boundary
grooving of the zirconia gains at the zirconia/MaCorTM interface. The gain boundary
gooves observed in the zirconia are roughly 0.25 um deep and 0.5 pm wide (Figure 3.5).

At some locations along the interface, porosity was observed in the MaCorTM near the

47

interface. Apparently, the joining process changes the surface gain morphology of the
zirconia near the interface, resulting in a non-planar surface in the zirconia gains at the 3
mol% yttria-zirconia/MaCorTM interface (for example compare figures 3.2 and 3.5).
Within the resolution limits of the SEM, no local modification of either the zirconia or
the MaCorTM was observed near the interface.

In addition to the SEM examinations, TEM analysis of the joined specimens was
carried out to provide a more detailed understanding of the interfacial and bulk
microstructures present. At the higher magnifications and resolution offered by TEM, no
silica bond interlayer was observed in the joined 3 mol% yttria-zirconia/MaCorTM
interfacial region (Figure 3.6, 3.7 and 3.8). Presumably the spin-on interlayer silica
and/or any other possible reaction product(s) diffused or dispersed into the surrounding
parent ceramics, leaving a perfectly bonded interface. Figure 3.6 shows an example of a
cross-sectional TEM image, which shows an interlocking microstructure resulting from
gain boundary gooving of the zirconia gains at the zirconia/MaCorTM interface.
Diffraction patterns of the MaCorTM matrix near the interface (Figure 3.7(b)) reveal an
amorphous structure, consistent with that expected for the borosilicate glass phase present

in the as-received MaCorTM specimens. A diffraction pattern from a mica plate near the

interface (Fig. 3.7(a)) shows that this particular plate is oriented to the [311]-zone axis of
the fluorophlogopite mica [Coming]. The 3-mol% yttria-zirconia near the interface has a
gain size about 0.5 pm (Figure 3.8).

The 3 mol% yttria-zirconia gains have been wet and dissolved along the gain
boundaries by the borosilicate glass matrix phase of the MaCorTM, resulting in intimate

bonding between the zirconia and MaCorTM and a non-planer interface with significant

48

interlocking (Figure 3.6, 3.7 and 3.8). The gain boundary gooves observed using TEM,
approximately 0.2 pm deep and 0.4 pm wide, are consistent with the SEM observations
(Figure 3.5).

The joining process does not appear to change the microstructure of either the
MaCorTM or the zirconia near the interface, other than the gain boundary gooves in the
zirconia gains at the zirconia/MaCorTM interface. The interfacial microstructures were
compared to the nricrostructures of the base materials examined at locations away from
the interface. The specimens were examined up to the electron transparency limits
imposed by the sample thinning process. The rrricrostructure at the interface was found
to be similar to the microstructure at these electron transparency limits, 10 um away from
the interface for the zirconia (Figure 3.9) and 100 um for the MaCorTM (Figure 3.10). In
the zirconia (Figure 3.9), the gain size remained consistent at approximately 0.5 pm,
with a distribution of monoclinic plates consistent with the base material. In the
MaCorTM (Figure 3.10), the size distribution and volume fraction of mica plates did not
change with distance from the interface. This shows that the microstructure of both the
zirconia and the MaCorTM near the joint region has not been significantly affected by the
joining process, except for the zirconia gain boundary gooves at the interface.
Microstructures similar to those shown in figure 3.5-3.9 were observed in other joined
zirconia/MaCorTM samples.

3.3.1.2 Indentation tests

In addition to microstructural analysis, Vickers indentation tests were carried out on

the joined zirconia/MaCorTM to estimate the interfacial fracture toughness of the joint at

the goss scale. By placing a series of Vickers indentations in the zirconia side of the

49

interface and comparing the crack lengths at different orientations of the far and near field
interface indentations, the residual stress and interfacial fracture toughness of the joint can
be estimated

Vickers indentation tests (Figure 3.11(a)) showed that the radial crack length near
the zirconia/MaCorTM interface in the zirconia was longer in the direction of parallel to
the interface than perpendicular to the interface. This thermal residual stress profile is
evidently caused by a thermal expansion coefficient difference and elastic mismatch

between the zirconia and MaCorTM. This means that the radial crack experiences a
compressive stress field in the direction of parallel to the interface and a tensile stress
field in the direction perpendicular to the interface. Zirconia, which has a lower thermal
expansion coefficient than that of MaCorTM, experiences a compressive stress in the
direction of parallel to the interface. Correspondingly, MaCorTM experiences a tensile
Stress in the direction of parallel to the interface. Furthermore, the porosity in the
L/IaCorTM near the interface did not appear to affect the crack propagation (Figure 3.11(a)
811d (b)). In addition, due to the Young’s modulus of zirconia being much higher than
that of MaCorm, the indentation cracks in the direction of parallel to the interface
(Figure 3.11(b)) propagated toward the interface [Lardner 1990].

Furthermore, a radial crack (Figure 3.11(c)) placed in the zirconia near the interface
propagated across the interface into the MaCorTM without deflection at the interface,
indicating the interfacial toughness was at least 60 percent of that of the bulk ceramics.
This suggests that this joining method, unlike others [Ureiia 1992, Ferraris 1998], can
result in reasonably strong interfaces without major modification of the parent ceramics.

3.3.2 Conventional joining of zirconia/MaCorTM

50

Similar to the microwave joining of zirconia/MaCorTM using a silica spin-on layer
bonding, conventional joining of polycrystalline 3 mol% yttria-zirconia and commercial
glass ceramics MaCorTM was carried out with and without a spin-on interlayer in relation
to a number of processing variables, such as joining temperature and time. Joining via
conventional versus microwave heating was assessed by comparing the experimental
temperatures and times required for successful bonding. Cross-sectional SEM, TEM, and
indentation tests were used to investigate the interfacial microstructure and interfacial

toughness development of MaCorTM and zirconia joined at different temperatures with or
without the aid of an amorphous interlayer, respectively.
3.3.2.1 Effect of heating mode on the joining of zirconia and MaCorTM
Although MaCorTM and zirconia were completely joined with a spin-on interlayer
by heating at 1020°C for 20 rrrinutes using microwave heating, successful joining of the
hn’IaCorTM and zirconia with the same silica spin-on interlayer was not achieved under the
Same conditions of joining time, temperature, and specimen size, when conventional
heating was employed. Instead, conventional heating at 1070°C for 60 rrrinutes was
I‘Bquired to achieve good bonding (Table 3.1). This difference in joining behavior is most
likely due to differences in the mechanisms involved in microwave and conventional
heating. Although the silica interlayer has a very low dielectric loss factor that makes it
essentially transparent to microwaves, the difference between the dielectric behavior of
the interlayer and surrounding base materials could result in preferential heating at the
interface, resulting in faster interface development during microwave heating [Loehman
1993, 1999]. In contrast, in conventional heating, the heat flows from the surrounding

bulk material to the interface without preferential heating at the interface.

51

3.3.2.2 Conventional joining of MaCorTM/zirconia at 1070°C for 60 minutes
using a spin-on interlayer

Similar to the studies of MaCorTM/zirconia joined with microwave heating, cross-
sectional SEM examination of the MaCorTM/zirconia specimens joined at 1070°C for 60
minutes using conventional heating revealed no observable reaction layers or interfacial
phases in the bond region (Figure 3.12). The MaCorTM/zirconia interface was wavy and

no porosity was observed in the MaCorTM near the interface.
In addition to the SEM examinations, TEM examination of the MaCorTM/zirconia
specimens joined at 1070°C for 60 rrrinutes using conventional heating revealed intimate
contact between the zirconia and MaCorTM without any observable reaction layers in the
j oint (Figure 3.13 and 3.14). This non-planer interface between zirconia and MaCorTM
appears to have resulted from the wetting and dissolution of zirconia along the gain
boundaries (Figure 3.13 and 3.14), similar to the interfacial microstructure of microwave
MaCorTM/zirconia joined at 1020°C for 20 minutes (Figure 3.7). Near the interface, the
mica plates were more rounded than those further from the interface, suggesting the mica
plate morphology changes. Despite this change in morphology, a diffraction pattern
(Figure 3.13(a)) from one of these rounded plates near the interface revealed the

fluorophlogopite (KMg3AlSi3010F2) mica structure, with the particular plate oriented to

[542]. As expected for the borosilicate glass matrix phase of MaCorTM, an amorphous
diffraction pattern is observed from the MaCorTM matrix (Figure 3.13(b)). The
microstructures of the MaCorTM (Figure 3.15) and zirconia (Figure 3.16) in the joined
MaCorTM/zirconia specimens were consistent with those observed in unjoined samples of

IVIaCorTM and zirconia heat treated at 1070°C for 60 minutes, respectively.

52

3.3.2.2.] Indentation tests

As in indentation studies of rrricrowave joined MaCorTM/zirconia, Vickers
indentation testing (Figure 3.17) of the zirconia/MaCorTM joined at 1070°C for 60
minutes using conventional heating shows that a radial crack placed in the zirconia near
the interface propagated across the interface into the MaCorTM without deflection at the
interface, indicating the interfacial toughness was at least 60 percent of that of the bulk
ceramics. This suggests that this joining approach, using either conventional or
Inicrowave heating, can result in reasonably strong interfaces. No significant differences
in the interfacial microstructure and interfacial toughness developed between

conventional versus microwave joined zirconia and MaCorTM.
3.3.2.3 Conventional joining of MaCorTMlzirconia at 1070°C for 120 minutes

using a spin-on interlayer.

Extending the time to join the MaCorTM to zirconia to 120 minutes of conventional
heating at 1070°C using the silica spin-on interlayer resulted in further microstructure
development at the interface (Table 3.1). As with samples joined for 60 minutes, cross-
sectional TEM images revealed no silica bond layer or reaction phases in the joint region
(Figure 3.18 and 3.19). Further rounding of the mica plates near the interface was
observed, forming a nearly continuous mica interfaces layer, with many mica plates now
incorporated into the interface. This morphology dominates the entire observable joint
region (Figure 3.18 and 3.19), suggesting more extensive interlocking. However, the
distorted plates continue to display the fluorophlogopite (KMg3AlSi3010F2) mica

structure, as shown in the figure 3.19. Again, except for the molding of the mica platelets

53

into the zirconia/MaCorTM interface, the morphology of the MaCorTM did not change near
or at a distance from the interface.

3.3.2.4 Conventional joining of MaCorTM/zirconia joined at 1070°C for 60
minutes without the use of a spin-on interlayer.

To determine the role of the spin-on interlayer, samples joined without an interlayer
were compared to samples joined with an interlayer. These comparisons were made
using the MaCorm/zirconia samples joined using conventional furnace heating at
1 070°C for 60 nrinutes (Table 3.1). Without the aid of a silica interlayer, MaCorTM and

zirconia were successfully joined at 1070°C for 60 minutes. Cross-sectional SEM
examinations revealed gain boundary gooving of the zirconia gains without any
observable interfacial phases (Figure 3.20), similar to the interfacial rrricrostructure of the
MaCormlzirconia joined using a spin-on interlayer.

TEM studies of the MaCorTM/zirconia joined without an interlayer showed intimate

bonding between the zirconia and MaCorTM without any interfacial phases (Figure 3.21).
However, near the interface in the MaCorm, no rounding of the mica plates was
observed, which is different than the interfacial microstructure of MaCorTM/zirconia
joined using spin-on interlayer. As expected, the microstructures of the base MaCorTM in
the zirconia/MaCorTM joined with and without a silica interlayer were the same,
consistent with that of unjoined MaCorTM annealed at 1070°C for 60 rrrinutes (Figure
3.22).

The above results show that the zirconia and MaCorTM can be successfully joined

with or without the aid of a silica spin—on interlayer. Thus, despite the fact that a spin-on

interlayer has been shown to geatly enhance the joining of crystalline ceramics [Lee

54

2000], the interlayer, which has an amorphous structure, is not necessary to successfully
bond the zirconia to MaCorTM. This is because MaCorTM is a glass-ceramic composite
containing a 45% volume fraction of amorphous matrix phase and this amorphous matrix
phase has the similar joining characteristics as the silica interlayer. Therefore, no
significant microstructural differences developed between using and without using a
silica spin-on interlayer in the joining of the zirconia and MaCorTM.
3.3.2.4.1 Indentation tests
Similar to the indentation studies of MaCorTM/zirconia joined using a spin-on
interlayer (Figure 3.11), Vickers indentation testing (Figure 3.23) canied out on
MaCorTM/zirconia specimens joined without the use of a silica interlayer showed that a
indentation induced crack placed in the zirconia propagated across the interface into the
I\’IaCorTM without deflection at the interface, indicating reasonably strong interfaces.
This suggests that the joining of zirconia and MaCorTM is not geatly enhanced or
hindered by the presence of a silica spin-on interlayer. Therefore, no significant
microstructural and interfacial toughness differences developed between joining using a
interlayer versus joining without the use of a silica interlayer of zirconia and MaCorTM.
3.3.2.5 Joining of MaCorTMlMaCorTM and zirconia/zirconia.

To further examine the rrricrostructure development of the early stages of the
zirconia and the MaCorTM joining process, zirconia/zirconia and MaCorTM/MaCorTM
specimens were joined using silica spin-on interlayers and conventional heating. When
conventionally heated at 1070°C, zirconia and zirconia failed to join for times as long as
60 nrinutes, but MaCorTM and MaCorTM were successfully joined using a silica spin-on

interlayer, in as little as 10 minutes (Table 3.1).

55

TEM observation of MaCorTM/MaCorTM joined at 1070°C for 10 nrinutes using a
spin-on interlayer revealed a varying microstructure along the interface. At some
locations, an interfacial phase was observed in the middle of the joint region (Figure
3.24). In these regions, the interface between this interfacial phase and the MaCorTM was
wavy and evidence of wetting and dissolution of the MaCorTM along the interface was
observed, suggesting an interfacial reaction has occurred between the MaCorTM and the
silica. The bond region was less than 0.5 pm thick and diffraction pattern analysis
showed that the interfacial phase is amorphous (Figure 3.24 inset). At other locations
along the joint, Mica platelets incorporated into the interface (Figure 3.25).
3.4 Joining of crystalline ceramics with crystalline ceramics
The nature of the interfacial nricrostructures of joined ceramics is key to
understanding the fundamental materials joining process. However, the cross-sectional
SEM, TEM, and indentation studies clearly indicated that the interfacial microstructural
and mechanical property differences between conventional versus microwave joined
zirconia and MaCorTM is trivial due to the glassy nature of MaCorTM. In addition, no
siglificant interfacial rrricrostructural and mechanical property differences developed
between joining with and without using interlayers in the joining of the zirconia and
MaCorTM. Thus, the final objective of current study was to focus on the interfacial
rrricrostructures of joined crystalline ceramics with crystalline ceramics. The nature of
microstructure and interfacial toughness development in the bond region of zirconia and
98% polycrystalline alumina, 98% polycrystalline alumina and 98% polycrystalline
alumina, sapphire and 98% polycrystalline alumina, and sapphire and 96%

polycrystalline alumina were assessed.

56

3.4.1 Effect of heating mode on the joining of alumina/alumina

98% polycrystalline alumina and alumina were joined with a silica spin—on
interlayer by heating at 1340°C for 20 minutes using microwave and conventional
heating under the same conditions (Table 3.2). The joint developed via microwave
heating was successful and robust. However, the joint developed via conventional
heating did not appear very robust, with the samples falling apart during attempts at SEM
sample preparation. Similar to the joining behavior of the zirconia and MaCorm, this
difference in joining behavior is most likely due to the preferential heating at the interface

and faster interface development during microwave heating.
3.4.2 The interfacial microstructure of polycrystalline alumina/alumina joined at

1340°C for 20 minutes

TEM examinations performed on the microwave joined 98% polycrystalline
alumina/98% polycrystalline alumina revealed the interface to be nearly indistinguishable
from the base polycrystalline alumina. The approximate location of the interface can
only be identified due to residual porosity in the interfacial region. Figure 3.26 shows the
interfacial microstructure of joined 98% polycrystalline alumina/alumina via microwave
joining. The gains along the joint were found to be equiaxial and interlocked, with no
evidence of the interlayer or any interfacial phases. This essentially perfect joining made
it impossible to identify the original interface (Figure 3.26). The alumina gains have
equilibrated along the joint, leading to a non-planar interface with a structure comparable
with the bulk gain boundary structure. In addition, no observable microstructural
difference is identified between the 98% polycrystalline alumina near the interface and

the parent ceramics up to the electron transparency limit, 10 um away from the interface

57

for both sides of alumina. This shows that the nricrostructure of the polycrystalline
alumina near the joint region has not been geatly affected by the joining process.
3.4.3 The interfacial microstructure of polycrystalline zirconia/alumina joined at
1500°C for 20 minutes
As in the studies of 98% polycrystalline alumina/alumina joined with microwave
heating, cross-sectional TEM examination of the 3 mol% yttria-zirconia/polycrystalline
alumina specimens joined at 1500°C for 20 rrrinutes using microwave heating revealed no
observable reaction layers or interfacial phases in the joint region. Presumably the
interlayer diffused into the surrounding parent ceramics, leaving a perfectly bonded
interface. Figure 3.27 shows a TEM microgaph of an interface in an alumina/zirconia
Specimen joined at 1500°C for 20 minutes using silica interlayer. In these images, the
interface is wavy and an interlocking microstructure is evident. The interlocking is on
the scale of gain sizes within the specimens. At the higher magrifications offered by
TEM, the interface boundaries appear to have taken up low energy orientations, with
morphologies similar to normal gain boundaries [Case 2001] (Figure 3.28). As before,
the joining process does not geatly affect the microstructures of the base ceramics.
3.4.4 Joining of sapphire and polycrystalline alumina
The work summarized in the following section was carried out to investigate the
interfacial rrricrostructural and interfacial toughness development in joined single crystal
and polycrystalline alumina. Single and polycrystalline alumina were chosen for this
study because the sapphire/polycrystalline alumina interface, unlike in the joining of like
polycrystals (Figure 3.26), will always be distinguishable, even for perfect bonding. In

addition, the effects of alumina purity, joining temperature, time, and the use of

58

interlayers on the interfacial microstructure evolution were assessed using a combination
of 0M, SEM and TEM.
3.4.4.1 Joining of sapphire and 98% polycrystalline alumina

Sapphire and 98% polycrystalline alumina were successfully joined in 180 minutes
at temperatures of 1400°C, 1450°C, and 1475°C with the aid of a silica interlayer. In
contrast, two pairs of sapphire and 98% alumina samples failed to join under the same
conditions without a silica interlayer (Table 3.3), indicating that the silica interlayer
substantially promotes the joining of sapphire and 98% polycrystalline alumina.

Figure 3.29 shows optical microscopy images of 98% alumina/sapphire joined at
1400°C, 1450°C, and 1475°C for 180 rrrinutes using a silica interlayer. With increasing
joining temperature, the thickness of the interfacial reaction layer decreased and joining
was more complete. At 1400°C (Figure 3.29(a)), a reaction layer approximately 5 pm
thick, characterized by regions of both light contrast and larger amounts of dark-gay
contrast, was found in the joint region. Note that this thickness is more than an order of
magritude geater than the spin-on interlayer thickness. Energy dispersive spectroscopy
(EDS) revealed similar spectra for the dark-gay contrast at the joint and the base
mounting material Bakelite. This mounting material most likely flowed into the joint
during the sample polishing and indicates the joining was not complete and voids or
porosity existed between the sapphire and alumina at this stage of joining. In addition,
the joint did not appear very robust, with the samples falling apart during attempts at
TEM sample preparation. Figures 3.29(b) shows cross-sectional optical microscopy

image of 98% alumina and sapphire joined at 1450°C for 180 rrrinutes. The joining was

more complete than in the specimen joined at 1400°C, and only small amount of residual

59

mounting material was observed. Cross-sectional optical microscopy images of 98%
alumina and sapphire joined at 1475°C for 180 minutes (Figures 3.29(c) and (d))
illustrate two slightly different interfacial rrricrostructures in the same joined bulk
specimen. The joining was more complete than in the specimen joined at 1450°C and no
evidence of residual mounting materials was observed. At some locations, the reaction
layer was found to be as thin as 1 pm (Figure 3.29(c)), while at other locations the
reaction layer was up to of 3 pm thick (Figure 3.29(d)).

In addition to the optical microscopy, cross-section TEM examination of the
sapphire/alumina interface joined at 1475°C for 180 rrrinutes was carried out to
investigate if any additional reaction phases were present on a finer scale. Consistent
with the optical microscopy observations (Figure 3.29(c) and (d)), TEM examination of
the sapphire and alumina joined at 1475°C for 180 minutes showed two different
interfacial microstructures in the same joined bulk specimen. At some locations (Figure
3.30(a)), the interlayer between the sapphire and alumina has apparently wet and
dissolved alumina along the grain boundaries, leaving an amorphous interfacial phase
(see inserted diffraction pattern in figure 3.30(a)) along the sapphire and alumina
interface and the near-interface alumina gain boundaries. The wetting penetrated the
alumina gain boundaries to a depth of approximately 1 pm. The amorphous interfacial
layer has a thickness around 0.5~1 um, consistent with some of the optical microscopy
observations (Figure 3.29(c)). The thickness is in the range expected from the
combination of two coated surfaces (each 250~320 nm thick), but may also be enhanced
somewhat by the additional dissolution of the base alumina into the amorphous phase. At

other locations along the interface, a very fine crystalline phase, with a gain size of about

100 nm, was observed (Figure 3.30(b)). In these regions the joint between the sapphire
and alumina has a thickness of approximately 3 pm, again consistent with optical
microscopy observations (Figure 3.29(d)). Diffraction patterns from the joint show fine
rings, typical of randomly oriented fine crystalline gains or particles (Figure 3.30(c)).
These ring patterns were found to be consistent with the (it-A1203 structure. As expected,
no sigrificant rrricrostructural differences were observed, beyond those immediately at
the 2~3 um interfacial region.

To investigate the microstructural evolution, sapphire and 98% alumina joined at
1475°C for 180 minutes were re-heated at 1475°C for an additional 300 minutes. Optical
microscopy revealed close contact between the sapphire and alumina, without any
observable residual reaction layer (Figure 3.31(a)). However, TEM examination (Figure
3.31(b)) found that the amorphous layer had decreased in thickness to approximately
0.1~0.5 pm. The sapphire interface remained smooth, with the alumina interface
becoming wavy, due to continued wetting by the amorphous phase. As before, the base
98% alumina and sapphire showed no microstructural changes for 180+300 minutes heat
treatment at 1475°C. Furthermore, no regions of fine polycrystalline alumina were
observed at this longer joining time.

3.4.4.2 Joining of sapphire and 96% polycrystalline alumina

Similar to the joining of sapphire with 98% polycrystalline alumina, sapphire and
96% polycrystalline alumina were successfully joined together using a silica interlayer at
1475°C for 180 rrrinutes (Table 3.3). However, unlike the joining of sapphire with 98%
polycrystalline alumina, sapphire and 96% polycrystalline alumina were also successfully

joined together without the use of a silica interlayer at 1475°C for 180 minutes. The

61

interfacial rrricrostructures developed in joining sapphire and 96% alumina varied
depending on if an interlayer was used (Figure 3.32). With the aid of a silica interlayer,
the amorphous reaction layer was approximately 0.5~1 pm thick (Figure 3.32(a)). In
contrast, when an interlayer was not used, the resulting amorphous interfacial phase was,
somewhat surprisingly, much thicker, on the order of 5~10 pm thick (Figure 3.32(b)).
The thicker interfacial phase appears to result from a capillary flow of the amorphous
gain boundary phase in the 96% alumina to the interface, driven by the tendency of the
system to restore the equilibrium when a higher energy interface is introduced. It appears
that the silica interlayer could inhibit this flow.

In addition to the SEM examinations, cross-section TEM examination of the
sapphire and 96% alumina interface joined at 1475°C for 180 minutes using a silica
interlayer revealed no silica bond interlayer or interfacial phases in the joint (Figure
3.33). Presumably the silica spin-on interlayer diffused into the surrounding parent
ceramics. In addition, the amorphous interfacial layer has a thickness around 0.5~2 um,
consistent with SEM observations (Figure 3.33). As before, the joining process does not
appear to change the microstructure of the polycrystalline alumina near the interface.

Reheating the interlayer joined sapphire/96% alumina samples for additional 300
minutes resulted in a decrease in the interfacial phase thickness. Figure 3.34 shows (a)
secondary electron and (b) backseattered electron SEM images of the sapphire and 96%
polycrystalline alumina joined at 1475°C for 180+300 rrrinutes using a silica spin-on
interlayer. With this longer joining time, the interfacial phase was no longer observable
and the base polycrystalline alumina and sapphire were in direct contact (Figure 3.34).

TEM analysis of the sapphire/96% polycrystalline alumina joined at 1475°C for 180+300

62

minutes (Figure 3.35) revealed a thin amorphous interfacial phase between the sapphire
and the alumina. The sapphire and the alumina interfaces remained wavy. At some
locations, the polycrystalline alumina gains and sapphire were in direct bonding without
any interfacial phase.
3.4.4.3 Indentation tests

Vickers indentation testing showed that the radial crack length near the 96%
polycrystalline alumina/sapphire interface joined at 1475°C for 180 rrrinutes using a silica
interlayer in the alumina equals in the direction of parallel to the interface and
perpendicular to the interface (Figure 3.36(a)). This means that the thermal residual
stress is negligible. In addition, an indentation induced crack (Figure 3.36(b)) placed in
the polycrystalline alumina near the joint propagated across the polycrystalline
alumina/sapphire interface into the sapphire without deflection, indicating the interfacial
toughness was at least 60 percent of that of the bulk cerarrrics. However, crack deflection
occurred at the polycrystalline alumina/sapphire interface joined at 1475°C for 180
minutes without the use of a silica interlayer, suggesting that the fracture energy of the
interface is less than about 10~20 % of the bulk materials [Lee 1996]. This difference
suggests that the silica interlayer dramatically promotes the joining of sapphire and 96%
polycrystalline alumina, resulting a relatively strong interface.

Reheating the interlayer joined sapphire/96% alumina samples for additional 300
minutes resulted in a similar behavior in the indentation tests. The indentation induced
crack propagated across the interface with no deflection (Figure 3.37). In addition, the

radial crack length near the polycrystalline alumina/sapphire interface in the alumina was

63

equal in the direction of parallel and perpendicular to the interface, indicating that the
thermal residual stresses due to joining are minimum.

Unfortunately, indentation tests were not carried out on the joined 98%
polycrystalline alumina/sapphire because the limited number of joined specimens were

instead sectioned for TEM analysis.

Table 3.1 The experimental condition and results for ceramics/ceramics joining via
conventional heating in the present study. Please note, all experiments were done twice
to verify joining results.

 

 

 

 

 

 

 

 

 

 

 

Specimens to be Joining Joining .
joined tempféature time (min) Coating Results
MaCorTM/zirconia 1020 20 SilicafilmTM Failed to join
MaCorm/zirconia 1020 40 SilicafilrnTM Failed to join
MaCorT'W/zirconia 1020 60 SilicaiilrnTM Failed to join
MaCorm/zirconia 1070 20 SilicafiltnTM Failed to join
MaCorm/zirconia 1070 40 SilicafilmT’.VI Failed to join
MaCorTM/zirconia 1070 60 SilicafilmTM Joined
MaCorTM/zirconia 1070 60 No coating Joined
MaCoF‘M/zirconia 1070 120 SilicafilmTM Joined
MaCormlMaCorm 1070 10 SilicafilrnTM Joined
zirconia/zirconia 1070 60 SilicafilmTM Failed to Join

 

 

 

 

 

 

 

65

Table 3.2 outlines the experimental conditions and results for the joining of 3 mol%
yttria-zirconia/polycrystalline alumina, and 98% polycrystalline alumina/98%
polycrystalline alumina. Please note, all experiments were done twice to verify joining
results.

 

 

 

 

 

 

 

 

 

 

Specimens to be Joining Joining

. . temperature time Coating Heating mode Results

joined 0 .

(C) (mm)

P°l¥°rysm"".‘° 1340 20 SilicaiilmTM Microwave Joined
alumrna/alumrna
Polycrystalline 1340 180 SilicafilmTM Conventional Joined
alurmna/alumrna
zirconia/alumina 1500 20 Silicafilm1M Microwave Joined

 

 

Table 3.3 The experimental conditions and results for the joining of sapphire and
polycrystalline alumina via conventional heating in the current research. Please note, all

experiments were done twice to verify joining results.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Joinin . . .
Specimens to be joined temperatire 1211123236 Coating Heating mode Results
(°-C)
Sapphire/98% alumina 1400 180 SilicafilmTM Microwave Joined
Sapphire/98% alumina 1475 180 SilicafilmTM Microwave Joined
Sapphire/98% alumina 1475 180 No coating Microwave Failed
Sapphire/98% alumina 1475 180+300 SilicafilmTM Microwave Joined
Sapphire/96% alumina 1475 180 Silicafilrnm Microwave Joined
Sapphire/96% alumina 1475 180 No coating Microwave Joined
Sapphire/96% alumina 1475 180+300 SilicafilmTM Microwave Joined

 

66

 

 

Figure 3.1 TEM images of the sintered as—received glass-ceramics MaCorTM (a) and
3 mol% yttria—zirconia specimens (b).

67

 

Figure 3.2 A cross—sectional TEM image showing the polished surface of the
s i ntered, and cleaned 3 mol% yttria-zirconia specimens prior to coating showing the
s urface gains morphology.

68

 

(b)

Figure 3.3 TEM images of the as-received microstructures of 98% (a) and 96% (b)
polycrystalline alumina. The commercial 96% polycrystalline alumina (b) consists of
platelike or elongated or-Al203 gains, with approximately 20% interganular gain
boundary phase. The diffraction pattern of the interganular gain boundary phase
reveals an amorphous structure, consistent with that expected for the aluminosilicate
glass [Brydson 1998]. However the 98% alumina (a) contains equiaxial or—Al203 gains
with little gain boundary phase.

69

 

Figure 3.4 SEM image of silica interlayer showing the interlayer has a thickness in
tl'ie range of 250~320 nm thick after spinning at 3000 rpm and curing at 200°C for 20
minutes.

 

Figure 3.5 An SEM microgaph of the microwave joined zirconia/MaCorTM
showing the bond interface. At the goss scale the interface is straight, locally the
zirconia contains grain boundary gooves. No interfacial phase(s) were observed.

70

fink-bee

    

milling hole _ - C?“

Figure 3.6 Low magnification TEM image of the interface between the 3 mol%
yttria-zirconia and the MaCorTM joined at 1020°C for 20 minutes using microwave
heating. Differential ion milling in the specimens results in the MaCorTM being thinned

back with respect to the zirconia, resulting in occasional perforation of the MaCorTM
specimen near the interface.

‘ in terfiac -

I
is,
M

 

Figure 3.7 A TEM image of the MaCorTM near the interface. Inserts showing (a) a

diffraction pattern from a single mica plate and (b) a diffraction pattern from matrix areas
between the mica plates.

71

interface
’/

/
//

 

Figure 3.8 A TEM image of the interface between the 3 mol% yttria-zirconia and
the MaCorTM showing that the zirconia has been strongly wetted and dissolved along
grain boundaries, resulting in a non-planer interface.

 

Figure 3.9 A TEM image of the microstructure of the zirconia in a joined 3 mol%
yttria-zirconia/MaCorTM specimen at a location at about 10 um away from the interface.
The nricrostructure of the zirconia at this location is similar to that of the zirconia
rrricrostructure near the interface (Figure 3.6).

72

 

Figure 3.10 A TEM image of the MaCorTM microstructure in a joined 3 mol%
y'tnria—zirconia/MaCorTM specimen at a location at about 100 um away from the interface.
'I‘lre microstructure at this location is similar to the rrricrostructure of the MaCorTM near

the interface (Figure 3.5). Due to a range of mica platelet orientations, the apparent
di mensions differ from platelet to platelet.

73

L/

if
'1
.. .
t:
..
=
_ ..
.17
._
..
, _.
.5.
, -
_ 5.
a...-
_.
.-
a.
4.
..
..
..

ii‘iliii

 

Figure 3.11 A Vickers indentation testing shows that the radial crack near the
microwave joined zirconia/MaCorTM interface with (a) and without (b) observable
porosity in the MaCorTM has different length on the vertical and horizontal directions, and
(c) propagates across the interface from zirconia into the MaCorTM without deflection

74

 

Figure 3.12 An SEM micrograph of the zirconia/MaCorTM joined at 1070°C for 60
njinutes using a silica spin-on interlayer showing the bond interface.

 

Figure 3.13 A TEM image of the zirconia/MaCorTM joined at 1070°C for 60
rrrinutes using a silica spin—on interlayer. Diffraction pattern from one of the rounded

plates near the interface revealed the mica structure (a), oriented to [542]. Diffraction
pattern from the matrix of MaCorTM revealed an amorphous structure (b).

75

 

Figure 3.14 The interfacial microstructure of the zirconia/MaCorTM joined at
1 (170°C for 60 minutes using a silica min-on interlayer showing intimate bonding
between zirconia and the matrix of MaCor .

 

Figure 3.15 A TEM image of the bulk MaCorTM microstructure. The center of the
image is about 5 um away from the zirconia/MaCorTM interface. The specimens were
joined at 1070°C for 60 rrrinutes using a spin—on interlayer.

76

 

Figure 3.16 A TEM image of the bulk zirconia microstructure. The center of the
image is about 5 pm away from the zirconia/MaCorTM interface. The diffraction pattern
is consistent with a two phases partially stabilized zirconia structure.

 

- Figure 3.17 A Vickers indentation testing shows that the radial crack near the joined
21 1‘ Conia/MaCorTM interface propagates across the interface from zirconia into the
aCorTM without deflection.

77

 

Figure 3.18 TEM brightfield microgaph of the zirconia/MaCorTM interface joined
at 1070°C for 120 minutes using a silica spin-on interlayer shows rrrica plates
incorporated into the MaCorTM/zirconia interface.

 

Figure 3.19 The interfacial rrricrostructure of MaCorTM and zirconia joined at
1 070°C for 120 rrrinutes using a silica spin-on interlayer. Insert showing that selected
area diffraction pattern of the distorted plates continue to display the mica structure.

78

erect/cw

 

Figure 3.20 An SEM microgaph of the zirconia/MaCorTM joined at 1070°C for 60
nrinutes without the use of a silica interlayer showing the bond interface.

 

Figure 3.21 A TEM image of the zirconia and MaCorTM joined at 1070°C for 60
minutes without the use of a silica interlayer. No interfacial phase(s) were observed at
ale interface.

79

 

Figure 3.22 A TEM microgaph of MaCorTM (unjoined control) that was heated at
1 070°C for 60 minutes. Insert shows the diffraction pattern from a single mica platelet
([241] orientation). MaCorTM heated at 1070°C for 120 minutes showed similar
rnicrostructure.

 

_ Figure 3.23 An SEM image of the Vickers indentation crack on the
,Z1I‘COnia/MaCorTM joined at 1070°C for 60 minutes without the use of a silica spin-on
l‘jferlayer in the zirconia side showing the radial crack propagates across the interface
“’1 thout deflection.

80

W

.2.

 

Figure 3.24 The interfacial microstructure of MaCorTM and MaCorTM joined at
1 070°C for 10 minutes using a silica interlayer showing an amorphous interfacial region,
as indicated by the inset diffraction pattern.

81

 

82

 

Figure 3.25 A series of images of the MaCorTM and MaCorTM joined at 1070°C for
l 0 minutes using a silica interlayer (a) optical microgaph of the TEM thin foil showing
the MaCorm/MaCorTM interface, (b) higher mggnification optical micrograph of the
TEM thin foil showing the MaCorTM/MaCor interface from box A, (c) low
magnification TEM image from the area indicated on (b) showing the foil edge and
interfacial region, (d) higher magnification TEM image from the area indicated on (c)
Showing the foil interfacial region, and (e) further higher magnification image showing
Inica plates incorporate into interface.

83

 

(b)

Figure 3.26 The interfacial microstructure of the microwave joined 98%
polycrystalline alumina/98% polycrystalline alumina using a silica interlayer silica
interlayer (a) optical microgaph of the TEM thin foil showing the alumina/alumina
interface, (b) higher magnification TEM microgaph showing the equiaxed alumina
gains cross the whole joint region, making the interface very difficult to distinguish from
the base material.

84

 

Ma©s

Figure 3.27 Low magnification TEM image of the interface between the 3 mol%
yttria-zirconia and the polycrystalline alumina joined at 1500°C for 20 minutes using
microwave heating.

 

Figure 3.28 A TEM image of the interface between the 3 mol% yttria-zirconia and
the polycrystalline alumina showing that no observable reaction layers or interfacial
phases in the join region.

85

 

86

@0313 [it‘ll ,r0‘

3* 41

 

(d)

Figure 3.29 Optical microscopy images of sapphire/98% polycrystalline alumina
joined at (a) 1400°C for 180 minutes, (b) 1450°C for 180 minutes, and (c, d) 1475°C for
180 rrrinutes using silica interlayer that illustrate two different interfacial microstructures
on the same joined bulk specimen. At some locations, the reaction layer is approximately
3 pm thick (c) while at other location reaction layer has a thickness only around 1 um (d).

87

3W

Sagmfl

 

88

 

 

 

Figure 3.30 TEM images of sapphire/98% polycrystalline alumina joined at 1475°C
for 180 minutes revealed two distinct interfacial microstructures at the joint. At some
locations (a), the interfacial phase has a thickness around 0.5~1 um, while at other
locations, a fine distribution of alumina crystalline phase occurred at the joint (b) with
indexed diffraction pattern (c).

89

 

Figure 3.31 (a) Optical microscopy and (b) TEM images of the sapphire/98%
polycrystalline alumina joined at 1475°C for 180+300 minutes using a silica spin-on
interlayer.

Sapphire

s.

M.

 

Figure 3.32 SEM images of the sapphire/96% polycrystalline alumina joined at
1475°C for 180 minutes (a) with and (b) without the use of a silica spin-on interlayer.

91

 

Figure 3.33 A TEM image of sapphire/96% polycrystalline alumina joined at
1475°C for 180 minutes using a silica interlayer. Insert shows the diffraction pattern
from the amorphous interlayer.

92

Sapphire

interface

 

(b)

Figure 3.34 (a) Secondary and (b) backseattered electron SEM images of the
sapphire/96% polycrystalline alumina joined at 1475°C for 180+300 minutes using a
silica spin-on interlayer. The bright contrast in the polycrystalline alumina in the BSEM
image (b) corresponds to gain boundary phases, the gay contrast corresponds to or-
Al203 phase, and the black contrast corresponds to porosity.

93

Sapphire

interface

 

Figure 3.35 A TEM image of the sapphire/96% polycrystalline alumina joined at
1475°C for 180+300 minutes using a silica spin-on interlayer.

94

 

interface
9'

Sapphire

Sap o hire

”a .
.. -. -__
:. .~ . 5‘: ~

,

_.v
fife-uran-

i‘

 

it”

(b)

Figure 3.36 An optical microscopy (a) and SEM (b) images of the Vickers
indentation crack on the sapphire/polycrystalline alumina joined at 1450°C for 180
minutes using a silica spin-on interlayer in the alumina side neat the joint showing the
radial crack propagates across the interface from polycrystalline alumina into the
sapphire without deflection.

95

 

Sapphire

P

 

Figure 3.37 An optical microscopy image of the Vickers indentation crack on the
sapphire/polycrystalline alumina joined at 1450°C for 180+300 minutes using a silica
spin-on interlayer in the alumina side neat the joint.

96

Chapter IV

Discussion

4.1 Spin-on interlayer joining mechanisms

The results presented in Chapter 111 show that joining involving MaCorTM is not
substantially enhanced or hindered by the presence of a spin-on bonding agent. This
effect is very different from the joining of crystalline ceramics, such as high purity 98%
polycrystalline alumina and sapphire, where bonding was not achieved without the use of
a silica spin-on interlayer. However, sapphire and low purity 96% polycrystalline
alumina can be successfully joined together with or without the use of a silica interlayer.
Thus, it is of interest to compare the nature of bonding in these two classes of ceramics in
relation to previous ceramic/ceramic joining studies, pertinent phase diagams, and
microstructure studies of other crystalline ceramics and glass ceramics that are similar in
composition to MaCorTM. In addition, it is critical to fully characterize the evolution of
the joining process from initial sputter-cure to final joining state in order to further the
understanding of ceramic joining mechanisms, reaction kinetics, and the development of
reliable ceramic joining approaches.

4.1.1 Surface and interface diffusion

Despite gowing interest in joining ceramics, there has been little research on the
detailed mechanisms or modeling of ceramic-ceramic joining. Consequently, process
conditions are often chosen by trial and error, rather than from scientific predictions that
ensure complete bonding. The reason for the trial and error approach may be related to

the complicated diffusion, mass transport, reaction, and interface formation mechanisms.

97

For example, Derby and Wallach [1982, 1984] developed a theoretical model of the
simplest joining process, metal-metal diffusion bonding without reaction phase
formation. They assumed that diffusion bonding has some similarity with powder
pressure sintering processes. The geometry of the contacting surfaces was simplified as a
series of asperities touching at their tips to form holes, voids and pores [Wallach 1989].
The difference in free energy and chemical potential between the contacting area and the
surface of the asperities fornring the voids provides a driving force which causes the
diffusion of materials, similar to mass diffusion in powder pressure sintering process
[Kingery 1976, Barsoum 1997]. Thus, they postulated that following three bonding
mechanisms could be initiated to close the interface porosities (Figure 4.1) [Derby 1982,
1984]:

0 Mechanism 1: Surface and volume diffusion from surface source to neck, driven
by differences in surface curvatures across the surface of an interfacial void.

0 Mechanism 2: Boundary diffusion and volume diffusion along the bond
interface from interface source to neck, driven by the chemical potential
differences along the bond line.

0 Mechanism 3: Local yielding or creep, driven by the applied force during
diffusion joining.

Figure 4.1 shows a schematic representation of the three mass transport
mechanisms for interface formation. However, for heterogeneous metal-ceramic or
ceramic—ceramic bond formation, additional processes certainly need to be considered.
For example, interdiffusion between bulk materials, a possible eutectic reaction at the

interface to form a liquid phase, wetting and dissolution of the ceramic or metal phase

98

into the liquid phase, diffusion activation energies, maximum solubility, and chemical
reactions at interface to form reaction products. Each of these mass-transport
mechanisms has to be taken into account to understand the detailed bonding interface
formation process. These factors suggest that diffusion is crucial to the understanding of
bonding, interface formation, and interface morphology.

In the current research, when two flat surfaces, such as zirconia and MaCorTM,
silica and MaCorm, and silica and alumina, were placed in contact, voids, porosity, or
channels were formed. This is consistent with what would form between ridges, and is
consistent with the original polished surface roughness being as a series of straight-sided
ridges [Derby 1982, 1984]. At elevated temperatures, driven by differences in surface
curvatures and chemical potential, mechanisms 1 and 2, surface and volume diffusion
from high radius ridge surface source to ridge neck area and/or along the bond interface,
can be initiated, leading to a gadual shrinking of these pores or voids (Figure 4.1).
However, the current spin-on interlayer joining approach uses very small nominal joining
pressure which is in the range of 2000~8500 Pa. Thus, mechanism 3, local yielding or
creep requiring a large stress concentration, is not likely to occur at the interface.
Therefore, as bonding progesses, the mass transport across the interface via mechanisms
1 and 2, surface and volume diffusion, resulted in the gadual shrinkage of interfacial
voids and interdiffusion between similar and dis-similar ceramics, such as silica and
MaCorTM, and silica and alumina.

4.1.2 Interfacial eutectic reactions
In the current study, the silica interlayer would not become liquid or spread through

wetting action along the interface, to assist the bonding at the joining temperature of

99

1020°C, due to its high melting point (1720°C) and high viscosity at this temperature
[Emulsitone]. However, the interfacial interdiffusion between similar and dis-similar
ceramics continues, and as such, eutectic liquid phase formation could occur at the
silica/MaCorTM and silica/alumina interfaces. The interfacial liquid phases may increase
interfacial wetting and dissolution, and assist the interface formation since wetting of
surface is a measure of the adhesion of the two joining materials [Burger 1998].
4.1.2.1 Silica and MaCorTM

According to the SiO2-MgO-K2O-A12O3 equilibrium quaternary phase diagam
[Schairper 1954], liquid is present down to the ternary eutectic temperature of 715°C in
the three phases field of K20-MgO-SSiO2\SiO2\K2O-4SiO2 (Figure 4.2). Likewise,
liquid is present down to the ternary eutectic temperature of 963°C in the three phase
field of K20-5Mg0-12SiO2\Si02\K20-MgO-5Si02, while liquid is also present to the
ternary eutectic temperature of 1013°C in the three phase field of
2Mg0-Si02\K2O-Mg0-SSiO2\K2O-MgO-5SiO2. MaCorTM contains 47.2% Si02, 16.7%
A1203, 9.5% K20, 14.5% MgO, and 8.5% B203 (wt%) [Pinckney 1991]. Thus, as the
silica interlayer reacts with the MaCorTM at the joining temperatures of 1020 or 1070°C,
it is likely that a liquid silicate phase will form at the silica/MaCorm interface.

4.1.2.2 Silica and alumina

Similar to the eutectic reaction between the silica spin-on interlayer and the
borosilicate glass phase that may result in liquid phase formation, a eutectic reaction
between the silica and the alumina phase (either polycrystalline or single crystal sapphire)
could occur at the silica/polycrystalline alumina interface. According to the equilibrium

A1203-Si02 phase diagram [Macdowell 1969], a liquid interfacial phase reaction between

100

SiO2 and high purity bulk A1203 should only occur at temperatures higher than 1595°C.
However, Sacks et al. [Sacks] found that fine micro-powders (around 200 nm) of alumina
and silica could react at much lower temperatures, leading to the formation of mullite
(3Al203-28i02) at temperatures as low as 1400°C, as a consequence of interfacial liquid
phase formation [Davis 1972]. Thus, a liquid interfacial phase reaction could occur at the
joining temperature of 1475°C used in the present study.
4.1.2.2.1 Effect of impurities

The temperatures at which liquid interfacial eutectic phase reactions can occur
between A1203 (either polycrystalline or single crystal sapphire) and the Si02 interlayer
depends on the particle size and the actual impurities present in the base component (such
as magnesium oxide, calcium oxide, and silicon oxide) [Sacks 1996, 1997, Nakajima
1998, Staley 1969]. Impurities in the alumina may further lower the interfacial liquid
phase formation temperature between the silica interlayer and the polycrystalline alumina
in the same manner as small quantities of CaO, Si02, and MgO have been found to lower
the liquid phase sintering temperature of alumina [Brydson, Nakajima 1998]. Because
CaO is typically present as an impurity in 96% alumina, a eutectic liquid phase could
form in the SiO2-Ca0-SiO2-Ca0-Al203'28i02 phase field at temperatures above 1170°C
or in the SiO2-3A1203-28i02-CaO-Al203-2Si02 phase field at temperatures above 1368°C
[Levin 1964]. Thus, as a result of impurities, the formation of a liquid phase at
silica/96% polycrystalline alumina might be expected.

4.1.3 Wetting and dissolutions
The eutectic liquid phase formation at the zirconia/MaCorTM and silica/alumina

interfaces could expedite the mass transfer across the interfaces and the formation of

101

thermal gooving. Generally, thermal gooving forms at the interface between a
polycrystalline solid phase and a liquid phase whenever a gain boundary meets the
interface [Mullins 1957, 1958, 1960]. The thermodynamic driving force for the
formation of thermal gooving is to develop an equilibrium condition of interfacial forces

at solid/liquid triple junctions or goove root (solid/liquid interfacial energy 7..., and gain

boundary energy 7gb) (Figure 4.3(a)). As investigated by Mullins [Mullins 1957, 1958,
1960], the formation and gowth of interfacial thermal gooving could take place through
the following three mechanisms: (1) interfacial diffusion, (2) volume diffusion, and (3)
interfacial reactions such as solution and precipitation, and evaporation and condensation
for solid/liquid and solid/gas interfaces, respectively. In an idealized case where
liquid/solid diffusion is dominated by one of the diffusion mechanisms, Mullins [1957,
1958, 1960] predicted that each thermal goove profile shape is determinative and
independent of time. Figure 4.3 summarizes the expected thermal goove shape based on
Mullin's analysis. When interfacial reactions are the dominant diffusion mechanism, the
goove surface lies everywhere below the original polished surface (Figure 4.3(a)).
However, in a case where the dominant mass transport mechanism for thermal gooving
is either interfacial diffusion or volume diffusion, a hump develops adjacent to the goove
that can lie above the original polished surface (Figure 4.3(b) and (c)). In addition, the
expected dependence of the goove depth and width with solutionization time is linear.
For example, the goove profile is proportional to t”2 for interfacial reactions mechanism,
to t"3 for volume diffusion, and to t"4 for interfacial diffusion, where t is the time

[Mullins 1957, 1958, 1960].

102

In the current study of joining of zirconia and MaCorTM via microwave (Figure 3.8)
and conventional heating (Figure 3.13), polycrystalline alumina and sapphire (Figure
3.27), cross-sectional TEM observations revealed that the gain boundary goove shapes
lack the humps above the original flat surfaces on either side, indicating mass-transport is
mostly controlled by interfacial reactions. However, it is impossible to calculate the
theoretical values of thermal goove depth since the physical and chemical characteristics
of eutectic liquid phases were unknown (Figure 4.2 and 4.3).

4.1.4 Interface formation

The wetting experiments performed by Saiz et al. [1998] with molten drops of
copper and nickel on polycrystalline alumina revealed mass transport involving gain
boundary gooves near the metal/alumina interfaces is two to four orders of magritude
faster than on the free surface of the alumina. In addition, many authors [Dennalay 1987,
Dalgleish 1994] illustrated theoretically and confirmed experimentally that there is a
direct correlation between a good wetting and a strong interfacial bonding that results in a
strong joint strength.

Thus, it is expected that the mass transport involving gain boundary gooves due to
eutectic liquid phase formation could accelerate the interfacial interdiffusion across the
interface. As many authors postulate the diffusion bonding process as a process analog to
powder pressure sintering processes, the spin-on interlayer joining approach used in the
current study can be viewed as a process analog to liquid phase sintering process. In
liquid phase sintering, the presence of liquid phase provides a capillary force that pulls
the solid particles together and promotes rapid densification in the compacts [Kingery

1976, Barsoum 1997]. However, for these effects to occur it is necessary to “have an

103

appreciable solubility and wetting of the solid in the liquid” [Kingery 1976]. The cross-
sectional TEM observations revealed that liquid phases apparently wet the zirconia and
alumina in the joining of zirconia and MaCorTM (Figure 3.8 and 3.13), and sapphire and
alumina (Figure 3.27 and 3.30), respectively. Therefore, it is likely that liquid phases due
to eutectic reactions will accelerate the ceramic-ceramic joining process.

Similar to sintering in the presence of a liquid phase [Kingery 1976, Barsoum
1997], the presence of a liquid phase during the joining could (I) generate an attractive
force between neighboring ridges across the joining interface due to a capillary force
resulting from surface tension, (2) introduce an additional mass transport mechanism
through the liquid phase via dissolution of small particles and gowth of large particles,
(3) filling of interfacial pores or voids. Thus, the presence of eutectic liquid phase at the
joint interfaces could expedite the mass transport across the joining interface, resulting in
a dramatic enhancement of interface formation. Therefore, the interfacial reactions or
wetting and dissolution that occurred during the joining of zirconia/MacorTM and
silica/alumina could result in a dramatic enhancement in the joining sapphire and
alumina, and alumina/alumina.

4.1.5 Phase development at the join interface

In the current spin-on interlayer joining approach, the silica interlayer thickness is
in the range expected from the combination of two coated surfaces (each 250~320 nm
thick). The presence of a eutectic liquid phase could expedite the materials transport
across the interfaces and bond formation. Thus, the additional dissolution of the base
ceramics into the amorphous eutectic liquid phase can broaden the reaction layer. For

example, optical microscopy revealed that a reaction layer approximately 5 pm thick was

104

found at the joint region in 98% alumina/sapphire joined at 1400°C for 180 rrrinutes using.
a silica interlayer (Figure 3.29(d)). This reaction layer thickness is more than an order of
magnitude geater than the spin-on interlayer thickness.

However, following cooling from the joining temperatures, a variety of interfacial
microstructures have been observed. For example, rrricrostructural analysis revealed that
no observable reaction phase or layer is observed for joining involving MaCorTM with
and without the use of a silica spin—on interlayer, while joining of sapphire and 98%
polycrystalline alumina, and sapphire and 96% polycrystalline alumina showed
interfacial layers and/or non—equilibrium alumina phase. Although the equilibrium phase
diagam illustrates the phases expected to develop under equilibrium conditions, the
joining in the current study may be far from equilibrium conditions. Thus, it is beneficial
to the development of joining procedures to investigate the details of the different joining
scenery based on previous ceramic/ceramic joining studies, pertinent phase diagams, and
microstructure studies.

4.1.5.1 Joining of zirconia/MaCorTM

Baik et al. [1995] reported that the microstructure of fluorophlogopite mica-based
glass-ceramics, containing 50% 8102, 20% A1203, 8% K20, 15% MgO, 0.9% U20 and
6.1% F (wt%) similar to MaCorm, is not affected by 2-wt% Zr02 addition, but at a 4%
addition level, crystallization is inhibited and spontaneous opalization occurs at 6%
addition level. Thus, the addition of zirconia to the glass ceramic can (1) leave the
nricrostructure unchanged, (2) inhibit crystallization, or (3) induce spherical precipitates
(opalization) [Baik 1995]. The TEM studies presented (Figure 3.9 and 3.10) indicate that

the microstructures of both the MaCorTM and the zirconia are unchanged near the

105

interface, which may indicate limited diffusion of zirconia into the liquid that is likely
formed (according to the equilibrium phase diagams) at the silica/MaCorTM interface.
Therefore, during the following cooling after joining, the liquid silicate phase can form
an amorphous matrix and fluorophlogopite mica platelet phase without additional
interfacial phases or reaction layers.
4.1.5.2 Joining of sapphire and polycrystalline alumina

In the joining of polycrystalline alumina and sapphire, equilibrium may not have
been reached due to slow reactions. Furthermore, impurities can shift the phase
boundaries significantly. Many researchers [Risbud 1977, 1978, Aksaky 1975] have
illustrated theoretically and confirmed experimentally the development of metastable
phases in the SiO2-Al203 system. This metastable system forms a simple eutectic of
Si02-Al203 (eutectic point: ~12 mol% A1203 at ~1260°C), without the development of
equilibrium mullite phase. This metastable behavior may explain why no mullite
formation has been observed at the SiO2-Al203 interfaces in the joining of sapphire and
polycrystalline alumina. Furthermore, the oxide impurities CaO and BaO, which are
present in the as-received 96% and 98% alumina, have been shown to inhibit the
devitrification of glassy phases during cooling [Macdowell 1969]. This may also be a
factor in inhibiting the formation of mullite at the interface. Instead of forming mullite, a
number of researchers [Risbud 1978, Aksaky 1975, Kennard 1972] have found that
alumina can precipitate from the glassy phase during cooling. This is consistent with the
observation of the fine alumina gains developed at some points in the joint region (figure
3.27).

4.2 Microwave vs. conventional joining

106

Microwave heating is fundamentally different from conventional heating. In
microwave heating, heat is generated from the electromagnetic energy losses associated
with the interactions between the internal electric fields and induced molecular motions.
This results in the heat being generated within the material, rather than flowing in from
the surface, as in the case for conventional heating. When compared to conventionally
heated materials, materials will be heated more uniformly, and typically more rapidly.
This heating results in reduced processing times and energy costs within the limits set by
microwave absorption, when heated using microwaves [Loehman 1999]. Thus, while
microwave heating results in joining at lower nominal temperatures in the present study,
the effective time at the joining temperature maybe longer than experienced with
conventional heating. However, further localized heating may be occuning at the joined
interfaces as a result of differences in the dielectric loss factor (tan 8) [Lee and Lin 1998].

4.2.1 Joining zirconia and MaCorTM

The available information on dielectric loss factors for the joining of zirconia and
MaCorTM using a silica interlayer used in this study is limited (See Table 4.1). However,
it is clear that the dielectric loss factor (tan 8) is much lower in the silica interlayer at
1020°C and 2.45 GHz than it is in the zirconia (a comparable value for MaCorTM is not
available) (Table 4.1). Given this, it might be expected that the silica interlayer would
not heat as rapidly as the neighboring zirconia.

4.2.2 Joining of 98% polycrystalline alumina and alumina

Alumina, as a low loss ceramic, is transparent to microwaves at ambient

temperatures [Sutton 1992]. Initially, the dielectric loss factor (tan 6) of the alumina rises

slowly with increasing temperatures. However, above a critical temperature which is

107

approximately 1000°C for alumina, tan 8 rises rapidly [Sutton 1992]. Thus, alumina can
absorb and couple more efficiently with microwave radiation, resulting in more effective
heating at high temperatures. The rapid increase of tan 8 at high temperatures is
associated with the softening of the gain boundary amorphous phase in alumina, causing
an increase in a local electrical conductivity, as reported by Ho [1988]. Thus, the tan 8
of higher purity alumina is less affected by temperature than that for the lower purity
alurrrina. Therefore, in the joining of 98% polycrystalline alumina and alumina using a
silica interlayer, the interganular amorphous phase in the 98% alumina could result in a
the rapid rise of tan 8, resulting in more effective heating at high temperatures. The
available information on the tan 8 for the 98% alumina is limited (Table 4.2). However,
it is expected that the tan 8 for the 98% alumina be between that of lower purity alumina
AD85 and higher purity AD99.5. Furthermore, the tan 8 is much lower in the silica
interlayer at high temperatures and 2.45 6112 than it is in the alumina (Table 4.2).
Therefore, it is expected that the silica interlayer would not heat as rapidly as the
neighboring polycrystalline alumina.
4.2.3 Microwave joining

In the microwave joining, it is known that large dielectric losses occur at interfaces
between materials with large differences in dielectric loss factors, leading to an additional
term in the total dielectric loss factor and large increase of tan 8 at the interface [Lee and
Lin 1998]. Thus, significant local heating might be expected at the silica/zirconia and
silica/MaCorTM interfaces. Similar differences and local heating might be expected at the

silica/alumina interface, which may further push up the local temperature at the joint

108

relative to the nominal joining temperature. As the interlayer is very thin (less than 1
pm), it should also absorb heat readily from the surrounding, rapidly heated materials.

A number of researchers [Palaith 1989, Janney 1991, Fanslow 1990] have reported
that microwave processing produces chemical and physical reactions that are different
from the reactions that would have occurred if the only effect of the microwaves were to
increase the temperature. For example, improved microstructures and mechanical
properties [J anney 1991, Fanslow 1990, Patterson 1992], enhanced chemical and physical
reactions [Fanslow 1990], increased diffusion coefficients [Janney 1991] and lowered
sintering activation energies [Janney 1988] have been observed in microwave heating.
These enhancements are known as the “microwave effect”. Janney et al. [1988] have
speculated that these enhancements occur as a result of the electromagnetic field
“coupling to lattice defects and other bulk crystalline effects”. The present results that
indicate joining of zirconia/MaCor1M and 98% polycrystalline alumina/alumina can be
achieved at shorter times and at lower nominal temperatures using microwave heating
may also be a result of such athermal microwave effects. Booske and co-workers
[Booske 1997, 1998, Freeman 1995, 1998] illustrated theoretically and numerically, and
confirmed experimentally that microwave fields induce an additional driving force
(ponderomotive force or pmf) that enhances the mass transport rate and ionic diffusion
compared to conventional heating mode. The pmf can occur at any abrupt
discontinuities, such as surfaces or interfaces. The authors [Booske 1997, 1998, Freeman
1995, 1998] state that the value of the pmf is higher than the solid-state diffusion driving

force for joining of two flat NaCl/NaCl surfaces. Thus, ceramic-ceramic joining

109

processes could be substantially enhanced since diffusion is crucial to the development of
eutectic liquid formation, wetting, dissolution, and interface formation.

Therefore, the additional pmf athermal driving force for mass transport across the
zirconia/silica, MaCorTM/silica, and silica/alumina interfaces, most likely enhances the
interfacial diffusion rates and mass transport across the interfaces to a large degee. In
addition, preferential interfacial heating can occur due to the large dielectric losses at
interfaces [Lee and Lin 1998]. These conditions can result in a lower nominal
microwave joining temperature, consistent with experimental results presented in this
thesis that show that microwave heating has geatly enhanced the joining of
zirconia/MaCorTM, 98% polycrystalline alumina/alumina although the silica interlayer is
essentially transparent to the microwave field. Furthermore, it is also expected that
microwave heating would assist the bonding of the zirconia and MaCorTM without the use
of a silica interlayer, due to increased mass transfer across the interface.

4.3 Joining without the use of interlayers

For direct ceramic-ceramic joining without the use of any interlayer, adhesion
generally takes place through high temperature and pressure diffusion bonding or as a
result of an interganular amorphous phase present in the base materials rrrigating to the
joint [Binner, Esposito 1998]. Ideally, diffusion bonding produces local deformation or
melting at the joint, promoting surface contact, and subsequently allowing atomic or ionic
Irrigation across the interface to produce a bond [Santella 1992, Sandhage 1996]. For
example, Glaeser et al. [1990] joined sapphire and polycrystalline alumina via hot

pressing at 1370°C for 60 minutes at 15 MPa pressure (as compared to the 8000 Pa in the

110

present study). However, such high pressures have the potential to lead to creep in the
base materials and a consequent change in shape of the joined components.

Because the adhesion and chemical compatibility between interganular glassy
phases and base cerarrrics is generally good, joining as a consequence of interganular
amorphous phase migation depends primarily on the existence and mobility of
interganular phases [Santella 1992, Sandhage 1996]. For example, Binner et al. [1998]
joined reaction-bonded silicon carbide (RBSC) to RBSC via microwave heating without
the use of an interlayer. During direct joining, the high temperature and pressure
promoted the motion and “bleeding” of liquid free silicon from the matrix to the interface
[Binner 1998], driven by the tendency of the system to restore the equilibrium when a
higher energy interface is introduced [Esposito 1998]. Subsequently, the liquid free
silicon spreads across the RBSC surface, generating a free silicon interlayer between the
two RBSC specimens which assist the bonding.

4.3.1 Joining of sapphire and polycrystalline alumina

The joining of sapphire and 96% polycrystalline alumina without the use of a silica
interlayer is similar to the joining of 92 to 96% purity alumina investigated by Fukushima
et al. [1988] and RBSC to RBSC investigated by Binner et al [1998]. The interface
between the two polished 96% polycrystalline alumina specimens put in contact can be
regarded as a macroscopic inhomogeneity since no interganular amorphous phase is
present on the surface of the gains at the interface. Thus, mass transport occurs, which
tries to minimize the macroscopic inhomogeneity between interfacial region and the base
alumina, via migating of the gain boundary amorphous phase through capillary flow to

the free surface. At high temperatures, the amorphous gain boundary phase in the 96%

111

polycrystalline alumina can melt and rnigate to the free surface, accounting for the
successful joining of the sapphire and polycrystalline alumina [Case 2001]. However,
due to the negligible amount of gain boundary phase in high purity 98% polycrystalline
alumina, sapphire and 98% alumina will not form a wetting phase without the use of a
silica interlayer.
4.3.2 Joining of zirconia and MaCorTM

Similar to the joining of sapphire and 96% polycrystalline alumina without the use
of a silica interlayer, the glassy matrix phase of the MaCorTM is sufficient to promote
bonding in the joining of zirconia and MaCorTM without the use of an interlayer. Such a
glassy borosilicate phase could become viscous or liquid at high temperatures and assist
the bonding of the zirconia and MaCorTM [Santella 1992]. As an example of glassy phase
assisted bonding, Fukushima [1988] directly joined 92 to 96% purity alumina via
microwave heating, presumably as a result of intergannular glassy phases being
preferentially heated and spreading through capillary wetting action along the joint
interface to assist the bonding. In the current study, evidence of the glassy nature of the
MaCorTM appeared with the development of ripples on the free surfaces of the MaCorTM
during the joining. Wunsche et al. [1999] reported similar ripples on the fracture surface
of borosilicate glass (the matrix of MaCorTM) that resulted from local melting. Thus,
when heated to 1070°C, it is likely that liquid phase(s) will form at the MaCorTM/zirconia
interface even without the use of a silica interlayer (Figure 4.2). In addition, similar to
the eutectic reactions that rrright be expected at the MaCorTM/silica interface [Schairper
1954], liquid silicate phase formation could also be expected at the zirconia/MaCorTM

interface or inside the MaCorTM since MaCorTM contains a total of 71% Si02, K20, and

112

MgO with additional amounts of A1203 and B203 [Corning]. Therefore, the interfacial
liquid that forms at the MaCorTM/zirconia interface wets and dissolves the zirconia along
the gain boundaries for temperatures at or below the joining temperature of 1070°C
(Figure 3.6), promoting a directly bonded interface between the MaCorTM and the
zirconia even without the use of a silica interlayer.
4.4 Effect of silica spin-on interlayers
4.4.1 Joining involving glassy ceramics

For the joining involving non-crystalline ceramics, such as glass-cerarrrics, or
crystalline ceramics containing fairly large amounts of amorphous interganular phase
under proper joining temperatures and low nominal pressures, adhesion could take place
as a result of glassy amorphous matrix or interganular amorphous phase present in the
base materials rrrigating to the joint [Binner, Esposito 1998]. The presence of an
amorphous phase appears to have two purposes: (1) it can directly hold the two joining
pieces together under proper temperature and pressure, acting as an adhesive medium,
similar to glass brazing, (2) it can allow the gains at or near the interface in base
materials to rotate and rearrange themselves at very high temperatures, which can lead to
the strengthening of the joint [Walls 1992, Binner 1998]. Therefore, the use of a silica
spin-on interlayer for the joining of ceramics containing large amounts of amorphous
phase, such as MaCorTM and 96% polycrystalline alumina, is not necessary.

4.4.2 Joining involving crystalline ceramics with little or no amorphous phase

In the joining of crystalline ceramics containing little or no amorphous phase using

low nominal pressures, successful bonding was not achieved without the use of a silica

spin-on interlayer. Since interfacial diffusion is critical to the development of bonding,

113

the interlayer should be properly chosen to assist the mass transport across the interfacial
region. The eutectic liquid phase formation between the chosen silica spin-on interlayers
and the base ceramics substantially enhanced the mass transport across the interfacial
region, resulting in successful bonding of crystalline ceramics containing little or no
amorphous phase. Thus, without the presence of a silica interlayer, liquid formation
should not occur between sapphire and 98% polycrystalline alumina below the melting
temperature of 98% alumina. Therefore sapphire and 98% polycrystalline alumina
cannot be joined without the use of a silica interlayer at the temperature and pressures
studied.

In addition, in the conventional joining of zirconia with zirconia using a silica
interlayer, the zirconia-silica equilibrium phase diagam [Curtis 1953] shows that the
zirconia-silica eutectic temperature is about 1680°C. Therefore, although interdiffusion
between zirconia and silica is likely to occur at the zirconia/silica interface, it is unlikely
that liquid formation will result at this side of the interface at the joining temperature of
1020°C. This is consistent with the experimental results that showed zirconia failed to
join with zirconia using a silica interlayer when heated at 1070°C for 60 minutes.

Thus, similar to liquid phase sintering where the compositions of sintering aids can
be chosen to result in an appreciable amount of liquid phase formation at elevated
temperatures [Kingery 1976, Barsoum 1997], the chemical compositions of the spin-on
interlayers could be chosen to assist the formation of a low melting point interfacial
glassy phase upon joining, such as the liquid phase formation due to eutectic reactions

between silica spin-on interlayer and the base joining ceramics in the present study.

114

Therefore, the wetting liquid phases formed upon heating could penetrate between the

two joining surfaces and enhance mass diffusion to assist the joining process.

115

 

(

mass transfer direction

o
”'
a
O
a
’o
‘ => ’
~
~~~
~
mass ‘5
~
~.

(C)

Figure 4.1 The schematic representation of the three mass transport mechanisms for
interface formation due to (a) surface and volume diffusion from surface source to neck,
(b) diffusion along the bond interface, and (c) bulk deformation mechanism after [Derby
1982, 1984]. The arrows denote the mass diffusion direction, while the dotted and bold
lines represent the interfacial void geometry before and after diffusions, respectively.

116

 

 

 

 

1| P=710°C
S=695°C
Q=725°C
Quartz R=810:C
K204120368i02 'b’l T=9180C
Potassium Tetrasilicate §V ‘ é , U=9050g
Potassium Disilicate t. 31.; V=923
'LuCite 4o
l<,OSio2 - T Th
V
40 KAlSiO4 A1203 60
A12 3
hduflflue
20 ‘ 80
K20 I 20 i 40 K10A110360 ' 80 " A1203

Figure 4.2 Equilibrium phase diagam of SiO2-K2O-Al203 [redrawn from Schairper
1954].

117

(a)

 

 

 

 

 

 

 

Figure 4.3 The expected thermal goove profile shapes due to (a) Interfacial
reactions with goove depth d =1.13m(At)°‘5 , (b) surface diffusion with depth

d =0.973m(B,t)°'25 and width 3 =4.6(B,t)°'25, and (c) volume diffusion with depth
d =1.01m(th)°'33 and width 3 = 5(B,,t)°’33 [After Mullins 1957, 1958, 1960]. Constants

= LCoyleZ ’ L = J ’3. = 60,2992 , and B, = xDvyslfl
kT c -— co kT

the solid/liquid interface at the goove root; Co is the initial concentration with a planar

interface; C is the concentration; S2 is the molecular volume and weight; M is the

molecular weight; x is the solubility of solid in the liquid; Dv and Ds are the volume and

surface diffusion coefficients, respectively; 8 is interfacial width; J is the flux, 'Ysl is the

solid/liquid interfacial energy, and ygb is the gain boundary energy.

A , where m is the slope of

118

Table 4.1 The dielectric properties of various ceramics used in the joining of
zirconia and MaCorTM using a silica interlayer.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Materials Frequency 8’ a” tan 8 Temgegrture Reference
MaCorTM 1 kHz 0.0047 25 [Coming]
MaCorTM 8.5 GHz 0.0071 25 [Coming]
Emmi“ 1 MHz 0.004 20 [Coming]
glass
B°r°s‘l‘°a‘° 1 MHz 0.11 350 [Coming]
glass*

. . [Walton
Slllca 8-10 GHz 0.0002 25 1970]
Silica 3 GHz 3'78 000023 0.00006 . 25 ”fig“?
Silica 0.5-5GHz 3.8 ~0.0001 25 [Arai 1993]

. . 4.2 0.0086 [Regnier
311103 0.1 MHz 0.002 25 1996]

. . [Hassler
Slllca 2.45 GHz 0.0032 1025 1988]

. . [Buchanan
Zircoma 1 MHz 0.01 25 1986]

. . [Smith
ercoma 5.4 GHz 0.0009 25 1992]
Zirconia 2.45 GHz 25 15.5 0.62 1020 [Arai 1993]

 

*Borosilicate glass is the matrix phase of MaCorTM [Corning]. However,
borosilicate glasses can have fairly broad composition ranges, and it is not clear if the
borosilicate glass noted in the table has a composition comparable to the MaCorTM
matrix. Thus, these compositions must be made with caution.

Table 4.2 The dielectric properties of polycrystalline alumina used in the joining of
alumina and alumina using a silica interlayer at 2.45 GHz.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Materials 8’ 8” tan 8 Tempegrture Reference
Alumina (AD85) 8.8 0.9 0.1 1000 [Arai 1993]
Alumina (AD85) 9.1 l 0.11 1200 [Arai 1993]

Alumina (AD99.8) 9 0.64 0.07 1000 [Arai 1993]

Alumina (AD99.8) 9.2 0.71 0.077 1200 [Arai 1993]
Silica 0.0032 1000 [Hassler 1988]
Silica 0.01 1200 [Hassler 1988]
Silica 0.016 1400 [Hassler 1988]

 

119

Chapter V

Conclusions and Future Studies

5.1 Summary and conclusions

The present study investigated the interfacial microstructure of ceramics joined
using spin-on interlayers. The fist key experimental result was the finding of no reaction
layers in the microwave joining of MaCorTM and zirconia, as described in detail in
Chapter III. In the microwave joined zirconia/MaCorTM specimens using a silica
interlayer, cross-sectional scanning and transmission electron microscopy (SEM and
TEM) reveal a non-planer zirconia/MaCorTM interface. This interlocking microstructure
appears to have resulted from the glass matrix of MaCorTM and silica strongly wetting
and dissolving the zirconia along gain boundaries.

However, because the glass-ceramic composite MaCorTM contains a glassy matrix
phase, it was not necessary to use a silica spin-on interlayer. Thus, the initial objective of
this study was modified to more comprehensively investigate the interfacial
microstructural development in the joining of zirconia and MaCorTM via conventional
versus microwave heating with or without a spin-on interlayer in relation to a number of
other processing variables. TEM analysis revealed similar interfacial nricrostructures
were developed with all of the interfaces, showing direct bonding between the zirconia
and MaCorm, without any additional interfacial or residual phases, regardless of joining
time or the presence or lack of an interlayer. In addition, successful interlayer joining

required higher temperatures for conventional heating than had been observed for

120

microwave heating, due to differences in the mechanisms involved in microwave and
conventional heating.

Because of the difficulties in identifying interfaces in joined like polycrystalline
ceramics/polycrystalline ceramics, the final objective of this research was to focus on the
nature of microstructure development in the interface region of polycrystalline alumina
bonded to sapphire. With the aid of silica interlayer, sapphire and 98% polycrystalline
alumina were successfully joined in 180 nrinutes at 1400°C, while samples without a
silica interlayer failed to join under these conditions, suggesting the silica interlayer
substantially promotes the joining of sapphire and 98% polycrystalline alumina. TEM
exarrrinations revealed the formation of non-equilibrium alumina and amorphous phases
at the joint interface in sapphire and 98% polycrystalline alumina joined at 1475°C.
However, sapphire and 96% polycrystalline alumina were joined with and without the
use of a silica interlayer, due to the presence of glassy gain boundary phases. SEM
analysis showed that an amorphous reaction layer approximately 0.5 pm thick formed in
sapphire/96% polycrystalline alumina samples joined with an interlayer. However, when
an interlayer was not used, the resulting amorphous interfacial phase was much thicker,
on the order of 5 pm thick, suggesting the silica interlayer may inhibit the flow of the
amorphous interganular phase. In both cases, higher joining temperatures and longer
processing times resulted in more complete joints and thinner interfacial regions.

A more comprehensive understanding of the fundamental processes related to
ceramic joining, such as reaction mechanisms, diffusion kinetics, wetting behavior, phase
behavior, and rrricrostructural development was achieved. The spin-on interlayer joining

approach can be envisaged as a process similar to liquid phase sintering processes. The

121

eutectic liquid phase formation between the chosen silica spin-on interlayers and the base
ceramics substantially enhanced the mass transport across the interfacial region, resulting
in successful bonding of crystalline ceramics. Since interfacial diffusion is critical to the
development of bonding, the spin-on interlayer should be properly chosen to assist the
mass transport across the interfacial region.

Unlike other joining processes using interlayers that are several microns thick or
more [Locatelli 1997, Ferraris 1998], the spin—on interlayer joining approach studied in
this thesis employs a thin interlayer to produce an imperceptible or thin bond layer
without the major modifications of the base ceramics. In addition, indentation tests
suggested that the resulting interfaces are strong, at least 60% of the strength of the base
ceramics.

5.2 Future studies

The nature of the interfacial microstructures of joined ceramics is key to
understanding the fundamental materials joining process. Factors such as thermal
expansion mismatch, interfacial roughness, surface energy, wetting, and adhesion will all
affect the resulting structures. This study shows that TEM is capable of providing
interfacial structural information at the micro level. However, to obtain information
regarding the atomic structure of the interface, the chemical nature of the interface, or the
mechanical properties of these joined ceramics, other techniques and further analysis are
required.

Thus, research in this joining area could be continued with the following studies:

(1) Thermal residual stress analysis of the joined ceramics systems could be carried

out using finite element analysis (FEA) to determine the optimal interfacial

122

microstructures and the related joining conditions. (2) High resolution TEM (HRTEM)
studies of the interfacial microstructures of the joined crystalline ceramics with
crystalline ceramics, to investigate atomic quality of the interfaces and lattice mismatch
between dissimilar materials. The HRTEM studies could allow assessment of the
interfaces at the micro and atomic level, the degee of joining, enhance our understanding
of interfaces and should also help in determining the joining mechanism. (3) Other
characterization techniques capable of obtaining additional information about the
interfacial rrricrostructures. Energy dispersive spectroscopy (EDS) and electron energy
loss spectroscopy (EELS) in TEM could be used to obtain the chemical distribution of
elements and to determine the elemental interdiffusion processes across the interfaces

during joining.

123

References
Ahn, B.G., Shiraishi, Y., High Temperature Materials and Processes (UK), 17(4) (1998)
209.
Aksaky, I.A., Pask, J .A., J.Am.Ceram.Soc., 58(11-12)(1975) 507.
Arai, M., Binner, J .G.P., Bowden, A.L., Cross, T.E., Evans, N.G., Hamlyn, M.G.,
Hutcheon, R., Morin, G., Smith, B., in: DE. Clark, W.R. Tinga, J .R. Laia Jr (Eds),
Microwaves: Theory and Application in Materials Processing 11. Proc. Symp. Cincinnati,
April 19-22, 1993, American Ceramic Soc. Inc.; Ceram. Trans. 36 (1993) 539.
Ashizuka, M., Ishida, B, Journal of the Ceramic Society of Japan, 104 (4)(l996)345.
Baik, D.S., No, K.S., Chun, J .S., J. Am. Ceram. Soc. 78 (5) (1995) 1217.
Barsoum, M.W., Fundamentals of ceramics, New York, McGraw Hill, 1997. p. 371-380.

Bates, C.H., Foley, M.R., Rossi, G.A., Sundberg, G.J., Wu, F.J., Am. Ceram. Soc. Bull,
69(3)(1990)350.

Binner, J .G.P., Femie, J .A., Whitaker, P.A., Cross, T.E., J.Mat.Sci. 33(12)(1998) 3017.
Binner, J .G.P., Femie, J .A., Whitaker, P.A., J.Mat.Sci. 33(12)(1998) 3009.

Boadi, J .K., Yano, T., Iseki, T., J. Mater. Sci. 22(7)(1997)2431.

Booske, J .H., Cooper, R.F., Freeman, S.A., Mater. Res. Innov., 1(2) (1997) 77.

Booske, J.H., Cooper, R.F., Freeman, S.A., Rybakov, K.I., Semenov, V.E.; Physics of
Plasmas, 5(5) (1998) 1664.

Brydson, R., Chen, S.C., Riley, F.L., Milne, S.J., Pan, X., Ruhle, M., J.Am.Ceram.Soc.,
81(2)(1998) 369.

Buchanan, R.C., Ceramic Materials for Electronics, Marcel Dekker, Inc., New York,
NY. 1986, PP. 33-47.

Burger, K., Ruhle, M., Ultramicroswpy, 29(1998)88.
Case, E. D., Lee, J. G., Lee, K. Y., “Joining of Optical and Infrared Materials Using Spin-

On Layers”, 1998, pp. 17 - 26 in Joining of Advanced and Specialty Materials, M. Singh,
J. E. Indacochea, and D. Hauser, eds., ASM International, Materials Park, OH. (Invited

paper)

124

Case, E. D., Lee, K. Y., Lee, J. G., “Joining of Polycrystalline Ceramics and Ceramic
Composites Using Microwave Heating”, 1997, pp. 17 - 20 in Proceedings of the 33'“
International Microwave Power Symposium, International Power Institute, Manassas,
VA.

Case, E. D., Lee, K. Y., Lee, J. G., Hoepfner, T., “Geometrical Stability of Holes and
Channels During Joining of Ceramics and Ceramic Composites”, 1998, pp. 27 — 34 in
Joining of Advanced and Specialty Materials, M. Singh, J. E. Indacochea, and D. Hauser,
eds., ASM International, Materials Park, OH. (Invited paper).

Case, E.D., Crimp, M.A., Advanced Engineering Materials (Germany), 3(6)(2001) 395.
Colombo, P., Sglavo, V., Pippel, E., Woltersdorf, J ., J. Mater. Sci. 33 (9) (1998)2405.
Corning information and Coming web site, h_ttp://www.coming.com.

Cross, T.H., Mayo, M.J., Nanostructured Materials, 3(1993) 163.

Curtis, C.E., Sowman, H.G., J. Am. Ceram. Soc. 36 (5) (1953) 190.

Dalgleish, B.J., Saiz, E., Tomsia, A.P., Cannon, R.M., Ritchie, R.O., Scripta Met, 31(8)
(1994)1109.

Davis, R.F., Pask, J .A., J.Am.Ceram.Soc., 55(10)(1972) 525.

Derby, B., Wallach, E.R., J. Mater. Sci., 19 (10) (1984)3140.

Derby, B., Wallach, E.R., Metal Science, 16 (1982)49.

Dennalay, F., Froyen, L., Deryttere, A., J. Mater. Sci, 22(1987)1.

Duvall, D.S., Owczarski, W.A., Paulonis, D.F., Welding Joumal, 53(4)(1974)203.
Elessner, G., Diem, W., Eallace, 1.8., in: J .A. Park, A.G. Evans (Eds), Surfaces and
Interfaces in Ceramics and Ceramic-Metal Systems, Plenum Press, New York, 1981,
pp629-939.

EMS, Electron microscopy image simulation online website, http://cimewww.epfl.ch.

Emulsitone Company Information and web site, http://www.emulsitone.com/sif.html.

Esposito, L., Bellosi, A., in A. Bellosi (Eds), Interfacial Science in Ceramic Joining,
Kluwer Academic Publishers, 1998, p. 211.

Fanslow, GB, in: W.B. Snyder Jr, W.H. Sutton, M.F. Iskander, D.L. Johnson (Eds),
Microwave Processing of Materials II. Proc. Symp. San Francisco, April 17-20, 1990,

125

Material Research Society, Pittsburgh, Pennsylvania, Materials Research Society
Symposium Proceedings, 189 (1991) 43.

Ferraris, M., Salvo, M., Isola, C., Montorsi, M.A., Kohyama, A., Journal of Nuclear
Materials, 258-263(l998) 1546.

Freeman, S.A., Booske, J .H., Cooper, R.F., J. Appl. Phys., 83(11) (1998) 5761.

Freeman, S.A., Booske, J .H., Cooper, R.F., Rev. Sci. Instrum., 66(6) (1995) 3606.
Fukushima, H., Yamanaka, T., Matsui, M., in W.H. Sutton, M.H. Brooks and 1.].
Chabinsky (Eds.), Microwave Processing of Materials, Mat. Res. Symp. Proc., Materials
Research Society, Pittsburgh, 124 (1988) 267.

Gale, W.F., JOM, 51(2) (1999) 49.

Geng, H., Ph.D dissertation, Michigan state university, 2003.

Grossman, D., American Machinist, May (1978) 21.

Grossman, D., J.Am.Ceram.Soc., 55 (9)(l972) 448.

Hanson, W.. Femie, J ., Materials World 16 (9) (1998) 534.

Hassler, Y., Johansen, L., in: W.H. Sutton, M.H. Brooks, I.J. Chabinsky (Eds),
Microwave Processing of Materials, April 5-8, 1988, Reno, Nevada, Material Research
Society, Pittsburgh, Pennsylvania, Materials Research Society Symposium Proceedings,
124(1988) 273.

Hauth, W.B., Am. Ceram. Soc. Bull, 58(1979)584.

Ho, W.W., in W.H. Sutton, M.H. Brooks and 1.]. Chabinsky (Eds.), Microwave
Processing of Materials, Mat. Res. Symp. Proc., Materials Research Society, Pittsburgh,
124 (1988) 137.

Janney, M.A., Kimrey, H. D., in: W.B. Snyder Jr, W.H. Sutton, M.F. Iskander, D.L.
Johnson (Eds.), Microwave Processing of Materials II. Proc. Symp. San Francisco, April
17-20, 1990, Material Research Society, Pittsburgh, Pennsylvania, Materials Research
Society Symposium Proceedings, 189 (1991) 215.

Janney, M.A., Kimrey, H.D., in: G.L. Messing, E.R. Fuller, H. Hausner (Eds.), American
Ceramic Society, Westerville, Ohio, 1988, Ceramic Powder Science 11, B, Ceramic

Transactions, 1(1988) 919.

Janney, M.A., Kimrey, H.D., Schmidt, M.A., Kiggans, J.O., J.Am.Ceram.Soc. 74(7)
(1991)]675.

126

Johnson, S.M., Rowcliff, D.J., J. Am. Ceram. Soc., 68(1985) 468.

Kennard, F.L., Bradt, R.C., Stubican, V.S., in J .S. Anderson (Eds.), Reactivity of Solids,
North Holland, Amsterdam, 1972, p. 580.

Kingery, W.D., Bowen, H.K., Uhlmann, D.R., Introduction to Ceramics, Second Edition,
John Wiley & Sons, Inc, New York, 1976, p. 466-513.

Kovalev, S.P., Miranzo, P., Osendi, M.I.,, J. Amer. Ceram. Soc, 81(9) (1998) 2342.

Lardner, T.J., Rotter, J .E., Shiao, M.L., Lin, M.R., International Journal of Fracture,
44(1990)l33.

Lee, C.C., Lin, P., Jpn. J. Appl. Phys. 37(1998) 6048.

Lee, J.G., Case, E.D., in: N.P. Bansal, J.P. Singh, E. Ustundag (Eds.), Advances in
Ceramic Matrix Composite V, American Ceramic Society, Inc., Westerville, OH, 2000,
Ceramic Trans. 103, pp. 571-581.

Lee, K. Y., Case, E. D., Reinhard, D., Ceramic Eng. And Sci. Proc., 18(1997)543.

Lee, K.S., Wuttiphan, S., Hu, X.Z., Lee, S.K., Lawn, B.R., J.Am.Ceram.Soc., 81
(3)(1998) 571.

Lee, W., Howard, S. J., Clegg, W. 1., Acta Mater. 44(10) (1996) 3905.

Levin, E.M., Robbins, C.R., McMurdie, H.F., Fig. 630 in Phase Diagrams for Ceramists,
Edited by MK. Reser. American Ceramic Society, Columbus, OH, 1964.

Locatelli, M.R., Dalgleish, B.J., Nakashima, K., Tomsia, A.P., Glaeser, A.M., Ceramic
International. 23 (1997) 313.

Loehman, R.E., in: DE. Clark, W.R. T inga, J .R. Laia Jr (Eds.), Microwaves: Theory and
Application in Materials Processing 11. Proc. Symp. Cincinnati, 19-22 April, 1993,
American Ceramic Soc. Inc.; Ceram. Trans. 36 (1993) 417.

Loehman, R.E., Key Engineering Materials. 161-163 (1999) 657.

Loehman, R.E., pp. 701-11 in Surface and Interfaces in Ceramic and Ceramic-Metal
Systems, Edited by J. Pask and A. Evans. Plenum Publishing Co., New York, 1981.

Macdowell, J .F., Beall, G.H., J.Am.Ceram.Soc., 52(1)(l969) 17.
Marshall, D.B., B.R. Lawn, J. Mater. Sci., l4(8)(l979) 2001.

Marshall, D.B., B.R. Lawn, J.Am.Ceram.Soc., 60 (l-2)(1977) 86.

127

Moorhead, A.J., Adv. Ceram. Mater., 2(1987)159.

Moorhead, A.J., Keating, H., Welding J., 65(1986)17.

Mullins, W.W., Acta Met., 6(1958) 414.

Mullins, W.W., J. Appl. Phys., 28(1957) 333.

Mullins, W.W., Trans. Met. Soc. AIME, 218(1960) 354.
Nakajima, A., Messing, G.L., J.Am.Ceram.Soc., 81(5)(l998) 1163.

Nakamura, M., Shigematsu, I., Nakanishi, M., Shigematsu, K., Yamada, Y., Shigegaki,
Y., J.Mat.Sci.Lett., 16(1997) 1654.

Nakamura-A, M., Shigematsu, I., Mabuchi, M., Yamada, Y., Shimojima, K., Saito, N.,
Nakanishi, M., Journal of the Ceramic Society of Japan, 105(1227)(1997) 1047.

Palaith, D., Silberglitt, R., Am. Ceram. Soc. Bull. 68(9) (1989) 1601.

Patterson, M.C.L., Apte, P.S., Kimber, R.M., Roy, R., in: R.L. Beatty, W.H. Sutton, M.F.
Iskander (Eds.), Microwave Processing of Materials III. Proc. Symp. San Francisco,
April 27-May 1,1992, Material Research Society, Pittsburgh, Pennsylvania, Materials
Research Society Symposium Proceedings, 269 (1992) 301.

Pinckney, L.R., Ceramics and Glasses, ASM International, Engineered Materials
Handbook. vol 4, ASM International, Materials Park, Ohio, 1991, p. 436.

Regnier, C., Tristant, P., Desmaison, J., Surface and Coating Technology, 80(1996) 18.
Richerson, D. W., Modern Ceramic Engineering: Properties, processing and Use in
Design, 2"d edition, p. 419-497, Marcel Dekker, Inc., New York (1992).

Risbud, S.H., Pask, J .A., Journal of Materials Science, 13(1978) 2449.

Risbud, S.H., Pask, J .A., Journal of Materials Science, 60(1977) 418.

Rode], J., Glaeser, A.M., J.Am.Ceram.Soc. 73(3)(l990) 592.

Rodriguez, A.D., Guiberteau, F., Melendo, M.J., Journal of Materials Research (USA),
13(6) (1998) 1631.

Rybakov, K.I., ,V.E., Phys. Rev. B 49 (1994) 64.

Sacks, M.D., Wang, K., Scheiffele, G.W., Bozkurt, N., J.Am.Ceram.Soc., 79(2)(l996)
571.

128

Sacks, M.D., Wang, K., Scheiffele, G.W., Bozkurt, N., J.Am.Ceram.Soc., 80(3)(1997)
663.

Saiz, E., Tomsia, A.P., Cannon, R.M., in A. Bellosi (Eds.), Interfacial Science in
Ceramic Joining, Kluwer Academic Publishers, 1998, p. 161.

Sandhage, K.H., Schmutzler, H.J., Wheeler, R., Frazer, H.L., J. Amer. Ceram. Soc., 79 (7)
(1996)1839.

Santellal, M.L., Am. Ceram. Soc. Bull., 71(1992) 947.

Schairper, J .F., J. Am. Ceram. Soc. 37 (l 1) (1954) 501.

Schwartz, M.M., Ceramic Joining. ASM International, Materials Park, OH, 1990.

Seiber, K. N., Lee, K. Y., Case, E. D., “Microwave and Conventional Joining of
Ceramics using Spin-on Materials,” pp. 941-949, Proceedings of the 12‘” Annual
Advanced Composites Conference, Technornic Publishing Co., Lancaster, PA., 1997.
Shalz, M.L., Dalgeish, B.J., Tomsia, A.P., Glaeser, A.M., Ceramic Transactions, 35
(1993) 301.

Singh, M.. 164-165(1999)415.

Smith, P.A., Davis, L.E., Electronics Letters, 28 (4) (1992) 424.

Staley, W.G., Brindley, G.W., J.Am.Ceram.Soc., 32(11)(l969) 616.

Suganuma, K., Okamoto, T., Koizumi, M., Shimada, M., J. Mater. Sci., 22 (4)(1987)
1359.

Sutton, W., Ceramic Bulletin, 68 (l992)376.

Tian, Y.L., Johnson, D.L., Brodwin, ME, in: G.L. Messing, E.R. Fuller, H. Hausner
(Eds.), American Ceramic Society, Westerville, Ohio, 1988, Ceramic Powder Science H,
B, Ceramic Transactions, 1(1988) 925.

Tinga, W.R., Mat. Res. Soc. Sym. Proc., Materials Research Society (USA), 124
(1988)33.

Urena, A., Gomez de Salazar, J .M., Quinones, J., J. Mater. Sci. 27(3) (1992)599.
Vegter, R.H., Ouden, G.D., J. Mater. Sci. 33(18)(1998) 4225.

Wallach, B.R., Hill, A., Acta Metall Mater, 37 (9) (1989) 2425.

129

Walls, P.A., Ueki, M., J. Am. Ceram. Soc. 75 (9) (1992) 2491.

Walton, J .D., Radome Engineering Handbook; Design and Principles: Inorganic
Randomes; Marcel Dekker, Inc., New York, 1970, pp. 229-344.

Weldon, L.M., Hampshire, 8., Pomeroy, M.J., J. Euro. Ceram Soc. l7(1997)l941.

Woods, M.E., Mandler, W.F., Scofield, T.L., Am. Ceram. Soc. Bull., 64 (1985)287.
Wunsche, C., Radlein, E., Frischat, G.H., Glass Sci. Technol. 72(2) (1999) 49.

Yano, T., Suematsu, H., Iseki, T., J. Mater. Sci. 23(3)(1998)3362.

Ye, J., Rodriquez, A. D., Scripta Metallurgica and Materialia, 33, (1994) 441.

Zdaniewski, W.A., J. Amer. Ceram. Soc. 58(5-6)(1975)l63.

Zdaniewski, W.A., Kirchner, H.P., Segall, A., Conway, J .C., J. Amer. Ceram. Soc. 70 (2)
(1987) 110.

Zdaniewski, W.A., Shah, P.M., Kirchner, H.P., Adv. Ceram. Mater., 2 (1987) 204.
Zeng, L., Case, E., Crimp, M.A., Mat. Sci. and Eng. A 307 (2001)74.

Zurbuchen, M.A., Carim, A.H., J. Am. Ceram. Soc., 82 (3)(1999) 705.

130

PART II
EFFECTS OF WORKING AND HEAT TREATMENT ON
MICROSTRUCTURAL AND CRYSTALLOGRAPHIC TEXTURE

EVOLUTION IN TI-6AL-4V

Chapter I

Introduction

1.1 Introduction

As a commercial alloy with 6 wt% aluminum and 4 wt% vanadium, Ti-6Al-4V is
currently the most widely used titanium alloy, accounting for more than 50% of all
titanium output in the world [Boyer-A 1994]. To date, no other titanium alloy can
replace its dominant position. Due to its good strength-to-weight-ratio or high specific
strength, Ti-6Al-4V is frequently used in aviation and aerospace industry that accounts
for more than 80% of this usage. The next largest application of Ti-6Al-4V is medical
prostheses, which accounts for 3% of the market [Boyer-A 1994]. The automotive,
marine, and chemical industries also use small amounts of Ti-6Al-4V [Boyer 1995]. The
initial application of Ti-6Al-4V includes compressor blades in gas turbine and fasteners
due to their good corrosion-resistance. However, the low modulus of elasticity and
thermal resistance of titanium alloys designed for high strength at low to moderate up to
400°C gives them new opportunities in turbine engine and air frame applications.

Unfortunately, cost continues to be a major factor for its use in industries where weight

131

and corrosion are critical considerations. In the aerospace industry, the superplastic
characteristics of fine grained, equiaxial Ti-6Al-4V are being used increasingly [Boyer-A
1994], such as fabrication of very complex structures for military airplanes, by combining
good diffusion bonding characteristics with superplastic forming.

In the medical industry, due to its biological compatibility, good tensile and fatigue
strength, wrought Ti-6A14V is used for surgical implants, especially for bone screws, and
for partial and total hip, knee, elbow, jaw, finger, and shoulder replacement joints [Ke 2001,
Orr 1992].

1.2 Microstructure evolution in Ti-6Al-4V

As an 0t+B alloy, Ti-6Al-4V has received extensive research and development
theoretically and experimentally since it was developed in the 19508. Much progress has
been accomplished through heat treatments, processing techniques, fundamental
knowledge of phase transformations and deformation mechanisms, and the relationship
between microstructure and the corresponding mechanical properties.

1.2.1 Phase transformations and heat treatments

Microstructure evolution control is necessary to change and control the physical and
mechanical properties of Ti-6Al-4V. More importantly, the effective controls of
microstructure evolution of Ti-6Al-4V mainly depend on the understanding of the phase
transformations occurring during heat treatment and processing. Table 1.1 lists the main
phase transformations and resulting phases that can occur in Ti-6Al-4V. Some of the
transformations are understood better than others. A complete review and research of all
phase transformations that occur in Ti-6Al-4V is beyond the scope of this research, so only

the subjects relevant to this dissertation will be reviewed.

132

As an or+B alloy, Ti-6Al-4V can have different microstructures having different
geometrical arrangements of many crystal phases, including on, B, a’, 0t”, a2, and (1)
phases, depending on the particular therrnomechanical treatment and interstitial
(primarily oxygen) content. These phases are all minor distortions of the bcc B or the hop
on phase, such that boundaries between them are typically coherent. The B isomorphous
phase diagram shown in figure 1.1 can be divided into three regions depending on the B
stabilizer V concentration in primary B grains [Koul 1970]. In region I where the B
stabilizer V concentration is approximately 4 wt%, the high temperature, bcc B phase can
completely transform to hexagonal close packed equilibrium 0: and/or martensitic or’

phases upon cooling. The needlelike or’ phase is obtained upon quenching and it is

frequently internally twinned on {1011}Cl planes. Slow cooling or furnace cooling from
above the B transus which is around 955~1010°C into the two phase (0t+B) region leads
to the formation of a-phase in plate form or plate-like alpha by nucleation and growth at
preferred sites, such as primary B grain boundaries [Boyer-A 1994], since the kinetics of
the on phase nucleation and growth reaction in region I that is lean in B stabilizers such as
V, is very fast [Chesnutt 1976]. Air-cooling results in a fine needle-like alpha or,

generally referred to as acicular alpha. Certain intermediate cooling rates develop

Widmanstaa tten structures (Figure l.l(b)) [Boyer-A 1994].

In region H where the B stabilizer V concentration is around 9~13 wt% V [Koul
1970], the beta phase can be retained at room temperature and/or transform to other non-
equilibrium orthorhombic 0t” martensite by quenching, or it can partially decompose to a

structure such as or phase during iso-thermal annealing at temperatures below 400°C

133

[Boyer-B 1994]. In Ti-6Al—4V, the a’ phase is obtained by quenching from above 900°C
and or” is a rather soft martensite that forms when quenching from temperatures between
800 and 900°C. Slower cooling leads to the formation of platelet (it-phase in B matrix
generally referred to as “transformed B”. In addition, stress induced on” martensite can
form under stress in this region [Koul 1970].

Upon aging the or’ and 0t” martensites can decompose into fine or phase since the
structure of a’ and a” is metastable. Just as in the case of ferrous martensites, the nature
of tempering reactions in Ti martensites is rather complex and depends on the

composition and crystal structure of the martensites. Several mechanisms have been

proposed to explain the age hardening effects that raise the yield strength:

(1) The decomposition of oc’ into fine or and B by nucleation and growth process.
During the precipitation of B phase which is preferentially oriented along the basal
directions within the Ot’ plates [Gil 1995], the composition of 0t’ approaches the
equilibrium composition of the or phase in equilibrium with the B phase at the tempering
temperatures.

(2) The decomposition of on” into fine or and B. The or” martensite tempers first by
spinodal precipitation of or phase with the attendant enrichment of the non-equilibrium
orthorhombic structure that eventually approaches the composition of B phase [Williams
1970, 1972], unlike the decomposition of or’. The overall tempering reaction can be
explained in terms of spinodal decomposition and reverse martensite transformation

represented by a”—->0t”.¢an+0t”,,ch—>a+B [Davis 1977].

134

(3) The formation of finely ordered equilibrium hexagonal a2 (Ti3Al) phase within
primary 0t phase. Due to the large, negative titanium and aluminum interaction energy,
the formation of tightly and directionally bonded a2 precipitates causes increases in
modulus and yield strength [Gehlen 1970, Welsch 1977].

(4) The ordering of interstitials (primarily oxygen) within primary or phase. Due to
the formation of short-range order of oxygen, oxygen induced internal friction and

promotion of glide band with increasing oxygen concentration produce large increases in

modulus and yield strength [Welsch 1977].

(5) The metastable B—m) transformation that is generally thought to cause hardening
and embrittlement of Ti-6Al-4V alloys. Generally the ordered hexagonal and/or cubic a)
phase does not occur in Ti-6Al-4V alloy of commercial purity unless the B phase is
highly enriched with vanadium (over 15 wt.%) that makes the formation of to phase
thermodynamically favorable instead of hexagonal or’ and orthorhombic a” martensites
[Fan 1994, Lasalmonie 1979]. The a) phase might occur during low temperature aging or

during cooling through the same temperature range, or be maintained throughout

processing due to its presence in as-received material [Boyer-A 1994].

In region HI where the V concentration is more than 15%, the high temperature B
phase can be completely retained or slowly transformed to more stable (1) phase during
cooling to room temperature [Koul 1970, Boyer—B 1994]. The B—xn transformation
kinetics in Ti-6Al-4V is very slow and thus this transformation takes very long time to

accomplish. Therefore, it is generally believed that the B phase remains stable with

135

increasing solute concentration and therefore no further transformations occur in this
region.
1.2.2 Crystallography of major phases in Ti-6Al-4V
Although the or, B, or’, on”, and (0 phases fall into different crystallographic
categories, their crystal structures are all really similar. Table 1.2(a) shows the main
crystallography relations of or, B, a’, or”, and (1) phases transformations in Titanium

alloys. The generally accepted lattice correspondence between B and on(or’) phases is the

Burgers relationship [Burgers 1934] in (1 10)B"(0001) a and <111>3||<1120>w Davis et

al. [1979] calculated the lattice strain and found the transformation between B and a(0t’)
phases can be achieved via 10% contraction along [100]5 which becomes [21I0]a, a 10%

expansion along the [011k that becomes [01 John and 1% expansion along [011]5 that

becomes [0001],» Similar to the B—>0t’ transformations, [Davis 1979] the equivalent
lattice correspondence of the B—MI” transformation is: [100]5 and [100]a», [011]]; and

[010]a», [011k and [001],» with reduced corresponding lattice strain (Table 1.2(b)),
suggesting the orthorhombic or” structure is an intermediate structure between B
(bcc)—)0t’ (hcp) transformations. The lattice correspondence between the B, or’ and on” is
shown in figure 1.2. It is generally known that the orthorhombic (1” structure is derived
from a distorted hexagonal lattice in which the c-axis of the or” corresponded to the c-axis
of the hexagonal oc’ lattice with the a and b axes of orthorhombic a” structure

corresponding to the orthogonal axes of the hexagonal lattice.

1.2.3 Solution quenching and aging (SQA)

136

Ti-6Al-4V alloy is most commonly used in the fully annealed conditions, often
referred to as nrill annealed. In an industry where high strength with adequate ductility is
required, such as for fasteners, it is necessary to solution treat at a temperature high in the
a-B field, normally 28~83°C (50~150°F) below the B-transus temperature which is
generally around 975°C [Boyer-C 1994]. If high fracture toughness or improved
resistance to stress corrosion is required, a solution treatment above the B transus that
causes fully transformed a’, may be desirable (Table 1.3(a)).

The microstructure obtained from solution treating in the oc-B field depends on the
degree of homogenization and cooling rate during the heat treatment. Mill annealed Ti-
6Al-4V alloy is a mixture of Al-rich 0t grains and V-rich B grains. At elevated
temperatures, or grains transform to B grains and upon subsequent quenching B may
transform into a’, or” or retained B, depending on the solution treating temperature, time
[Boyer-D 1994] and oxygen concentration [Kahveci 1986]. The higher the solution
treating temperature in the or-B field, the greater the amount of or grains that can
transform to B (figure 1.3), which causes lower V content in the equilibrium B phase
[Maykuth 1971] and higher strength after aging [Lasalmonie 1979]. Quenching from
above 900°C, the V content in the equilibrium B phase is approximately 4 wt%, so a’
phase is obtained, and soft or” martensite is formed when quenching from between 800
and 900°C, when the V content is around 9~13 wt%. However, when quenching from
below 800°C, when 15 wt% or higher V content is in the primary B phase, the B phase is
retained without transformation.

The effects of solutionization time and temperature on microstructural changes

depend on diffusion coefficients [Boyer-D 1994]. Therefore, approximate assessment of

137

the extent of possible nricrostructure changes by solutionization or aging can be made on
the basis of the magnitude of the Ti element self-diffusion, Al and V alloy elements
diffusion coefficient in or and B grains, respectively. The available information on
diffusion coefficients for the Ti-6Al-4V used in this study is limited (See Table l.3(b)),
so data comes from similar structure of different Titanium alloys. A rough estimate of
the average distance x traveled by an atom in one direction can be obtained from the
random walk equation ,1; =2Dt, where D is the diffusion coefficient and t is the diffusion
time (Table l.3(b)). For example, at normal aging temperature 500°C, the average
diffusion distance of Titanium for 4 hours are approximately 0.5 and 0.25 pm in the or
and B phase, respectively, suggesting that the aging process can only lead to local
modification of rrricrostructure and no obvious microstructure changes could occur when
examined using optical and scanning electron microscopy. However, for 1 hour at
1000°C, they are 3 and 30 pm, respectively, suggesting homogenization of the
nricrostructure occurs rapidly enough to facilitate observable microstructure changes.

In addition, oxygen, carbon and nitrogen interstitial impurities can increase the B-
transus temperature, lattice parameters [Anderson 1957], notch sensitivity [Maykuth
1971], strength and a/B phase ratio. Kahveci et al. [1986] found oxygen significantly
decreased the volume fraction of B phase (Figure 1.3) and raised the B-transus
temperature in the following relation:

R-M.,,“ (°C) = 937 + 242.7 x 0

Where 0 is the oxygen content in wt%.

138

Recently, Boswell [1995] et al. researched the effects of solutionization temperature
and time in B phase field on grain growth, texture and grain rnisorientation behaviors.
With the higher solutionization temperature and longer solutionization time in B phase
field, the preferred orientation or texture increases and the average degree of grain
boundary rnisorientation decreases, leading to a decreasing grain growth rate at larger
grain sizes. The authors suggested that slowing down grain growth and deviations from a
classic grain growth law may be due to the decrease in grain boundary rrrisorientation,
occurrence of local grouping of similar boundaries and solute segregation at grain
boundaries. However, Boswell et al. [1995] failed to correlate the ou—aB phase
transformation mechanisms to the observed phenomena.

1.2.4 Deformation and processing
1.2.4.1 Deformation mode of a phase

The deformation of Titanium alloys has been the subject of intensive study over the
last 40 years. An understanding of deformation mechanisms is essential for the solution
of problems such as the development of rolling texture, creep, and fracture toughness.
Deformation in the (1 phase can occur by either slip or twinning. The slip and twinning

modes reported in the literature are listed in table 1.4. At room temperature, the or phase
slips along the basal {0001}, prism {1010}, and pyramidal {1011} planes by the
motions of 1/3<11i0>, or Z dislocations, as well as the {1010} and pyramidal {112 i}
systems with the 2+2 dislocations l/3<1123> (Figure 1.4) [Jones 1981, Boyer-B

1994]. c dislocation <0001> is restricted to only prismatic planes and always sessile

and generally does not occur. The relative ease of activation for different slip systems is

139

controlled by the core structure of dislocations, impurity content, and temperature [Jones

1981]. Ti-6Al-4V can also deform by twinning (Figure 1.4). It has been reported to
occur on a variety of pyramidal planes. {1123}, {1012} and {1121} twins allow an
extension, while {11 i 2}, {101 1} and {11 i4} twins permits a contraction [Rosi 1953].

In polycrystalline Ti-6Al—4V, basal and prism slip of I; dislocations do not support
sufficient slip modes to satisfy von Mises criterion that every grain should be able to
plastically deform generally to meet the shape changes imposed by its neighbors.

Therefore, at least five independent slip systems are required, whereas basal and prism

->-9

slip can only provide four, so twinning and c +a dislocation slip must accommodate

shear strain and displacement in the c directions [Boyer-B 1994].
The deformation modes of or phase as a function of stress, temperature, and the

impurity level have been widely explored. Compression perpendicular to the c-axis, or
along the <1010>, is accommodated by the prismatic {1100} slip of a screw
dislocations, either 1/3<11i0> or 1/3<i110> [Jones 1981] and {1012} [Boyer-B 1994]

or {1121} twins [Paton 1969, 1970]. However, Jones et al. [1981] found no twins were

ever observed due to (< 10 um) fine grain size for compression normal to the c-axis, so

—o-r

c +a dislocations on {1101} satisfied the von Mises criterion. TEM analysis revealed
dislocations arranged in planar glide. Although Jones et al. [1981] did not specify the
oxygen content, a transition from planar to wavy glide was observed with decreasing
temperature and increasing oxygen content [Williams 1972]. In the planar arrays, most

of dislocations are jogged screws, suggesting edge dislocations move more quickly, and

140

the average distance between jogged screw dislocations is less within planer bands,
leading to increased internal stress [Williams 1972], as compared to tangled dislocation

arrangements.

Compression [Jones 1981] along the c-axis is mainly accommodated by the {1 101}

—r—)

slip of c + a dislocations with extensive cross-slip on second {1101} and dragging out
edge dipoles until the neighboring segments also cross-slip and join them. In contrast,
with c-axis tension, slip planes are clean without cross-slip, leading to increased CRSS in

compression. However, Paton et al. [1969, 1970] found compression along the c-axis is

accomplished only by wide and lenticular shaped {1122} twins for single crystal
Titanium via metallographic examinations, while increasing aluminum content tends to

suppress the occurrence of twinning. For example, compression along the c—axis even at

—r-)

77°K resulted in very limited amount of twinning for 6% Al addition and extensive c + a

was always seen. In addition, almost all dislocations [Jones 1981] are of screwa type

gliding on (0001) and (0111) planes for compression along <1011>, implying edge
components are more mobile.

TEM analysis also showed that the screw dislocation components are less mobile
than edge components. N aka [1994] has proposed a complex core structure for screw
dislocations. The screw components have a core spread on different planes, while the
edge components have a planer core. Below the athermal temperature (~500°K), the
motion of screw dislocations is controlled by the thermally activated recombination of the
core and therefore the screw components move slower and thus control the deformation

rate .

141

To explain the hcprcc transformations, a few authors [Akhtar 1975, Regnier 1970]

proposed a dissociation of a dislocations on {1101} plane is similar to bcc structure based

on the correlation between the ease of hcp transformations and a tendency for prismatic

dslip. However, Jones et al. [1981] found no justification for such correlations and
proposed that the hcp(—>bcc transformations are a separate phenomenon influenced by the
c/a ratio [Spedding 1916].
1.2.4.2 Deformation mode of B phase

In addition to the research of deformation mode in the or phase, the B phase has also
received much attention. It is expected that deformation in single crystal B phase is the
same as for other bcc metals, i.e. {110}, {112}, and {123} planes with <111> slip
directions [Barrett 1966]. Studies on commercial binary and ternary B Titanium alloys in
which the B phase is metastable or stable at room temperature revealed {332}<113>
twinning or <111> slip as the dominant plastic deformation mode [Hanada 1986, 1987].
{332}<113> twinning appears in the most unstable B phase alloys where the diffraction
pattern of the B phase reveals sharp to phase reflections. As the stability of the B phase
increases, the deformation mode changes from twinning to slip, and correspondingly, the
diffraction pattern of the B phase changes from sharp to reflections to diffuse streaking
[Hanada 1987]. In addition, modeling of the high temperature B phase in a Ti-6A1-4V

showed that rrrixed slip on {110}<111> and {112}<111> are the dominant deformation

modes [Gey 1996, 1997].

142

1.3 Texture evolution in Ti-6Al-4V

Texture or preferred orientation plays an important role in the mechanical properties
of Titanium and its alloys because of the highly anisotropic crystalline structure of the or
phase, leading to pronounced anisotropy in stress, strain and strength. It is almost
impossible to eliminate texture or preferred orientation in commercial manufacturing
processes [Fujishiro 1980]. Further, texture strengthening or hardening can be beneficial.
For example, it has long been realized that strong basal plane texture of Ti-6Al—4V is
desirable for applications whenever the stress field is biaxial [Babe], Murayama 1968].
Therefore, it is necessary to study the formation and evolution of Ti-6Al-4V textures
induced by different thermomechanical treatments since better adaptation of the
mechanical properties for a given applications requires good control of the texture.

Based on the formation mechanisms, three kinds of textures are found in Ti-6Al-
4V, deformation, recrystallization and phase transformation textures. Deformation
texture, including fiber and sheet texture, is formed due to the tendency of polycrystalline
grains in an aggregate to rotate during plastic deformation. For fiber texture, the
individual grain in a wire is oriented so that the same crystallographic direction [uvw] in
most grains is parallel or nearly parallel to the wire axis. Materials having a fiber texture
have rotational symmetry (about [uvw] parallel to wire axis) such that all crystal rotations
about the wire axis are equally probable. Therefore, fiber texture is expected in materials
formed by forces that have rotational symmetry about an axis, such as, in wire drawing,
swaging or extrusion [Cullity 1978]. Sheet texture implies that most of the grains are
oriented with a certain crystallographic plane (hkl) roughly parallel to the sheet normal

(SN), and a certain crystallographic direction roughly parallel to the rolling direction

143

(RD). Such a texture is described by the notation (hkl)[uvw] [Cullity 1978].
Recrystallization or annealing textures are formed when annealing cold or hot worked
Titanium to initiate new grain structure that generally has a preferred orientation,
possibly different from the deformation texture. A phase transformation texture is
formed through following three steps: (1) development of transformation texture from
a—>B phase, (2) development of B phase texture at elevated temperatures usually above
850°C, and (3) development of transformation texture from B—>0t phase during cooling
[Tanabe 1980]. In industry, texture of Ti-6Al-4V usually originates from hot working
processes, which produce a combination of deformation, recrystallization and phase
transformation textures [Fujishiro 1980]. Therefore, the overall influential factors on the
texture formation are mainly: (1) deformation mode, (2) prior texture, (3) deformation
temperature, and (4) subsequent processing after deformation.
1.3.1 Deformation and recrystallization texture

Due to its poor cold forming properties which are generally associated with
cracking or excessive spring back problems, Ti-6Al-4V typically is formed using hot
forming above 540°C [Boyer-F, G 1994]. At temperatures below 850°C, the hot rolled
texture [Tanabe, Peters 1980] is composed of transverse, split RD, and split transverse
direction (TD) components. The split TD texture due to hot rolling at 500°C with (0002)
pole concentrated at 20~30° from the SN toward the TD is generated with a spread of
(0002) poles. With increasing rolling temperature, an additional (0002) pole peak
concentrated at 60~80° from the SN in the TD is generated, called transverse texture

[Tanabe 1980]. This can be interpreted as a-rolling texture because the volume fraction

144

of the primary B phase is in the range of 25~30% from 500~850°C, so the newly formed
or phase after the phase transformation cycle a—>B——>a is less than 10% (Figure 1.3).

Upon further annealing after hot rolling in the temperature range of 500~850°C, the
split RD and transverse texture will not change [T anabe, Peters 1980]. For example,
Peters et al. [1980] investigated the hot rolling and annealing texture at 800°C starting
from randomly orientated Ti-6Al-4V specimen. After hot rolling at 800°C followed by
quenching to avoid recrystallization, two different types of texture, split RD and
transverse, are developed (Figure 1.5(a)). Further annealing (Figure 1.5(b)) at 800°C for
up to 96 hours will not change the texture significantly [Peters 1980], except for a slightly
increased maximum intensity, even though the microstructure changes from deformed to
equiaxed rrricrostructure, suggesting that recrystallization texture is essentially the same
as deformation texture for Ti-6Al-4V alloy.

The similar deformation and recrystallization texture behaviors are different from the
high purity Titanium, where recrystallization due to selective growth of the newly formed
nuclei [Hu 1968] leads to sharpening of texture and rotation of the basal plane [Keeler
1956]. Figure 1.5 shows the schematic representation of cold rolling deformation and
recrystallization texture in pure Titanium sheet [Keeler 1956]. In addition, hot rolling at
700°C of randomly oriented pure Titanium yields a split RD and TD texture; while further
annealing at 700°C leads to a crystal rotation with disappearance of the split RD texture
[Tanabe 1980]. These different behaviors between pure Titanium and Ti-6Al—4V are
generally considered to be caused by alloying elements.

1.3.2 Transformation texture and variant selections

145

At elevated temperatures above the ala+B transus, the low temperature hcp or phase
transforms to the bcc B phase and the subsequent cooling results in the B phase
transforming back to the hcp or phase, leading to the formation of many new orientations,

or the phase transformation texture. This 0t<—-)B transformation behavior is based on the

Burgers relationship in (110)3||(0001).ll and <111>Bll<11é0>w Thus, upon heating, each
on phase orientation can produce up to six cubic orientations with a (110)B plane parallel

to the original (0001),,l plane and with the six possible combinations of parallel

orientations, including two <lll>p directions in the (110)}; plane and three <1120>(ll
directions in the (0001).,l plane. With cooling to room temperature after solutionization,
each one of the six cubic orientations can generate another twelve orientations of the hcp
0t phase, with two different ways to align the hcp cell with reference to the [110]
directions on each of the six (110)B planes. Therefore, one orientation of or phase could
generate an a phase texture with potentially 6x12=72 orientations on heating to above the
transformation temperature followed by cooling to room temperature [Keeler 1956].
Figure 6 schematically shows the (1—96—90 transformation texture cycle from an initial or
single variant to the three possible variants of transformation B texture, and the six
possible variants of room temperature on phase from single orientation of high
temperature B phase. Due to crystal symmetry, only three variants of high temperature B
phase and six variant of room temperature or phase are shown (Figure 1.6). However,
many authors found each or phase orientation may not be able to equally provide 72

orientations after the transformation cycle 0t——>B—)0t. Based on the three steps of

146

formation of transformation texture, some important aspect of variant selection will be

discussed below.

First, each 0t phase orientation at room temperature will probably not be able to
equally provide 6 orientations of B phase on heating to high temperatures [Gey 2000,
Moustahfid 1997]. For example, heating initially randomly orientated 0t+B phases should
be able to generate randomly orientated high temperature B phase without any significant
texture if variant selection does not apply. However, Gey et al. [Gey 2000] simulated the
high temperature B phase texture, based on the (1) measurement of room temperature
retained B phase texture, and (2) simulation with a rate sensitive pancake Taylor model
assuming mixed slip on {110}+{ll2}<111> [Toth 1998]. They found the high
temperature B phase texture is essentially similar to room temperature retained B phase
texture which has two distinct and strong fiber texture components characterized by the 7
fiber and Eular angles ((pt, 0, (p2) in bcc. Furthermore Moustahfid et al. [1997] confirmed
the relationship between high temperature B texture and the retained B phase texture,
indicating valiant selections can apply during a—)B transformation on heating, resulting
in some preferred B phase orientations.

Second, the complicated physical mechanisms that occur upon heating and high
temperature solutionization that develop high temperature B phase texture remain
unexplained [Moustahfid 1997]. Generally, high temperature solutionization will lead to
B grain growth which could further modify or sharpen the existing B texture [Boswell
1995], and thus the subsequent transformation or texture. For example, Boswell [1995] et

al. found that a higher solutionization temperature and longer solutionization time caused

147

a larger B grain size, stronger transformation on texture, and a smaller grain boundary
misorientation angles. Apparently, all the factors that affecting the metallurgical state of
the B phase can also affect the subsequent transformation texture [Keeler 1956, Gey
1997]. Therefore, Moustahfid et al. [1997] even thinks the physical mechanisms
operating upon heating and solutionization might be more complex than the ones
occuning during the B—>0t transformation.

Lastly, many authors found that each B phase orientation formed at high
temperatures from the original or phase may not be able to equally provide 12 orientations
of on phase upon cooling. In other words, it is highly possible that B grains can prefer one
of these 12 0t orientations. It has long been realized that constraint, such as residual
stress, crystalline inclusion and an oriented overgrowth, can affect the B—->0t
transformation texture, causing only one of the six (110)5 planes to be favored in
transforming to the (0001)., with two strongly preferred 0t orientations [Keeler 1956].
This explanation is in agreement with the result that upon heat treatment in the B phase
field without deformation, then 0t phase texture was developed from the high temperature
B phase without variant selection. However, heat treatment in the or+B phase field or hot
rolling in the B phase field with more than 30% reduction results in an a phase texture
with variant selection [Gey 1996, 1997, Moustahfid 1997]. These observations can be
correlated to residual stress generation since solution treating in the 0t+B phase field and
deformation in the B phase field results in higher values of internal friction and/or
residual stress than solution treating in the B phase field and undeformed B phase,

respectively [Ouchi 1999]. In addition, if the B phase hot rolling was allowed to

148

recrystallize first then the subsequent or phase texture would have all six (110)B plane
equally favored in transforming to the (0001),, [Peters 1980]. Therefore, Gey et al.
[1996, 1997] think the metallurgical state of the B phase, especially high temperature
texture, is the key to the variant selections and affect the subsequent transformation
texture. Noticeably, a model to simulate the transformation texture that correlated the
favored variants and the most active slip system was developed, based on the
assumptions that during high temperature rolling deformation the most active slip or the
highest shear directions of B phase can be correlated to high defect density sites that are

known to be nucleation sites of the new on phase, and further determine the later favored

transformation texture. For example, (0001) [2110]0L 0t variant can be generated due to

shear along the (110)[i1i]g or (lI2)[IlI]5 slip systems (Figure l.7(a)). Therefore,
based on the high temperature B phase texture and calculation of the highest resolved
shear of all B slip systems, favored variants can be determined, so the simulated or phase
transformation texture from the high temperature B phase texture is in general agreement
with experimental texture. However, many authors reported that stress could speed the
phase transformation by several orders of magnitude [Boyer-H 1994], such as
dramatically reducing the time needed to complete the eutectoid reaction in Ti-12Cr.
Furthermore, just as the growth of the pre-existing B nuclei inherited from high
temperature forging determines the 0t—>B transformation mechanism and the high
temperature B phase texture upon heating, the “continuous dynamic recrystallization”
with the appearance of small equiaxial grains during B phase hot rolling can also affect or

even determine the subsequent B—>0t transformation [Gey 1996, 1997, 2000]. Therefore,

149

does hot rolling in the B phase field with more than 30% reduction lead to the or phase

texture with variant selection by a stress, recrystallization or transformation texture

effect, or a combination of these three effects?

In general, the variant selection rules for the transformation cycle (at—)B—m remain
unanswered, especially the development of high temperature B phase texture upon
heating and solutionization. The texture change due to B—ra transformation has been
widely studied and a little work has been done on the (1—>B transformation upon heating
[Gey 2002]. As in the case of ferrous alloys, the martensite nucleation process involves
the interaction of different defects to produce a configuration resembling the martensitic
structure [Cahn 1996], so it is expected that different arrangements of defects, especially
dislocations, may speed up or slow down the Ti-6Al-4V 0t—)B and B—-)0t transformations.
Recently, Gey et al. [2002] found that cold rolling prior to the transformation cycle
a—eB—mt leads to stronger high temperature B phase texture and stronger variant
selections in pure Titanium sheet, suggesting strong initial or texture resulting in strong
variant selections on both a—->B and B—roz transformations and vise versa.

Therefore, it is reasonable to assume that defects of certain arrangements in the
initial or phase due to initial deformation can affect or even determine the a—>B phase
transformation texture and be kept or stored in the high temperature B phase to further
influence the subsequent B—)0t transformations texture. In addition, it is also expected
that not only the metallurgical state of high temperature B phase as discussed by Gey et

al. [1996, 1997, 2000], but also the initial or phase texture prior to deformation.

150

1.4 Research strategy

The aim of this study is to achieve a better understanding of the relationship
between texture and processing associated with the high temperature heat treatment in
axisymmetric sample geometries. This research focuses on the effects of working and
heat treatment on microstructural and crystallographic texture evolution in Ti-6Al-4V.
The a—)B—>0t transformations, especially the texture variation, variant selections, and
predictions were investigated using available analytical equipment. The effects of heat
treatment, including solutionization and aging, on characteristics of variant selections of
crystallographic texture evolution, major phases present, lattice parameters, grain sizes,
and volume fractions of major phases were examined. These variables affect the
deformation mode and phase/texture transformation. By deforming a textured sample in
different deformation directions to introduce different internal defects or dislocations, the
effects of crystallographic dislocations on the variant selections of transformation cycle
a—>B—->a were investigated. These results provide theoretical and experimental basis for
extending the ability to manipulate Ti-6Al-4V into new and innovative textures that enable
new applications by proper control of the texture and microstructure. This understanding
will be critical in developing, commercializing and enhancing current or future
applications.

In chapter three, crystallographic texture and microstructural features of Ti-6Al-4V
were measured through several stages of processing, from as-received wire, extrusion, to
solution heat-treating and then aging. In addition, a natural aging effect was found for

the first time in Ti-6Al-4V. The main objective of this research is to gain comprehensive

151

understanding of the physical metallurgy of manufacturing processes that introduce
variability in Ti-6Al-4V products such as inter-related effects of

- Intrinsic variability in source material

Texture evolution,

Phase transformation,

Microstructure/property development.

In chapter four, effects of solution temperature and time on crystallographic texture
and mechanical properties in Ti-6Al—4V were examined. The main purpose of this
investigation is to gain a phenomenological understanding of the physical metallurgy of
high temperature solution treating and water quenching that is widely used in industry on
the following parameters and the correlation between them:

0 Grain sizes

0 Volume fraction of primary 0t and B

0 Diffusion of Ti, Al and V in primary or and B grains

0 Variant selections of B—mt transformation texture

- Texture

0 Mechanical properties at different sample orientations

In chapter five, the a/B texture transformations mechanisms in Ti-6Al-4V were

investigated. By deliberately deforming samples in different directions, different
deformation modes and hence different dislocation densities and defect structures can be
generated, so therefore different transformation texture might result. The main objective
of this research is to investigate the effects of dislocation structure on the transformation

texture and correlations of the following factors:

152

Deformation mode

Dislocations

Variant selections of on—>B and B—)0L transformation texture
Development of high temperature B phase

a/B phase transformation mechanism

153

Table 1.1 Main phase transformations in Ti-6Al-4V.

 

 

Transformations Process Reference
Martensitic . . , ,, [Williams 1967,
transformation Ma“°“3"°° 5")“ ’ °‘ 1970]

 

Stress-induced
transformation

Martensite: B—>0t”

[Davis 1979]

 

Atherrnal diffusionless (r) phase:
(it—m)

[Rabinkin 1981]

 

 

 

 

 

 

 

 

 

 

Decomposition of B Iso-therrnal (1) phase: B—>(l) mggfiiigfiéw]

fgiieation and ow th) Acicular or basketwave 0t platelet: [ Rhodes 1975,
gr B—Xx or B—)Type1a, Typezot, 1977]

Ordering of oz phase 0t; phase precipitation [Welsch 1977]

Decomposition of Hexagonal 0t’—)0t+B [Williams 1967,

martensites Orthorhombic a”—)a+B 1970]

Spin odal decomposition Phase separation: B—)B’1£,,+B’,M_I [Fan 1994]

 

Phase separation: a”—>a”1£an+0t”ficr_,

 

[Davis 1979]

 

154

 

Table 1.2 The (a) main crystallography relations of 0:, B, a’, (1”, and (1) phase

transformations in titanium alloys, the (b) lattice principle strains for B—)0t’ and B—>0t”

transformations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Process Orientation relations Reference
Obe 'n Bur ers relations: (110) (0001)
iii-m (Type y‘ g g _ 5" °‘ [Rhodes 1975}
100 <111>515112 0>a
Not obeying Burgers relations but follows the
B—m (Type - -
2a) {1012}<101 1> twin orientation with respect [Rhodes 1975]
to the Burgers relations 0t
B—mt’ Obeying Burgers relations: (110)5||(0001) a- [Burgers 1934’
<1 1 l>g||<11 2 0>ar Rhodes 1975]
B-m” (110)},j[(001),,» & <1 i 1>B||<110>an [Davis 1979]
B-ND (110)p||(0001),, & <111>gjj<11 i0>.,, [Silcock 19551
0H“) (0001)a}}(1 i 10),, & <1120>a||<0001>9 [Rabinkin 1981]
(3)
Alloy Transformations 81(%) 82(%) e3(%)
Ti B—)0t’ -10.1 10.1 0.9
Ti-8Mo B—oa” -8.5 7.8 0.37

 

 

The calculations were made from the relations ei=‘ni-l, n1=ala3, n2=bl J2 a5,

n3=cl J2 as, where a, b, c are the lattice parameters of orthorhombic 0t” and at}
corresponding to the lattice parameter of B phase [from Davis 1979].

(b)

155

 

 

Table 1.3 The (a) relationship between nricrostructure characteristics, mechanical
properties, and processing conditions [Boswell 1995, Boyer-B, C, E 1994], and the (b)
diffusion coefficients in Titanium alloys.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Microstructure Enhanced Degrades Processing
Strength
Ductility Fatigue crack growth STA in
Bi-Modal on Fatigue initiation resistance resistance the a—B
Low-cycle fatigue Fracture toughness field
resistance
Fatigue crack growth .
, . . . STA in
or martensrte Fatrgue performance resrstance
F the B field
racture toughness
Fatigue crack growth Ductility
resistance Strength
Widmanstatten Fracture toughness Fatrgue rmtratron B field
ngh-cycle fatrgue resrstance al
or resistance Low-cycle fatigue anne
Aqueous stress corrosion resistance
Creep
(a)
Temperature or Phase B Phase
(°C) El°m°m° D (mzls) Materials D (ml/s) Materials
1'3“”‘2 x r4 1A1 ‘x T4 55 Al
5004250 Ti ”P" 0 1hi 1980 l'iggtllgomr 0111 hi 1980
74500/RT) [ “C ] exp(- ) [ C ]
-13
5004250 A] 87:30 X Ti-4.1Al 2.9x10‘x Ti-4.55 A1
70700/RT) [Ouchr 1980] exp(-180000/RT) [Ouchr 1980]
Ti
3.1x10°x
[Murdock
900~1550 v °"P('3222°’RT)+ 1964. 1968.
l.37x10 xexp(- In ouye
57200/RT) 1980]
Ti
700~950 O 0.45xexp(-48000/RT) [David 1980]
Titanium
. 6.4x10"2xexp(- . .
690~850 Tl [Lrbanatl --- -..-
29300/RT) 1963]
(b)

156

Table 1.4(a). Slip and twin modes in or phase Titanium

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Slip mode Reference
{0001 }<11 50> [Jones 1981]
3 Slip {1010}<1120> [Jones 19811
{101 i }<11éo> [Jones 1981]
-’ a . {115 2 }<1183> [Boyer-B 1994]
c M Slip {101 1}<11:3.3> [Jones 1981]
{112 3} [Rosi 1953, Churchman 1954]
{1012} [Rosi 1953]
Twin {112 1} [Rosi 1953]
{ 11 2 2} [Rosenbaum 1964]
{101 l} [Paton 1969, 1970]
{11 i4} [Rosenbaum 1964]
—» _ (1180}<0001> [Boyer-B 1994]
c Slip {1010}<0001> [Boyer-B 1994]

 

 

 

157

 

       

Temperature

at

 

"\

9
/
A

 

'-‘-‘-17-
/
/’\§

V

 

 

 

Figure 1.1 (a) Beta isomorphous phase diagram for Ti-6Al-4V showing three
regions that undergo different transformation behaviors [After Koul 1970] and (b) a
schematic diagram showing different microstructures under thermal treatment with three
different cooling rates [Maykuth 1971, Boyer-B 1994].

 

(a)

Primary (1+

 

158

 

    
  

egg»

Solute (%)

Water Quenched Air Cooled Furnace Cooled
aciculara+B Plate-like a-l-B

Eqiuaxed

aciculara+B a+B

Primarv (1+5 Eqiuaxed (1+

intergranular B

.

)
‘
a

an- a;
v.7 -
371‘ .

3.3.».

‘I
'8

y:
I
.

3"”

T '5 '.
no
‘-

 

[00011all 111018" 1001],,»
C

A
>§

 

110010;" f

 

 

 

 

 

 
  

 

 

 

 

lllllslltllimalltuon»

 

[010]a»

[010]5 (\[10013

Figure 1.2 The lattice correspondence for B, a’ and 0t” transformations [From
Davis 1979].

 

1m Oxygen Content (wt%)

0 0.09
o 0.2
90 . n26
‘ 0.56
0 1.0
370-
/
g .
l
Q30, l1 1.0
§ (’6
O
>

5L.
5

.//// ..

%/ m

/

 

 

 

l A A

400 ‘ 600 1 800 ‘ 1000 1200
Quenching Temperature (C)

Figure 1.3 Primary B phase volume with different oxygen content as a function of
temperature [After Kahveci 1986].

159

(1122)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ bc=[0001] _
(0001) A ba=12 1101
>< b...=111 2 31
// \ — -
(011 1}
s‘ {1 I. l bc+a= (11 i O)
\ L r,”
....... ‘1’ ’: 32
Il’l”:><:’a
(0110)
al
(a)
_ c .. ._ _
(1123) ‘ (1122) (1121) (1011)
\ 1 \ "
x 1 \ .
1‘1. \ \\ “
3 ‘1 3,
: \\ \ \
l ‘\ \ o
g \ 1‘
i 1“. \‘i.
s : ’1 \‘. ..
“ (1124
..---..-- .---:.;~.:‘: ........
Ill, “~ 32
4
(b)

Figure 1.4 Major (a) slip and (b) twinning planes in on phase Titanium alloys.

160

 

 

As-deformed Recrystalllzed

(a) (b)

 

 

 

 

  
 

7 8
Alledllolratmc 42 0
500C 35 14
700C 35 17

Figure 1.5 (0002) pole figures of Ti-6Al-4V starting from initially randomly
oriented specimen rolled at 800°C followed by water quenching (a) and after further
annealing at 800°C for 1 hour (b) results in minor texture variations [After Peters 1980].
Two major (002) pole intensity maxima are observed. One major component, transverse
texture, belongs to basal plane normals in the TD aligned perpendicular to the SN and
planes parallel to the RD. The other one splits RD texture belongs to basal planes in the
RD align about 20° with the SN. The schematic representation of cold rolling deformation
(c) and recrystallization texture (d) in pure Titanium sheet [After Keeler 1956], results in
slightly different crystal rotations of the basal planes, indicated by Y and 8.

161

 

 

 

 

 

 

 

 

 

 

 

 

+ (002)
* /\
k
R/
/ (a)
(110) (MON

 

 

 

 

 

 

 

 

 

 

 

 

'l \

4 0

4 )

1 It

NI!

1. ‘2‘
I

r

'o ::a
CW-
'\ ,17

\‘ : vr‘ll *1
‘ l A"
’7

 

 

 

‘(b)///0 (d)

(i) (i)

 

 

 

 

 

 

(e) (f) (8)

Figure 1.6 Schematically represent the 0t—>B->0t transformation texture cycle from
(a) initial on single variant with crystal orientation and the (002) pole figure to the possible
(b) (c) (d) variants of transformation B texture with the corresponding with crystal
orientation between or and B and the (110) pole figures. (e) (f) (g) (h) (i) (j) are the
possible six pole figures of room temperature or phase variants from a single variant of
high temperature B phase (c) after the B—>ot transformation. Due to crystal symmetry,
only three variants of the high temperature B phase and six variants of the room
temperature or phase are shown.

162

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a Variants Slip systems in {110}<111> Slip systems in {112}<111>
1 135°/90°/355° (110)[111] (11 2)[111]
2 225°l90°l346° (110)1111] (112)[111]
3 315°/90°/355° (110)[111] (112)[111]
4 45°/90°/355° (1 10)[11 i] (112)[111]
5 90°/45°I85° (101)[11 1] (121)[111]
6 270°/45°/264° (101)1111] (121)[111]
7 270°/135°/265° (101)}111] (121 )[111]
8 90°/135°I84° (101)1111] (121)[111]
9 180°/45°/85° (011)11111 (21 1)[111]
10 0°l45°/265° (01 i)[111] (21 1)(111]
ll 360°ll35°l265° (011)11111 (211 )[111]
12 180°ll35°l85° (011)[111] (211)[111]

 

 

 

 

 

Figure 1.7 (a) Relationship between the slip systems in {110}<1ll> and
{112}<110> and the preferred 0t variants and the simulated (b) and experimental (c) on
phase texture after GAB—rot the transformation cycle with 75% reduction deformed
specimen [After Gey 1997].

163

Chapter 11

Experimental procedures

2.1 Research strategy

The overall objective of this research is to investigate the effects of working and
heat treatment on crystallographic texture and microstructural (CTM) evolution during
typical manufacturing processes in Ti-6Al-4V. The CTM features of Ti-6Al-4V were
examined through several stages of the manufacturing processes, including as-received
wire, extrusion, to solution heat-treating and then aging. Preliminary texture
measurements showed that after solutionization and quenching, a new or phase

orientation developed with a fiber texture with the crystal basal plane (0001) normal

direction aligned with the wire axis, in addition to the (1010) prism plane normal
direction along to the wire axis, which was the only texture present before
solutionization. The new orientation is rotated 90° from the prior texture due to a
preferred variant selection transformation.

Because of the 90° texture rotation due to heat treatment, the specimens’
mechanical strength changed, so the objective of this study was further focused to
investigate the effects of solution temperature and time on crystallographic texture
rotation and the associated mechanical properties, denoted as the SQA (solution
quenching and aging) project. However, as the research progressed, it became clear that
90° texture rotation is clearly associated with primary or and B phase initial texture,
working history, and defects generations and structures. Thus, a further project focused

on the correlation between variant selections and odB phase transformation mechanisms,

164

denoted as the variant selection (VS) project. Therefore, through this progression of
research projects, better understanding of the mechanisms of microstructure and texture
evolution associated with heat treatment, deformation modes, and working history in Ti-
6Al-4V have now been successfully achieved.
2.2 Materials and fabrication

2.2.1 Materials and processing for the Crystallographic Texture &
Microstructure project

The Ti-6Al-4V alloy used in the present study was all provided by Dynamet (195
Museum Road, Washington, PA 15301). The material came as 6.5 mm diameter
cylindrical wire (rod), which had been forged above the B-transus temperature and
subsequently cooled to room temperature. The chemical composition of this alloy is
shown in Table 2.1. Figure 2.2 shows the specimen preparation procedures and nanring
nomenclature, from as-received wire, hot heading or extrusion, to solution and aging
thermal treatment, simulated from typical industrial manufacturing processes (Figure
2.1). The samples were cut into approximately 20 mm long cylinders from the beginning
of the as-received wire, middle of as-received wire, and end of as-received wire to
evaluate metallurgical variations inside the as-received wire. Cylinders from the
beginning of coil, middle of coil, and end of coil were partially extruded, so as to have
samples with undeformed and deformed nricrostructures from each part of the coil. The
extruded area was reduced to 5.7 mm in diameter, corresponding to ~30% area
reductions. Extrusion was done in an industrial machine at a strain rate of approximately

1000/8, and followed by air cooling.

165

Since an equiaxed 0t phase in a transformed B matrix microstructure is generally
preferred for aerospace applications where high strength with adequate ductility is
required and desired, conventional solution treatment of the both extruded and the un-
extruded cylinder pairs cut from middle of coil was done in the two-phase region at
920°C for 2.4 and 3.6 ks in an industrial resistance furnace with an argon atmosphere,
and rapidly quenched in water at room temperature. To avoid diffusion bonding between
Titanium specimens at elevated temperatures, the furnace floor was shaken periodically,
which also moved them from the input end to the output end. During the solution
treatment, the dew point was less than -40°C to rrrinirnize surface contamination from
possible chemical reactions between Titanium and water vapor originating from the
quench end of the furnace. Aging treatments at 500°C in air for 14.4 ks at temperature,
and then air-cooled.

2.2.2 Materials and processing for the SQA project

The Ti-6Al-4V alloy used for the STA project was also provided by Dynamet. This
lot of material was processed similarly as above, but the wire diameter was 4.9 mm in
diameter (T able 2.1). The samples were cut to about 15 mm long cylinders from as-
received wire, followed by partial extrusion of one end to 4.3 mm in diameter. The
solution treatment of the partially extruded cylinders was done in two-phase region at
temperatures of 905, 925, 940 and 960°C for of 1.2, 2.4, 3.6, or 4.8 ks in a resistance
shaker furnace as described above. Aging treatments in air at 500°C were subsequently
applied for 14.4 ks and then air-cooled. After solution treating and aging, the specimens

were cross-sectioned to further microstructural, texture, and mechanical tests.

166

In addition, X—ray diffraction studies of certain selected specimens in a direction
perpendicular to the wire axis via continuously gradual sectioning were used to examine
the possible alloy elements local grouping or residual stress. Three representative parts
were chosen, first one solutionized at 905°C for 1.2 ks and aged at 500°C for 14.4 ks,
second one solutionized at 905°C for 4.8 ks and aged at 500°C for 14.4 ks, and the last
one solutionized at 960°C for 4.8 ks and aged at 500°C for 14.4 ks. Initially, the chosen
specimens were cross-sectioned and finely polished until 0.05 um, followed by X-ray
diffraction study, and then remove the surface layer of approximately 500 um and finely
polished, and followed by another X-ray diffraction study under the exact the same
conditions as before, and so on (Figure 2.3). After a series of sectioning and diffraction
studies and analyzes, the corresponding texture variation and lattice strain due to micro-
residual stress, macro-residual stress, or solid solution of alloy or impurities can be
determined.

2.2.3 Materials and fabrications for the Variant Selection project

For the VS project, the material came as cylinder rods of 10 mm in diameter, with
the same type of prior processing as indicated above. (Dynamet) (Table 2.1). The
samples were cut into approximately 10 mm long cylinders from as-received wire. This
study included two parts of experimental work. The first part, samples were grouped into
pairs that were adjacent to each other in the parent wire (Figure 2.4): one specimen from
each pair was stress relief annealed at 525°C for 28.8 ks on a copper plate in a resistance
furnace with a thermocouple attached to the copper plate near the location of the sample,
followed by furnace cooling to room temperature which took approximately 40 ks.

Thereafter, the stress relief annealed and un-annealed specimen pairs were flash heat

167

treated at a temperature of 920°C for either of 0.06 or 2.4 ks on the copper plate with
thermocouple in the resistance furnace, and rapidly quenched in water to room
temperature. After surface oxide removal, and light polishing, texture measurements
were made on the surfaces of the specimens in contact with the copper plate.

In another part of the designed experiment, the 10 mm long cylinder wire sample
pairs were further machined into approximately 6.8x4.4x5.0 rectangular shape specimens
using low speed diamond saw, followed by metallurgical polishing. Figure 2.5 shows the
specimen preparation procedures. Samples were grouped into pairs that were adjacent to
each other in the parent wire. The texture measurements were always done on the
transverse surface (perpendicular to the original wire axis). After texture measurements,
one specimen of each pair was compressed along the transverse direction, the other along
the longitude direction. These specimens were deformed at room temperature to
approximately 5% strain level (See compression test description, below for details),
followed by texture measurements on the exact same planes as before. Then, these
deformed rectangular specimen pairs were put in a sealed quartz tube under a vacuum of
10‘5 Torr to avoid surface oxidation at elevated temperatures. Solution heat treatments
were done in the two-phase region at a temperature of 920°C for a 2.4 ks on a copper hot
plate with a calibrated thermometer in the resistance furnace, and rapidly quenched in
water at room temperature (the quartz did not break, so the quench was less severe).
After solutionization and quenching, the rectangular specimen pairs were taken out from
the quartz tube to measure the texture and rrricrostructure on the exact same planes as
before. Table 2.2 shows the detailed nomenclature used for this project. For this project,

the texture measurements were made on the exact same surface after different processing

168

conditions, from as received, deformed, and solution heat-treated to monitor the surface
texture evolution.
2.3 Mechanical tests
2.3.1 Compression tests

All the mechanical and static compression tests were conducted using a screw
driven testing machine (Instron Model 4206, Canton, MA). All cylinder test specimens
were prepared with equal length and diameter using electric discharging machining
(EDM). In addition, all specimens, cylinder and rectangular shapes were cleaned using a
50:50 mixture of acetone and alcohol for 20~40 minutes in an ultrasonic bath prior to
compression test. Figure 2.6 shows the representation of the experimental set up used for
normal compression test and related specimen orientations and dimensions. No
lubrication was used during the mechanical compression test. The compression tests
were conducted at a crosshead speed of 2.54 mm/min. The cylinder or rectangular shape
specimens with longitude direction (labeled as L) means the wire axis is parallel to the
compression axis, while the specimens in transverse direction (labeled as T) means the
wire axis is perpendicular to the compression axis. Young’s Modulus, which represents
the material’s resistance to elastic deformation under external load, was characterized
from the slope the elastic region in the stress-strain curve (though elastic deformation of
the platen was not accounted for in post-test analysis that used the machine compliance to
remove the effects of elastic deformation of the machine). The yield stress corresponds
to the stress of the 0.2% permanent deformation from the stress-strain curve.

2.3.2 Micro-hardness tests

169

To evaluate micro-hardness variations in samples, Vickers hardness measurements
were performed by applying a load of 500 g for 30 seconds on selected specimens. Prior
to a micro-hardness test, specimens were cross-sectioned using a low speed diamond saw,
followed by mounting and mechanical polishing to a 0.05 um surface finish.

2.4 Optical microscopy and scanning electron microscopy (SEM)

For optical microscopy and scanning electron microscopy (SEM) analysis,
specimens were cross-sectioned using a low speed diamond saw. A portion of each
sectioned specimen was mounted in diallyl phthalate powder (LECO Corporation, MI).
The specimens were then mechanically polished to a 0.05 pm (alumina powder) surface
finish, and etched using Kroll’s reagent (2 ml HF, 6 ml HNO3, and 92 ml H20) using
standard metallographic polishing and etching techniques. The primary 0t and B grains
sizes were determined by the planimetric grain counting methods or linear intercept
technique along both longitudinal and transverse directions, which effectively gives the
mean areas of grains and mean section diameters in the selected areas. The intercept
method was done by drawing a straight line of a specific length on a micrograph and
counting the number of grains intercepted by the line. Volume fraction of the primary 0t
and B phases were determined by point counting analysis along both longitudinal and
transverse directions. The point counting was done by randomly distributing a grid of
points over an optical, SEM or TEM micrograph and counting the fraction of total point
count that falls in the given phase, which gives the volume fraction of the phases directly.
The aspect ratio of primary 0t and B grains were determined by the intercept method,
which is done by calculating the ratio of the grain size along the length and along the

width using oriented lines.

170

The optical microscopy was done using a Nikon optical microscope (Model
EPIPHOT 200, Japan). SEM observations were performed using a CamScan 44 FE SEM
or Hitachi S-2500C equipped with energy dispersive spectroscopy (EDS) operated at 15
kV. In addition, all specimens had been cleaned using a 50:50 mixture of acetone and
alcohol for 20~40 minutes in ultrasonic bath prior to optical or SEM examinations.

2.5 TEM sample preparation and analysis

Cross-sections for transmission electron microscopy TEM samples were sliced from
the bulk specimens using a low-speed diamond saw, followed by gluing the two-mm by
two-mm squares cut from the bulk specimens to 3-mm diameter molybdenum rings with
epoxy resin. The orientation of the foil with respect to the specimen was always known,
along the longitude specimen directions. All cross-sections were then polished, pre-
thinned using a VCR Model D500i electronically damped dimpler, and low-angle ion
milled to electron transparency using a Gatan Precision ion polishing system (PIPS).
TEM observations were performed at 200 kV using a Hitachi H-800 with double-tilt
specimen stage, using conventional amplitude contrast image techniques.

TEM, as one of the most powerful techniques to perform very fine rrricrostructure
analysis, was used to characterize the overall crystal structure of the thin foils Ti-6Al-4V.
Both brightfield and darkfield imaging were carried out in conjunction with local selected
area diffraction pattern (SAD) to provide unique and direct information on the local
microstructure features, such as primary 0t and B grain sizes, a’ and on” platelet sizes, the
orientation and texture of the primary or and B grains, the internal structures of primary or
and B phases, and orientation relations.

2.6 X-ray techniques

171

2.6.1. X-ray diffraction techniques

X-ray diffraction was used to determine the “unknown” spacing and phases present.
Samples for phase and lattice measurements were prepared by sectioning to obtain a
surface normal in a direction along the transverse directions (Figure 2.7). Prior to X-ray
diffraction measurements, the X-ray diffractometer was precisely calibrated using
polycrystalline quartz to insure good alignment of the instrument. In addition, all
specimens analyzed were polished using standard metallographic polishing technique to a
0.05 um (alumina powder) surface finish. Furthermore, all specimens surface normal
were adjusted to coincide as closely as possible with the diffractometer axis. X-ray
normal diffraction was made using CuKa radiation on a Scintag XDS-2000 system with
Ni filter. The X-ray diffraction measurements were performed at 35 kV and 25 mA with
a continuous scanning mode at 02° per minute. A 2 mm tube divergent and 3 mm tube
scatter slits, and 0.5 mm detector scatter and 0.3 mm detector reference slits were used to
control the spot size and sharpness of the X-ray peaks.

For simplicity, lattice parameter calculations were based on the 20 locations of
maximum peak intensities without using least squares method. The lattice parameter of
B phase was calculated based on averaging the d-spacing of (110)}3 and (200)p peaks;
while the lattice parameters of the on phase were calculated based on the (002),, and
(100),, peaks, and 0t” based on the (110),,rr, (020),,rr, and (0002) peaks.

Analysis of the diffraction pattern was made using the software package
Diffractometer Management System for windows NT (DMSNT). The measured raw
diffraction patterns were processed to obtain normalized diffraction patterns using a

series of programs, including Background removal with Box Car Curve Fit with filter

172

width 1.5° which determines how closely the background curve follows the ups and
downs of the raw data files, followed by smoothing using a simple box car averaging
algorithm with 3 points, and to remove the Kong using Kozz Stripping. The peaks and
maximum intensities were located using Lorenz correction analysis to correct the raw
data for the polarization effects without fitting Kaz.
2.6.2. Texture

Samples for texture measurements were prepared by sectioning to obtain a surface
normal in a direction along the longitudinal or wire axis directions. This same specimen
orientation was used to measure peak intensities in the wire axis direction, corresponding
to the 90° texture rotation direction (Figure 2.7). Therefore, this sample orientation can
provide direct measurement and evidence of 90° texture rotation without any tilting of
pole figures, superior to the other sample configurations with the surface parallel to
longitudinal directions. Texture measurements were made using CuKa radiation on a
Scintag XDS-2000 system equipped with a Huber ring goniometer and Ni filter without
oscillation. A 1 mm collimator and 2 mm detector slit were used for the measurement of
pole figures. During the pole figure measurements, 5° increments scan along the pole
distance and azimuth for all pole figures were followed (to a tilt angle of 80°), resulting
in an array of 72x16=1152 measurements that were subsequently used for processing and
analysis. (1010) and (0002) pole figures were measured for the primary on phase, and
(110) pole figures were measured for the primary B phase.

The procedure for reflection pole figure measurements was the following.

173

1. Mount the specimen in the texture goniometer. The specimen surface normal should
be at the correct height and exactly in the reflection plane for incoming and
diffracting X-ray beam when no tilt of specimen, or oc=0°.

2. Perform a 0-20 scan on the goniometer over the 20 range of interest and record it.
For (110),,», (020),,--, (110)g, (1010),, and (0002),, pole figures, the 20 scan ranged
from 30~45° in step of 005° to find all possible 0t, 0t” and B peaks. For (200) pole
figure, the 20 scan ranged from 50~65° to find the possible B peak.

3. Repeat the 0-20 scan at different tilt angle, such as 0t=30° or a=60° over the same
angular range in step of 005° to find all peaks of interest and to locate the angular
center of the peaks.

4. Locate the background 20 angles on both sides of the peak for all tilts. For (110),,»,
(020),,», (110)p, (1010),, and (0002),, pole figures, the experimentally measured
intensities at 33° and 45° were determined as backgrounds (since these peaks were
too close together to measure background close to each peak). For (200)}; pole
figures, intensities at 50° and 65° were determined as backgrounds.

5. Start the pole figure scan in a defined scanning pattern, such that net intensity is

calculated by subtracting a weighted average of the two background intensities from

the peak intensity for all tilts, followed by data processing.

Analysis of the pole figures was made using the preferred orientation package from
Los Alamos (popLA). The measured raw texture data was processed to obtain
normalized experimental pole figures using the program UNRAW which uses defocusing

correction data generated by program MAKEDFB. The theoretical MAKEDFB files

174

 

were created according to the detector slit width used, the 0 value of each
crystallographic plane, and the width of each peak in degrees (1°). Then experimental
raw pole figures were smoothed by 2.5° using a Gaussian filter and minor rotations were
made to maximize the symmetry in the pole figure, followed by a harmonic analysis to
complete the rim of pole figures using popLA software package without imposing any

assumed symmetry. A maximum scale of 400 and lowest contour value of 3 were

consistently chosen for final plotting of pole figures.

175

Table 2.1 Chemical compositions of as-received Ti-6Al-4V alloys in weight

 

 

 

 

 

 

 

 

 

 

 

 

 

 

percentage.
Diameter Al V Fe C O N H Y Ti
(nun)
4.9 6.12 3.86 0.18 0.02 0.16 0.1 0.0029 <0.005 Bal
6.5 6.01 3.71 0.18 0.06 0.14 0.06 0.0037 <0.005 Bal
10 6.09 4.16 0.16 0.08 0.16 0.06 0.0046 <0.005 Bal

 

 

Table 2.2 Detailed nomenclature used in the Variant Selection project to evaluate
the influence of different deformation mode on variant selections of 0t—>B-)0t

transformation cycle.

 

 

 

 

 

 

 

Pairs Name Processing conditions
Pair 1 SlQ As-received+Heat treated 920°C for l rrrin
Ann+S1Q Annealed+Heat treated 920°C for 1 rrrin
Pair2 S40Q As-received+Heat treated 920°C for 40 rrrin
Ann+S40Q Annealed+Heat treated 920°C for 40 min
L Wire As-received
L def Deformed along longitudinal direction from L Wire
L SQ Heat treated 920°C for 40 rrrin :11 fquartz tube after deformation L
. e
Parr3 T Wire As-received
T def Deformed along transverse direction from T Wire
T SQ Heat treated 920°C for 40 min in quartz tube after deformation T

 

 

 

 

 

 

 

 

def

 

176

 

 

 

As-received wire

it

 

 

 

 

 

 

 

 

 

Extrusion
Solution at 900~955°C for 20~60 min -_.
As-received Large Extruded small
Diameter diameter

 

Aging at 500~550°C for 4~6 hr

1

Further mechanical processing

 

 

 

 

 

 

 

Figure 2.1 Schematical representation of typical industrial manufacturing flow
chart.

177

 

C B (As-received wire) 0

Beginning of coil Middle End of coil

(0( O ( O( O ( O( O

l Extrusionl lExtrusionl lExu-ugionl

 

 

 

 
 
     

Solution at 920°C

 

for 2.4 ks
( {)(_O
M-sq20 Mex-sq20 M-sq60 Mex-sq60
Aging at 500°C for
14.4 ks
M-sq20-a Mex-sq20-a M-sq60-a Mex-sq60-a

Figure 2.2 Schematic representations of the specimen preparation procedures and
nomenclatures, from as-received wire, extrusion, to solution and aging thermal treatment.

178

K X-ray

>811) (11(1)

4
Longitude direction & wire axis

 

Figure 2.3 Schematic representations of the sample orientation for serial sectioning
X-ray diffraction techniques used in the current study.

 

 

Annealing at 525°C
for 6 hours.

As-received

 

 

 

Solutionization at 920°C for either
1 or 40 rrrinutes followed by water

Figure 2.4 Schematic representation of the specimen preparation procedures, from
as-received wire, to solution treated conditions. See table 2.2 for detailed nomenclatures.
The arrow mark referred to specimens’ orientations for texture measurements.

179

—lfi

Quartz tube

 

 

Heating at 920°C for 2.4 ks in
quartz tube under vacuum
followed by water quenching.

 

‘ Ldef (5%) Tdef (5%)

XRD XRD

 

Lheat

Figure 2.5 Schematic representation of the specimen preparation procedures, from
as-received wire, deformed state, to solution treated conditions. All the texture
measurements were taken on the exact same surface plane after different processing and
deformation conditions, from as received, deformed, to heat-treatment to monitor the
surface texture evolutions. See table 2.2 for detailed nomenclatures. The arrow mark
referred to specimens’ orientations for texture measurements.

180

 

\\\
a ms? (0 i O

Longitude direction & wire axis (L)

 

 

 

 

 

 

(a) (b)

Figure 2.6 Representation of the experimental set up used for normal compression
test (a) and related specimen orientation (b). The two cylinders inside the wire represent
the actual orientation corresponding to the original as-received wire and how they were
sectioned using EDM. The cylinder with longitudinal direction (labeled as L) means the
as-received wire axis is parallel to the cylinder axis, while the cylinder in transverse
direction (labeled as T) means the as-received wire axis is perpendicular to the cylinder
axis. The compression axis was always parallel to the specimens’ axes.

 

X-ray
X-ra
a K ,
4 >
Longitude direction & wire axis Longitude direction & wire axis
(a) (b)

Figure 2.7 Schematic representations of the sample orientation for (a) X-ray
diffraction techniques and (b) texture measurements used in the current study.

181

 

Chapter 111

Effects of working and heat treatment on microstructural evolution and

crystallographic texture of a and 0t” phases in Ti-6Al-4V

Overview

Crystallographic texture has been measured on titanium at several stages of a
production process. As-received wire had a moderately strong fiber texture with prism
plane normals aligned with the wire axis. The wire was worked by extrusion, solution
heat-treated and water quenched, then aged. Extrusion strengthened the as-received
texture. After solutionization and quenching, a secondary fiber with basal plane normals
aligned with the wire axis emerged at the expense of the initial texture. This 90° rotation
is similar to rotations observed in a rolled plate subjected to a super-B transus anneal,
indicating strong variant selection occurred after solutionization and quenching. The
development and selection of a particular variant after or—-)B—>0t transformation cycles is
related to the least activeness of high temperature B phase slip system. Natural aging for
a year led to the or” and B to transform back to the or phase. The relationship between the
orientations of 0t, 0t”, and the processing history are discussed in the context of the

known physical metallurgy of titanium alloys.
3.1 Experiment results

In Ti-6Al-4V, a wide variety of phases can be generated at different stages of heat

treatment and processing history, such as a’, or”, B, to, 012 and or phases, depending on the

182

particular thermomechanical treatment and interstitial (primarily oxygen) content [Boyer-
A 1994]. From the measured and published crystal lattice constants the similarity
between or”, B and 0t phases, are illustrated in figure 3.1, which shows the coherent
relationship between these phases. Although all the crystal structures of a’, 0L”, B, (1), a2
and 0t phases are really similar, they fall into different crystallographic categories.

In this study, a set of specimens was collected at different stages of a typical
manufacturing process. These specimens are identified in table 3.1, with Mex-sq60-a
stands for middle of coil, followed by extrusion, solutionization and quenching at 920°C
for 3.6 ks, and aging at 500°C for 14.4 ks. All specimens were examined using texture,
x-ray diffraction, micro-hardness, and electron microscopy techniques.

3.1.1 As-received wire (B, M, and E)

Figure 3.2 shows the primary on phase pole figures of as-received wire from middle

of the coil and the dominant crystal orientations within the specimen. The (1010),, and

(0002),, (1 phase pole figures show that the primary 0: phase orientation is a strong fiber

texture with 5.37 x random with the axisymmetric crystal orientations about the wire

axis. The crystal prism plane normal direction, (1010),, is aligned with the wire axis,
and the (0001),, plane normal direction is in the radial direction.

To evaluate the intrinsic texture variation in the original as-received wire, texture
measurements were made on the wire at different locations. The overall texture in
beginning, middle and end of the as-received wire is essentially the same with a strong
fiber texture parallel to the wire axis. However, the intensity of preferred crystal

orientation in the radial direction is higher in the ends of the coil than in the middle.

183

Figure 3.3 reveals the overall relationships between major texture intensities and the
experiment processing steps.

In addition to texture evaluations, X-ray diffraction analyses performed on as-
received specimens made from middle of the coil show primary or and B phases (Figure
3.4). In addition, TEM analysis on the M specimen revealed the same phase
combinations, equilibrium 0t and metastable B phases (Figure 3.5). Besides the strong B
phases reflections, the diffraction pattern from primary B phases shows typical (1) phase
diffuse streaking which is a typical characteristics of early B—->(r) transformations [Fan
1998], indicating that the primary metastable B phase has an ordering tendency to
transform to a thermodynamically more stable to phase. However, this B—>(1)
transformation is not complete, otherwise, sharp (1)-reflections would be revealed (Figure
3.5). Similar to the rrricrostructure of M, X-ray diffraction and TEM analysis performed
on B and E sectioned from beginning and end of coil revealed essentially the same phase
combinations, phase morphology and spatial distributions.

3.1.2 Extruded specimens (Bex, Mex, and Eex)

The two or phase pole figures (1010)., and (0002)., of extruded Mex sectioned from
middle of as-received wire show that the texture is the same as the as-received wire, but
stronger in intensity (Figure 3.2(b) and 3.3) due to the ~30% reduction of area.

TEM examination of the extruded Mex material showed different microstructures in
different specimens, one only contained primary or grains and the other only primary B
grains (Figure 3.6). This may be due to non-uniform element and/or phase distribution in
different regions, since the region examined in the two foils were originally 2 mm apart

from each other. In some locations after the extruding operation, the specimen shows a

184

high dislocation density obscuring grain or phase boundaries in prior B regions (figure
3.6(a)). The inset diffraction patterns show streaked diffraction spots that imply a
severely distorted crystal. At other locations, only non-equiaxed primary or grains
(Figure 3.6(b)) were observed with high dislocation densities with an identifiable grain
size of about 1 pm. The high dislocation density is expected from the extrusion process.
3.1.3 Quenching (M-sq20 & Mex-sq20 and M-sq60 & Mex-sq60)
After quenching, texture measurements show that a new 0t phase orientation

developed; a fiber texture with the crystal basal plane (0002),, normal direction aligned

with the wire axis and the (1010),, prism plane normal direction perpendicular to the wire

axis (Figure 3.7). This introduction of a new orientation weakened the prior fiber texture

(Figure 3.3). Figure 3.7(a) shows the or phase (1010),, and (0002),, pole figures of M-
sq20 and related schematic showing the orientations. The new orientation is rotated 90°
from the prior texture, suggesting that a transformation occurred during the solutionizing
process. This transformation may be explained by the prior or phase transformed to B
phase, and then retransformed back to or based upon a different {110}5 plane 90°
nrisoriented from the prior crystal while cooling, but in a favored way to cause a well
defined (0001),, plane normal aligned with the wire axis (Figure 3.3). In addition, it is
apparent that a shorter solutionization time does not affect the texture significantly.
However, the major texture intensities weakened. In contrast to the extruded condition
where extrusion strengthened the major texture intensities, the texture intensities of the
Mex-sq20 was weaker than that of M-sq20. A similar trend was observed for both short

time solution and long time solution.

185

After quenching from 920°C, the as-received and extruded specimens show similar

phases present, a mixture of primary or and hexagonal a’ martensites which cause

broadening of the (1010),, and (0002),, peak (Figure 3.4). This peak broadening was
evidently caused by the presence of a’ martensite, with lattice parameters that differed
only slightly from the lattice parameters of the or phase due to compositional variations in
these phases [Boyer-A 1994]. No orthorhombic 0t” martensites or retained B were
observed. The TEM observations were consistent with the X-ray diffraction results with
the presence of many needlelike hexagonal 01’ martensitic platelet with many finely
divided interfaces (Figure 3.8). The diffraction pattern of a primary B grain showed both

the primary or and B spots in figures 3.8(b) which shows the indexed Bloc phases in the

Burgers orientation relationship of <111>B||<1120>,, and (110)3"(0001),,. Figure 3.8

shows the TEM diffraction pattern from a primary B grain with [111] B zone pattern and

three on <11 20> patterns superimposed.
3.1.4 Ageing
Texture measurements were made from specimens aged at 500°C for 14.4 ks, and
summarized in figure 3.3. Similar to the as-quenched specimens, pole figures of aged
specimens showed the secondary texture component with the crystal (0002),, plane

normal (c-axis) aligned with the wire axis, but the intensities of the components aligned

with the wire axis were significantly reduced. The (1010),, and (0002),, pole figures for
the M-sq60-a are shown in figure 3.7(b). The systematically decreased texture implies

that aging caused a randomizing effect on the texture due to the decomposition of

186

hexagonal oc’ martensites which favors a decrease of existing orientations to form more
randomly oriented new orientations. In addition, this texture decrease was more
pronounced in the extruded specimens than in the as-received wire.

In addition to texture measurements, X-ray diffraction scans on the M-sq20-a, Mex-
sq20-a, M-sq60-a, and Mex-sq60-a specimens revealed similar phase combinations,
having a mixture of primary on phase with hexagonal on’ and orthorhombic or.” martensites
with small amounts of retained B phase after aging (Figure 3.9). The TEM analyses show
typical duplex microstructures with primary 0t and B grains (Figure 3.10(a)). The
primary (1 and B phases have a grain size of approximately 2~3 run. The complex
diffraction pattern from a primary B grain of the aged M-sq60-a specimen shows distinct
diffraction peaks from orthorhombic 0t” martensites, in agreement with X-ray diffraction
analyses, and the orientation relationship between a and 0t” phases. The appearances of
orthorhombic 0t” martensites and retained B phase after aging suggested a new
transformation phenomenon occurred, a’—>a”+B, assuming no phase transformation
happened in the primary 0t phase.

Approximately one year later, similar X-ray diffraction were made on the same bulk
M-sq20-a, Mex-sq20-a, M-sq60-a, and Mex-sq60-a specimens, and they revealed
different phase combinations; a primary on phase without orthorhombic 0t” and retained B
phases (Figure 3.9). On the normal X—ray diffraction scan, all or” and B diffraction peaks
had consistently disappeared. During the prior year, all bulk and thin foil specimens were

separately packed in a labeled envelopes without any mechanical, chemical, thermal

processing, or damage.

187

The natural aging effect was also observed in the thin foil Ti-6Al-4V TEM
specimens. TEM analyses were performed on the same M-sq20-a, thin foil specimens
made one year previous. The diffraction patterns from the primary B phase showed only
or phase spots without the prior orthorhombic 0t” and retained B phases (Figure 3.10).
Similarly, TEM analyses of the thin foil Mex-sq20-a, M-sq60-a, and Mex-sq60-a
specimens revealed the same effects.

The natural aging was also examined using micro-hardness tests. It is generally
known that natural aging in Aluminum and Copper alloys leads to mechanical
strengthening and a hardness increase. Micro-hardness measurements were performed on
the same bulk M-sq20-a specimens using the same hardness tester, measurement
conditions, and specimen orientations to compare micro-hardness values done one year
previous. Using Student t test statistics, there is clearly a hardness increase from the
mean 375 HV to 379 Hv with 90% confidence interval. Therefore, the X-ray bulk
specimens’ diffraction, TEM thin foils analyses, and rrricro-hardness testing all indicate
that natural aging did occur in Ti-6Al-4V, leading to the transformation of orthorhombic
a” martensite and retained B phases transform to the on phase.

In summary, from the current study, a series of phase transformation phenomenon
can be proposed as follows:

9 Quenching: B——>or’,

0 Artificial aging at elevated temperatures: a’——>0t”+B and Ot’—)0L+B,

o One-year natural aging at room temperatures: 0t”+B—>0t.

3.1.5 Crystal lattice measurements

188

Crystal lattice constants of major phases or, on” and B phases were carefully
measured and calculated to determine the vanadium and oxygen concentration in the B
and (1 phases, respectively (Table 3.2). Increasing vanadium concentration decreases the
lattice parameter of B phase, while increasing interstitial elements, especially oxygen,
increase the lattice parameters of the or phase by occupying a fraction of octahedral
interstitial sites [Boyer-A 1994]. Initial vanadium concentration in the B phase was
probably high and steadily decreased with further processing since the vanadium can
diffuse into or phase at elevated temperatures. This is consistent with the TEM
observation of diffuse (1) streaking only in as-received wire since high concentration of
vanadium in the B phase makes the formation of (1) phase thermodynamically favorable.
3.2 Discussion

The texture and microstructural analyses show a new transformation phenomenon
or mechanism occurred and 90° crystal orientation rotation from previous texture during
the experimental processing. In Ti-6Al-4V, texture and microstructure can be adjusted
and controlled separately [Peters 1980]. Therefore, it is of interest to investigate the
microstructural transformation behavior and texture rotation separately in relation to
previous Tr-6Al-4V alloys studies, physical metallurgy principles, and microstructure
studies of other similar Titanium alloys.

3.2.1 Phase transformations in Ti-6Al-4V

A variety of phase transformations can occur in titanium alloys, depending on the

particular therrno mechanical treatment and interstitial content. Some of the

transformations are now better understood, while some others are not. A complete

189

review of all phase transformations that occur in Ti-6Al-4V is beyond the scope of this
research, so only the most relevant findings are considered.

At temperatures above the (x/a-I-B transus, the equilibrium or phase can partially
transform to a high temperature B phase. The subsequent quenching would lead to the
formation of hexagonal a’ phase instead of orthorhombic or” since the V stabilizer
concentration in primary B grains was lean at high solutionization temperature [Koul
1970]. Generally, upon aging, hexagonal or’ decomposes into fine a and B by nucleation
and growth process through the precipitation of B phase which is preferentially oriented
along the basal directions within the (1’ plates [Gil 1995]. Davis et al. [1979] calculated
the lattice strain between B—)0t’ and B—m” transformations, and suggested the
orthorhombic or” structure was an intermediate structure between B->0t’ transformation.
However, the principle lattice strain between 0t’—90t”+B is significantly larger than
a’—->0t+B due to large 2-0 difference between a’ and a” peaks in the X-ray scan,
suggesting the transformation 0t’—)0t”+B is thermodynamically unfavorable.

However, many authors reported that orthorhombic 0t” structure could form by the
spinodal decomposition of retained B phase during aging into Bv lid), and By [an phase,
B—9B1m+Brich—>a”+5rich. Therefore, it is certainly possible that the B phase precipitated
from the decomposition of or’ could further decompose into the a” and Bfich phases,
a’—->a+B—>01+B,ean+B,,ch—->0t+0t”+B,,,h.

The disappearing of or” and Bfich phases might be caused by the following reasons:
(1) Room temperature diffusions of oz” and B phases. Generally, upon aging, the

orthorhombic 0t” martensite can decompose via spinodal precipitation into 0t+B through

190

01”—)0t”}¢,n+0t”,,ch—->a+B [Davis 1979], or transform directly to or phase, depending on the
concentration of B stabilizer [Boyer-I 1994]. Thus, during the approximate one-year
period, the CL” may further decompose into 0t+B or or phase. To the author’ knowledge, no
room temperature V diffusion data is available. The typical diffusion distance for V in a
year is less than 10’5 um based on extrapolation of high temperature diffusivity and
activation energy, indicating diffusion cannot account for the aging. In addition, the B
phase obtained from artificial and/or natural aging is metastable at room temperature
assuming the V concentration in the B phase is less than 15-wt%. Thus, the B phase may
eventually transform back to stable 0t phase, leading to the natural aging effect. (2)
Hydrogen absorption. Hydrogen absorption can lead to the shifting of 0:” peaks to a
phase and disappearing of B phase due to eutectoid decomposition [Fang 1998, Qazi
1965]. (3) Oxygen absorption. The absorption of oxygen in the B phase can reduce the B
phase stability and promote the B phase transformation [Boyer-B 1994]. (4) Other

unknown mechanism. In general, additional work is neededed to fully investigate the
natural aging mechanism.
3.2.2 90° texture orientation rotation
3.2.2.1 Variant selections
According to the Burgers relationship (Figure 3.11) [Burgers 1934], upon heating
each indivial or phase orientation can produce six cubic orientations with a (110)p plane

parallel to the original (0001),, plane and with the six possible combinations of parallel

orientations, including two <111>g directions in the (110),; plane and three <1120>,,

directions in the (0001),, plane. On cooling to room temperature after solutionization,

191

each one of the six cubic orientations can generate another twelve orientations of the hcp
(1 phase, only one of which is the original on phase orientation. Therefore, one orientation
of Cl phase could ideally become 6X12=72 orientations of on phase texture, suggesting that
a nearly random texture could be developed after the a—aB—mt transformation cycle. In
contrast, the present study revealed a strong 90° rotation orientation with (0001),, plane
normal direction aligned with the wire axis, suggesting that a strong variant selection
occurred during the solutionization and quenching. This is consistent with the prior result
that upon heat treatment in the B phase field, the or phase texture developed without
variant selection, while heat treatment in the 0t+B phase field, the resulting on phase
texture had a strong variant selection at room temperature [Gey 1996, Moustahfid 1997].
3.2.2.2 Stress

It has long been realized that constraints, such as residual stress and oriented
overgrowth, can affect the B—>or transformation texture, resulting in only one of the six
(110),; planes being favored while transforming to the (0001),, with two strongly
preferred orientations [Keeler 1956]. Upon heat treatment in the 920°C oc+B phase field,
approximately 40~60% of the primary or grains would transform to high temperature B
grains [Kahveci 1986]. This Ot—>B transformation leads to large internal stress build up
between the B grains and the remaining or grains to avoid cracking due to changes in the
atomic volumes of or and B phases [Rudrnan 1965]. This large internal stress can be
reduced by volume contraction by as much as 0.55~2.6% through annihilation of
transformation dislocations [Fujishiro 1980]. Thus the magnitude of internal stress would
depend on the progress of a——>B transformation and dislocation movement and recovery,

which in turn depends largely on the solutionization temperature, time, solute

192

concentration, spatial phase distribution, initial stress state, high temperature texture
evolution, grain sizes, grain growth behaviors and impurities. To the author’ knowledge,
the complicated physical metallurgy operating upon heating and solutionization is not
available in open literature, and this might be more complex than the transformation
process occurring during the B—90t transformation [Moustahfid 1997].

In addition, the on grain has a large anisotropic thermal expansion coefficient. For
example, the linear thermal expansion coefficient along the c-axis is 20% higher than
perpendicular to the c-axis [Boyer-J 1994]. Upon cooling prior to martensitic
transformation, the large anisotropic thermal expansion of the existing or grains can lead
to internal stress between different orientations of or grains, as well as with B gains that
would shrink by transformation, assuming the stress state at the end of solutionization is
zero. Prior to the beginning of B-)0L’ transformation at approximately 800°C, the B and
a grains would shrink, with or grains shrinking preferentially along the c-axis. Figure
3.12 shows a schematic representation of in plane (radial and hoop) stresses that develop
upon cooling prior to B—>or’ transformation. The stresses in the B gains could be tensile,
zero or negative magnitudes, depending on the exact arrangements of the or and B grains,
in planes perpendicular to the wire axis. However, in the present research, the majority
of on gains in as-received wire were orientated with the c-axis along the radial direction,
suggesting that internal stresses might be highest in the plane perpendicular to the wire
axis in the high temperature B gains (Figure 3.12), prior to transformation.

Based on this assumption, heat treatment in the B phase field would result in
uniformly distributed B grains which have isotropic thermal expansion, leading to

minimal internal stress between each B grain upon cooling prior to martensitic

193

transformation. This is consistent with the results that solution treating in the 0t+B phase
field results in higher value of internal friction than solution treating in the B phase field
[Ouchi 1999].
3.2.2.3 Texture of high temperature B phase

Unfortunately, the direct measurement of high temperature B texture was not
possible with our texture system. However, the high temperature B phase texture can also
be inferred from the a-—>B transformation based on the Burgers relationship without
valiant selection relationship (Figure 3.13(a)). It is expected that the majority of the
(110)}; plane normal angles with the original wire axis would be 0, 30, 60, or 90°,
assuming variant selection does not apply (Figure 3.13(b)). Furthermore, the angle of 0,
60, or 90° could come from the same B orientation. In addition, the high temperature B
phase texture that resulted in the 90° texture rotation can be inferred based on Burgers
relationship (Figure 3.13(c)). The particular (110),; plane that can become the (0002),,
plane aligned along wire axis in the B—)0t transformation was always parallel to the radial
directions, suggesting majority of (110),} planes normals aligned with the wire axis.
Thus, the B phase (110),, plane normal having 30° angles with wire axis, did not exist
after the (at-9B transformation due to variant selections, or it disappeared due to B gain
g'owth resulting in shrinking of the non-favored B orientations. Therefore, the majority
of the B phase is orientated with one of their six (110),; planes normals aligned with the
wire axis. However, it is almost impossible to determine if variant selection occured in
the a—iB transformation.

3.2.2.4 Variant selections and activeness of slip system

194

As mentioned above, the internal stresses most likely build up along the radial

directions. These radial internal stresses could potentially and locally deform the existing
high temperature B phase prior to transformation. Most importantly, from the lack of
change in the original a component of the texture, the change must have occurred in the
part of the microstructure that transformed from a—rB—ra such that the majority of the B
phase was orientated with one of their six (110)}; planes normal aligned with the wire
axis. Given this B fiber texture, then there would be little if any shear stresses from
expansion mismatch resolved on the (110),} plane that became the transformation planes
to form 0!. Thus, based on the proposed model by Gey et al. [1996, 1997] who conclude
that the most active slip system correlates to a high density of defect sites that would
favor the transformation texture, the other four (110),; planes would have had shear
stresses resolved on them to generate dislocations that would favor them becoming the
(0002),, plane after the B-)0t transformation. However, this was not the case. In the
current study, the data suggests that B—)0t transformation preferentially concentrated on
the particular (110),; — (0002),, plane with its normal parallel to the wire axis. The
differences might be explained by the following reasons:

(1) Different processing history. In the experiment developed by Gey et al. [1996,
1997], high temperature “continuous dynamic recrystallization” with small equiaxed B
grains was observed before further cooling down to room temperature. In addition, the
cooling rate was chosen to give micro-sized (21 pm thick) or platelets (as compared to
~0.05 pm thick in the present research), suggesting that direct quenching was not used.

Thus, upon cooling, those newly nucleated equiaxed B grains can further grow,

195

precipitate on platelet via nucleation and growth, and affect or even determine the
subsequent variant selections [Keeler 1956]. However, in the current research, direct
quenching was employed to eliminate any possible diffusion or nucleation prior to the
martensitic transformation (which is not thermally activated) and takes place at high
(temperature-independent) speed. Therefore, hot rolling in the B phase field with more
than 30% reductions leading to the on phase texture with variant selection might be better
categorized as a recrystallization texture instead of transformation texture [Gey 1996,
1997].

(2) Internal stresses level. The internal stresses that developed in B grains may not
have been high enough to generate a large number of defects on (110),; planes, thus, there
may be other criteria to determine the favored transformation texture. The present studies
suggest that the other five (110),; planes under local stress that can be activated for slip
were not activated to induce the formation of the or phase upon cooling, and in contrast to
prior work, the development of a preferred variant is apparently related to the least active
slip system. Further investigations are needed to determine if there was any reason to
expect pro-existing dislocations to be present on the (110),} planes that became the basis
for the observed transformation, that may have resulted from variant selections from
a—>B, development of high temperature B phase texture, and thus affected the B—-)0t
transformation.

3.3 Conclusions

The present study investigated the effects of working and heat treatment on

microstructural evolution and crystallographic texture of on phase in Ti-6Al-4V. Texture

analysis revealed a 90° rotation from the prior texture emerges with the (0001) c-axis

196

aligned with the wire axis after quenching, which can be correlated to the least active slip
system within the high temperature B phase as it accorrunodated differential thermal
contraction in or during cooling.

Microstructural analysis showed a’ martensitic plates formed after solutionization
and quenching. The subsequent artificial aging process led to the formation of metastable

0t” and B, but natural aging for a year caused 0t” and B to transform back to the or phase.

197

 

I .alpha" A alpha x beta

 

 

 

 

 

 

 

 

 

6
a It Ix ‘-..
4 _

Viewed
< 2 - I I I, .X along the
.. c—aro's

:
2 or beta
§ 0 i I .< .X l X (110)
.5 plane
.0 .2 - normal
I i 3‘ 1"
.4 u
x
i i‘ 1" r-
.6 i I T I i
-2 0 2 4 6 8 10
a direction, A
6 [ lalpha' eaipha xbeta
I I Ix I x
4 4
Viewed
2 q x x x x along the
<. b-axis,
S prism
§ 0 a I I: .x AI x plane
.= normal,
1:
o _2 . orbeta
x x x x (110)
normal
.4 _
I 8 IX AI X
.6 l I f i J
-2 0 2 4 6 8 10
a direction, A

Figure 3.1 Crystal phases and their coherency relationship to each other.

198

Table 3.1 Specimens examined using X-ray diffractometer, texture measurements,
and TEM technologies. Note, M stands for middle of the coil, ex for extruded, sq for

solutionization and quenching, a for aging.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Process ID Phase Processing conditions
constituents
B or, B Beginning of Coil
Bex or, Beginning of Coil, Extrusion
M or, Middle of Coil
Mex or, Middle of Coil, Extrusion
E or, End of Coil
Eex or, B End of Coil, Extrusion
M-sq20 on’, 0t Middle of coil, 920°C for 20 minutes,
Mex-sq20 or’, on Middle of coil, Extruded, 920°C for 20 rrrinutes
M-sq60 or’, on, Middle of coil, 920°C for 60 minutes,
Mex-sq60 a’, 0t, Middle of coil, Extruded, 920°C for 60 minutes
M-sq20-a a” a’ a 3 Middle of Coil, 920°C for 20 rrrinutes, Aged at
’ ’ ’ 550°C for 4 hours
Mex-sq20-a a” or’ or Middle of coil, Extruded, 920°C for 20 rrrinutes,
’ ’ ’ Aged at 550°C for 4 hours
M-sq60-a or” a’ 01 Middle of Coil, 920°C for 60 minutes, Aged at
’ ’ ’ 550°C for 4 hours
Mex-sq60-a a” oc’ on Middle of coil, Extruded, 920°C for 60 nrinutes,
’ ’ ’ Aged at 550°C for 4 hours

 

 

199

 

As-received wire Extruded wire
(9)

Figure 3.2 shows the primary or phase pole figures of (a) the as-received middle of
wire M with wire axis out of page and (b) the extruded Mex with wire axis out of page
sectioned from middle of as-received wire, and (c) the corresponding crystal orientations

with specimens. The crystal prism plane normal direction, (1010),,“ is aligned with the
wire axis, and the (0001),, plane normal direction is in the radial direction in the as-
recived and extruded specimens.

200

 

 

 

Ends of cor] 5 Middle of coil I 10-10 H wire
: l 0002 || wire
after quench El 11 radial
___l0002
after aging

L0g(Intensity x random)

 

=5

e
e
o
e
e
.
.
- e
- e
- u
u u
e .
u
u
.
e
-
-
l l =
K \

6’9 to"? a?" «25' g5" ‘80' 0° 09 e” e” g”
sf egg 9g+§

9" sit

 

 

Figure 3.3 The overall texture variations from as-received wire through extrusion,
solutionization and quenching, and aging.

 

XRD of Mas and M-st20
1800
1600 - a(0002)
1400 —
1200 —
m1000 _ ot(10-10)
91
o 800 — l
600 r
400 A
200 -
0 2A

34 35 36 37 38 39 40 41 42 43
2-Theta

 

   
    
    
  

0t(10-11)

 

As-received

Solutionized &
quenched

3(110)

 

 

 

 

 

Figure 3.4 X-ray diffraction patterns of the wire M cut from middle of as-received
coil showing primary 0t and B phases, and as-quenched M-sq20 specimens showing the
peaks broadening due to the formation of (1’ martensites.

201

 

 

Figure 3.5 shows the TEM image of as-received rrriddle of wire M and the related
indexed diffraction pattern from primary B phase.

202

 

(b)

Figure 3.6 TEM rrricrographs of (a) the primary B phase in the extruded Mex
specimens showing high dislocation densities and (b) the primary a grains in another thin
foil. Inserts show the corresponding diffraction pattern from (a) primary B and (b) 0t
phases.

203

H129 Convicted by HEW
(aquel—u'ee proJ.) nex.- 5.51

4."
2.83
LU
1.11

 

(a)

Int-291 Cmplntnrl bu m
(equal-Iron proJ.) tux.- 4.97

[0001]

 

 

[IO-10] [0001]

(C)

Figure 3.7 shows the primary on phase pole figures of (a) the as-quenched nriddle of
wire Mex—sq60, (b) the aged Mex-sq60-a with wire axis out of page sectioned from
middle of as-received wire, and (c) the corresponding crystal orientations with
specimens. The pole figures indicated that a new fiber texture with the crystal basal

plane (0002),, normal direction aligned with the wire axis and the (1010),, prism plane
normal direction perpendicular to the wire axis after solutionization and quenching. This
introduction of a new orientation weakened the prior fiber texture having (0002),, normal
direction aligned along radial directions.

204

 

 

(b)

Figure 3.8 shows (a) the TEM image of as-quenched M-sq60 sectioned from middle
of wire (b) the related diffraction pattern from primary B phase showing a [111] B zone

pattern and three or <1120> patterns superimposed, and (c) is the schematic of (b).

205

 

 

XRD of M-st60 and M-st60-a

 

      
 
    

 

 

 

 

 

2500
“’107-0) ottooom
2000 a (1(11-20)
- _ (1110-12)
“(1010) WHO-ll) M
06’ ,,
L91000 1 M-sq60-a \ l
‘ Qie-mcasmgL - a - A 1
my 1 / i .
0 AIA Af l l ’1 l r i IA fl l l 1 _1A‘ I l r

 

33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69

 

 

Figure 3.9 X-ray diffraction patterns of the as-quenched M-sq60, M-sq60-a and re-
measured M-sq60-a (year later) specimens, showing the formation of orthorhombic 0t”
martensites and B phases due to artificial aging, and disappear of on” martensites and B
phases due to natural aging effect. Note, the M-sq60-a and re-measured M-sq60-a (after
a year later) are the specimen measured at different time periods.

206

 

(b)

Figure 3.10 shows the TEM image of (a) the aged M-sq60-a specimens, and (b) the
aged M-sq60-a specimens one year later. Inserts show the corresponding diffraction
pattern from the corresponding indexed diffraction pattern from primary B phase

207

Table 3.2 Lattice parameters of Ti-6A1—4V specimens at various experimental

processing conditions.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

or (A) or” (A)

Process a c B (A) a b c
M 2.9131 4.6572 3.2415 ------------------
Mex 2.9164 4.654 3.2463 ------ ------ ------
M-srL60 2.9223 4.669 ------------------------
Mex-sq60 2.9217 4.6676 ........................
M-sq60-a 2.9214 4.6672 3.2552 3.0327 4.9238 4.6672
Mex-sq60-a 2.9212 4.6676 3.2553 3.0334 4.9246 4.6676
M-sq60-a (new) 2.9247 4.6708 ------------------------
M°"'Sq°°'° 2.9251 4.6708 ............ -..--- ......

(new)

 

208

 

[0001],,|| [110]p

 

 

 

 

 

 

 

 

 
 

 

‘ C
5 /\

.4'1 """""" 7%.

{I t“ I” t“

 

 

‘{ ’ [lll]p||[1120],,

Figure 3.11 The lattice correspondence for B, and a’ (or) transformations.

209

 

 

 

 

 

 

 

 

 

E:.'>

or a In plane
Stress built up

 

 

 

 

 

 

(a) Initial two 01 grains prior to cool down (b) In plane stress built up upon cooling

 

(I

(1

 

 

 

 

 

 

 

 

 

 

 

(c) In plane stress develops due to B grain shrink upon cooling prior to B—m’
transformation

Figure 3.12: Schematic representation of internal stress build up between different

orientations of or gains, as well as with B gains upon cooling prior to transformation to
maintain compatibility due to large anisotropic thermal expansion difference between the

a and c-axis. Initially, (a) assuming these two hexagonal or gains are well annealed at
high temperature, and have a 90° mis-orientation. The (b) large anisotropic thermal
contraction along c-axes leads to the development of in plane or radial stress between or
grains upon cooling. In addition, if a B grain is located between these two or gains, the
in plane tensile stress could develop in the B grain due to cooling. Note: the a-axes of the
two 0t grains drawn are out of the page or radial directions are in plane.

210

4- >
E a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘xdl 10]};
V lllllsllllliol ‘
01
[l 1 0113 [1 10k}
(a)
Relative ‘
intensity
Angle between (110),;
and wire axis
b
0 30 60 90
(b)
l (0002),,
->

 

 

 

 

 

(C)

211

A

Wire axis

 

 

 

0
L2
V «19%
radial
(91)

Figure 3.13 Schematic represent the possible (a) high temperature B phase texture,
and (b) the angles of (110),; with wire axis based on the Burgers relationship [Burgers]
from 0L—->B, (c) high temperature B phase texture could lead to the formation of or phase
texture with the c-axis normal along the wire axis upon quenching, and (d) the
corresponding crystal orientations with specimens.

212

Chapter IV
Effects of solution temperature and time on crystallographic texture

and mechanical properties in Ti-6Al-4V wire

Overview

Effects of solution temperature and time on crystallographic texture of the (X phase
and mechanical properties in Ti-6Al-4V were systematically and comprehensively
examined using a series of microstructural and texture examinations and mechanical
property tests. The wire was worked by extrusion, solution heat-treated and water
quenched, then aged. As-received and extruded wire had a moderately strong fiber
texture with prism plane normal aligned with the wire axis. After solutionization and
quenching, a secondary fiber with the basal plane normal aligned with the wire axis
emerged and consistently increased with solutionization temperatures and times. This
implies more favored 90° mis-orientations developed from prior texture after higher
solutionization temperatures and longer time heat treatments. Microstructural analysis
revealed that the B phase volume fraction increased with solutionization temperature
independent of solutionization times. The 0.2% yield strength increased consistently in
both wire and radial directions with increasing solutionization temperatures and times,
while Young’s Modulus was consistently higher in the radial direction than that in the
wire direction. The relationship between the variant selections of or phase, and the effects
of dislocations are discussed in the context of the known physical metallurgy of titanium

and other alloys.

213

4.1 Introduction

In chapter III, the effects of working and heat treatment on crystallographic texture
of the or phase revealed that strong variant selections or 90° texture rotations from the
prior texture emerged with the (0001) c-axis aligned with the wire axis after
solutionization and quenching. However, many questions about the underlying
transformation texture mechanisms and microstructure development remain unanswered.
First, despite the fact that solutionization and quenching has been shown to cause variant
selections, it is not clear if a higher solutionization temperature would cause a stronger
90° texture rotation with c-axis normal along the wire axis. This is because higher
solutionization temperature in the or-B field could potentially generate increased amount
of high temperature B phase, leading to different or phase orientations upon subsequent
quenching and aging. Second, although a longer solutionization time and quenching has
been shown to enhance the preferred variant selection, it is not clear if this strong 90°
texture rotation is caused more by the B—->0t transformation variant selection, or whether
it depends on the complicated physical metallurgy involved with the oc-B solutionization
process. Third, it is not clear if a different 90° orientation rotation from the prior
orientations having the (0001) c-axis along the radial directions could occur which would
maintain the prior texture. This is because there are two perpendicular orientation
changes with 90° rotation from a prior orientation in 3-D space. From a review of the
literature, the physical mechanisms of the 0t—>B transformation that operates with heating

and solutionization may be more complex than the process occuning during B—>0t

transformation, because the 0t—>B transformation is not well documented. Last, the

214

effects of texture rotations on the mechanical property development have not been fully
investigated.

Therefore, the purpose of this set of experiments was to comprehensively and
intensively assess the effects of process variables on the transformation texture,
nricrostructure, and mechanical property development in Ti-6Al-4V wire after
solutionization, quenching, and aging. The as-received wire specimens were partially
extruded, solutionized in the two-phase region at temperatures of 905, 925, 940 and
960°C for either of 1.2, 2.4, 3.6, or 4.8 ks in an industrial shaker furnace designed to
solutionize and quench specimens in a continuous process, followed by aging treatments
in air at 500°C for 28.8 ks and air-cooling. Cross-sectional OM, SEM, and TEM were
used to investigate the microstructure development, including primary a and B grain
sizes, aspect ratio, and volume fractions. Mechanical compression tests were conducted

with different specimen orientations, with the wire axis parallel or perpendicular to the

compression axis. (1010),, and (0002),, pole figure measurements were used to
investigate the effect of heat treatment variations on the texture after solutionization and
transformation. X-ray diffraction of selected specimens with the surface normal parallel
to the wire axis was assessed with serial sectioning to examine the effects of alloy and
microstructure variations that could cause differences in peaks (phases), peak shape
(defects), peak position (stress), and intensity (texture).

4.2 Experiment results

4.2.1 Texture of as-received and extruded specimens

As in previous studies of Ti-6Al-4V wire in chapter IH, two on phase pole figures

(1010),, and (0002),, of the as-received wire in diameter of 4.9 mm revealed that the

215

primary or phase orientation is a strong fiber texture with the axisymmetric crystal

orientations about the wire axis (Figure 4.1(a)). After extrusion, the primary or phase

texture was strengthened. The crystallographic direction [1010],, in primary 0t gains is
parallel or nearly parallel to the wire axis, and the [0001],, direction is in the radial
direction, similar to the texture of as-received wire (Figure 4.1(b)).

4.2.2 Texture of solution treated, quenched, and aged (SQA) specimens

Figure 4.2(a) reveals the overall relationships between major texture intensities and
the solutionization conditions. After SQA, the B phase orientations could not be
measured due to very small amount of B phase existing. The texture measurements show
that the or phase orientation consisted of two fiber textures; one newly emerged fiber
texture with the crystal basal plane (0001),, normal direction aligned with the wire axis
and the other prior fiber texture orientation with the basal plane (0001),, normal aligned
perpendicular to the wire axis (Figure 4.1(c)). The new fiber texture has a 90° rotation
from the prior texture, suggesting that a transformation texture and variant selection
occurred, leading to a crystal orientation with (0001),, plane normal aligned with the wire
axis after the transformation cycle a—rB—mt. Therefore, the presence of 90° mis-
orientations from the prior texture may be a transformation related texture rotation, or
possibly the result of a recrystallization process that causes a preferred orientation.

The intensities of the newly emerged fiber texture change dramatically with
different solutionization times and temperatures. The intensities of the original fiber
texture with the basal plane (0001),, normal aligned perpendicular to the wire axis are

essentially the same for all SQA specimens (Figure 4.2 (b)). However, the intensities of

216

the newly emerged fiber texture with the crystal basal plane (0001),, normal aligned with
the wire axis is strengthened with increasing solutionization times and temperatures
(Figure 4.2 (c)), indicating that special 90° misoriented orientations relative to the prior
as-received crystal orientation are more favorable at higher solutionization temperatures
and longer times.
4.2.3 Microstructure of SQA specimens

In addition to texture examinations, OM, SEM, and TEM were done to provide
detailed microstructural analysis of SQA specimens. After SQA, all specimens show
typical duplex nricrostructures, consisting of equiaxed primary or and transformed B
grains. The transformed B grains reveal martensites with an intertwined mixture of
needle or platelet shaped domains with many finely divided interfaces or twins, each
having a different orientation variant (Figure 4.3). Diffraction patterns taken from one of

the primary B grain regions reveals the Burgers orientation relationship of

<lll>g||<1120>,, and (110)B||(0001),, between or and B phases, consistent with other
studies in bcc<—->hcp transformation systems. No orthorhombic 0t” martensites or to
phases were observed. A diffraction pattern of a primary 0: grain (Figure 4.3) reveals this
particular on grain is oriented to the [221]-zone axis of the hexagonal on phase without
extra superlattice reflections due to the precipitation of a2 phase.

In addition, grain size, aspect ratio, and volume fractions of the primary or and B
phases were examined using cross-sectional OM and SEM observations since large areas
can be chosen and are therefore more representative. After SQA, the primary or and B

grains size were essentially the same and equiaxed. No abnormal grain growth was

217

observed. Figure 4.4(a) shows the overall relationships between the grain sizes, prior B
phase ratio, and the experiment solutionization conditions. As expected, with increasing
solutionization times and temperatures, the primary or and B grain size and volume
fractions of prior B phase increased consistently. However, Figure 4.4(a)) shows that the
volume fractions of the primary 0t and B phases are independent of the solutionization
time in the present studies, suggesting the kinetics of (it—9B transformation is very fast.
Therefore, with increasing solutionization temperature, more primary on phase
transformed quickly to high temperature B phase, so more retransformed back to a phase
upon cooling, resulting in more of the variant that is misoriented 90° from the prior
crystal orientation. In summary, the effects of solutionization time and temperature
increase the amount of texture rotations, the gain size and the volume fraction of prior B
grains.
4.2.4 X-ray diffraction of serially sectioned SQA specimens

Three representative parts were chosen for X-ray diffraction studies of serially
sectioned specimens; a sample solutionized at 905°C for 1.2 ks and aged at 500°C for
14.4 ks, a sample solutionized at 905°C for 4.8 ks and aged at 500°C for 14.4 ks, and a

sample solutionized at 960°C for 4.8 ks and aged at 500°C for 14.4 ks. The d-spacing of

(1010),, and (0002),, on phase peaks, d(1010)a and d(ooo|),, values changed with measurement

locations (500 um apart), indicating that lattice parameters vary with positions (Figure

4.5). However, both the or phase (1010),, and (0002),, lattice spacing decreased by about
0.2% after solutionization for 1.2 to 4.8 ks at 905°C, indicating that changes occurred in

dislocation density, internal strain, substitutional Al and V atom concentrations, or

218

impurities (primary 0) with time. The short time solutionization (905°C for 1.2 ks) led to
an increase of the unit cell size and volume and a decrease of the c/a ratio from the
extruded condition. The extrusion showed a larger c/a ratio (Figure 4.6(b)) which was
evidently caused by a macro residual stress from extrusion which generated a tensile
stress along wire axis direction and compression in the radial directions. After SQA the
variation in the ratios of the d spacings (Figure 4.5) were smaller by an order of
magnitude than variations in the d spacings (Figure 4.6(b, c)), indicating that the
proportionality of the on phase unit cell did not change, though the d-spacings varied with
time and position. In comparison, the variation in the d spacings and the ratios of the
planes were consistently higher in the extruded wire than those of SQA specimens,
indicating more consistency in the c/a ratio and d spacings in the SQA specimen than in
the extruded wire.
4.2.5 Mechanical properties of SQA specimens

Figure 4.7 shows the (a) overall relationships and the related statistical analysis
identifying the effects of solutionization temperature (b) and time (c) on 0.2% yield stress
and the apparent Young’s modulus along directions of parallel and perpendicular to the
wire axes for compression tests. Clearly, after SQA the 0.2% yield stress at room
temperature increased consistently in the wire directions with increasing solutionization
temperatures and times due to the increasing B volume fraction during the or—)B—>0t
transformation, but in the radial direction, yield stress minima were observed at 940°C.
However, the yield stresses in the radial direction are significantly higher than in the wire
direction at lower solutionization temperatures where there was less material with the c-

axis aligned with the wire axis (Compare Figures 4.2(a) and 4.7(a)). The apparent

219

Young’s Modulus was consistently higher in the radial direction than that in the wire
direction. Young’s modulus along both wire and radial directions goes through a
maximum point for the present alloys quenched from approximately 925°C and aging.
The existence of this maximum in stiffness and minimum in yield stress has not been
reported before.
4.3 Discussion

The present study investigated the effects of solution temperature and time on
crystallographic texture and microstructure evolution such as 90° texture rotations, grain
size, volume fraction of prior B grains, and the associated mechanical properties. Two
main questions arise about the relationship between transformation texture,

recrystallization, and microstructure development. First, it is not clear why the texture

intensities of crystal orientation of (1010),, aligned along wire axis were essentially the

same since less primary 0t phase was retained with increasing solutionization temperature

(Figure 4.2(b) and 4.4(b)). For example, the specimen texture intensity of (1010),, along

wire axis after solutionization at 960°C had less than 10% volume fraction of residual

primary or phase, but the intensity of the (1010),, along wire axis is the same as that of
specimens solutionized at 905°C which had around 50% on phase. Secondly, despite the
fact that the volume fraction of the primary B phase is independent of the solutionization
time, it is not clear why the texture intensities of crystal orientation of (0001),, aligned
along wire axis were consistently increasing with time (Figure 4.2(c) and 4.4(a)). It is
assumed that the longer time solutionization may lead to the further B grain re-orientation

or recrystallization or grain growth of B grains, making the 90° mis-orientation with

220

(0001),, along wire axis thermodynamically more favorable upon cooling. Therefore, it is
of interest to concurrently investigate the microstructural transformation and texture
rotation behaviors in relation to previous Ti-6Al-4V alloys studies, physical metallurgy
principles related to recrystallization, grain growth and phase transformations, and similar
microstructure studies of Titanium alloys.

4.3.1 Variant selections and mechanical properties

The present studies indicated that, with increasing solutionization temperature, the
resulting intensity of the or phase orientations with (0002),, along the radial directions was
the same (Figure 4.2) even with much less a phase existing at high temperature,
suggesting that some of the B phase transformed epitaxially back onto the existing parent
or phase variants with (0002),, along radial directions. If this were not so, after cooling
the resulting on phase texture with (0002),, aligned along radial directions would decrease
with increasing solutionization temperatures. This assumption is based upon experimental
results from rolling where one single or phase variant can generate many on phase variants
after 0t<—)B transformation cycles, including two perpendicular variants both having 90°
nus-orientations from prior orientation (Figure 4.8) [Moustahfid 1997]. Therefore, it is
expected that at elevated temperatures the or phase transforms to B phase, followed by
transformation to two or more favorable 0t variants occurring during quenching, based on
the Burgers relationship [Burgers 1934].

With increasing solutionization temperatures and times, the yield stress difference
between wire and radial directions decrease consistently (Figure 4.7) and can be
explained qualitativly: When the solutionization temperatures are increased, the volume

fraction of high temperature B phase becomes higher, leading to more 90° 0t phase

221

rotation from the dominant prior crystal orientation, so the (0001),, hard orientation of the
oc phase is aligned along the wire axis rather than only along the radial direction. This
development of nearly equal fractions in the two orientations has a homogenization effect
on the modulus and microyielding processes, leading to the almost equalized yield stress
in both wire and radial directions. It is assumed that this larger difference in the effect of
solutionization on Young’s modulus is due to differences in the effects of microstructure
and crystal orientation on bond strength. Generally, Young’s modulus of polycrystalline
materials is mainly determined by a weighted average of the modulus for the constituent
crystals, which depends on the crystal orientation. Comparison of Figures 4.2(a) with
4.7(a) show a strong correlation between modulus values, and texture measurements,
indicating that texture measurements are qualitatively representative of the volume
fraction of crystal orientations. However, yield stress is more strongly affected by the
material’s microstructures than the crystal orientations.
4.3.2 Grain growth behaviors

At elevated temperatures, grain gowth occurred after the a—)B transformation. For
many materials, the isothermal grain growth can be represented by equation of the form
d“ — do“ =Kt,

Where d denotes the mean grain diameter, d,, represents the starting diameter, t
denotes the time, K denotes the constant, and n denotes the grain gowth exponent.

The driving force for the Cl and B gain growth is a reduction of grain boundary
energy. This process occurs by an exchange of elements or fluxes of the constituents
such as Ti, Al and V between different phases. The phase boundary interfaces move by

the process of conversion of one phase to another. Thus, regions of or phase convert to B

222

phase with diffusion of V from or to B gains, and Al away from the B phase gains,
resulting in overall a and B gain growth while maintaining a constant volume ratio
[Grewl 1989, 1990]. Since the rate-controlling mechanism requires exchange of V and
Al between phases, this process is the same for both a and B gains, so the gowth
behaviors of both or and B phases should be the same, consistent with the experimental
results. The average diffusion distances of Ti, Al and V in the B phase under
experimental solutionization conditions were faster than those in 0t phase (Figure 4.4(c)),
and the diffusion distances in the OL phase are comparable to the rate of boundary
migation shown in figure 4.4(a). However, the gain growth exponents at different
solutionization temperatures were significantly different, ranging from 7.5 at 900°C, to
2.5 at 960°C, indicating different diffusion rate controlling mechanisms, though the
trends of average diffusion distances of Ti, Al and V in 0t and B phases were similar at
different solutionization temperatures.

Furthermore, the gain growth behavior strongly affects the texture. A strongly
textured material contains low angle gain boundaries, having reduced gain boundary
mobility and driving force for gain gowth. During the current research, increasing gain
sizes leads to the slowing down of gowth rate based on gain gowth exponential
exponent, suggesting decreasing in gain boundary mis-orientations or local grouping of
similar gain boundaries [Boswell 1995]. Therefore, it is highly possible that gain
gowth by gain rotation or preferential migation of certain boundary misorientations
could decrease gain boundary area and strengthen the pre-existing B phase fiber textures,
resulting in preferred or phase orientations with (0002),, along wire axis upon quenching.

4.3.3 Dislocations and (re-9B transformation

223

The changes in the unit cell size, c/a ratio and d-spacings after short time
solutionization were due to changes in residual stress and dislocation density, and not due
to compositional changes. The lattice parameter of the CL phase is unaffected by heat
treatment because the composition is very stable [Boyer-A 1994]. Since Al atoms are
bigger and V atoms are smaller than Ti atoms, an increase of Al and a decrease in V in
the or phase should increase the size of the unit cell slightly with increasing
solutionization time, in contrast to the observations in figure 4.6(a). Since the a—B
transformation occurred quickly (Figure 4.4(a), and since gain gowth occurred at a rate
similar to Al and V diffusion in 0t (Figure 4.4(c)), this implies that the dislocation density
increased due to the transformation, such that the unit cell was still large after the short
time solutionization after extrusion process. With longer time and higher temperature
solutionization, a reduction of dislocation density occurred that is shown by the reduced
unit cell volume and the reduced and well established c/a ratio. Thus, the high
temperature solutionization process can be viewed as the combinations of 0t—)B
transformation and primary recrystallization processes which resulted in a reduction of
dislocation density.

Because the or—>B transformation proceeds with the superimposed recrystallization
process [Divinski 1998], the 0t——>B transformation and differential thermal expansion
mismatch stresses could generate dislocations at or near the ovB phase boundaries,
resulting an increasing of dislocations densities after short time solutionization (Figure
4.5 and 4.6). This internal strain/stress can be reduced by volume contraction through
annihilation of nearby transformation dislocations [Fujishiro 1980], but the subsequent

gain gowth by 0t<—>B phase transitions can continually exert internal stress as long as

224

gain gowth continues. Thus, it is expected that micro-strains would always exist at Ol/B
phase boundaries; while their magnitude would decrease due to the slowing down of the
gain growth and recrystallization behaviors, consistent with our serial sectioning X-ray
diffraction studies and internal friction measurement that higher solutionization in the
oc+B phase field results in lower internal friction [Ouchi 1999].

Cahn [1953] demonstrated that the dissipation of strain associated with dislocations
via the formation of a secondary phase is facilitated by preferential phase nucleation on
slip planes with dislocations. It is expected that the B—m transformation could start from
the 011B boundaries. The alternative (transformation starting from BIB boundaries) is not
reasonable, because if the transformation started from BB boundaries, similar to
solutionization in the Ti-6Al-4V B phase field, the resulting 0t phase texture would
develop without variant selection [Gey 1996, 1997]. The most pronounced difference
between (ll/B and BB boundaries is that strains evolve at ovB boundaries, accommodated
by dislocations, suggesting that dislocations near boundaries could facilitate the
transformation. Hence, the dislocations on the {110}5 planes with high dislocation
densities could provide the martensite nucleation sites [Furuhara 2001].

Thus, the most favorable variant is likely to be the one which has the resulting
transformation strain accommodated most effectively with the pre-existing dislocations to
minimize the required nucleation energy for transformation [Humbert 2002]. For

example, if specific {110}B planes have lattice strains of approximately 1.38%, 9.9%, or -

3.4%, then these specific (110),; planes are more likely transform to basal (0002),,,

(1100),, or pyramidal (1011),, planes, respectively, during the subsequent B—)0t

transformation since the required lattice distortion and energy is minimized with this

225

correlation (see Table 4.1). In addition, the direction of the maximum misfit strain field

between the parent and the secondary phase (<111>B||<1l20>,,) tends to develop along
the direction of Burgers vector of dislocations, located either at the boundary or inside the
B crystal [Furuhara 2001]. In addition, the habit plane of the newly formed or phase tends
to be parallel to the dislocation line [Luo 1987], which cannot be determined due to the
lack of measureable B phase after quenching in this study.

Furthermore, Furuhara et al. [2001] studied the gain boundary 0t precipitates from
B matrix via low temperature aging of B titanium alloy. The 0t precipitates maintained

Burgers orientations with their parent B gains, and coherency relations with nearby B

gains via maintaining the maximum misfit direction, <111>B||<1120>,,, parallel to the
closest B/B grain boundary plane. Thus, it is expected that coherency could also be
maintained between the newly formed or and the pre-existing neighboring or gain, in
addition to the Burgers orientations of the newly formed or and parent B gain. Therefore,
it is expected that the metallurgical state of high temperature or and B phases, and
boundary dislocations could affect or determine the subsequent variant selections.
4.3.4 Conceptual model for the B—ra transformation

The dislocation arrangement could strongly affect the B—>0t transformation. The
gain gowth behaviors via a<—>B boundary movement resulted in the same gain sizes of
a and B, independent of solutionization temperature and time, implying the
transformations of a—>B and B—>0t were essentially the same. Thus, it is likely that
overall, 0t—>B and B—>0t boundary misfit dislocations were similar at all times, although

dislocations inside gains and ovB boundary reactions could affect the misfit boundary

226

dislocations. The texture analyses revealed the presence of 90° mis-orientations from the
prior texture, implying that a common texture rotation occurred, suggesting that the a/B
boundary dislocations have an intrinsic feature which could result in the 90° mis-
orientations. This is consistent with results that the development of the new on texture
component is independent of cooling rate or B—NX transformation mechanisms, either
martensitic or diffusional, as pointed out by many authors [Fujishiro 1980, Bieler 2000].
In the current study where longer time solutionization resulted in more strongly
favored variant selection can be explained by a dislocation reaction effect. Upon
solutionization, B phase boundary sweeping across on gains containing defects could
maintain existing dislocation structures, since the boundary is coherent, or generate
transformation-induced dislocations. In addition, dislocations inherited from primary on
phase or caused by transformation could disappear by recrystallization in the high
temperature B phase, leaving near-perfect B crystals, as indicated by the reduction in the
unit cell dimension in figures 4.5 and 4.6. Thus, for short solutionization times, the
dislocations in the boundary may have more variations, as indicated by the larger unit cell
dimension with 20 minutes than 80 minues in figure 4.6, resulting in more variant
selections of the B-oc transformation, making the resultant texture is less pronounced.
Thus, the longer the solutionization time, only the most stable boundary dislocations
would remain, leading to a more favored 90° variant selection. This indicates that the
variant selection process is strongly affected by B phase primary recrystallization
processes that reduce the dislocation density during solutionization. However, further

investigations are needed to fully characterize the change in metallurgical state of a and B

227

gains, especially regarding the relationship between crystal orientations, dislocation
orientations and variant selections.
4.4 Conclusion

The present study systematically investigated the effects of heat treatment on
mechanical, microstructural and crystallographic texture evolution of the or phase in Ti-
6Al—4V using compression tests, OM, SEM, TEM and X-ray diffraction techniques.
Texture analysis reveals 90° rotation from the prior texture after solutionization and
quenching with the c-axis aligned with the wire axis. The 90° rotation consistently
increased with solutionization temperatures and times, in a manner consistent with
compression tests showing that the yield strength difference between wire and radial
directions consistently decreased. The transformation texture is mainly determined by
the orientation of dislocations, located either at phase boundary or inside the gains. The
presence of 90° mis-orientations, independent of solutionization temperatures and times,
can be intepreted as a common transformation related texture rotation due to the intrinsic

OVB boundary dislocations.

228

Table 4.1 lattice strain or linear dimensional changes in B—ra transformation.

 

 

 

 

 

 

 

B phase Lattice plane or phase Lattice plane dirrfiggilornal
distance (A) distance (A) changes (%)
(110),, 2-3 ——> (0001) 2.33 1.38
(110),, 2.3 ——> (1100) 2.53 9.9
(110),, 2.3 —_> (0111) 2.22 -3.4
(011)!5 2.3 ——> (1011) 2.22 -3.4
(101% 2.3 —-> (0111) 2.22 -3.4
(011), 2.3 __> (1011) 2.22 -3.4

 

 

 

 

 

 

 

Convicted by HE“
(equal—em proJ.)

 

(a)

229

Canister! MI
(equal—area proJ.)
. r . .

 

New

orientations “01 0]

(C)

Figure 4.1 The primary 0t phase (1010) and (0002) pole figures and schematic
crystal orientations of (a) as-received wire, (b) extruded specimens, and (c) extruded
specimens after solutionization at 925°C for 20 nrinutes and aging (SQA) with wire axis
out of the page.

230

 

 

Effects of solution temperature and time on crystal orientations

 

12

 

 

 

 

 

 

 

 

 

 

I10olO|| wire

E 10 1 I0001 || wire
-8 8-
E
’1 6 ~
.9
E:

2 -

O ..

9f .89 0 90 90 9° 3 .90 ref
9950 913‘} 9134s 995$, 9'11)“ 9'19 qfiyyye§$ efi 9'9\ $29 (19:? (9099 9Q k9991911119
Solutionization conditions
(a)

 

Effects of solution temperature on crystal
orientations (average of 4 times)

 

Intensity, x random

 

 

 

I 1010 || wire
I0001||wire

 

 

 

 

 

905 925 940 960
Temperatures
(b)

231

 

 

 

 

 

 

Effect of solution time on major crystal
orientation (average of 4 temperatures)
7-00 ‘ I 10-10 H wire
.0001 wire
E 61” "
'8 5 oo -
5 r
h
a: 4.00
E
g 3.00 -
a
5 2.00 -
1.00
0.00
20 40 Time 60 8o

 

 

 

(C)

Figure 4.2 The overall relationships (a) between major texture intensities and the
experiment solutionization conditions, and (b) the effects of solutionization temperature
and (0) time on the major texture intensities using statistical analyzes via averaging the
corresponding texture intensities of the same temperature of different times and same
times of different temperatures, respectively.

 

Figure 4.3 TEM image of specimen after solution treated at 925°C for 2.4 ks and
aging at 500°C for 28.8 ks showing typical duplex nricrostructure, consisting of equiaxed
primary or and B gains. Diffraction patterns taken from one of the transformed B gain

regions reveals the Burgers orientation relationship of <111>g||<1120>,, and

(110)g||(0001),, between at and B phases, consistent with other studies in bcc<—>hcp
transformation systems.

232

 

 

Grain sizes and prior Beta phase ratio

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

233

1.00
14-00 9 —I—Grain size (um) 9 095
~- 0.90 .°
+ ' 5
12.00 ,_ Beta phase ratro d? 0.85 E
0
g N i 0.80 g
.5 10.00 1r 4- 075 9'
° 8
65 8.00 4 " ”-70 .8
-_ 0.65 :5
6.“) irM “ 060 E
-- 0.55
4.00iii4‘iilfiflifi‘ii‘..r0.50
gofifiefffi goflfiéfey ”90% figy $6194;er
$$®$ 099$ 9999 33 §
(a)
Effect of temperature on the grain sizes and prior
beta phase ratio (average of 4 times)
12.00 1" +Grain size (um) .- 33(5)
“'00 1’ +Beta phase ratio -- 0.90 g
8 10.00 4 9 0.85 a
'5 9.00 -- t 980 .3
a 4 0.75 g a
O 8.00 ._ .- 0.70 I5
7.00 .. -~ 0.65 E
.. 0.60
6'00 ‘” 4 0.55
5.00 ‘n i 1‘ 0.50
905 925 940 960
Solutionization temperature
(b)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Average diffusiond’mtances of Ti,Al,Vina 3rgrrdbphrirses
33-.
~30

A 29""
s i
g 25 4 ~25§
0 ' ..
i 21-- .a e A +11%“
:5 E g +Trrnb

17 ~- -
.g d 155V —.—Vmb
g 13.- > +Alina
b 9 dog +Alinb
1': 2'

5~~ “5

hiflmflflfln

e9 sf?“ 89 89,89 boggy
a9$9§§§° “n9“9, 98323:“: .9 93$“ $69,994:“ ,
Solutionization
(C)

Figure 4.4 The overall relationships (a) between the gain sizes, prior B phase ratio,
and the experiment solutionization conditions, (b) the effects of solutionization
temperature on the gain sizes and prior B phase ratio using statistical analyzes via
averaging the corresponding gain sizes and prior B phase ratio of the same temperature
of different times, and (c) the average diffusion distances of Ti, Al, V in or and B phases
under the experimental solutionization conditions.

234

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2537 11 ‘” 4'675 Variation of d-spacing
.. 4.67

m 2.532 __ ._ 4.665 +d(10-10) extruded
g i +d(10-10) 905C120m
E. . " 4'66 E- —e—d(10-10)905Cl80m
'9 2.527 ~—
'§ ’— 9- 4.655 g —e—d(10-10)960C180m
Q
5 ._ 4,65 2 -e-d(0001)cxtruded
§ 2.522 “ t-i (1 ml
2'. -- 4.645 g ‘9‘ ( ”050”“
~75 “=1 —e—d(0001) 905C180m

2517 “ ._ 4'64 +d(0001) 960C180m

-- 4.635
2.512 t l l l l l l 4 4.63 Vertical scale
1 2 3 4 5 6 7 8 9 variation = 1%
Serial sectioning positions
(a)
Mean Signa
Temperature/time _ _
(1010) (0001) (1010) (0001)
Extruded 2.5228 4.6642 5.46E-03 7.16E-03

905°C/20m 2.5345 4.6670 1.55E-03 2.68 E-03

905°C/80m 2.5253 4.6515 2.30 E-03 3.68 E03

960°C/80m 2.5273 4.6556 1.19 E-03 1.95 E-03

(b)

Figure 4.5 shows the gadual sectioning of X-ray diffraction studies of specimens.
The (a) zigzag relationship between d-spacing of d,mio),, and d(0001)ar and (b) the related

mean, and standard deviation (sigma) of the d(1010)a and d(0001)a under different
solutionization conditions were identified.

235

 

 

Volume of unit cell, A3

Volume of hexagonal alpha unit cell vs. processing
conditions

 

 

 

 

+ extruded

—I— 905C/20m
-O— 905C/80m
+ 960C/80m

 

 

 

2.5% variation
on vertical scale

 

 

r I T r

3 4 5

I

6

7

Serial sectioning positions

 

 

(a)

 

 

Lattice parameter ratio vs. positions

 

 

 

 

 

 

 

 

 

236

1.85 -- 0.887 +d((XX)l)/d(10-10) extruded
~~ 0.886
1.848 9- +d((XX)l)/d(10-10) 905C120m
9 0.885
8 m 1346 " 8 —e—d((X)01)/d(10-10)905C/80m
#1211844“ "0.884q-.i.2
2 E ' claratio= 1.595 , 0883 a 3 +d(0001)Id(1010)960080m
V 1.842 ~~ - v
E m g m -e-d(10-ll)/d(10-10) extruded
“51.84“ ‘r0.882=5
g 5 1838 __ ,_ 0.881 g 5 —a-—d(10-11)/d(10-10)905C/20m
e ' :5
1.836 -_ 1, 0,33 —e—-d(10-11)/d(10-10) 905C180m
1.834 9-W-9 0.879 —O—d(10-11)/d(10-10)9600&)m
1-832 'r . i i i ' i i 0378 Vertical scale variation = 1%
l 2 3 4 5 6 7 8 9 corresponding c/a ratio scale
Serial sectioning positions ‘5 1586 to L602
(1))

 

 

 

 

 

 

 

 

 

 

 

 

Temperature/time Mean Srgn a
d(0001)/d(1010L d(10-ll)/d(10-10) d(0001}/d(10-10) damn/(10040)
Extruded 1.849 0.881 1.44E-03 5.67E-04
905°C/20m 1.841 0.879 2.57E-04 0.93E-04
905°C/ 80m 1. 842 0.8795 3.16E-04 1.60E-04
960°C/ 80m 1. 842 0. 8793 6.6E—08 0.56E-04

 

Figure 4.6 The serial sectioning X-ray diffraction studies show, (a) the volume of
hcp unit cell vs. processing conditions, (b) the variations of d spacings d(0001)a/d(1010)a and

duping/dado)“, (c) the related mean and signa of the d(0001)o/d(1010)o and d(1oi1)o/d(1010)a
with solutionization temperatures and times. For comparison, the large variations of d

spacings for as-received and extruded wire suggested that macro-residual stress might

cause the lattice strain variations.

(C)

237

 

 

 

Yield Strength (Mpa)

Yield strength and Apparent Young'e Modulus of SQA Parts

 

 

 

 

 

Solution Temperature and Time

 

Apparent Young's
Modulus (Gpa)

 

(a)

 

 

 

Effect of Solution Temperature on Mechanical

 

 

 

 

 

1370 - em - 65
_ 60 .Ui
- 55 30$
,2 Q
— 50 .. g
t:
'5
. 45 E 'g
E z
— 40 <
— 35
— 30
905 925 940 960
Average Solution Temperature
(b)

238

 

 

Effects of solutionization time in mechanical

 

 

 

 

 

 

 

properties
60
1330 9—
p
g 1320 . ’5’ 8
O
o _ >‘ a
U)
’2 1300 ~ 8. E
.2 n.
F 1290 ~ 4
1280 -
20 40 60 80
Average solutionization time
(9)

Figure 4.7 The overall (a) relationship between solutionization conditions and
Young’s Modulus and yield stress in the radial and wire directions, the (b) effects of

solutionization temperature and (c) time on the mechanical properties using “39311031
analysis. The black line indicates the yield strength along radial directions and gay line
indicates the yield strength along wire directions. The black column indicates the
Young’s modulus strength along wire directions and gay line indicates the Young’s
modulus along radial directions.

239

[1001(RD)

ék

[0011(ND)

 

[001]
[0011(RD)

 

 

 

 

 

 

 

[0011mm
<63

[0011013)
[001]

(b) Final a phase orientations after solutionization and quenching

Figure 4.8 Pole figures and schematic crystal orientations representations of the Ti-
6Al-4V plates showing that one single or phase variant can generate three or four a phase
variants after ST or (nu—>8 transformation cycles in the (1+6 phase field, including two
perpendicular variants both have 90° mis-orientations from prior orientations, and one

variant having 30° unis-orientations from prior orientation in the RD plane [After
Moustahfid 1997].

240

Chapter V

Variant selections during 0t<—)B transformations in Ti-6Al-4V

Overview

The effects of low temperature annealing and different deformation directions on

crystallographic texture evolution after solutionization and quenching of the on phase in

Ti-6Al-4V wire specimens with <10l0> axisymetric texture were investigated using X-
ray diffraction techniques. As-received specimens were either annealed or compressed
5% along the wire or radial directions. The specimens were solutionized in a sealed
quartz tube, followed by water quenching. Texture analysis reveals residual B phase and
90° (1 phase nus-orientations from the prior as-received texture that are lowest for
annealed specimens, and highest for specimens deformed along radial directions. Variant
selection criteria, based on dislocation transformations between 0t(—>B phases, and
dislocation reactions related to a/B phase boundary transformation dislocations, are
discussed in detail. Apparently, the dislocations generated by room temperature
deformation are fairly stable during heating prior to the oc->B phase transformation,
resulting in dislocation assisted high temperature B phase variant selections. Thus,
dislocations in the on or B phase could still exist in B or or phase, respectively, after the
one—>B transformation based on the fact that the or and B phase boundaries are coherent. In
addition, the dislocation reactions and residual stress in the B phase lead to prefened

concentration of dislocations on certain (I 10)5 planes that cause preferred variant selection

241

that depends on details of initial dislocation population and recovery and recrystallization in

the B phase during solutionization.

5.1 Introduction
The study in chapter IV indicates that of 90° nus-orientations from the prior texture

are due to a common texture rotation related to intrinsic dislocations. The Burgers vector

of the <111>B||<11i0>CL dislocations that promote nucleation of or phase in B grains
should be on planes of maximum shear. Due to the anisotropic thermal expansion of the
or phase, it is expected that specific {110}5 planes of high temperature B phase parallel
and perpendicular to radial planes could have a significant amount of dislocations present
that facilitates formation of or phase orientations with the c-axis parallel to the wire axis
or radial directions that are observed experimentally. However, the detailed underlying
transformation texture mechanisms, especially the effect of either prior or thermal
expansion induced dislocations on variant selections remains unanswered. Thus, the
purpose of this chapter is to assess the role of dislocations on the odB texture
transformation mechanisms in Ti-6Al—4V wire during solutionization and quenching.

To accomplish this goal, two sets of experiments were performed. First, as-
received wire and annealed wire (525°C 6 hrs, furnace cooled) were subsequently
solutionized for either 1 or 40 minutes and quenched to evaluate the effects of prior
annealing and solutionization time on transformation texture changes. Secondly, a
different set of as-received Ti-6Al-4V wire specimens were deformed in different
directions, resulting in operation of different deformation modes, to determine if this

would cause different variant selections after solutionization and quenching.

242

Compression in the wire direction would activate mainly prism slip, since the dominant
orientation has the c-axes nearly perpendicular to the compression direction, and since
prism slip has the lowest CRSS [Jones 1981]. Compression in the radial direction would
cause a variety of slip responses; some grains would have the c-axis nearly aligned with
the compression direction, resulting in very little strain due to it being an orientation
where only <c+a> slip can be activated; some grains would be favorably oriented for
only prism slip, where the c-axis is nearly perpendicular to the compression axis, while
grains with orientations in between would be favorably oriented for a mixture of basal
and prism slip. The texture measurements were made on the exact same specimen
surfaces after each processing condition to eliminate the effects of intrinsic texture
variation. Therefore, the objective of these experiments is to investigate the effects of
dislocations on the transformation texture and how they are correlated with:

o Deformation mode

- Dislocation reactions

0 Variant selections of 0t—-)B and B—>0t transformation texture

0 Development of high temperature B phase

0 (XHB phase transformation mechanism.
5.2 Experiment results

5.2.1 Texture of as-received and annealed specimens

As in Chapters III and IV, solutionization and quenching caused two or phase fiber

texture components, one newly emerged fiber texture with the crystal basal plane (0001),,

normal aligned with the wire axis and the as-received fiber texture with the (10l0)a

normal along wire axis (Figure 5.1(b)). The length of solutionization time had a much

243

larger effect than annealing, where the new orientation had about twice the intensity with
40 minutes as at 1 minute at the solutionization temperature. This shows that even 1
rrrinute is enough time to establish a significant amount of the new orientation, but that a
longer time causes more of the B phase transformation, recovery, and/or recrystallization.
The effect of annealing was less dramatic, but the intensity of the new (0001),, orientation
in the annealed + solutionized specimens was 70-80% of the as-received specimens
(Figure 5.1(a)). Clearly, annealing must have decreased point defects and dislocation
densities to cause a lesser amount of the new component, since annealing at 525°C
should not change the microstructure and texture significantly.
5.2.2 Texture of solutionized specimens after deformation

Figure 5.2 shows that 5% compression deformation in the radial direction caused
much more of the new 90° rotated texture component than deformation in the
longitudinal direction, after solutionization for 40 minutes at 920°C and quench in a
sealed quartz tube. Compression alone along the longitudinal wire direction (L def)
strengthened the (0001),, fiber texture component along the wire axis by a factor of 5, but
weakened the other main components, but solutionization and quench increased the initial
deformation texture component back to 80-90% of their undeformed values. The new
component was 10 times stronger (2.5 x random), and the retained B peak was 50%
stronger (5.5 x random). The specimens deformed in the radial direction followed by
solutionization and quenching showed a slightly larger reduction in the initial
deformation texture components, and a 46 times stronger new component (13.4 x
random), and a 3.7 times stronger retained B peak (9.2 x random). The crystal orientation

intensities of (0001):: and (110)B fiber texture components aligned along wire axis were

244

5.3 and 1.7 times higher than the corresponding texture intensities of specimens deformed
along the wire axis and heat treated under the exact same conditions.
5.3 Discussion

The present study investigated the effects of dislocation defects on crystallographic
texture evolution. Therefore, it is of interest to investigate the texture development in
relation to previous Ti-6Al-4V alloys studies, physical metallurgy principles related to
deformation modes, dislocation theories, and phase transformations.

5.3.1 Effects of Dislocations

The difference in 90° mis-orientations between annealed and as-received specimens
followed by solutionization and quenching (Figure 5.1) indicates that defects, especially
dislocations, affect the variant selections, assuming the same initially low density of
(0001),, orientations were aligned with the wire axis prior to heat treatment. The
difference in texture development after heat treatment was most likely due to the
difference in deformation defect structures along radial and wire directions (Figure 5.2).
It is assumed that deformation in the as received wire caused microstructural changes
such as residual stress, point defects and dislocations. More importantly, different
deformation directions led to different residual stress and dislocation
orientation/arrangement that could assist or hinder the subsequent transformation texture.
A few conclusions about the effects of dislocations on the development of 90° texture
mis-orientations can be identified as follows:

(1) The dislocations were kept without annihilation during the 0t—>B phase
transformation and the subsequent grain/phase growth (described in Chapter IV). If this

were not the case, the high temperature B phase and or phase (0001)a texture along wire

245

axis would be the same, independent of deformation orientations. Also, annealing and
solutionization time would not have had an effect on transformation textures. Although
the a—>B transformation and grain growth were accomplished through movement of pre-
existing grain boundaries [Ivasishin 1999], the pre—existing dislocations inherited from
room temperature could also be kept through the transformation and at least part of the
solutionization process, or they would react with the moving phase boundary to form new
dislocations even after other grain or phase boundaries swept across the grain.

(2) The dislocations are fairly stable even at high ~ temperatures, as shown by the
effect of solutionization time in figure 5.1. The movements of dislocations are fairly
slow even at high solutionization temperatures, resulting in stable dislocation
arrangements and the subsequent dislocation assisted variant selections [Ivasishin 1999].

5.3.2 Deformation modes in B phase

It is expected that the deformation systems in B along the wire and radial directions
are the same as other bcc metals, i.e. {110}, {112}, and {123} slip planes with <lll>
slip directions [Barrett 1966]. The diffraction studies of the initial B phase in as-received
specimens, shown in Chapter III, revealed diffuse streaking that was different from to
phase reflections, in addition to sharp B reflections, suggesting <111> slip rather than
twinning in B was the dominant deformation mode [Hanada 1986, 1987]. Therefore, in
the current study the <111> dislocations in the B phase help accommodate compression
along wire and radial directions.

5.3.3 a—>B transformation and dislocation-interface interactions
The cold deformation-induced dislocations could affect the development of high

temperature B phase and the subsequent 0t transformed texture. The 0L—>B transformation

246

and grain growth were accomplished by odB boundary movement. Thus, it is expected that
the existing dislocations in primary or and B phases could affect the or/B boundary
movement, and thus affect texture development. This assumption is further supported by
the titanium (nu—9B transformation theory of Ivasishin [1999]. The lattice reconstruction
stress of the ore-9B transformation is close to a simple shear of low values that would not
cause recrystallization in the transformed high temperature B phase, especially with initially
low dislocation densities. Thus, the required transformation stresses are too low to destroy
the orientation relationship between a and B phases, in contrast to the fcc —) bcc
transformation in steel, where the reconstruction stresses are too high to maintain the
orientation relationship. In addition, the or phase is more resistant to relaxation and
recrystallization due to the lower symmetry of the hexagonal a phase [Ivasishin 1999].
Therefore, it is possible that room temperature deformation induced dislocations could be
kept or partially kept through the transformation, and that it may be easier for the or phase
dislocations to be transferred to B than from B to or phase. Also, since diffusion is much
faster in the B phase, climb/recovery reactions are likely to be faster in the B phase.
Furthermore, Priester [2001] stated that matrix dislocations could be directly
transferred from one phase to the neighboring phase without annihilation at the phase
boundary. This direct dislocation transfer across the phase boundary requires that the
traces of slip planes from each side of the grain boundary are parallel at the grain
boundary plane. Thus, if the Burgers vectors of the incoming and outgoing dislocations
are parallel (bin=bom), no residual dislocation is left at the grain boundary. Otherwise, a
residual dislocation is left with bin=bom+bmidw Also, in-situ TEM studies on stainless

steel, titanium and Ni3Al alloys performed by Lee et al. [1990] indicated that direct

247

dislocation transfer requires (1) the angle between the slip plane traces in the grain
boundary plane should be as small as possible, (2) minimum residual dislocation products
are left at the phase boundary, and (3) the resolved shear stress on the outgoing
dislocation must be high. However, the stress state near the a/B phase boundary is

unknown.

Thus if we neglect the stress requirements, the basal Z dislocations in figure 5.5 in
the primary or phase are most likely to transform to the 1/2<111> type dislocations in a
horizontal (110)B plane in the B phase upon the 0t—)B transformation. This transformation
will accomplish 1.38% strain in <lll> directions on the (110)p plane. In the specimen
deformed radially, many grains oriented favorably for basal slip will transform that are
full of basal plane dislocations, while only a few grains with only c+a dislocations will
transform, although the c+a dislocations will not survive the transformation. Only half

of the prism dislocations are likely to survive the transformation, since only one of the
prism planes will be parallel to a <112>5 plane (see Figure 1.2). The third (1010),,l

“(11mg prism plane had very low resolved shear stress and no likely dislocation activity,
and even if there were some dislocations, the burgers vector would be the less mobile
<001>5, but they would be less likely to survive since the two planes are not parallel.
The grains with predominantly c+a and prism dislocations will have a lower density of
surviving dislocations in the B phase than the grains with basal dislocations. The grains
with lower dislocation density will thus have a growth advantage and may consume the

grains having the inherited basal dislocations by a primary recrystallization process.

248

In the current study, OL—>B transformation after compression along the wire

direction (approximately normal to the page in figure 5.5 for example) could generate
l/2(l 12);;[11 I] type dislocations. Although a screw dislocations could cross slip onto
(lOl)g, (101),; planes via a—>B transformation, it is not expected that edge dislocations

could directly transfer from the prism plane of the or phase onto (101)5, (101)5 planes.

On the other hand, transformation after compression along radial directions would
generate wide variety of dislocations, such as 1/2[l ll] or l/2[ll I] type dislocations on

the (110), planes that correlate with basal (0001)., planes, and (1i2)31/2[1i i]. Thus,
different dislocations evolved after the high temperature a—>B transformation, which
could further affect the subsequent grain growth. Since the kinetics of grain growth
depends on the chemistry of high temperature phases [Ivasishin 1999] and the diffusion
of Ti, Al and V [Grewl 1989, 1990], the development of high temperature B phase
texture by grain growth could be enhanced depending upon the arrangements of
dislocations.
5.3.4 Slip activity and dislocation reaction mechanism

Dislocation reactions in the B phase could also determine the subsequent 0t
transformed texture. In ferrous alloys, the dislocation model criteria for variant selection
during the austenite to martensite transformation can be summerized as: (1) variant
selection according to slip activity on appropriate corresponding active slip system, and
(2) dislocation reaction, which corresponds to certain variants related to dislocations with

inactive Burgers vectors or a Burgers vector on a plane that has no slip activity at all

[Wittridge 2001, Butron-Guillen 1996, 1998, Sum 1999]. However, no experimental

249

work proves that dislocations are actually present at the austenite/martensite interface and
thus plays the roles proposed above [Wittridge 2001]. However, in bcc alloys, although a
1/2[111] dislocation spreads into three 1/6 [111] cores, these partial dislocations do not
bound a stable stacking fault, unlike Shockley partial dislocations in fcc alloys [Hull
1986]. Thus, the available dislocation dissociations in bcc B phase are very limited.
Therefore, the dislocation and interface reactions after radial compression

(compression along any line roughly in the plane of the page in figure 5.5) and heat
treatment could result in the formation of l/2(110)5[111], 1/2(110)B[1i i], 1/2

(112)5[11 I] or l/2(1I 3);;[111] dislocations, while compression in the axial direction
(out of page in Figure 5.5) would not provide dislocations on (110)5 planes. None of
these dislocations could assist transformation leading to the 90° nus-orientation from

prior as-received texture, since this would require that mobile dislocations are present on

(i 10) planes. The 1/2(110)B[1i i] and 1/2(110)n[1i i] dislocations could only lead to

the formation of the original as-received fiber texture. In addition, the B—m
transformation could occur with shear movements of atoms on the (112)B plane with

[11 l] direction (Figure 1.7), resulting in as-received fiber texture ((1 12)3 is
perpendicular to (110)B plane) [Gey 1997]. Thus, the system with the most slip activity
would not be associated with the newly formed or variants in the current study.

However, the perfect dislocation reactions in common bcc metals, such as the B phase

can have the following types [Zhi]:

(l) %<111>+%<111>————)a<001>

250

(2) %<111>+%<111>——>a<110>

(3) a<llO>+a<OlO>——>a<110>

(4)%<110>+—:—<110>——->a<110>
a _ a - "
(5)3<Oll>+§<lOl>—->a<110>

(6) g—<110>+%<001>—-—9a<111>

All of the dislocation reactions listed above can occur theoretically in either co-planner
or a non-coplanar manner [Zhi 1992], suggesting the same type of dislocation reactions
could occur in wide varieties of different combinations of planes (Figure 5.6). Zhi [1992]
calculated 864 combinations of planes for just the first type of dislocation reactions. For
example, the following type of dislocation Burgers reaction could occur in any of the

conelated planes listed beneath the Burgers reaction (Figure 5.6):

%[lll]+%[lll]——)a[001]

(101) (101) (100)
(011) (011) (010)
(110) (110) (110)
(211) (210) (110)

It is possible that multiple dislocation reactions under residual stress or 0t(—>B

transformation stress could lead to appreciable dislocation concentration on (110»; planes

that are necessary to stimulate a dislocation generated 90° this-orientation from prior as-
received texture. The necessary dislocations have to be 1/2(ll 3);;[111], 1/2(I 10)p[l 1 1],

or 1/2(l 10);; [111], which were different from the dislocations expected from the prior

251

deformation; 1/2(1i2)p[1ii], l/2(110)B[ili] or 1/2(110)p[1ii]. However, it is
possible that the 1/2(110)B[111], 1/2(110)B[111], and 1/2(112)B[1ii] dislocations

could cross slip or move onto 1/2(11 i )51111] and l/2(11i)p[111] planes via dislocation
reactions under residual stress or ore—>B transformation stress. As mentioned in chapter
III, the a<—>B transformation could result in volume contraction by as much as
0.55~2.6%. The transformation stress could be roughly estimated by the equation

E AV
0 = 3|!
3(1— 2v) V o

 

[Barsoum 1997],

where E denotes Young’s modulus (IOOGPa for Ti-6Al-4V), v denotes Poisson’s ratio
(0.264136), AVNo volumetric change in materials.

Thus, the residual stress in titanium alloys is on the order of 0.03~3.1 GPa, which
potentially could push dislocations onto certain planes. In addition, the thermal

expansion differences between on and B phases could generate strain on the level of ~0.15

% upon heating and cooling process. Furthermore, l/2(110)p[i 1i] or 1/2(110)p[1i i]
dislocations resulting from basal slip could easily survive through the solutionization;
while prism slip less is likely to survive the transformation. Therefore, the dislocations in
B phase after wire axis deformation and followed by heat treatment could assist

transformation to the same on phase orientation as the original fiber orientations.
Even though the (l 10);; plane is properly oriented to facilitate the observed 90°

transformation based upon 1/2[111] or 1/2[111] dislocations, there appears to be a

relatively low probability of developing such dislocations on this plane via dislocation

252

reactions. Instead, an epitaxial retransforrnation appears to be favored using the existing

dislocations. However, the fact that the 90° this-orientation occurred implies that (1)

1/2[111] or 1/2[111] dislocations somehow got onto the (110)3, and that even fewer
dislocations existed on the (110)13 plane that favored the epitaxial reverse transformation,
or (2) variant selection was associated with the planes having the least slip and
dislocation density that is opposite of the currently accepted theory. However, two other
possibilities to consider are (1) a recovery/recrystallization process that removes
dislocations from the grains with high dislocation density, which is required to reduce the

epitaxial retransforrnation possibility, and (2) dislocations may have been generated on

(I 10);; planes during the heating and cooling process due to thermal stresses.

Primary recrystallization occurs by motion of grain boundaries through grains
having a high dislocation density, from grains with a preferred nucleation orientation or
grains with a preferred boundary mobility. If an existing minority orientation that was
not oriented for easy slip existed, then there is little expectation that a highly preferred
recrystallized orientation would result, per the observation. In contrast, an existing
orientation that had a low dislocation density could have a growth advantage into grains
with a high transformed dislocation density, e.g. grains with primary prism slip would
have fewer dislocations than grains with primary basal slip dislocations. Such grains
would have a misorientation of about 45° about (100)B axis, and growth of these B
orientations would still result in (110)5 type planes rotated about the wire axis. Given

this possibility, this still requires that remaining dislocations on {112} planes would have

to cross slip onto (i10)B planes, but the burgers vector would still be out of the (I 10);;

253

plane. Thus recrystallization seems unlikely to provide a basis for dislocation-assisted

transformation using mobile dislocations on (110)g planes. However, recrystallization
does cause disappearance of dislocations that would favor epitaxial retransforrnation.

Next we consider the effect of thermal expansion and contraction stresses during

heating and cooling that could generate dislocations on (110),; planes. Since thermal

contraction is greatest along the c axis, this provides a basis for generation of shears on

(110)5 planes shown in figure 5.7. Shears on the (i10)B planes can be envisioned in B
grains between or grains aligned along the wire axis with different c-axis orientations, but
not within B grains between or grains along the radial direction. The resulting stress and
deformation could be substantial due to small amount of prior B phase upon heating
(Figure 5.7(a)). Thus the generation of dislocations could be kept without annihilation
during the subsequent grain growth. In addition, similar stress and dislocations could be
generated upon cooling in the B phase (Figure 5.7(b)). Therefore, the generation of
dislocations that cause preferred variant selection may occur upon heating and cooling
rather then during the B phase solutionization part of the process. This would imply that
some of the B transforms epitaxially in some regions, but transforms by 90° in others.
However, further investigations including analytical instruments and finite element
analysis are needed to characterize the distribution and magnitude of thermal stresses
upon cooling.

Therefore, as pointed out in Chapter IV, the lattice similarities of a specific plane of
prior phase with a certain dislocation arrangement with a correlating plane of the

subsequent phase determines the latter transformation variants. Thus, local

254

concentrations of reacted or freshly generated dislocations on the (I 10);; plane (parallel to
the radial plane) determined the new transformation variants since the B—>0t
transformation starts from the B/0L gain boundaries and high dislocation density sites inside
the B gains.
5.4 Conclusion

The present study systematically investigated the effects of deformation direction
and dislocations on crystallogaphic texture evolution of or phase in Ti-6A1-4V using X-
ray diffraction techniques. Deformation along different orientations resulted in different
dislocation types, leading to different texture development after the 0t—>B—>0t
transformation cycle. Texture analysis revealed 90° mis-orientations from the prior
texture that were consistently higher for specimens deformed along radial directions, than
specimens deformed along wire directions, respectively, followed by the same
solutionization and quenching. The required low stress for lattice reconstruction of a—)B
transformation suggests that the CL and B phases are reluctant to recrystallize, leading to
high dislocation stabilities. The variant selection mechanisms based on dislocation
reactions and thermal residual stresses during cooling were discussed to show that
generation of dislocations that facilitate the 90° misorientation observed are not easily

envisioned.

255

 

Effects of annealing on the alpha phase texture development

 

 

 

 

I10-10|| wire —
7 - 50001 || radial —
6 _ C1000] I] wire

 

 

 

 

 

 

Texture intensity (x random)
A

 

 

 

Wire SlQ Ann+SlQ S40Q Ann+S40Q
Processing history (Solution if minutes, Quenched)

 

100011 ‘fl— " D

Wire axis
[IO-10] [0001] 4———>
(C)

Figure 5.1(a) reveals the overall relationships between major texture intensities and
the experiment conditions, (b) the experimental (1010)“, (0002)“, and (110)g pole figures
with wire axis out of the page for specimen S40Q after solutionization without prior
annealing, and (c) the schematic acrystal orientations. The levels of 90° mis—orientations
from prior texture were consistently lower in the prior annealed and followed by
solutionization and quenching specimens than that of the as received one without
annealing, indicating the effects of defects on texture development.

 

256

 

Effects of Longitudinal or Radial 5% compression
defomration and solutionization on texture development

 

 

 

 

 

l4
I10-10||wire

Alz‘ I0(X)l||radial *“ Ivleasurementsareallonthesamesurface
glow, IOWIIIWire __ perpendiculartothewireaxis
.6 .

=1 I110||w1re

a:

5.8- .e—fii
5
:6-
E

2-

0‘

 

 

 

 

 

Lwire Rwire Ldef Rdef LSQ RSQ
As-ReeeivedwireDefornnd,920°C40ninuteSolufion-l-Quench

 

 

 

Figure 5.2 reveals the overall relationships between major texture intensities and the
experiment processing conditions after solutionization for 40 minutes at 920°C.

Different deformation directions, followed by the same heat treatment resulted in
different texture development.

257

 

  

[10-10] *-¢e\ (D

Wire axis
[0001] ‘——'>

(b)

Figure 5.3 showing (a) the (1010)“, (0002)“, and (110)5 pole figures of as—received
wire with wire axis out of the page, and the (b) schematic a crystal orientations. The a
phase consisted of one fiber texture with (0002)OI plane normal along radial directions,
The B phase consisted of a fiber texture with the axisymmetric (110)g crystal orientations
about the wire axis, and a few of rotated or tilted (110)g fiber texture aligned
approximately 30° from the wire axis.

258

 

259

(equal—am proJ.) am.- 8.95

 

 
  

[0001] [IO-10]

Wire axis

[10-10] [0001] <——>

(C)

Figure 5.4 showing the (1010)“, (0002),,, and (1 10);; pole figures with wire axis out
of the page of heat-treated specimens (a) L heat and (b) T heat after solutionization for 40
minutes at 920°C, followed by quenching in sealed quartz tube, and (c) the schematic a
crystal orientations for both cases. Note, the crystal orientation of (0001),I and (110)3
plane normal along wire axis of T heat (deformed along transverse direction and heat
treated) is 13.42 and 9.21 X random, respectively, which is 5.3 and 1.7 times higher than
the corresponding textures of specimens L heat deformed along longitude direction and
heat treated under the exact same conditions.

260

[110113 I] [0001]“
1010 _

    
  
   
   

\(1121ell(1100)a
b3: [1 120]a ‘ bc+a=[ll 2 3]

b,= [21 10],,

(110)11" (0001),,
[111]p||[2110],,

[lillpllllléma

[owls [0011.311 112101.

[10018

Figure 5.5 showing dislocations mutations based on the geometric and Burgers
relationship between or and B phases. Compression along the wire axis is roughly out of
the page. Compression in radial directions imposes compressive stresses in directions
roughly in the plane of the page. Neither would activate m_uch prism slip on (1010),,
planes. The prism a dislocations transform into [111](112)p dislocations, but they
cannot mugte onto (110)5 planes after 0t—)B transformation. However, either basal or
prismatic a screw dislocations could cgossqslip onto the two inclined (101)p, (101)B
planes after (Jr—)B transformation. The c +a dislocations could be directly transferred
onto B phase (101),}, (101)3 planes, but would have a large irrational Burgers vector.

261

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 . C
If] \\\ l .
\ I
° ,_- _._- f l /
B .
(a)
B a[110] f :— C
t. /l\ e
’9 109‘ 2: \ 8
__ /§? 5 ro¢\ '= g?
(112) / ea :‘3 J \"/\ T ‘é
/ “g z a 3‘9 \"b (010)
/ e e <
e é_$lill_ ___"’2 [11.11. \ g (001) J “[0011
\39 ¢ .J- a)! e°// (100)
a \ a? 3 :1 / (112) g?
\ ’0‘
\§ - \ J |g /$\ ’0’ H
(112) \ As? 9
g ‘* / alfiol
A h 'D
(b)

Figure 5.6 shows the spatial construction of <100>, <110>, and <111> directions and
{100}, {110}, and {112} planes in bcc metal, and (b) planer representation for projections
of the ABCDX pyramids [After Zhi 1992]. In the ABCDX pyramids, sides of hexahedron
represents {100} planes, sides of the dashed line pyramids represents {110} planes, and

sides of the solids line pyramids represents {112} planes.

262

A Wire axis Aa

 

0t : 4—
Grain with: ‘-
c-axi
4—-’> :
1 ‘-

 

 

d

B gain(l 10 lane normal || wire axis

 

 

 

 

 

 

 

 

 

 

 

otGrainwith gaff; " I a
c-axis 1 d _. :Grain with
O reso ve ' c-axis
stress on .
(110) plant?> '<—>
I
Aa
(a)
A Wire axis
A01
I I
I
a I <-
Grain with :
C-aXifi | ‘—
1
1 ‘-
. ‘r
B gain 1
- “ Aa
(110)01ane normal || wire axisl I
I
aGrainwith Bgain +1 0t
c-axis with no + :Grain with
resolved I c-axisI
0 stress on 1
(110) plane“H E

 

 

 

 

 

 

(b)
Figure 5.7 Schematic descriptions of effect of differential thermal expansion and

contraction of primary or gains (shaded boxes) on state of shear in B gains during (a)
heating and (b) cooling process.

263

References
Akhtar, A., Metallurgical Transactions, 6A(1975) 1105.
Anderson, 8., Acta Chem. Scand, 11(1957) 1641.
Babel, H.W., Frederick, S.F., JOM, October 20 (1968) 32.
Barrett, C.S., Massalski, T.B., Structure ofMetals, 3rd Ed, McGraw, NY. 1966.
Barsoum, M.W., Fundamentals of ceramics, New York, McGraw Hill, 1997. p. 502.

Bieler, T.R., Serrriatin, S.L., J.J. Jonas symposium, COM 2000, proceeding, CIM, Montreal,
2000, pp. 253-64.

Boswell, E., Strangwood, M., Pate, P.S., Titanium ’95 Science and Technology, 1995, p.
2446.

Boyer, R.R., Titanium '95 Science and Technology, 1995, p. 41.

Boyer-A, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 483-491.

Boyer-B, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 1053-1054.

Boyer-C, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 1113.

Boyer-D, R., Welsch, (3., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 611-612.

Boyer-E, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 606-607.

Boyer-F, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 263.

Boyer-G, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 599.

Boyer-H, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, p. 62-63.

Boyer-I, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM International, Materials Park, OH, 1994, pp. 56.

264

Boyer-J, R., Welsch, G., Coiling, E.W., Materials Properties Handbook: Titanium Alloys,
ASM Intemationai, Materials Park, OH, 1994, pp. 516.

Burgers, W.G., Physica, 1(1934)56l.

Butron-Guillen, M.P., Da Costa, Viana, C.S., Jonas, 1.1., in 2D. Liang, L. Zuo. Y.Y. Chu
(Eds.), Proceeding of the Eleventh International Conference on Texture of Materials,
ICOTOM-ll, Xian, China, Sept. l6-20, 1996, International Academic Publishers, 137
Chaonei Dajie, Beijing 100010, China.

Butron-Guillen, M.P., Da Costa, Viana, C.S., Jonas, 1.1., Materials Science Forum, 273-275
(1998) 151.

Cahn, J .W., Acta Metall., 5(1953) 169.

Cahn, R.W., Haasen, P., Physical Metallurgy, Fourth Edition, Elsevier Science B.V., North-
Holiand, 1996, pp. 1531.

Chesnutt, J .C., Rhodes, C.G., Williams, J.C., Fractogaphy—Microscpic Cracking Processes,
ASTM STP 600, p. 99, 1976.

Churchman, A.T., Proc. Roy. Soc., A226(1954) 216.

Cullity, B.D., Elements of X-ray Diffraction, Second Edition, Addison-wesley publishing
company, Inc., October, 1978, pp. 295-300.

David, D., Garcia, E.A., Beranger, 6., Titanium ’80, Science and Technology, H. Kimura
and O. Izunri (Eds.), TMS/AIME, 1980, p. 537.

Davis, R., Flower, H.M., West, D.R.F., Journal of Materials Science, 14(1979) 712.
Davis, R., Flower, H.M., West, D.R.F., Acta Metallurgica, 27(1979) 1041.

Divinski, S.V., Dnieprenko, V.N., Ivasishin, O.M., Materials Science and Engineering,
A243 (1998) 201.

Fan, 2., Scripta Metallurgica et Materialia, 31(1 le994) 1519.
Fang, T.Y., Wang, W.H., Materials Chemistry and Physics, 56(1998) 35.

Fujishiro, S., Nadiv, 8., “Effect of Transformation on the Texture of Titanium”,
Titanium ’80, 1980, p. 899.

Furuhara, T., Maid, T., Materials Science & Engineering A, 312(2001) 145.

265

Gehlen, P.C., The Science, Technology and Applications of Titanium, p. 349, Pergamon
Press, 1970.

Gey, N., Humbert, M., Acta Materialia, 50 (2002) 277.
Gey, N., Humbert, M., Moustahfid, H., Scripta mater., 42(2000)525.

Gey, N., Humbert, M., Philippe, M.J., Combres, Y., Materials Science and Engineering,
A219 (1996) 80.

Gey, N., Humbert, M., Philippe, M.J., Combres, Y., Materials Science and Engineering,
A230 (1997) 68.

Gil, F.G., Manero, J.M., Plannell, J.A., Titanium ’95 Science and Technology, 1995, p. 2454.
Grew], G., Ankem, S., Acta Met. Mater.,38(9)(1990) 1607 .

Grew], G., Ankem, S., Metallurgical Transactions A, 20A (1989) 39.

Grewl, G., Ankem, S., Metallurgical Transactions A, 21A (1990) 1645.

Hanada, S., Izumi, O., Metallurgical Transactions, 18A(l987) 265.

Hanada, S., Izumi, O., Metallurgical Transactions, 17A(1986) 1409.

Hu, H., Cline, R.S., Transactions of The Metallurgical Society of AIME, 242 (1968) 1013.

Hull, D., Bacon, D.J., Introduction to Dislocations, Third Edition, Pergamon press, 1986, p.
121-126.

Humbert, M., Gardiola, B., Esling, C., Flemming, G., Hensger, K.E., Acta Materialia, 50
(2002)]741.

Inouye, K., Iijima, Y., Hirano, K., Titanium ’80, Science and Technology, H. Kimura and O.
Izumi (Eds.), TMS/AIME, 1980, pp. 569-76.

Ivasishin, O.M., Teliovich, R.V., Materials Science and Engineering, A263 (1999) 142.
Jones, I.P., Hutchinson, W.B., Acta Metallurgica, 29(1981)951.
Kahveci, A.I., Welsch, G.E., Scripta Metallurgica, 20(9)(1986)1287.

Ke, H., Ning, C., Jia, D., Zhou, Y., Song, G., Nonferrous Metals (China), vol. 53, no. 4, pp.
8-14, Nov. 2001.

Keeler, J .H., Geisler, A.H., JOM, February, 8 (1956) 80.

266

Koul, M.k., Breedis, J .F., Acta Metallurgica, 18(1970) 579.
Lasalmonie, A., Loubradou, M., Journal of Materials Science, 14(1979) 2589.

Lee, T.C., Robertson, I.M., Bimbaum, H.K., Philosophical Magazine A, 62(1) (1990)
131.

Libanati, C.M., Dyment, F., Acta Met., ll(1963)1263.

Luo, C.P., Weatherly, G.C., Acta Metall., 35(1987) 1963.

Maykuth, D. J ., Monroe, R.E., Favor, RJ., Moon, D.P., Titanium base alloys: 6Al-4V;
processes and properties handbook, Columbus, Ohio : Battelle Memorial Institute, Defense
Metals Information Center, 1971, Chapter 1: Physical metallurgy, Page, I-II.

Moustahfid, H., Gey, N., Humbert, M., Philippe, MJ., Metallurgical and Materials
Transactions A, 28A (1997) 51.

Murayama, Y., Suzuki, Y., Shimura, M., Titanium ’80, 1980, p. 965.
Murdock, J.F., Lundy, T.S., Stansbury, E.E., Acta Met., 12(1964)1033.
Murdock, J .F., McHarque, CJ., Acta Met., l6(l968)493.

Naka, S., Phil Mag A, 57(5)(1994)717.

Orr, R.D., de Bruijn, J.D., Davies, J.E., Cells and Materials (USA), vol. 2, no. 3, pp. 241-
251, 1992.

Ouchi, C., Fukai, H., Hasegawa, K., Materials Science and Engineering, A263 (1999) 132.

Ouchi, K., Iijima, Y., Hirano, K., Titanium ’80, Science and Technology, H. Kimma and O.
Izumi (Eds.), TMS/AIME, 1980. Pp. 559-68.

Paton, N .E., Backofen, W.A., Trans. T MS—AIME, 245(1969) 1369.

Paton, N.E., Backoffn, W.A., Metallurgical Transactions, 1A(l970) 2839.
Peters, M., Luetjering, G., Titanium ’80, 1980, p 925.

Priester, L., Materials Science and Engineering, A309-310 (2001) 430.

Qazi, J.I., Senkov, O.N., Rahim, J., Genc, A., Froes, F.M., Metallurgical and Materials
Transactions A, 32A (2001) 2453.

267

Rabinkin, A., Talianker, M., Botstein, O., Metallurgical, 29(1981) 691.

Regrier, P., Dupouy, J .M., Physica status solidi, 39(1970)79.

Rhodes, C.G., Paton, N.E., Metallurgical Transactions, 8A(1977) 1749.

Rhodes, C.G., Williams, J .C., Metallurgical Transactions, 6A(1975) 2103.
Rosenbaum, H.S., Deformation twinning, Gorden and Breach, New York, 1964, p. 43.
Rosi, F.D., Perkins, F.C., Seigle, L.L., Trans. TMS-AIME, 197(1953) 257.

Rudman, P.S., Trans. AIME, 233(1965) 864.

Silcock, J .M., Davis, M.H., Hardy, H.K., A symposium, Monograph and Report Series No.
18, p. 93, Inst. Of Metals, London (1956).

Spedding, F.H., Hanak, J.J., Daane, A.H., J. Less common Metals, 3(1961)1 10.
Sum. M., Jonas, 1]., Textures Microstr., 31(1999) 187.

Tanabe, A., Nishimura, T., Fukuda, M., Titanium ’80, 1980, p. 938.

Toth, L.S., Gilorrnini, P., Jonas, J .J ., Acta Metall., 36(12)(1988) 3077.

Welsch, G., Lutjering, G., Gazioglu, K., Bunk, W., Metallurgical Transactions, 8A(1977)
169.

Williams, J.C., Blackburn, M.J., Metallurgical Transactions, 60 (1967) 373.
Williams, J.C., Hickman, 13., Metallurgical Transactions, 1(1970) 2648.
Williams, J.C., Sommer, A.W., Tung, P.P., Metallurgical Transactions, 3(1972) 2979.

Wittridge, NJ ., Jonas, J.J., Root, J.H., Metallurgical and Materials Transactions A, 32A
(2001) 889.

Zhi, Z.F., Phys Status SolidiA 130(1): Kl9-K24 MAR 16 1992.

268

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

' 1111111ilgiiljiiliwl