geisha.“ . . . armies" . V a5?
IE. :I« . xii:

. , vyza

(3...:

s 17“.}.
~ .1.

:11 ‘ ..: Int .

.71.... £155

‘ L I... .

13.63.,
Livia, .

e.
... I.
..x.,..... .

.
z. t

.. L. .:

Lit:
. .

. . :.

 

“3:9

 

LIBRARY
Michigan ’ State
University

 

 

 

This is to certify that the
dissertation entitled

The Physical and Mechanical Metallurgy of Advanced O+BCC

PhD.

Titanium Alloys

presented by

Christopher John Cowen

has been accepted towards fulfillment
of the requirements for the

 

 

 

 

degree in Materials Science and
EngineerinL
Major Pro ssor's fléignature
i7/\l \il 0(0
Date

MSU is an Affirmative ActiorVEquaI Opportunity Institution

_.—.—.-o-.-.-.--—-.-

—-—.—-—.---.—.-.--—4-._._.-.--_ .—

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 plelRC/DaIeDue‘indd-pfi

 

THE PHYSICAL AND MECHANICAL METALLURGY OF ADVANCED
O-l-BCC TITANIUM ALLOYS

By

Christopher John Cowen

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

2006

ABSTRACT

THE PHYSICAL AND MECHANICAL METALLURGY OF ADVANCED
O+BCC TITANIUM ALLOYS

By
Christopher John Cowen

This thesis comprises a systematic study of the microstructural evolution, phase
transformation behavior, elevated-temperature creep behavior, room-temperature and
elevated-temperature tensile behavior, and room-temperature fatigue behavior of
advanced titanium-aluminum-niobium (Ti-Al-Nb) alloys with and without boron
additions. The specific alloys studied were: Ti-5A1-45Nb (at%), Ti-15Al-33Nb (at%),
Ti-lSAl-33Nb-O.SB (at%), Ti-15Al-33Nb-5B (at%), Ti-21Al-29Nb (at%), Ti-22Al-26Nb
(at%), and Ti-22Al-26Nb-SB (at%). The only alloy composition that had been previously
studied before this thesis work began was Ti-22Al-26Nb (at%). Publication in peer-
reviewed material science journals of the work performed in this thesis has made data
available in the scientific literature that was previously non-existent. The knowledge gap
for Ti-Al-Nb phase equilibria over the compositional range of Ti-23Al-27Nb (at%) to Ti-
12Al-38Nb (at%) that existed before this work began was successfully filled. The
addition of 5 at% boron to the Ti-15A1-33Nb alloy produced 5-9 volume percent boride
phase needles within the microstructure. The chemical composition of the boride phase
measured by electron microprobe was determined to be approximately BzTiNb. The

lattice parameters of the boride phase were simulated through density functional theory

calculations by collaborators at the Air Force Research Laboratory based on the measured
composition. Using the simulated lattice parameters, electron backscatter diffraction
kikuchi patterns and selected area electron diffraction patterns obtained from the boride
phase were successfully indexed according to the space group and site occupancies of the
827 orthorhombic crystal structure. This suggests that half the Ti (c) Wyckoff positions
are occupied by Ti atoms and the other half are occupied by Nb atoms in the boride phase
lattice. Creep deformation behavior is the main focus of this thesis and in particular
understanding the dominant creep deformation mechanisms as a function of stress,
temperature, and strain rate. Microstructure-creep relationships for Ti-Al-Nb—xB alloys
were developed through the understanding gained. A rule-of-mixtures empirical model
based on constituent phase volume fractions and strain rates was developed to predict the
minimum creep rates of two-phase O+BCC microstructures. The most innovative results
of this thesis were produced through the development of an in-situ creep testing
methodology. The creep defamation evolution was chronicled in-situ during high
temperature creep experiments, while creep displacement versus time data was
simultaneously obtained. The in-situ experiments revealed that prior-BCC grain
boundaries were the locus of damage accumulation during creep deformation. A
methodology that allows in-situ observation of surface creep deformation as a function of

creep displacement has yet to be presented in the literature.

Dedicated To

The memory of my fraternity brother Benjamin Klein and my uncle David Root

iv

ACKNOWLEDGMENTS

I first and foremost thank my thesis adviser Dr. Carl Boehlert. He gave me the
opportunity to perform this work, and he turned me into a metallurgist. I was his
graduate student at Alfred University, and when he took a new job at Michigan State half
way through my graduate career, he gave me the opportunity to come to Michigan State
with him and finish my Ph.D. research. He was an excellent adviser and friend and I
thank him for the many opportunities that were provided to me during my time as his
student, and I look forward to future collaboration.

I thank my thesis committee Dr. Tom Bieler, Dr. Marty Crimp, Dr. Jim Lucas,
and Dr. Patrick Kwon. I thank Dr. Tom Bieler in particular for his always constructive
comments and for working with me long—distance fiom Germany while he was on
sabbatical leave my last year of graduate school.

I thank my mother Darlene, my father John, and my sister Marissa for their
constant love and unwavering support. It goes unspoken that without them, I would not
be where I am now. My fiancee Nicole deserves the most thanks because without her
love and support I simply would not have been able to complete this work, or get by in
day-to-day life. I thank my friends Aaron and Crystal for giving me another good reason
to come home and visit. My last thanks go to my graduate student colleagues Ke Bin
Low and Jeff Quast for being great friends and coworkers, and for making my time at

Michigan State much more enjoyable that it would have been without them.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... ix
LIST OF FIGURES ............................................................................... xii
CHAPTER 1
INTRODUCTION .................................................................................. 1
A. RATIONALE ............................................................................ 2
B. CONVENTIONAL TITANIUM ALLOYS .......................................... 3
C. THE ORIGIN AND EVOLUTION OF ORTHORHOMBIC
TITANIUM-ALUMINIDES ............................................................... 4
D. THE WORK PERFORMED AND SPECIFIC AIMS ............................. 11
CHAPTER 2
BACKGROUND AND LITERATURE REVIEW .......................................... 16
A. Orthorhombic T1tamum-Alum1mdesl7
1. Constituent Phases and Crystal Structures .......................................... 17
2. Microstructural Evolution .............................................................. 21
2.1 Phase Field Ranges ........................................................... 21
2.2 Phase Transformation Mechanisms ........................................ 25
3. Deformation Behavior of the Constituent Phases .................................. 27
4. Creep Deformation Behavior .......................................................... 32
4.1 Nabarro-Herring Creep ....................................................... 32
4.2 Coble Creep ................................................................... 33
4.3 Harper-Dom Creep .......................................................... 34
4.4 Dislocation Climb ............................................................ 36
4.5 Grain Boundary Sliding ...................................................... 37
4.6 Creep Stress Exponents and Apparent Activation Energies ............ 38
4.7 Creep Behavior of Ti-Al-Nb Alloys ....................................... 42
4.7.1 Creep Behavior of dig-based Ti-Al-Nb Alloys ................ 42

4.7.2 Creep Behavior of the O-phase in Ti-Al-Nb Alloys. . . . ......44
4.7.3 Creep Behavior of the BCC phase in Ti-Al-Nb Alloys. . ....47

4.7.4 Creep Behavior of O+BCC Ti-Al-Nb Alloys ................. 48
5. Tensile Deformation Behavior ........................................................ 56
6. Fatigue Deformation Behavior ........................................................ 60
B. Boron Modification of Titanium Alloys ............................................ 62
CHAPTER 3
EXPERIMMENTAL PROCEDURES ........................................................ 63
A. Alloy Processing ........................................................................ 63
1. Hot-Forged and Hot-Rolled Materials ................................................ 63

vi

1.1 Ti-15A1-33Nb and Ti-21Al-29Nb Materials .............................. 63

1.2 Ti-15Al-33Nb—O.5B Material ................................................ 65
1.3 Ti-5Al-45Nb Material ........................................................ 65
2. Hot Isostatically Pressed Material .................................................... 65
2.1 Ti-15A1-33Nb-5B Material .................................................. 65
2.2 Ti-22Al-26Nb and Ti-22Al-26Nb-SB Materials ......................... 67
B. Microstructural Evaluation ............................................................ 68
1. Chemical Composition Analysis ...................................................... 68
2. Heat Treatments ........................................................................ 68
3. Sectioning, Cleaning, and Metallography ........................................... 71
4. Microscopy .............................................................................. 71
5. Phase Volume Fractions ............................................................... 73
6. Grain Size Analysis ..................................................................... 74
7. Differential Thermal Analysis ......................................................... 74
8. X-ray Diffraction Analysis ............................................................ 75
C. Creep Testing ........................................................................... 75
D. Tensile Testing .......................................................................... 77
E. Fatigue Testing .......................................................................... 80
F. In-situ Tensile and Creep Testing ..................................................... 82
CHAPTER 4
RESULTS ........................................................................................... 89
A. Microstructure Results ................................................................. 89
1. Bulk Chemical Compositions ......................................................... 89
2. Differential Thermal Analysis ......................................................... 9O
3. As-Processed/Solution—Treated and Water Quenched/Solution-Treated and
Aged Microstructures ..................................................................... 95
4. Phase Volume Fractions .............................................................. 114
5. Grain Size ............................................................................... 116
6. BCC Phase Ordering .................................................................. 118
7. Equilibrium Heat-Treated Microstructures ......................................... 123
8. The Effect of Boron Modification on Microstructure ............................. 129
B. Creep Behavior Results ............................................................... 141
1. Minimum Creep Rates, Creep Stress Exponents, and Apparent Activation
Energies .................................................................................... 141
2. Proposed Deformation Mechanisms ................................................ 157
3. Creep Deformation Behavior ......................................................... 159
4. The Effect of Boron on Creep Behavior ............................................ 180
4.1 Minimum Creep Rates, Creep Stress Exponents, and Apparent
Activation Energies for the Boron-Modified Alloys ........................ 180
4.2 Proposed Deformation Mechanisms for the Boron-Modified
Alloys .............................................................................. 189
4.3 Creep Deformation Behavior of the Boron-Modified Alloys ......... 189
C. Tensile Behavior Results ............................................................. 198
l. Room-Temperature Tensile Behavior ............................................... 198
1.1 Stress versus Strain Behavior at RT ....................................... 198

vii

1.2 Room—Temperature Tensile Deformation Behavior .................... 203
1.3 The Effect of Boron Modification on Room-Temperature Tensile

Behavior ........................................................................... 210
2. Elevated-Temperature Tensile Behavior ........................................... 214
2.1 Stress versus Strain Behavior at 650°C .................................. 214
2.2 650°C Tensile Deformation Behavior .................................... 215
2.3 The Effect of Boron Modification on 650°C Tensile Behavior ...... 221
D. Fatigue Results ........................................................................ 228
1. S-N Behavior ........................................................................... 228
2. Fatigue Deformation Behavior ...................................................... 233
CHAPTER 5
DISCUSSION .................................................................................... .243
A. Microstructure Discussion ........................................................... 243
1. Phase Equilibria and Phase Field Ranges .......................................... 243
1.1 Ordering of the BCC Phase ................................................ 249
1.2 The BCC Phase Field ....................................................... 250
1.3 The BCC+a2 Phase Field ................................................... 255
1.4 The O+BCC+a2 Phase Field .............................................. 257
1.5 The O+BCC Phase Field ................................................... 260
2. Phase Transformation Mechanisms ................................................. 261
3. The Effect of Boron Modification on the Microstructure of Ti-Al-Nb
Alloys ....................................................................................... 264
B. Creep Discussion ...................................................................... 267
l. Microstructure-Creep Relationships ................................................ 268
2. Prior-BCC Grain Boundary Cracking and Sliding/Upheaval .................... 273
3. Creep Deformation Mechanisms .................................................... 282
3.1 The Apparent Activation Energy for Creep .............................. 282
3.2 Creep Deformation Mechanisms Based on Pure Metal Theory ...... 284
3.3 A Composite Creep Model Based on Constituent Phase Data ........ 291
4. The Effect of Boron Modification on Creep Deformation ....................... 303
C. Tensile Behavior Discussion ......................................................... 306
l. Monolithic Alloy Room-Temperature Tensile Discussion ....................... 3 06
2. Boron-Modified Alloy Room-Temperature Tensile Discussion ................ 310
3. Monolithic Alloy 650°C Tensile Discussion ....................................... 312
4. Boron-Modified Alloy 650°C Tensile Discussion ................................. 315
D. Fatigue Discussion .................................................................... 316
CHAPTER 6
SUMMARY AND CONCLUSIONS ........................................................ .320
A. Summary .............................................................................. 320
B. Conclusions ........................................................................... 328
C. Recommendations for Future Work ................................................ 332
BIBLIOGRAPHY .............................................................................. .333

viii

LIST OF TABLES

Table I. Dislocation Climb Controlled Creep Parameters ..................................... 43
Table II. Self-Diffusion and Inter-Diffusion Data for Ti3Al .................................. 46
Table 111. Various Orthorhombic Alloy Creep Parameters .................................... 49
Table IV. Creep Parameters for et-Ti, org-based, and O-based Alloys ....................... 55
Table V. RT Tensile Properties for Various Ti-Al-Nb Alloys ............................... 57

Table VI. RT Tensile Properties and Microstructural Features for Various Ti-Al-Nb
Alloys ................................................................................................. 58

Table VII. Bulk Chemical Compositions of the Ti-Al-Nb Alloys ........................... 90

Table VIII. Measured Phase Volume Fractions for Ti-15A1-33Nb Microstructures. . ...l 14

Table IX. Measured Phase Volume Fractions for Ti-21Al-29Nb Microstructures ...... 114

Table X. Measured Equiaxed Grain Size for Ti-15Al-33Nb and Ti-21Al-29Nb
Microstructures .................................................................................... 116

Table XI. Measured Phase Volume Fractions and Phase Chernistries for Selected Ti-Al-
Nb Microstructures ................................................................................ 125

Table XII. Measured Bulk Chemical Compositions of the Ti-Al-Nb—xB Alloys. . . . . ....130

Table XIII. Measured Phase Volume Fractions and Grain Sizes of the Ti-Al-Nb—xB
Alloys ................................................................................................ 130

ix

Table XIV. Measured Minimum Creep Rates and Grain Size for the Ti-5Al-45Nb
Alloy ................................................................................................ 145

Table XV. Measured Minimum Creep Rates and Grain Size for Heat Treated Ti—21Al-
29Nb Microstructures ............................................................................. 145

Table XVI. Measured Minimum Creep Rates and Grain Sizes for Heat Treated Ti-15Al-
33Nb Microstructures ............................................................................. 146

Table XVII. Measured Creep Exponents and Apparent Activation Energies for the Ti-
5Al-45NB, Ti-15Al-33Nb, and Ti-21Al-29Nb Heat Treated Microstructures ............ 147

Table XVIII. Measured Minimum Creep Rates and Grain Sizes for the Ti-AI-Nb-xB
Microstructures ........................................................................... ' ......... 182

Table XIX. Measured Creep Exponents and Apparent Activation Energies for the
HT :1005 Heat-Treated Ti-15Al-33Nb-xB Materials .......................................... 182

Table XX. Room-Temperature Tensile Properties of the Ti-SAl-45Nb, Ti-15Al-33Nb Ti-
21Al-29Nb, and Ti-22Al-26Nb Microstructures .............................................. 202

Table XXI. Room-Temperature Tensile Properties of the Ti-Al-Nb-xB Alloys. . . . .......211

Table XXII. 650°C Tensile Properties of the Ti-15Al-33Nb and Ti-21Al-29Nb
Microstructures .................................................................................... 215

Table XXIII. Difference In Tensile Properties at 650°C Compared to Room-
Temperature ........................................................................................ 215

Table XXIV. 650°C Tensile Properties of the Ti-Al-Nb-xB Alloys ....................... 222

Table XXV. Room-Temperature Tensile UTS Values of the AP Samples that Exhibited
Run-Out ............................................................................................. 233

Table XXVI. Minimum Creep Rates Determined by Experiment and Empirical
Model ................................................................................................ 302

Table XXVII. Room-Temperature Tensile Properties for a Ti-22Al-27Nb Alloy With
and Without 6.5mass% TiB ...................................................................... 312

xi

LIST OF FIGURES
Images in this Dissertation are presented in color.

Figure 1.1. The compositions of near or-Ti, a2, 82, y, and O-alloys. The directional
arrow indicates the direction in which (12 compositions evolved to orthorhombic
compositions [Banerjee (2006)] ..................................................................... 8

Figure 1.2. A vertical section for Ti-25Al-be alloy compositions [Sagar et al. (1996)...9

Figure 1.3. A radar diagram comparing the properties of titanium aluminides with the
Ni-based superalloy IN 718 and conventional Ti alloy [MI 834. The low cycle fatigue
(LCF) comparison is for IMI 834 at 600°C, IN 718 and the y alloy Ti-46AI-2Nb at
650°C, and the O alloy at 550°C. The crack growth comparison is for [M] 834 at 550°C,
and IN 718, the y alloy Ti-47Al-2Cr, and the O alloy at 650°C. The densities used to
normalize the data are 8.2 g/cm3, 5.5 g/cm3, 4.8 g/cm3, and 3.8 g/cm3 for IN 718, the O
alloys, [MI 834, and the y alloys, respectively [Baneljee (2006)] ............................ 10

Figure 2.1. Crystal structures of the constituent phases: (a) the ordered B2 unit cell, (b)
one third of the or; unit cell, and (c) the 01 unit cell [B. Mozer et al. (1990), K.
Muraleedharan et al. ( 19953)] ..................................................................... 19

Figure 2.2. The similarities between the basal planes of the or; phase and the 02 phase
shown in an [001] projection ..................................................................... 20

Figure 2.3. Pseudobinary diagram based on TiAl and TiNb, for Ti = 50 at%, with
differing ratios of Alsz, first developed by Bendersky and coworkers [Bendersky et al.
(1991)] and later modified by Boehlert and coworkers [Boehlert et al. (1999)] ............ 24

Figure 2.4. Possible Burger’s vectors in (a) the a; phase containing Nb and (b) the 02
phase [Banerjee (1997)] ............................................................................ 29

Figure 2.5. Schematic illustrating the creep stress exponent values for the diffusion— and
dislocation-controlled creep regimes for pure metals .......................................... 41

xii

Figure 3.1. (a) The as—forged Ti-15Al-33Nb and Ti-21Al-29Nb ingots. The Ti-15Al-
33Nb ingot is labeled 005 is and the Ti-21Al-29Nb ingot is labeled 006. Note that the Ti-
21Al-29Nb ingot exhibited more extensive cracking than the Ti-lSAl-33Nb ingot. (b) An
as-processed Ti-21A1-29Nb sheet. Note the elimination of the cracking that occurred
during forging ....................................................................................... 66

Figure 3.2. Schematic depicting the heat treatment schedule for the HT: 1005 heat
treatment ............................................................................................. 70

Figure 3.3. A schematic depicting the three different sheet orientations from which
samples were taken for microstructural characterization ....................................... 72

Figure 3.4. Variable environment thermomechanical servohydraulic test machine used in
this thesis. Image (0) depicts the inside of the test chamber including the friction grips
and extensometer .................................................................................... 79

Figure 3.5. (a) A schematic of the fatigue specimen geometry used. All values are given
in millimeters. (b) Image of the fatigue set-up at the Fraunhofer Institute. A pulsating
sinusoidal load (tension—tension mode) was used for fatigue testing performed ........... 81

Figure 3.6. Tensile and creep samples (approximately 38 mm long) glued to a metallic
platen for metallographic polishing prior to evaluation of the in-situ deformation. . . . . ....86

Figure 3.7. A digital image of the in-situ tensile stage. The sample is gripped between
the two fixtures shown in the middle of the tensile stage. The 6 mm diameter tungsten-
based heating element is visible directly beneath the gage section of the sample and is
surrounded by the rectangular ceramic sheath ................................................... 87

Figure 3.8. Samples creep tested for up to 162 hours within the SEM using the tensile
stage are shown in (a). Note the preservation of the sample surfaces and the localization
of the deformation in the gage section to the portion of the sample gage section that sat
directly above the 6 mm diameter tungsten-based heating element. The typical placement
of the thermocouple attached directly to the sample gage section is illustrated in the BSE
image in (b). The white circle below the specimen in (b) is the tungsten-based heating
element ................................................................................................ 88

xiii

Figure 4.1. Differential thermal analysis curve for a Ti—SAl-45Nb specimen. The sample
was equilibrated at 600°C and heated to 1200°C at a rate of

Figure 4.2. Differential thermal analysis curve for a Ti-15Al-33Nb powder specimen.
The sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The
temperatures indicated on the curves represent the peak value of the
exotherm/endotherm ................................................................................ 92

Figure 4.3. Differential thermal analysis curve for a Ti-21Al-29Nb powder specimen.
The sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The
temperatures indicated on the curves represent the peak value of the
exotherrn/endotherm ................................................................................ 93

Figure 4.4. Differential thermal analysis curve for a Ti-22Al-26Nb specimen. The
sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The
temperatures indicated on the curves represent the peak value of the
exotherm/endotherm ................................................................................ 94

Figure 4.5. Backseattered electron images of the AP microstructures: (a) Ti-15Al-33Nb
and (b) Ti—21Al-29Nb. The a2 phase is the dark equiaxed phase and the BCC phase is the
continuous matrix phase in each alloy ............................................................ 99

Figure 4.6. Backseattered electron images of Ti-15Al-33Nb solutionized and water
quenched microstructures: (a) 855°C and (b) 910°C .......................................... 100

Figure 4.7. Backseattered electron images of Ti-ISAl-33Nb solutionized and water
quenched microstructures: (a) 960°C and (b) 990°C. The pitting present in (b) occurred
due to over-etching the sample... .................................................................................... 101

Figure 4.8. Backseattered electron images of Ti-15A1-33Nb solutionized and water
quenched microstructures: (a) 1005°C and (b) 1050°C ....................................... 102

Figure 4.9. Backseattered electron images of Ti-15Al-33Nb solutionized and water
quenched microstructures: (a) 1075°C and (h) 1105°C ....................................... 103

xiv

Figure 4.10. Backscattered electron images of Ti-21A1-29Nb solutionized and water
quenched microstructures: (a) 855°C, and (b) 910°C ......................................... 104

Figure 4.11. Backscattered electron images of Ti-21Al-29Nb solutionized and water
quenched microstructures: (a) 960°C and (b) 990°C .......................................... 105

Figure 4.12. Backscattered electron images of Ti-21Al-29Nb solutionized and water
quenched microstructures: (a) 1005°C and (b) 1050°C ....................................... 106

Figure 4.13. Backscattered electron images of Ti-21Al-29Nb solutionized and water
quenched microstructures: (a) 1075°C, and (b) 1105°C ...................................... 107

Figure 4.14. Backscattered electron images of HT:960 microstructures: (a) Ti-15Al-
33Nb and (b) Ti-21Al-29Nb ..................................................................... 108

Figure 4.15. Backscattered electron images of HT :1005 microstructures: (a) Ti-15Al-
33Nb and (b) Ti-21Al-29Nb. The arrows in (a) point to some of the prior-BCC grain
boundary triple-points ............................................................................. 109

Figure 4.16. Backscattered electron images of Ti-lSAl-33Nb (a) HT:650°C/100h/FC and
(b) HT:1105 microstructures. Note the extremely fine O-phase laths present in (a) and
the arrows in (b) indicate prior-BCC grain boundary triple-points ......................... 110

Figure 4.17. Backscattered electron images of (a) AP Ti-15A1-33Nb and (b) Ti-15Al-
33Nb HT:650°C/100h/FC. Note the presence of a large volume fraction of O—phase laths
in (b) with no O-phase present in (a) ............................................................ 111

Figure 4.18. Stereomicroscope image of the surface of the AP Ti-5Al-45Nb rolled sheet
is shown in (a) and a BSE image of the As-HIPed Ti-22Al-26Nb alloy is shown in

(b) .................................................................................................... 113

Figure 4.19. O-phase volume fraction versus solutionizing temperature for the Ti-ISAl-
33Nb and Ti-21Al-29Nb alloys .................................................................. 115

Figure 4.20. Measured equiaxed grain size versus solutionizing temperature for the Ti-
15Al-33Nb and Ti-21Al29Nb alloys ............................................................ 117

XV

Figure 4.21. Long count time scans between 60-65° 20 for (a) the Ti-15Al-33Nb alloy,
and (b) the Ti-21Al-29Nb alloy. Note the presence of the superlattice (210) reflection in

(b) .................................................................................................... 120

Figure 4.22. (a) A [001] zone axis SADP and (b) its associated BF image obtained from
a Ti-21Al-29Nb 910°C/3h/WQ sample ......................................................... 121

Figure 4.23. A [00]] zone axis SAD pattern obtained from a Ti-5A1-45Nb sample. The
lack of <100> superlattice reflections confirm that this alloy has the disordered BCC
structure ............................................................................................. 122

Figure 4.24. Backscattered electron images of 855°C/200h/WQ microstructures: (a) Ti-
15Al-33Nb and (b) Ti-21Al-29Nb ............................................................... 126

Figure 4.25. Backscattered electron images of 910°C/200h/WQ microstructures: (a) Ti-
15Al-33Nb and (b) Ti-21Al-29Nb ............................................................... 127

Figure 4.26. Backscattered electron images of 960°C/200WQ microstructures: (a) Ti-
15Al-33Nb and (b) Ti-21Al-29Nb ............................................................... 128

Figure 4.27. Differential thermal analysis curve for a Ti-22Al-26Nb-5B specimen. The
sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The
temperatures indicated on the curves represent the peak value of the
exotherm/endotherm 131

Figure 4.28. Backscattered electron images of Ti-22Al-26Nb-xB gas atomized powders:
(a) Ti-22Al-26Nb (b)Ti-22Al-26Nb-SB ........................................................ 132

Figure 4.29. Backscattered electron images of Ti-22Al-26Nb-SB rnicrostructures at (a)
low magnification and (b) high magnification ................................................. 133

Figure 4.30. High-magnification BSE images of the borides in (a) Ti-22Al-26Nb-5B and
(b) Ti-15A1-33Nb-SB. Faulting appears to exist across the boride cross-sections. The
cracking within the borides in is due to in-situ tensile testing, and is not an artifact of
processing ........................................................................................... 134

xvi

Figure 4.31. Backscattered electron images of incompletely homogenized gas atomized
powders evident in the microstructure of the Ti-22Al-26Nb-SB alloy after

Figure 4.32. Backscattered electron images of the Ti-15Al-33Nb-O.5B microstructures:
(a)APand (b) HT:1005...... ........................................................................................... 137

Figure 4.33. Backscattered electron images of The Ti-lSAl-33Nb-5B microstructures:
(a) As-HIPed and (b) HT :1005 ........................................................................................ 138

Figure 4.34. EBSPs obtained fi'om the boride phase in Ti-15A1-33Nb-5B. The patterns
in (c) and (d) indicate the same patterns in (a) and (b) indexed according to the
orthorhombic 827 TiB phase structure ............................................................................ 139

Figure 4.35. SADPs obtained from a boride needle within a Ti-15A1-33Nb-5B sample:
(a) [130] zone axis and (b) [121] zone axis. The SADPs in (a) and (b) are fi'om the same
boride, along the same kikuchi band, but are not from the boride in the upper left-hand
corner of (c), which is a BF image of a boride within a Ti-15A1-33Nb-SB HT:1005
sample. Note what appears to be coring or faulting within the boride ..................... 140

Figure 4.36. Creep strain versus time data obtained from a Ti-lSAl-33Nb HT:1105
sample during a load-jump experiment ......................................................... 142

Figure 4.37. Creep strain versus time data obtained from a Ti-lSAl-33Nb HT:1105
sample during a temperature-jump experiment ................................................ 143

Figure 4.38. Plot of log Em... versus log 0 for Ti-15A1-33Nb microstructures creep tested
at T = 650°C ........................................................................................ 148

Figure 4.39. Plot of log 2...... versus log a for Ti-21Al-29Nb microstructures creep tested
at T = 650°C ........................................................................................ 149

Figure 4.40. Plot of log Em versus log 0 for the Ti-5A1-45Nb alloy creep tested at
T = 650°C ........................................................................................... 150

xvii

Figure 4.41. Plot of log E... versus log 6 for the Ti-5Al-45Nb, Ti-15Al-33Nb, and Ti-
21Al-29Nb alloys creep tested in this thesis ................................................... 151

Figure 4.42. Activation energies determined at o = 34 MPa and o = 75 MPa for the Ti-
5Al-45Nb alloy .................................................................................... 152

Figure 4.43. Activation energies determined at 0' = 150 MPa for the Ti-lSAl-33Nb
HT:1005 and HT:1105 heat-treated microstructures .......................................... 153

Figure 4.44. Activation energies determined at o = 275 MPa for the Ti-15Al-33Nb
HT: 1005 and HT:1105 heat-treated microstructures .......................................... 154

Figure 4.45. Activation energies determined at o = 48 MPa and o = 148 MPa for the Ti-
21Al-29Nb HT:960 and HT:1005 heat-treated microstructures ............................. 155

Figure 4.46. Creep strain versus time plot for heat treated Ti-15Al-33Nb microstructures
creep tested at 172 MPa/650°C .................................................................. 156

Figure 4.47. Displacement versus time data collected during insitu creep testing of the
Ti-15Al-33Nb HT: 1005 samples. The black curve represents a stress of 225 MPa and the
red curve represents a stress of 300 MPa ....................................................... 163

Figure 4.48. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-15A1-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-
(n). The loading direction is horizontal in each image ................................. 164-170

Figure 4.49. Backscattered electron images showing the most extensive grain boundary
cracking/sliding/upheaval observed in the Ti-15Al-33Nb HT:1005 sample creep tested at
225 MPa/650°C. The total displacement in mm is indicated in the lower right-hand
comer of micrographs (a)-(h). The loading direction is horizontal in each
image ................................................................................................ 171

Figure 4.50. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-15Al-33Nb HT:1005 sample creep tested at 300 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-
(h). The loading direction is horizontal in each image ................................. 172-175

xviii

Figure 4.51. Backscattered electron images of (a) Severe environment-assisted edge
cracking and (b) internal cracking at prior-BCC grain boundaries observed within the
bulk of a Ti-lSAl-33Nb HT :1005 sample which failed at a creep strain of 1.7% after 820
hours of creep deformation under conditions of 172 MPa/650°C ........................... 177

Figure 4.52. Backscattered electron images of internal cracking at prior-BCC grain
boundaries observed within the bulk of a Ti-15A1-33Nb HT:1005 sample which failed at
a creep strain of 1.7% after 820 hours of creep deformation under conditions of 172
MPa/650°C .......................................................................................... 178

Figure 4.53. Backscattered electron images of internal cracking at prior-BCC grain
boundaries observed within the bulk of a Ti-15Al-33Nb HT:1005 sample which failed at
a creep strain of 1.7% after 820 hours of creep deformation under conditions of 172
MPa/650°C .......................................................................................... 179

Figure 4.54. Creep strain versus time data obtained from a Ti-15Al-33Nb-0.5B HT:1005
sample during a load-jump experiment ......................................................... 183

Figure 4.55. Creep strain versus time data obtained from a Ti-lSAl-33Nb-0.5B HT:1005
sample during a temperature-jump experiment ................................................ 184

Figure 4.56. Plot of log 2...... versus log a for the Ti-5Al-45Nb alloy creep tested at T =
650°C ................................................................................................ 185

Figure 4.57. Activation energies determined at 0 =150 MPa, 0 = 250 MPa, and o = 275
MPa for the Ti-15Al-33Nb-0.5B and Ti-lSAl-33Nb-SB heat-treated microstructures...186

Figure 4.58. Creep strain versus time behavior under test conditions of 250-
275MPa/I=650°C in air for the Ti-15AI-33Nb-xB materials ............................... 187

Figure 4.59. Displacement versus time behavior for Ti-Al-Nb-xB materials under creep
testing conditions of 400 MPa/650°C in a vacuum atmosphere acquired using the in-situ
tensile stage ............................................................................................. 188

xix

Figure 4.60. Backscattered electron images illustrating the evolution of creep
deformation in at Ti-22Al-26Nb-5B sample creep tested at 375 MPa/650°C. The total
displacement in mm is indicated in the lower right-hand comer of micrographs (a)-(h).
The loading direction is horizontal in each image ....................................... 191-194

Figure 4.61. Backscattered electron images illustrating extensive surface cracking
exhibited by a Ti-22Al-26Nb-5B sample creep tested at 375 MPa/650°C. The loading
direction is horizontal in each image ............................................................ 195

Figure 4.62. Backscattered electron images taken during a 650°C creep experiment at
400 MPa for a Ti-22Al-26Nb sample. Both images were acquired at a creep
displacement of 0.41 mm. The white arrows indicate some of the numerous prior-BCC
grain boundary cracks. The loading direction is horizontal in each
image ................................................................................................. 196

Figure 4.63. Backscattered electron images of a post-mortem Ti-15Al-33Nb-5B sample
creep tested at 400 MPa/650°C. Note that the majority of the boride needles remain

uncracked while the cracking and sliding/upheaval of prior-BCC grain boundaries
accounted for the majority of the deformation ................................................ 197

Figure 4.64. Room-temperature tensile behavior for the Ti-5Al-45Nb, Ti-15A1-33Nb, Ti-
21Al-29Nb, and Ti-22Al-26Nb microstructures ............................................... 201

Figure 4.65. Secondary electron images of RT tensile fracture surfaces: (a) Ti-15Al-
33Nb AP and (b) Ti-21Al-29Nb AP ............................................................ 204

Figure 4.66. Secondary electron images of RT tensile fracture surfaces: (a) Ti-15Al-
33Nb HT:960, (b) Ti-21Al-29Nb HT:960 ...................................................... 205

Figure 4.67. Secondary electron images of RT tensile fracture surfaces: (a) Ti-15Al-
33Nb HT:1005 and (b) Ti-21Al-29Nb HT1005206

Figure 4.68. Secondary electron images of RT tensile fracture surfaces: (a) Ti-15Al-
33Nb HT:650/100h/FC and (b) Ti-lSAI-33Nb HT :1 105 .................................... 207

Figure 4.69. Secondary electron images of surface slip observations taken from a fully-
BCC Ti-SAl-45Nb sample tensile tested to failure at RT. The sample failed at a strain of
24.9% at RT ........................................................................................ 208

Figure 4.70. Secondary electron images of surface slip and cracking exhibited by a Ti-
15A1-33Nb sample tensile tested to failure at RT. The sample failed at a strain of 16.8%

at RT ................................................................................................. 209
Figure 4.71 . Ti-15Al-33Nb-xB RT stress versus strain behavior ........................... 212
Figure 4.72. Ti-22Al-26Nb-xB RT stress versus strain behavior ................................... 213

Figure 4.73. 650°C tensile behavior of the Ti-15Al-33Nb and Ti-21Al-29Nb alloys. . .216

Figure 4.74. Secondary electron images of 650°C tensile fracture surfaces: (a) Ti-lSAl-
33Nb AP and (b) Ti-21Al-29Nb AP ........................................................... 217

Figure 4. 75. Secondary electron rmages of 650°C tensile fracture surfaces: (a)WTi-15Al-
33NbHTz96O and (b) Ti-21Al-29Nb HT: 960.. . ...218

Figure 4. 76. Secondary electron images of 650°C tensile fracture surfaces: (a)m Ti-15Al-
33NbHT: 1005 and (b) Ti-21Al-29NbHT: 1005.. . 2.19

Figure 4.77. Secondary electron images of 650°C tensile fracture surfaces: (a) Ti-15Al-
33Nb HT:650/100h/FC and (b) Ti-15Al-33Nb HT:1105 .................................... 220

Figure 4.78. Ti-15Al-33Nb-xB 650°C stress versus strain behavior ....................... 223

Figure 4.79. Ti-22Al-26Nb-xB stress versus strain behavior at 650°C. These
experiments were performed using the tensile stage inside the SEM chamber and accurate
strain measurements were not available, though it is clear that the Ti-22Al-26Nb sample
underwent significantly more displacement and strain than the Ti-22Al-26Nb-SB
sample ............................................................................................... 224

Figure 4.80. Backscattered electron images of boride cracking at different locations on
the surface of a Ti-lSAl-33Nb-5B HT tensile sample tested to a strain to a strain of 1.7%
at 650°C in vacuum ................................................................................ 225

Figure 4.81. Backscattered electron images of surface damage observations made behind
the fiacture surface after tensile testing a Ti-22Al-26Nb sample to failure at 650°C. ...226

Figure 4.82. Backscattered electron images of surface damage observations made behind
the fi'acture surface after tensile testing a Ti-22Al-26Nb-SB sample to failure at 650°C.
Part of the fracture surface is visible in the upper right-hand corner of (a) ................ 227

Figure 4.83. Maximum cyclic stress versus fatigue life for the AP alloys tested at the
Fraunhofer Instrtute230

Figure 4.84. Maximum cyclic stress versus fatigue life for the all alloys tested ......... 231

Figure 4.85. Maximum cyclic stress versus fatigue life for the AP and HT :1005 alloys
tested at the Toyohashi University of Technology ............................................ 232

Figure 4.86. Secondary electron images of the fatigue fiacture surfaces of (a) an AP Ti-
15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles and (b) an AP Ti-21Al-29Nb sample tested at a maximum cyclic stress of
930 MPa which failed after 19,087 cycles ...................................................... 234

Figure 4.87. Secondary electron images of a fatigue crack initiation site in an AP Ti-
15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles ....................................................................................... 235

Figure 4.88. Secondary electron images of a fatigue crack propagation region in an AP
Ti-15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles ....................................................................................... 236

Figure 4.89. Secondary electron images of a fatigue crack overload region in an AP Ti-
15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles ....................................................................................... 237

xxii

Figure 4.90. Secondary electron images a fatigue crack initiation site in an AP Ti-21Al-
29Nb sample tested at a maximum cyclic stress of 930 MPa which failed afier 19,087

cycles ................................................................................................ 238

Figure 4.91. Secondary electron images of a fatigue crack propagation region in an AP
Ti-21Al-29Nb sample tested at a maximum cyclic stress of 930 MPa which failed after
19,087 cycles ....................................................................................... 239

Figure 4.92. Secondary electron images of a fatigue crack overload region in an AP Ti-
21Al-29Nb sample tested at a maximum cyclic stress of 930 MPa which failed after
19,087 cycles ....................................................................................... 240

Figure 4.93. Backscattered electron images of surface slip traces exhibited by an AP Ti-
lSAl-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles ....................................................................................... 241

Figure 4.94. Backscattered electron images of surface slip traces exhibited by an AP Ti-
21Al-29Nb sample tested at a maximum cyclic stress of 930 MPa which failed after
19,087 cycles ....................................................................................... 242

Figure 5.1. Pseudobinary diagram based on TiAl and TiNb, for Ti = 50 at%, with
differing ratios of Al:Nb, frrst developed by Bendersky and coworkers [Bendersky et al.
(1991)] and later modified by Boehlert and coworkers [Boehlert et al. (1999)] .......... 246

Figure 5.2. An isopleth based on Ti = 50 at%, with differing ratios of Al:Nb. This
diagram was constructed from the phase equilibria results of this thesis added to the work
of [Bendersky et al. (1991)] and [Boehlert er al. (1999)]247

Figure 5.3. A 900°C ternary slice of the Ti-Al-Nb system [Rowe et al. (1993)]. The
black and white data points represent EMPA measurements from Boehlert et al. (1999).
EMPA measurements obtained from the BCC phase in Ti-15A1-33Nb and Ti-21Al-29Nb
samples solutionized at 910°C for 200 hours followed by water quenching are shown on
this diagram. The red point is the nominal composition of the Ti-15A1-33Nb alloy and is
also the composition of the BCC phase in this alloy at 910°C. The blue point represents
the nominal composition of the Ti-21Al-29Nb alloy and the orange point is the
composition of the BCC phase in the Ti-21Al-29Nb alloy at 910°C ....................... 248

xxiii

Figure 5.4. BCC-phase grain growth kinetics for the Ti-15A1-33Nb and Ti-21Al-29Nb
samples heat-treated in the BCC phase field for three hours followed by water
quenching ........................................................................................... 254

Figure 5.5. Maximum grain boundary displacements measured for the grain boundaries
labeled 1, H, and III in Figure 5.5 as a function of the measured total creep displacement.
The maximum grain boundary cracking displacement for the grain boundaries shown in
Figure 5.6. The grain boundaries shown in Figure 5.6 displayed the largest extent of
grain boundary cracking/sliding out of all the grain boundaries contained within the gage
section of the Ti-15A1-33Nb HT:1005 sample creep tested under conditions of 225
MPa/650°C .......................................................................................... 275

Figure 5.6. Backscattered electron images of grain boundaries 1, II, and III at creep
displacements of (a) 0 mm and (b) 0.41 mm. This figure shows the grain boundaries
from which the displacement measurements in Figure 5.6 were measured. The loading
direction is horizontal in each image ............................................................ 276

Figure 5.7. Backscattered electron images obtained at a creep displacement of 0.41 mm
for (a) the most extensive amount of grain boundary cracking/sliding/upheaval and (b)
the second most extensive amount of grain boundary cracking/sliding/upheaval. The
loading direction is horizontal in each image .................................................. 277

Figure 5.8. Backscattered electron images obtained from the Ti-15Al-33Nb HT:1005
sample creep tested at 225 MPa/650°C: (a) Pretest and (b) at a total creep displacement of
412 11m. The loading direction is horizontal in each image. The boxed-in area illustrates
the same area shown in Figure 5.6 ................................................................ 281

Figure 5.9. The measured minimum creep rate versus OZ/d for the Ti-15Al-33Nb and Ti-
2lAl-29Nb microstructures creep tested within the 11 value regime of 1.5-3.2 at
650°C ................................................................................................ 287

Figure 5.10. The measured minimum creep rate versus 62/(1 for the Ti-SAl-45Nb alloy
creep tested in the n=2.6 regime at 650°C ...................................................... 288

Figure 5.11. The stress dependence of the minimum creep rate at 650°C for fully-BCC,
O+BCC, and fully-O Ti-Al-Nb alloys .......................................................... 296

xxiv

Figure 5.12. The minimum creep rates determined experimentally and by the model
given in equation (6) versus applied stress. The solid lines represent the model
calculation and the data points are fiom experiment .......................................... 299

Figure 5.13. The minimum creep rates determined experimentally and by the model
given in equation (6) versus applied stress. The model values obtained fiom equation (6)
were then normalized by the average equiaxed grain size. The solid lines represent the
model calculation and the data points are from experiment ................................. 301

Figure 5.14. Room-temperature tensile stress versus strain curves for fully-B Ti-5Al-
45Nb and Ti-l2Al-38Nb [Boehlert (1999)] alloys. The Ti-12Al-38Nb alloy did not fail.
It was unloaded afier reaching 27% strain ...................................................... 309

Figure 5.15. Yield strength versus temperature for selected O+BCC Ti-Al-Nb
microstructures ..................................................................................... 314

Figure 5.16. Maximum cyclic stress versus fatigue life for the Ti-15A1-33Nb and Ti-
21Al-29Nb alloys in comparison to other Ti alloys to be used for biomedical
applications ......................................................................................... 319

XXV

CHAPTER 1

INTRODUCTION

This thesis comprises a systematic study of several aspects of the physical and
mechanical metallurgy of advanced titanium-aluminum-niobium (Ti-Al-Nb) alloys with
and without boron additions. In particular, the microstructural evolution, phase
transformation behavior, elevated-temperature creep behavior, room-temperature and
elevated-temperature tensile behavior, and room-temperature fatigue behavior were
evaluated. The specific alloys studied were: Ti-5Al-45Nb (at%), Ti-15A1-33Nb (at%),
Ti-15Al-33Nb-0.SB (at%), Ti-15Al-33Nb-SB (at%), Ti-21Al-29Nb (at%), Ti-22Al-26Nb
(at%), and Ti-22Al-26Nb-SB (at%) [henceforth, it will be assumed throughout the
remainder of this thesis that all alloy compositions are reported in atomic percent unless
specifically stated otherwise]. The only alloy composition that had been previously
studied before this study began was Ti-22Al-26Nb. Thus, this thesis presents needed
information on several new alloys and fills the knowledge gap that existed before this
work was completed for O+BCC Ti-Al-Nb alloys targeted by the Air Force for structural
applications. Also presented in this thesis is a methodology that was developed for
identifying the deformation evolution in-situ during high-temperature creep experiments.
Such a methodology has yet to be presented in the literature. The creep deformation
behavior was the main focus, and in particular understanding the dominant creep
deformation mechanisms as a function of stress, temperature, and strain rate in addition

to understanding the affect of microstructure on creep deformation.

The focus of this Chapter is three-fold. First, the rationale for performing this
work is stated. Second, a brief overview of conventional titanium alloys is given,
followed by the origin and evolution of the class of titanium alloys studied in this work.
The specific aims of this thesis are given third in order to lead into the relevant
background and literature review of orthorhombic-based Ti-Al-Nb alloys presented in

Chapter 2.

A. Rationale

Within the aerospace industry there is a continuing need for the development of
materials that can withstand the harsh environments found in jet turbine engines. The
ongoing drive to ever higher temperatures exists due to the fact that increasing engine
operating temperature increases engine efficiency, with the simultaneous cost benefit of
decreasing fuel consumption and the environmental benefit of decreasing the amount of
pollution generated. Ti-Al-Nb alloys are prime candidates for use in the last stages of
high-pressure compressors or in the early stages of low-pressure turbines that are found
in gas turbine engines that operate in the temperature range of 550-800°C.

Turbine blades generally experience longitudinal stresses up to approximately 140
MPa and temperatures in the range of 650-1000°C in their airfoil section [Reed-Hill and
Abbaschian (1994)]. The blade root is subjected to much higher tensile stresses, on the
order of 280-560 MPa, but the temperature does not exceed 760°C [Reed-Hill and
Abbaschian (1994)]. Therefore, the blades must not only be strong enough to withstand
the high stresses, but must also have adequate creep resistance [Reed-Hill and

Abbaschian (1994)]. Creep is defined as the time-dependent deformation experienced by

a material under a constant stress, and can occur at stresses well below the yield stress of
the material. Creep deformation typically becomes a concern when temperatures
approach and exceed half the absolute melting temperature of the material. For materials
used in jet engine environments, the ability to resist creep deformation is one of the main
design limiting mechanical properties. Therefore, sufficient creep resistance combined
with a thorough understanding of the micromechanisms by which creep deformation
occurs in the material system of interest are required before the material can be

considered for use in jet engine applications.

B. Conventional Titanium Alloys

After the Second World War, titanium-based alloys gained consideration as key
components for aircraft engines. Today, the aerospace industry is still the prime
consumer of titanium and its alloys, but titanium has gained acceptance in architecture,
chemical processing, medicine, power generation, transportation, marine and offshore,
and sports and leisure markets. The reason titanium alloys stand out as prime candidates
for these applications is due to two properties: high specific strength and excellent
corrosion resistance [Peters et. al. (2003)].

Pure titanium is an allotropic element, meaning that it can exist in more than one
type of crystal structure. The two crystal structures that pure titanium exists in are the
hexagonal close-packed structure (HCP), referred to as or, and the body-centered cubic

structure (BCC), referred to as B. At high temperatures B-titanium is the stable phase and

at low temperatures a-titanium is the stable phase. The temperature above which a

titanium alloy is completely BCC is referred to as the beta-transus temperature (BCC-

transus), and for pure titanium this transformation occurs at 882 i 2°C [Peters et. al.
(2003)].

Conventional titanium alloys are classified as either or alloys, B alloys, or or+B
alloys and can be further divided into near-or and metastable B alloys. The amount and
type of alloying elements determine the type of classification a particular titanium alloy is
given. Elements such as carbon, oxygen, nitrogen, and aluminum stabilize the or phase
and raise the BCC-transus temperature. Chromium, nickel, molybdenum, and niobium
are B stabilizers and therefore lower the BCC-transus temperature. The or alloys are
generally categorized by good strength, toughness, creep resistance, and weldability.
These types of alloys cannot be strengthened by heat treatment, but can be used for
cryogenic temperature applications due to the absence of a ductile-to-brittle transition.
The B alloys are characterized by high hardenability and forgeability, but are susceptible
to a ductile-to-brittle transition due to the strong temperature dependence of the thermal
component of the flow stress that is associated with the BCC crystal structure. In the
solution-treated condition, B alloys have good ductility, toughness, excellent forrnability,
and can be age hardened through heat treatment to produce a two-phase microstructure

with well balanced mechanical properties [Reed-Hill and Abbaschian (1994)].

C. The Origin and Evolution of Orthorhombic Titanium-Aluminides

lntermetallic alloys in the titanium-aluminum system are a revolutionary class of
materials that have the potential to replace dense nickel-based super alloys that are
currently used in the last stages of high pressure compressors or in the early stages of low

pressure turbines found in gas turbine engines that operate in the temperature range of

550-800°C. Ti3Al ((12) alloys with a hexagonal crystal structure and P63/mcm space
group and TiAl (y) alloys with a face-centered tetragonal (FCT) crystal structure and
P4/mm space group are the two intermetallic alloys that have received the most attention.

The problem with the majority of structural intermetallics is their lack of ductility,
especially at room temperature. Blackburn and Smith were able to realize engineering
plasticity in Ti3Al by alloying with Nb, which incorporated the BCC phase into the
microstructure [Blackburn and Smith (1978)]. This led to the emergence of the Ti-24Al-
lle alloy in the 1970’s [Marquardt et al. (1989)]. Blackburn and Smith then explored
higher Nb additions in the range of 15-20 at% which resulted in the Ti-25Al-10Nb-3V-
1M0 super (12 alloy and various Ti-Al-Nb alloys with a fixed Nb composition of 17 at%
and varying Al and Mo contents, such as Ti-24Al-17Nb-xMo alloys [Blackburn and
Smith (1989)].

In the late 1980’s and through the 1990’s Rowe and coworkers showed that alloys
containing up to 25 at% Nb displayed improved combinations of specific strength,
toughness, and creep resistance [Rowe (1991a), Rowe et al. (1991), Rowe and Larsen
(1996)]. At the same time as Rowe’s investigations, Banerjee and coworkers discovered
that the orthorhombic (0) phase, which is a ternary intermetallic based on the
composition TizAle with the cmcm space group, was a major constituent phase of the
Ti-Al-Nb alloys in Rowe’s studies [Banerjee et al. ( 1988)].

The orthorhombic titanium-aluminides are a relatively new class of alloys that
evolved from the development of engineering alloys based on T13Al-Nb. Figure 1.1
depicts O-alloy compositions in relationship to near a—Ti, (12 alloys, and y alloys. Figure

1.2 displays a vertical section of the Ti-Al-Nb system which shows the influence of Nb

content and temperature on the type of constituent phases present. A radar diagram that
compares the properties of O-alloys, y-TiAl, the nickel-based super alloy Inconel (IN)
718, and the conventional Ti alloy [MI 834 is given in Figure 1.3. The realization that
TizAle alloys displayed superior elevated temperature strength and creep resistance to
conventional Ti alloys while simultaneously displaying greater elongation-to-failure at
room temperature than both their y—based and org-based counterparts led to a wide-ranging
alloy development effort that lasted through the end of the 20‘h Century. This effort was
led by researchers at the Air Force Research Laboratory (AF RL) (Dayton, OH, USA) and
the Defense Metallurgical Research Laboratory (DMRL) (Hyderabad, India). However,
many investigators from several different organizations were involved such as Rowe and
Woodfield from General Electric Research and Development (GERD) (N iskayuna, NY) ,
researchers at Allison Aerospace Engines (Indianapolis, IN), Hagiwara and Emura from
the National Institute for Materials Science (N IMS) (Tsukuba, Japan), Yang fiom the
Korean Advanced Institute of Science and Technology (KAIST) (Daejeon, South Korea),
Popille and coworkers from Laboratoire d’Etude des Microstructures (Chatillon Cedex,
France) and Centre d’Elaboration des Materiaux et d’Etudes Structurales (Toulouse
Cedex, France), and Bendersky and Boettinger from the National Institute of Standards
and Technology (N IST) (Gaithersburg, MD).

The O-phase is the main constituent phase of the alloys studied in this thesis. The
BCC phase is the second main constituent phase. The alloys studied in this work differ
from all other O-based alloys that have been previously researched in that they contain
less Al and more Nb. As will be discussed in Chapter 2, this decrease in Al content

generally leads to significant improvement in room-temperature elongation-to-failure. It

is important to state that decreasing the Al content in this alloy system does not
necessarily lead to increased room-temperature elongation-to-failure, because the effect
of microstructure can never be neglected.

The desire for a structural intermetallic alloy that possesses both elevated-
temperature strength and creep resistance, while at the same time maintaining room-
temperature ductility, has always existed. In the course of this thesis work, an
intermetallic alloy (Ti-15Al-33Nb) with an exceptional balance of elevated-temperature
creep strength and room-temperature elongation-to-failure has been discovered.
Understanding the creep deformation behavior of this alloy and others in this alloy
system as a ftmction of stress, temperature, strain rate, and microstructure constituted the

main focus of this thesis.

 

 

 

 

 

 

 

Figure 1.1. The compositions of near or-Ti, a2, BZ, y, and O-alloys. The directional
arrow indicates the direction in which 012 compositions evolved to orthorhombic
compositions [Banerjee (2006)].

 

 

1400 "‘

 

 

 

 

 

 

 

 

l3
g 132

1200
g.) .
n5
5
a 1000
E) 04412

800

0+0
TI-25A1 10 20

ATOMIC PERCENT NIOBIUM

Figure 1.2. A vertical section for Ti-25Al-be alloy compositions [Sagar et al. (1996)].

Burn Resistance

Larson-Miller Parameter da
for rupture at 650°C at a \ Isothermal Oxi tion
specific stress of hi air 8‘ 6~ 0° C ,

to MPwp mar/rmz
20.5 2 ,
20
9.5 3
SW“ “a" l n [A mum
\2‘3‘ ' '

 

  

 

 

suTss at 6500C. ~ ‘ at room-temperature.
MPa/p '\ 10 % elongation
- ' ' 1000
l ’ ‘\
Specific Modulu. ‘ Specific Low Cycle Fatigue
at 650°C, r Life, in cycles at a plastic strain
MPatp amplitude of +1- 0.030 otp
Specific Stress Intensity, AKpo/p for
Fatigue Crack Growth at 3:11?4 mmtcycle
IN] 834 l 0 Alloys y-TiAl IN 718

 

 

 

 

 

 

 

 

Figure 1.3. A radar diagram comparing the properties of titanium aluminides with the
Ni-based superalloy IN 718 and conventional Ti alloy [MI 834. The low cycle fatigue
(LCF) comparison is for IMI 834 at 600°C, IN 718 and the y alloy Ti-46A1-2Nb at
650°C, and the O alloy at 550°C. The crack growth comparison is for IMI 834 at 550°C,
and IN 718, the y alloy Ti-47Al-2Cr, and the O alloy at 650°C. The densities used to
normalize the data are 8.2 g/cm3, 5.5 g/cm3, 4.8 g/cm3, and 3.8 g/cm3 for IN 718, the O
alloys, IMI 834, and the y alloys, respectively [Banerjee (2006)].

10

D. The Work Performed And Specific Aims

This thesis project began in June 2003 with the acquisition of the Ti-15Al-33Nb
and Ti-21Al-29Nb alloys, both of which had never been examined previously. The Ti-
15A1-33Nb alloy became the main focus of this thesis, as there was no literature available
for alloys based on a nominal Ti content of 50 at% and Al contents ranging between 13-
21 at% and Nb contents ranging 29-37 at%. This will be explained in more detail in
Chapter 2.

The first step was to characterize the microstructures of these alloys through a
solution-treatment study, in order to determine what microstructures were obtainable.
The second step involved devising heat treatment schedules that would bring out the
desired features of interest within the microstructure. Mechanical testing commenced
after an understanding of the microstructural evolution and phase transformation behavior
was obtained. The mechanical testing was focused on understanding microstructure-
tensile relationships and microstructure-creep relationships. Tensile testing was
performed at room-temperature (RT) to evaluate both tensile strength and elongation-to-
failure as a function of microstructure. Tensile testing was performed at 650°C in order
to evaluate elevated-temperature strength and obtain yield strength values in order to
intelligently design creep experiments at the appropriate stress levels.

The RT fatigue life behavior of the Ti—15Al-33Nb and Ti-2lAl-29Nb alloys was
characterized from the fall of 2003 through the fall of 2004 through collaborative efforts
with Fraunhofer Institut fiir Werkstoffrnechanik IWM (F reiburg, Germany) and with

Toyohashi University of Technology (Toyahashi, Japan).

11

Creep experiments were conducted continuously over the course of approximately
two and half years, from the spring of 2004 through the fall of 2006. Creep stress
exponents and apparent activation energies were calculated, and the stress and
temperature dependence of the minimum creep rate was determined over the stress range
of 10-400 MPa and temperature range of 650—710°C.

An in-situ creep testing methodology/technique was developed from the summer
of 2005 through the fall of 2006. This technique uses a specially designed stage to
mechanically test metallographically prepared samples of a special geometry within the
chamber of a Scanning Electron Microscope (SEM). With this technique, damage
evolution as a function of applied stress, displacement, and time can be tracked, allowing
acquisition of direct microstructural evidence of the creep deformation processes
occurring on the sample surface.

During the course of this study it became evident that in order to try and
model/predict the minimum creep rates of O+BCC microstructures, the constituent
properties of a fully-BCC alloy would be required. This led to the processing of a Ti-
5Al-45Nb alloy at AFRL during the summer of 2005. In addition, new discoveries
regarding the effect of boron additions on conventional titanium alloys, mainly Ti-6Al-
4V (wt%), were made at AF RL during the course of time the research for this thesis was
being performed. In particular, boron additions were found to increase the strength and
elastic modulus by approximately 20%, without any adverse effect on elongation-to-
failure. Discussions with collaborators at AFRL led to the processing of Ti-15A1-33Nb-
0.5B, Ti-15Al-33Nb-SB, Ti—22Al-26Nb, and Ti-22Al-26Nb—5B alloys in order to

evaluate the effect of boron additions on the tensile and creep behavior of the Ti-Al-Nb

12

O+BCC alloy system, with the hope of achieving similar property enhancements as those
achieved for the Ti-6Al-4V-xB (wt%) alloy system. A similar research methodology as
outlined above was applied to the Ti-22Al-26Nb alloy and the boron-modified alloys
after their acquisition. A more detailed description of why these compositions were

chosen will be given in Chapter 2.

The specific aims of this thesis were divided into four categories:

(1) Microstructu re

1. To quantify microstructural features such as grain size, phase chemistry, phase volume
fraction, and phase morphology as a function of thermomechanical processing and heat
treatment. Determination of these parameters enabled the construction of microstructure—
property relationships.

2. To determine the temperature ranges of the respective phase fields, and further modify
the available pseudobinary phase diagram (discussed in Chapter 2) for alloys bridging the
compositional gap between Ti-23Al—27Nb and Ti-12Al-3 8N1).

3. To determine the phase transformation behavior that occurs for the Ti-Al-Nb alloys
studied in this thesis.

4. To understand the effect of boron on the microstructural evolution and phase

transformation behavior of the TH 5A1-33Nb and Ti-22Al-26Nb alloys.

13

(2) Creep Behavior

1. To measure creep strain rates as a function of stress, temperature, and microstructure
over the stress range of 10-400 MPa and 650-710°C.

2. To calculate creep stress exponent values and apparent activation energies based on the
measured creep strain rates, and compare these values to those obtained in a similar
matter for other Ti-Al-Nb alloys found in the literature. Accomplishing this objective
allowed performance evaluation and also provided initial direction in estimation of the
potential deformation mechanisms based on pure metal theory.

3. To identify the dominant secondary creep mechanisms as a function of stress,
temperature, and strain rate using SEM deformation observations and in-situ SEM
experiments. A portion of this objective was accomplished through obtaining direct
microstructural evidence of the evolution of creep deformation on the sample surface as a
function of creep displacement through carefully selected in-situ creep experiments, with
a particular focus on observing grain boundary cracking and sliding deformation.

4. To identify microstructure-creep relationships by completing the three aforementioned
objectives, the microstructure objectives, and by applying the understanding of physical
and mechanical metallurgy gained through undergraduate and graduate course work.

5. To determine the minimum creep rates as a function of stress at 650°C for the Ti-SAI-
45Nb alloy in order to obtain the creep behavior of the constituent BCC-phase for
modeling purposes. As will be shown, this was the first time that BCC-phase constituent
creep properties have been obtained for a single phase BCC Ti-Al-Nb alloy.

6. To understand the effect of boron additions on the minimum creep rates and the creep

deformation behavior of the Ti—15A1-33Nb and Ti-22Al-26Nb alloys.

14

(3) Tensile Behavior

1. To determine tensile properties at room-temperature and 650°C in order to construct
microstructure-tensile relationships, especially for the alloy compositions between Ti-
23Al-27Nb and Ti-12Al-3 8Nb that were nonexistent in the literature.‘

2. To obtain constituent BCC-phase tensile data from a Ti-5Al-45Nb alloy.

3. To identify tensile deformation behavior features through fiacture surface observations
and surface damage observations from metallographically prepared samples in order to

aid in the development of microstructure-property relationships.

(4) Fatigue Behavior

1. To evaluate the room-temperature fatigue behavior of the Ti-lSAl-33Nb and Ti-21Al-
29Nb alloys through the construction of maximum cyclic stress versus number of cycles
to failure (S-N) curves.

2. To characterize the fatigue crack initiation and propagation behaviors of the TH 5A1-

33Nb and Ti-21Al-29Nb alloys through fatigue frature surface observations.

15

CHAPTERZ

BACKGROUND AND LITERATURE REVIEW

This chapter presents background information and a review of the literature
published to date that is pertinent to the areas studied in this thesis. The phases present in
this alloy system are described first. An overview of previously determined phase field
ranges is given for alloys processed in a similar manner to those in this study. The three
possible phase transformation mechanisms that have been observed to occur in Ti-Al-Nb
alloys are reviewed. The deformation behavior of the constituent phases is discussed
next, with an emphasis on the similarities and differences in dislocation activity observed
in org-based and O-based alloys. Mechanisms by which creep deformation occurs are
reviewed. The review of literature available on the creep behavior of org-based Ti-Al-Nb
alloys, fully-O Ti-Al-Nb alloys, and multiphase Ti-Al-Nb alloys is provided. The effect
of microstructure on tensile properties is discussed, and tensile properties for several Ti-
Al-Nb alloys are presented. Fatigue failure is defined and a brief overview of the
requirements needed for biomedical fatigue-driven applications is given. The benefits of
boron additions are discussed with respect to conventional Ti alloys and the few works

that have been performed on orthorhombic-based Ti-Al-Nb alloys.

16

A. Orthorhombic Titanium-Aluminides

The alloys that comprise this work differ from conventional titanium alloys in that
they contain an orthorhombic (O) intermetallic phase based on the composition TizAle.
This phase was first discovered in a Ti-25Al-12.5Nb alloy using Transmission Electron
Microscopy (TEM), convergent beam electron diffraction (CBED), and channeling
enhanced microanalysis [Banerjee et a1. (1988)]. The atomic coordinates of the Ti, Al,
and Nb atoms in the 0 structure were later determined using neutron diffraction [Mozer

et al. (1990)].

l. Constituent Phases and Crystal Structures

Ti-Al-Nb alloys can posses up to three constituent phases: (1;, O, and BCC. The
or; phase is an intermetallic hexagonal phase based on the composition Ti3Al with D019
symmetry. As previously mentioned, the O-phase is an intermetallic phase based on the
composition TizAle. This phase has cmcm orthorhombic symmetry and has been found
to exist in two forms, denoted 01 and 02. The Nb atoms occupy specific atom positions
in the 02 structure, as opposed to randomly occupying Ti sites in the 01 structure
[Banerjee (1997)]. The BCC phase is designated as either ordered BZ or disordered B,
and the ordering is dependent on alloy composition and heat treatment [Kestner-
Weykamp et al. (1989), Bendersky et al. (1991), Rhodes et al. (1993)]. The B2 structure
is ordered and based on the CsCl crystal structure and the TizAle composition. The B
structure is disordered BCC and is typically depleted in Al and enriched in Nb. The

crystal structures of the constituent phases are presented in Figure 2.1.

17

Figure 2.2 shows the similarities between the 02 phase and the or; phase. These
phases differ from each other by their arrangement of Ti and Nb atoms. In particular, the
ordered arrangement of Nb atoms in the 02 phase destroys the hexagonal symmetry
present in the or; phase. The combination of the constituent phases gives rise to
interesting mechanical properties, and the amount, type, and morphology of the phases

present in the microstructure control the mechanical behavior.

18

   

    

a = 0.323 nm a = 0.578 um
i Atoms .Al or Nb Atoms .Al Atoms GTi or Nb Atoms

c = 0.460 nm

0 Ti Atoms
.Al or Nb Atoms

 

(C)

Figure 2.1. Crystal structures of the constituent phases: (a) the ordered B2 unit cell, (b)
one third of the 012 unit cell, and (c) the 01 unit cell [3. Mozer et a1. (1990), K.
Muraleedharan et al. (1995a)].

 

 

 
 
  

AI Atoms

:Ti or Nb Atoms

8 Ti Atoms

Nb Atoms

 

 

 

 

 

 

 

Orthorhombic

Figure 2.2. The similarities between the basal planes of the or; phase and the 02 phase
shown in an [001] projection.

20

2. Microstructural Evolution

2.1 Phase Field Ranges

Since the discovery of the O-phase, numerous studies have been performed on the
phase equilibria of the O-phase [Muraleedharan et al. (1995), Muraleedharan et al.
(1990), Muraleedharan et al. (1992a), Muraleedharan er al. (1992b), Muraleedharan et al.
(1995b), Bendersky et al. (1991), LA. Bendersky and WJ. Boettinger (1993)]. Several
studies have focused on the microstructures that result from different processing methods
[Li and Boehlert (2005), Boehlert (2000), Emura et al. (1996), Yolton and Beckman
(1995), Peng et al. (2001), Sagar et al. (1996), Narayana Murty and Nageswara Rao
(2000), Zhang et al. (2000)]. Since the alloys in this work were processed at subtransus
temperatures, which significantly influences the range of microstructures that can be
produced through post-processing heat treatment, an example of previously determined
phase equilibria will be given for subtransus-processed Ti-25Al-24Nb and Ti-23Al-27Nb
alloys [Boehlert er. al. (1999)]. Solutionizing heat treatments above the BCC-transus
(1030°C and 1070°C, for Ti-25Al-24Nb and Ti-23Al-27Nb, respectively) followed by
water quenching led to fully-82 microstructures containing large equiaxed grains. When
solution treated and water-quenched at temperatures between 1000°C and the B2 transus,

equiaxed a2+B2 microstructures resulted. The [1 l-20]ot2//[1-11]BZ and
(0001)a2//(011)B2 orientation relationship (OR) was observed between neighboring 82
and 012 grains, which is the classic Burger’s OR originally determined for the or and B
phases in zirconium [Burgers (1934)]. A narrow three phase O+B2+orz regime was found
to lie below the 012+B2 phase field over the temperature range of 975-1000°C. The

microstructures that resulted from solutionizing and quenching within this temperature

21

range contained equiaxed grains of the three phases, with the O-phase preferentially
forming as a rim around the or; phase. “Rim” O-phase surrounding the or; phase is a
typical feature observed in three-phase Ti-Al-Nb microstructures, and results fi'om the
following peritectoid phase transformation: BCC+a2 —) O [Banerjee et al. (1988),
Muraleedharan et al. (1992a), Gogia et al. (1992), Banerjee ( 1997)]. The “rim” of 0-
phase adds difficulty to the removal of the 012 phase fi'om the microstructure through heat
treatment by acting as a diffusion barrier. The O+BZ phase field for these alloys occurred
in the temperature range of 875—975°C. Microstructures containing equiaxed BZ and
equiaxed O grains, with no or; phase present, were produced after solutionizing and water
quenching in this phase field. The OR observed between the B2 and O grains is given by
[-lll]B2//[2-10] and (110)BZ//(001)O, which is the same OR observed for O platelets
within B2 grains [Banerjee et al. (1988), Muraleedharan et al. (1992b)]. Increasing the
solutionizing temperature within this phase field led to increased 32 phase volume
fraction within the microstructure at the expense of O-phase volume fiaction.

Three important microstructural observations arose from this phase equilibria
study. First, fine grained equiaxed microstructures are only obtained when hot working
is performed below the BCC-transus temperature. This was verified by solutionizing and
quenching samples above the BCC-transus, followed by aging in the O+BZ phase field.
After aging, the microstructures consisted of O or or; precipitates within the large retained
prior-BCC grains. Therefore, the only way to produce equiaxed fine-grained multiphase
microstructures is to perform hot-work below the BCC-transus. Second, the temperature
of 875°C constitutes a transition point for O-phase morphology. Solutionizing above

875°C produced microstructures containing equiaxed grains of both phases.

22

Solutionizing below 875°C resulted in Widmanstatten precipitation of O-phase plates
within the B2 grains. Widmanstatten precipitation occurs when plate, or even needle-
shaped precipitates grow in such a manner that they are aligned along specific
crystallographic planes or directions of the matrix crystals [Reed-Hill and Abbaschian
(1994)]. For the Ti-25Al-24Nb alloy, solutionizing at 875°C for 100 hours followed by
water quenching produced a fully-O microstructure. The fact that a fully-O
microstructure could be produced for Ti-25Al-24Nb and not Ti-23Al-27Nb under
identical heat treatment schedules illustrated the effect of increased Al content increasing
O-phase volume fractions [Boehlert et al. (1999)]. Third, a fully-lenticular O+B2
microstructure was observed in the Ti-25Al-24Nb alloy when the fully-equiaxed O
microstructure of this alloy was resolutionized at 900°C and water-quenched. This
illustrates that the B2 phase can form by Widmanstatten precipitation fiom a parent 0-
phase when the temperature is raised, as opposed to lowered, from the single-phase field.
The pseudobinary phase diagram first constructed by Bendersky and coworkers
[Bendersky et al. (1991)] and modified due to the work of the phase evolution study

described above [Boehlert et al. (1999)] is depicted in Figure 2.3.

23

  

 

 

 

 

\\
14m _ 293$ E [:1 _
DE - \ A
_ fl 2 s s a s a
'3‘” 00+ “-i, s E \ 3 s -
00+2 §:§\ s;
,0 A0 a t s \ s a
if." "(mm 3 E E \ E E B
:1 1100 -- ‘0+°‘2+Bz la B2\ CI --
I— 012+B2 ‘.
E O+a2+B23~FK
E O
E 900 — .3 —
r- A
700 - —
mu Ti21l1N_b TiNb

COMPOSITION

Figure 2.3. Pseudobinary diagram based on TiAl and TiNb, for Ti = 50 at%, with
differing ratios of Al:Nb, first developed by Bendersky and coworkers [Bendersky er al.
( 1991)] and later modified by Boehlert and coworkers [Boehlert et al. (1999)].

24

2.2 Phase Transformation Mechanisms

The BZ phase has been found to transform to the 012 and/or 0 phases through three
possible mechanisms.

The first mechanism is Widmanstatten precipitation of the (12 and/or 0 phases
from the parent BCC phase. For Ti-25Al-24Nb and Ti-23Al-27Nb alloys,
Widmanstatten precipitation of the O-phase from the parent BZ phase only occurred
below 875°C in the O+BZ phase field [Boehlert et al. 1999)]. Upon cooling fiom 975°C
to 875°C within the O+B2 phase field, the reduced volume fraction of B2 phase and
corresponding increased volume fraction of O—phase occurs through the growth of
equiaxed O grains at the expense of equiaxed 32 grains. Therefore, the change in
equilibrium volume fiaction requirements dictated by the change in temperature is
controlled by the grain grth of the constituent phases. Below 875°C, grain boundary
diffusion and grain growth kinetics are more sluggish, and Widmanstatten precipitation
of the O-phase becomes the dominant phase transformation mechanism. This
transformation mode requires lattice reconstruction along with thermally activated
diffusion of atoms across the O and BZ phase boundaries. High concentration gradients
and large lattice mismatches occur at the tips of the growing 0 platelets, and assist in
driving this phase transformation [Boehlert et al. (1999)].

The second phase transformation route that occurs in superuansus heat treated
microstructures is a composition invariant mechanism that rapidly transforms the parent
B2 into metastable oz for low Nb compositions or metastable O for high Nb
compositions [Muraleedharan et al. (1992b), Bendersky et al. (1991)]. A composition-

invariant transformation of the B2 phase to 0 phase after short aging times at low

25

temperatures has been observed; further aging resulted in the reprecipitation of B2 phase
platelets within the transformed BZ matrix in a Ti-24Al-16Nb alloy [Muraleedharan et al.
(1992b)] and a Ti-24Al-25Nb alloy [Bendersky et al. (1991)]. For Ti-25Al-24Nb and
Ti-23A1-27Nb alloys that were supertransus solutionized and quenched, followed by
aging at 650°C, martensitic-type O grains precipitated at transformed B2 grain
boundaries and proceeded to coarsen [Boehlert et al. (1999)]. This provided evidence of
this type of phase transformation occurring for near-stoichiometric TizAle alloys when
directly aged at 650°C after supertransus solutionizing and quenching.

Discontinuous cellular precipitation of the a2+BCC or the O+BCC phases is the
third transformation mode and it replaces fine intragranular matrix precipitation [Rowe
and Hall (1991), Rowe (1991b), Rowe and Larsen (1996), Boehlert et al. (1997a), Rowe
and Gigliotti (1990)]. When subtransus solutionized and water quenched, followed by
aging at 750°C, discontinuous precipitation of the B and 0 phases from the parent B2
phase was observed for Ti-25Al-24Nb and Ti-23Al-27Nb alloys [Boehlert et a1. (1999)].
Solutionizing at high temperature and quenching, followed by low temperature aging,
produced the cellular precipitation reaction for the alloys in the above-mentioned study.
The occurrence of this mechanism was suggested to be driven by the large compositional
gradient observed between the high temperature BZ phase and low temperature B phase.
The Al content in the B2 phase is typically 17 at% and greater, while the Al content in
the B phase is typically 12 at% and lower [Boehlert et al. (1999)]. When aged at 650°C
after being quenched from high in the O+B2 phase field, the degree of supersaturation of
the B2 phase is greatly increased when compared to aging at higher temperatures and/or

solutionizing at lower temperatures. Diffusion within the bulk of the B2 grains to

26

produce Widmanstatten O-phase is suggested to not be solely possible with such a steep
concentration gradient. Therefore, a cellular precipitation reaction observed to initiate at
O/BZ grain boundaries is driven by the steep concentration gradients within the BCC
phase and utilizes increased grain boundary diffusion at lower temperatures.

The first objective of the microstructure portion of this thesis was to quantify
microstructural features such as grain size, phase chemistry, phase volume fractions, and
phase morphology in order to construct microstructure-property relationships. After this
objective was met the temperature ranges of the respective phase fields were then
determined. The available pseudobinary phase diagram for alloys bridging the
compositional gaps between Ti-23Al-27Nb and Ti-12A1-38Nb was then modified. The
phase transformations that occurred in the alloys in this thesis were then determined. The
final objective of the microstructure portion of this thesis was to understand the effect of

boron additions on the microstructure of Ti-Al-Nb alloys.

3. Deformation Behavior of the Constituent Phases

The constituent phase with the most limited elongation-to-failure (8f) is the or;
phase, due to the inherently low number of available slip systems in its hexagonal
structure at both ambient and elevated temperature. Slip in hexagonal metals occurs most
easily on the (0001), {IO-10}, and {10-11} planes in the 1/3 <11-20>, or <a>, directions
which yields a total of three slip directions per unit cell, only two of which are
independent on each family of planes [Reed-Hill and Abbaschian (1994)]. This is
contrary to von Mises Criterion, which states that five independent slip systems must

operable in order for the grains in a polycrystalline metal to undergo a general plastic

27

shape change [Hull and Bacon (2004)]. While these potentially 6 slip systems are
available, none allow a shape change in the <c> direction, so that <c+a> slip is required,
which has a much larger Burger’s vector and hence, a much larger critical resolved shear
stress (CRSS). Alternatively, mechanical twinning can permit strains that change the
dimension of a crystal in the <c> direction, but they also have a higher CRSS. It has
been found that adjacent 112/012 grain boundaries provide preferential crack initiation sites
that lead to failure and are detrimental to both strength and ductility [Akkurt et al. (1991),
Gogia et al. (1992), Boehlert et al. (l997b)]. Therefore, to produce microstructures with
adequate elongation-to-failure, the or; phase is undesirable, especially when (12 grains are
located adjacent to one another.

Figure 2.4 depicts an illustration of the similarities and differences between the
possible slip vectors of dislocations in the or; and the 02 phases. Dislocations with the
possible Burger’s vectors depicted in Figure 2.4 are not perfect dislocations, and
therefore antiphase boundaries, self intrinsic stacking faults, and complex stacking faults
result from their motion. The 1/3 <11-20> dislocations that glide in the basal and
prismatic planes of the (12 phase become equivalent to the crystallographically distinct
[100] and 1/2 <110] dislocations in the O-phase. It is convenient for comparative
purposes to collectively refer to these dislocations as <a> dislocations. More specifically,
the [100] dislocations are often referred to as ‘a’ dislocations and the 1/2 <110] prismatic
dislocations are often referred to as ‘a*’ dislocations. A similar relationship also exists
between the 1/3 <11-26> a2 dislocations and the [102> and 1/2 <114] O dislocations. It
is convenient for comparative purposes to collectively refer to these pyramidal

dislocations as <c+al2> dislocations [Banerjee (1997)].

28

1 611156]

 

   

. Ti or Nb Atoms
.Al Atorm

 

Ti Atam

Al Atoms

 

 

 

Nb Atarrs

 

(b)

Figure 2.4. Possible Burger’s vectors in (a) the (12 phase containing Nb and (b) the 02
phase [Banerjee (1997)].

29

Although the 0:2 and 0 structures are similar, they exhibit different deformation
behavior, which was demonstrated in a study comparing the dislocation structures
observed in the rrricrostructures of an O-based Ti-26Al-21Nb alloy and an org-based Ti-
27Al-16Nb alloy [Banerjee (1995)]. The composition of the 012 phase in the Ti-27Al-
l6Nb alloy was Ti-27Al-10Nb, which at that time was the expected limit of Nb solid
solubility in Ti3Al. Recent results have shown higher Nb solubility with increased
nominal Nb content in Ti-23Al-27Nb and Ti-15Al-33Nb alloys [Boehlert er al. (1999),
Cowen and Boehlert (2006a)]. Cylindrical samples of the O alloy were deformed in
compression to strains of 1.5-2% at both room-temperature and 650°C, and samples of
the or; alloy were deformed in tension to a strain of 2% at room-temperature. Foils for
TEM analysis were prepared from sections cut at 45° to the stress axis. In the O alloy,
basal slip was found to occur for [100](001) and prismatic slip for 1/2 [110](l-10). Near-
screw oriented [100] dislocations were observed to locally cross-slip onto (041) planes,
rendering them relatively immobile on (001) planes, and cross-slip of 1/2 [110]
dislocations onto the basal plane was also observed. The [102] and 1/2 [114] pyramidal
dislocations were found to glide on {221} planes. At room temperature the <c+a/2>
dislocations in the O alloy were pinned by interactions with <a> dislocations, while these
locks were not observed after deformation at 650°C. In the or; alloy, the dislocation
structure after deformation was found to consist of junctions of the three types of 1/3
<11-20> dislocations. These three different types of <a> dislocations were observed to
glide extensively in the basal plane. This finding is significant because in both
stoichiometric and non-stoichiometric Ti3Al, extensive basal slip of <a> dislocations is

only observed when the stress axis is oriented at 45° to the basal plane and the Schmid

30

Factor is at its maximum [Inui et al. (1994), Minonishi (1991)]. Ifthe effect of having a
polycrystalline microstructure is ignored, this suggests that the critical resolved shear
stress for basal slip and the critical resolved shear stress for prismatic slip are closer in
magnitude for the high Nb containing (12 alloy [Banerjee (1995)]. Pure <c> dislocations
with a [0001] Burgers vector were not observed and <c+al2> dislocations were rarely
observed which suggests an extremely high critical resolved shear stress for the 1/3 <11-
26>{2-201} slip system as is also observed in Ti3Al [Inui et al. (1994), Minonishi
(1991)]. Banerjee concluded that the increased elongation-to-failure of the O-phase alloy
over the or; phase alloy was due to the substantial operation of <c+a/2> pyramidal slip on
{221} planes in the O-phase [Banerjee (1995)].

The BCC structure is characterized by four close-packed <111> directions and
the lack of a truly close-packed plane such as the octahedral plane of the face-centered
cubic (FCC) lattice or the basal plane of the hexagonally close-packed (HCP) lattice.
Slip in BCC metals occurs in the <111> directions, however the slip plane is not well
defined. The {110}, {112}, and {123} planes have all been identified as slip planes in
BCC crystals, which yields 48 slip systems [Reed-Hill and Abbaschian (1994)]. The B2

and B phases will tend to provide this alloy system with greater elongation-to—failure due

to their larger number of available slip systems when compared to the or; and 0 phases.

31

4. Creep Deformation Behavior

Since studying creep deformation behavior was the main focus of this thesis, a
concise review of the theoretical creep deformation mechanisms that are suggested to be
dominant during secondary creep deformation behavior will be presented first, followed
by a review of creep literature focused on Ti-Al-Nb alloys.

The creep deformation behavior of pure metals and single phase alloys has been
extensively researched, and at this point in time an adequate understanding of the various
microstructural features that control creep deformation is known. In particular, research
has been mainly focused on the deformation mechanisms controlling the secondary
steady-state creep behavior of pure metals and single phase alloys because the secondary
steady-state strain rate can be measured experimentally and used to suggest the dominant
deformation mechanism. The current lack of understanding in the creep deformation
behavior of metals lies in the area of multiphase alloys in which both phases plastically
deform during creep. The presence of more than one crystal structure, with a different
cherrristry, in the microstructure results in the different phases creeping at different rates,

under the same applied stress at the same temperature.

4.1 Nabarro-Herring Creep

The deformation of a metal due to the stress directed flow of vacancies through
the crystal lattice was proposed as a deformation mechanism by Nabarro [Nabarro
(1948)] and Herring [Herring (1950)]. When an external stress is applied, an excess of
vacancies accumulate along grain boundaries experiencing a tensile stress and a

corresponding depletion of vacancies occurs along grain boundaries experiencing a

32

compressive stress [Langdon (2000)]. Therefore, driven to restore the equilibrium
concentration of vacancies, a net flow of vacancies from grain boundaries in tension to
those in compression occurs through the lattice. A flux of atoms in the direction opposite
to vacancy flow arises, which Nabarro and Herring proposed would result in grain
elongation along the tensile axis, and a permanent creep strain. A standard form of the
Nabarro-Herring (N -H) equation for strain rate is given by:

° 100D L!)

g :

dsz (1)

where o is the applied stress, DL is the lattice diffusion coefficient, 0 is the vacancy

 

volume, d is the grain size, k is Boltzmann’s constant, and T is the absolute temperature
[Herring (1950)]. Nabarro-Herring deformation is believed to be the rate controlling
mechanism at high temperatures and relatively low stresses, i.e. o/G < 10“, where G is

the shear modulus [Dieter (1986)].

4.2 Coble Creep

Upon analyzing creep data for polycrystalline A1203, and realizing that oxygen
(the least mobile species) diffusion would be the rate controlling process over Al
diffusion, Coble proposed that the deformation process consisted of Al diffusion through
the lattice and oxygen along the grain boundaries [Coble (1963)]. This raised the
following question for Coble: “If oxygen diffusion through the grain boundaries is
required for deformation, why couldn’t grain boundary diffusion be the dominant
deformation process [Coble (1963)]7” Coble derived a model for creep deformation due

to vacancy diffusion through the grain boundaries, rather than through the crystal lattice.

33

The driving force for N-H creep is the same as that for Coble, where the maximum
change in vacancy concentration on boundaries perpendicular to the stress axis produces
lattice expansion parallel to the applied stress, and a net flow of vacancies occurs to
restore the equilibrium vacancy concentration [Coble (1963)]. The equation Coble
derived for the strain rate occurring from vacancy diffusion through the grain boundaries
is given by:

- 14801), W!)

8 : d3kT ‘2)

where 0' is the applied stress, D3 is the grain boundary diffusion coefficient, W is the

 

width of the grain boundary, (2 is the vacancy volume, (1 is the grain size, k is
Boltzmann’s constant, and T is the absolute temperature [Coble (1963)]. Coble creep is
believed to be the strain rate controlling process at stresses on the same order of
magnitude as N-H creep, but at lower temperatures [Dieter (1986)]. It is important to
note that both of the above mechanisms are highly dependent on the grain size of the

microstructure, with N-H creep scaling as (1'2 and Coble creep scaling as d'3.

4.3 Harper-Dom Creep

When plotting experimentally determined steady state strain rate values versus
applied stress on a log-log scale, a linear relationship between stress and strain rate
should arise when N-H and Coble creep deformation is occurring, as predicted by
equations (1) and (2). The mechanism of Harper-Dom (H-D) creep was first proposed by
Harper and Dom in 1957 due to the results of their creep studies at low stresses and high

temperatures on coarse-grained Al [Harper and Dorn (1957)]. They obtained a linear

34

relationship between stress and strain rate, but the strain rates they acquired were
significantly higher than those predicted by the N-H and Coble creep equations. Thus,
they concluded a different deformation mechanism was controlling the creep behavior of
A1 at low stresses and high temperatures. The equation given to describe H-D is as
follows:
' _ A HD D L b 0'
g __

[CT (3)

 

where Ann is a material constant, DL is the lattice diffusion coefficient, b is the Burger’s
vector, 0 is the applied stress, k is Boltzmann’s constant, and T is the absolute
temperature [Harper and Dorn (1957)].

The microstructural mechanism involved in the H-D deformation processed is
widely believed to be diffusion-controlled dislocation motion that occurs intragranularly
with the dislocation density remaining constant [Ruano et al. (1988)], but has not been
elucidated exactly. Proposed dominant mechanisms in which intragranular diffusion
controlled dislocation motion occurs during H-D creep are as follows: the climb of edge
dislocations under conditions of vacancy saturation [Langdon and P. Yavari ( 1982)],
cyclic changes in the equilibrium concentration of vacancies due to low amplitude
thermal cycling [Weertrnan and Blacic ( 1984)], and the coarsening of dislocation
networks when the glide and climb of edge dislocations is restricted [Ardell and Lee,
(1986)].

Unlike N-H and Coble creep, in theory H—D creep has no dependence on the grain
size of the microstructure; samples of different grain sizes should show no deviation in

strain rate when tested at the same stress and temperature. Owen and Langdon have

35

proposed three requirements that must be firlfilled in order to establish the H-D process:
(1) the creep stress exponent must equal unity, (2) the experimentally acquired creep rates
must be significantly faster than those predicted by the N-H equation and the Coble
equation (if the creep temperatures are lower and within the range where Coble creep
might be expected), and (3) identical creep rates must determined over a wide range of

grain sizes and may include experiments on single crystals [Owen and Langdon (1996)].

4.4 Dislocation Climb

Creep deformation due to the glide and vacancy aided climb of dislocations is
referred to as dislocation climb creep [Dieter (1986)]. Steady state creep occurs during
this process when the rate of strain hardening due to the glide of edge dislocations is
balanced by the annihilation of edge dislocations due to thermal recovery [Dieter (1986)].
It is noted that during this deformation process, the glide of the dislocations accounts for
almost all of the strain, but the rate controlling process is the diffusion of vacancies which
controls the speed with which the dislocations can surmount obstacles [Dieter (1986)].
The tendency for this deformation mechanism to be dominant increases with increasing

stress, and the equation that describes the strain rate dependence is:

8: Ba" exp(——— (4)

- __Q__app)
RT

where B is a constant, a is the applied stress, 11 is the creep stress exponent, Qapp is the

activation energy for creep (in kJ/mol) , R is the gas constant, and T is the absolute

temperature [Dieter (1986)]. In this case the Qapp value should be on the order of

magnitude of that for lattice self-diffusion, since the diffusion of vacancies through the

36

crystal lattice to allow climb is the rate controlling process. It is noted that in theory, this
mechanism should not display a grain size dependence. This dislocation creep process
dominates the elevated temperature response of many engineering alloys at temperatures

greater than 0.5TM, where TM is the melting temperature [Hertzberg (1996)].

4.5 Grain Boundary Sliding

At elevated temperatures, the grains in a polycrystalline metal are able to move
relative to one another, and the phenomenon of this shearing process is referred to as
grain boundary sliding [Dieter (1986)]. Grain boundary sliding is typically thought of as
a process which accommodates deformation due to a diffusional- or dislocation-
controlled deformation mechanism [Dieter (1986), Hertzberg (1996), Reed-Hill and
Abbaschian (1994)]. When N-H or Coble creep occur due to the stress directed diffusion
of atoms from grain boundaries in compression to grain boundaries in tension, the grain
boundaries will separate from one another and sliding will occur in order to maintain
grain contiguity [Hertzberg (1996)]. Intragranular dislocation climb within two adjacent
grains can lead to deformation occurring by the grains shearing past one another by
sliding, rather than deformation by grain elongation [Hertzberg (1996)].

Grain boundary sliding can be observed on the sample surface by first
metallographically preparing a specimen to reveal the grain structure and scribing a grid
of fiducial marks on the sample surface [Reed-Hill and Abbaschian (1994)]. The sample
is then creep tested at the appropriate stress and temperature where grain boundary
sliding is suggested to occur, in an appropriate environment to preserve the integrity of

the surface, and if grain boundary sliding occurs the offset in the fiducial marks can be

37

measured. The main flaw in this technique is that only surface measurements can be
made, and there is uncertainty in whether the interior of sample is deforming in the same
manner. Rachinger made an attempt to estimate the contribution of strain due to grain
boundary sliding to the total strain in the sample by noting the shape and aspect ratio of
the grains before and after creep testing [Rachinger (1952-1953)]. Rachinger proposed
that if all the deformation occurs due to grain boundary sliding, and not slip within the
grains, then the grain size and aspect ratio should remain the same after testing. This type
of grain boundary sliding has been referred to as Rachinger Sliding.

Decreasing the grain size in a given specimen should promote grain boundary
sliding because a smaller grain size should allow grains to slide past one another with
more ease, in addition to the simple fact that decreasing the grain size increases the total

number of grain boundaries in the specimen.

4.6 Creep Stress Exponents and Apparent Activation Energies

The major focus of the creep behavior study of this thesis was to determine the
deformation mechanisms and the most important microstructural feature that controls the
secondary creep behavior of Ti-Al-Nb alloys. The values of creep stress exponents (n)
and apparent activation energies (Qapp) give insight into the deformation mechanisms
that occur during the secondary stage of creep. These parameters are typically calculated
using equation (4) after the minimum creep rate is determined directly from strain versus
time data collected from actual creep experiments. The n values are determined by
applying a power law curve fit to the log-log plot of minimum creep rate versus stress

and the Qapp values are determined from the slope of the least squares linear curve fit to

38

the plot of the natural log of the minimum creep rate versus III. The n and Qapp values
determined are used to predict the active secondary-creep deformation mechanisms.

Based on creep theories for pure metals and disordered alloys, n values close to
unity suggest stress directed atomic diffusion of vacancies as the deformation mechanism
[Evans and Wilshire (1985)]. This is evident upon examining the linear relationship
between applied stress and minimum creep rate portrayed in equations (1), (2), and (3)
for Nabarro-Herring, Coble, and Harper-Dom creep, respectively. Activation energies
determined when Nabarro-Herring and Harper-Dom creep deformation are active are
typically close to those for lattice self-diffusion. When Coble creep is the dominant
deformation mechanism, activation energies are typically 0.5-0.6 of those for lattice self-
diffusion, due to diffusion of vacancies through the grain boundaries.

Grain boundary sliding deformation should have Qapp values typical of lattice-
self diffusion and n values (approximately 2) between those for difiusion and dislocation
controlled creep [Garofalo (1965), Hertzberg (1996)].

For n values over the range of 3.5-6, deformation typically occurs by a dislocation
climb mechanism [Hertzberg (1996), Evans and Wilshire (1985)]. The activation energy
for dislocation climb is associated with lattice self diffusion.

Therefore, based on previous secondary creep theories of pure metals and
disordered alloys, by combining n and Qapp values the deformation mechanism at a
particular stress range and temperature may be suggested. A change in 11 value typically
corresponds to a change in deformation mechanism. Therefore, by determining n values
over a range of stresses at a constant temperature, the stress ranges where particular

deformation mechanisms are active at that temperature can be estimated. Figure 2.5

39

illustrates the dependence of n on applied stress. In several cases alloys have also
exhibited similar secondary creep behavior as pure metals [Eylon et al. (1976), Miller et
al. (1987), Mishra and Banerjee (1990), Mishra et al. (1995), Worth et al. (1995),
Hertzberg (1996), Evans and Wilshire (1985), Malakondaiah and Rao (1981), Nandy et
al. (1993), Boehlert et al. (1999), Boehlert (1999), Takeuchi and Argon (1976),
Mukherjee et al. (1969), Woodard et al. (1996), Smith et al. (1993), Woodard and
Pollock (1997), Hayes (1996), Rowe and Larsen (1996), Nandy et al. (1995), Boehlert
and Bingert (2001)], and therefore the secondary creep theories based on pure metals
have been applied to alloys.

It is extremely important to note that n and Qapp values alone are not sufficient
for conclusively verifying the dominant secondary creep deformation mechanism. In
order to state that grain boundary sliding is the dominant secondary creep deformation
mechanism, surface observations of metallographically prepared and fiducially marked
samples that have been creep tested in a vacuum or inert environment are necessary.
TEM observations of post creep deformed microstructures are typically performed in
order to observe dislocation density and activity within the microstructures. Combining n
and Qapp values and surface and bulk deformation observations with dislocation activity
observations provides a thorough methodology for suggesting the dominant secondary
creep mechanism over an applied stress range at a given creep temperature. It will be
shown in Chapter 4 of this thesis that the in—situ creep testing methodology developed in
this thesis allowed for direct observation of grain boundary cracking and grain boundary

sliding as a function of creep displacement.

4o

 

High Stress
n=5 dislocation controlled

_ \ _.

/.

Low Stress
n=1 diffusion controlled

1

Steady-State Creep Rate (log scale)

 

 

 

Stress (log scale)

Figure 2.5. Schematic illustrating the creep stress exponent values for the diffusion- and
dislocation-controlled creep regimes for pure metals.

41

4.7. Creep Behavior of Ti-Al-Nb Alloys
4.7.1 Creep Behavior of az-based Ti—AI-Nb Alloys

The creep deformation behavior of an org-based Ti-24Al-11Nb alloy was studied
by Mishra and Banerjee [Mishra and Banerjee (1990)]. The Ti-24Al-11Nb alloy was
heat treated to produce three microstructures which consisted of 90% equiaxed 02-10%
BCC, 40% equiaxed org-60% transformed BCC, and 100% transformed BCC. Over the
stress range of 30-400 MPa at 650°C, all three microstructures displayed a transition in 11
value from one in the low stress regime, to values geater than unity in the high stress
regime. Qapp values of 107 kJ/mol and 120 kJ/mol were obtained in the n=1 stress
regime for the 40% equiaxed a2 and 100% transformed BCC microstructures,
respectively. A Qapp value of 260 kJ/mol was obtained for the 40% equiaxed a2
microstructure in the high-stress regime.

Based on n and Qapp values, Coble creep was suggested to be dominant in the
low stress regime, and dislocation climb in the high stress regime. In order to support the
suggestion of Coble creep deformation, post creep-deformed specimens were examined
by TEM. A lack of sigrificant dislocation activity within the or; gains was observed in
samples creep tested in the suggested Coble creep regime. In addition, the activation
energies obtained were close to those obtained for grain boundary diffusional creep in
pure Ti (104-121 kJ/mol) [Malakondaiah and Rao (1981)] and also close to those
measured for gain boundary self-diffusion in zirconium (112 kJ/mol) [I.M. Bernstein
(1967)]. Using the gain size of the 90% equiaxed a2 microstructure, the measured
strain rates and activation energies, and data available for pure Ti, an estimate for D0 in

the Coble creep equation [Coble ( 1963)] resulted in a value of 3.0x10'7 mzs". The D0

42

value for pure Ti is 7.51110'5 mzs’l [Malakondaiah and Rao (1981)]. The authors
concluded that in the suggested Coble creep regime the Ti-24Al-11Nb alloy has greater
creep resistance due to the lower gain boundary diffusion coefficient frequency factor.
This decrease in gain boundary diffusion is attributed to the ordered crystal structures
present in the Ti-24Al-11Nb alloy.

In the high stress regime, the creep stress exponents and Qapp values suggested a
dislocation climb controlled deformation mechanism. Table I gives 11 and Qapp values
determined for the Ti-24Al-11Nb alloy [Mishra and Banerjee (1990)], (12 alloys
[Mendiratta and HA. Lipsitt (1980)], and pure Ti [Malakondaiah and Rao (1981),

Conrad et al. (1973)].

Table I. Dislocation Climb Controlled Creep Parameters

 

 

 

 

 

 

 

Creep
Parameter Ti-24AI-11Nb or; or-Ti
242 [Conrad et al. (1973)]
Qapp 285 (Ti3Al+5Nb) 241 [Malakondaiah and Rao
(kJ/mol) 260 206 (Ti3Al) (1981)]
4.5 [Conrad er al. (1973)]
6 (Ti3Al+5Nb) 4.3 [Malakondaiah and Rao
11 5 4.3 (Ti3Al) (1981)]

 

The authors inferred that the increased creep resistance of the Ti-24Al-11Nb alloy
is not due to an increase in activation energy for bulk diffusion, but arises again from a
lower fi'equency factor for diffusion that they associate with the alloy’s ordered crystal
structure. TEM observation of samples deformed within the suggested climb-controlled
regime revealed a dislocation structure dominated by pile-ups of dislocations, and the
density of pile-ups was observed to be highest at the primary org-transformed BCC

interfaces.

43

 

4.7.2 Creep Behavior of the O—phase in Ti-Al-Nb alloys

The creep behavior of the O-phase in a Ti-27Al-21Nb alloy was initially
characterized by Nandy, Mishra, and Banerjee [Nandy et al. (1993)]. After solution
treatment and aging, their alloy consisted of equiaxed fully-O gains with an average
gain size of 88 um. Constant-load compression creep tests were conducted in the
temperature range of 650—750°C and the stress range of 240-500 MPa. Creep stress
exponent values ranging between 5.7 and 7.5 and a Qapp value of 340 kJ/mol were
obtained over the examined stress and temperature range. Therefore, dislocation climb
was suggested as the dominant deformation mechanism.

The effect of Al content on the creep behavior of the O-phase was evaluated after
confirming that the O-phase possessed desirable creep resistance [Nandy er al. (1995)].
Ti-24Al-15Nb and Ti-26Al-15Nb alloys were produced and creep tested in compression
in the temperature range of 700-750°C and the strain rate range of 108-105 8". Stress
exponent values (approximately 4) and a Qapp value (3 70 kJ/mol) were obtained for both
alloys over the stress and temperature range studied. The higher Al containing alloy
possessed steady state strain rates an order of magritude lower than the low A] containing
alloy. This result is interesting due to the fact that increasing the Al content sigrificantly
improved the creep resistance, but the fact that the n and Qapp values did not change
suggests the deformation mechanism was the same for both alloys. The microstructure of
both alloys was very similar and consisted of O-phase platelets surrounded by films of
retained B2 phase. The average O-phase platelet size was 2 urn for both alloys. The Ti-

24Al-15Nb alloy possessed 5% BZ phase by volume while the Ti-26Al-15Nb alloy

possessed 1% 32 phase by volume. Three possibilities were given for the proposed

increase in creep resistance due to the higher Al concentration: (1) microstructural
differences, (2) an effect of Al on the fi‘equency factor for diffusion (Do), and (3) an
effect of Al on the term A in the power-law dislocation climb controlled equation for
steady state strain rate. A four percent difference in B2 phase volume fraction was the
only difference in measurable microstructural features, so choice (1) was ruled out. This
conclusion was further supported by a volume fiaction weighted rule of mixtures
calculation under the assumption that the O-phase takes all the load. The calculation
predicted the strain rate of the Ti-24Al-15Nb alloy to be 1.3 times geater, when in
actuality is was an order of magnitude geater. The authors conclude that the increased
Al content increased the degee of order in the O-phase. This would impose a
requirement of geater correlation atom jumps in diffusion, and therefore lower the
diffusion coefficients and affect the pre-exponential ADo term in the power-law equation
for creep [Nandy et al. (1995)].

Since the Ti-27Al-21Nb alloy exhibited superior creep properties to the Nb—lean
org-based Ti-24Al-11Nb alloy [Mishra and Banerjee (1990)], a study was conducted
evaluating the effect of Nb content on creep behavior [Nandy and Banerjee (2000)],
including microstructural examination of post-creep deformed samples. Three alloys
were produced with a constant Al concentration of 27 at% and varying Nb
concentrations: Ti-27Al-18Nb, Ti-27Al-20Nb, and Ti-27Al-25Nb alloys. The gain size
of these alloys ranged between 157-166 um. Compression creep tests were conducted in
the temperature range of 700-750°C and the strain rate range of 109-107 s". This strain
rate regime corresponds with the applied stress regime evaluated in the previous study

[Nandy et al. (1993)]. The creep behavior of these alloys corresponded to that of class II

45

pure metals [Takeuchi and Argon (1976), Mukherjee er al. (1969)] in that the strain rate
decreased with increased strain after stress jumps and well-defined steady state behavior
was observed at the stress levels and temperatures investigated. Creep stress exponents
ranged between 5-7 and Qapp values ranged between 294-327 kJ/mol for the three alloys
investigated. At the time this study was performed no diffusion data was available for the
O-phase, and to the author’s knowledge this data is still unavailable to date. Diffusion

data is available for Ti3Al and this data is given in Table II [J . Rusing and C. Herzig

 

 

 

 

(1996)].
Table II. Self-Diffusion and Inter-Diffusion Data for Ti3Al
Element 1), (11129.4) Q (kJ/mol)
r1 in mm 2.44x10'5 288
Al in Ti3Al 2.32x10’l 374
Ti3Al 1x10'5 312

 

 

 

 

 

Comparing the Qapp values determined from the creep experiments to the Qapp values
available (See Tables I and 11) suggests that the Qapp values determined for the O-phase
are consistent with lattice self-diffusion.

TEM observations were made from foils taken from Ti-27Al-25Nb samples creep
tested well into the steady state regime under testing conditions of 300 MPa/725°C. The
dislocation structure was observed to consist of a 3-D network of attractive junctions of
[100], 1/2 <110], and [001] dislocations. No pyramidal dislocations were observed. The
size of a dislocation network is refined when the applied stress causes dislocations to
escape from pinning points at the network nodes and then interact to from attractive
junctions. Glide and climb of the network dislocations causes coarsening of the network,

which leads to a decrease in the dislocation density. Steady state is achieved when

46

hardening due to network refinement is balanced by recovery due to network coarsening
[Mishra and Banerjee (1990)].

Combining n and Qapp values, along with dislocation observations led to the
conclusion that dislocation climb is the rate controlling process during the steady state
creep deformation of these alloys. The varying Nb content in the three alloys
investigated produced no effect on the steady-state creep behavior. Similar n and Qapp
values obtained from the three different alloys also suggest that varying the Nb content

also does not have an effect on the dominant deformation mechanism.

4.7.3 Creep Behavior of the BCC phase in Ti-Al-N b alloys

To the author’s knowledge no creep studies have been performed on a fully-BCC
Ti-Al-Nb alloy at 650°C. This is due to the fact that the O+BCC phase field exists at
650°C and alloys with a low enough Al composition to remain fully-BCC at this
temperature have not been produced [Boehlert et al. (1999)]. The second lowest Al
containing Ti-Al-Nb alloy that has been studied to date is Ti-12Al-38Nb, and 28% 0
phase precipitated within the rrricrostructure after direct aging at 650°C for 55 hours
followed by water quenching [Boehlert (1999)]. In this work, a Ti-5Al-45Nb alloy was
produced, and was verified to remain fully-BCC at 650°C. Therefore, the constituent
BCC phase creep behavior at 650°C has been characterized and will be presented in the

subsequent Chapters of this thesis.

47

4.7.4 Creep Behavior of O+BCC Ti-Al-Nb alloys

The creep behavior of Ti-25Al-24Nb, Ti-25Al-23Nb, Ti-23Al-27Nb, and Ti-
12Al-38Nb O+BCC alloys was investigated by Boehlert and Miracle [Boehlert and
Miracle (1999)]. Table III displays 11 and Qapp values determined for several different
Ti-Al-Nb alloys over a stress range of 30 to 660 MPa and a temperature range of 650 to
760°C. The values given in Table III [Boehlert and Miracle (1999)] show indications of
Coble creep at low stresses, gain boundary sliding at intermediate stresses, and
dislocation climb at high stresses, based on the creep stress exponents and apparent
activation energies. The objective of their study was to understand creep mechanisms
occurring in these alloys with an emphasis on obtaining microstructural observations that
support the suggested deformation mechanisms, in particular verification of surface gain
boundary sliding. The effect of tension versus compression creep testing was evaluated,
along with relative effects of microstructmal instability.

Constant load tensile creep experiments were conducted in the stress range of 15-
450 MPa and the temperature range of 650-760°C. The creep strain versus time behavior
for all the alloys in this study resembled that for pure metals and alloys [Evans and
Wilshire (1985)].

The effect of microstructural instability on creep behavior was evaluated by
comparing the creep strain versus time behavior under the same testing conditions for Ti-
23Al-27Nb samples that had been aged at 650°C prior to creep testing versus Ti-23Al-
27Nb samples that were solutionized and water quenched without aging. Aging resulted
in lower primary creep rates and primary creep strains half those of the unaged samples.

During the secondary stage of creep this effect lessened, and the unaged samples

43

produced a minimum creep rate 1.3 times higher than the aged samples. Selected
samples were unloaded after 44 hours and 66 hours of creep deformation in order to
determine the effect of load on microstructural stability. No siglificant differences in
measured phase volume fractions and gain sizes were obtained for the creep deformed

samples compared to samples heat treated for the same amount of time without load.

Therefore, the effect of load on microstructural stability was deemed negligible.

Table 111. Various Orthorhombic Alloy CreeLParameters

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Testing Conditions off, Qapp,
Alloy Composition Heat Treatment MPa,°C n kJ/mol
Ti-21Al-21Nb
[Woodard et al. (1996)] NA 69-172/650 vacuum 2.4 NA
Ti-2 lAl-22Nb
[Smith et al. (1993)] NA 30-90/650-760 air 1.4 NA
Ti-21Al-22Nb
[Smith et al. (1993)] NA 90-172/650-760 air 8.2 197
Ti-22AI-23Nb
[Woodard and Pollock (I99fl1 NA 69-110/650 air + vacuum 1.3 187
Ti-22Al-23Nb
[Hayes (1996)] 996°C/lh/Ag 69-172/650-760 air + argon 2.8 327
Ti-22Al-27Nb
[Rowe and Larsen (1996)] 815°C/ 1h 310-380/650 argtl 5.3 NA
Ti-22Al-27Nb 1090°C/ 1 Wm +
[Rowe and Larsen (1996)] 815°C/1h 175-310/650 argon 5.3 NA
Ti-24AI-16Nb 1150°C/1h/CC +
[Nandy er al. (1995)] 750°C/24h/AQ 150-540/700-750 air 4.24.3 371
Ti-25Al-23Nb
[Rowe and Larsen (1996)] 815°C/lh 175-310/650 m 2.8 NA
Ti-25Al-23Nb 1065°C/1h/Ar +
[Rowe and Larsen (1996)] 815°C/lh 310-380/650 argm 2.8 NA
Ti-27Al-16Nb 1 170°C/ lh/CC +
[Nandy et al. (1995)] 750°C/24h/AQ 240—660/700-750 air 4.2-4.3 376
Ti-27Al-21Nb l 170°C/OQ +
[Nandy et al. 0993)] 900°C/AQ 240-500/650-750 air 5-6 340

 

 

AQ: air quench, 0Q: oil quench, Ar: cooled in static Ar gas, CC: control cool at 2.5°C/s, NA: not
available

Tensile and compressive creep testing at 250 MPa and 650°C produced almost
identical creep strain versus time behavior for a Ti-23Al-27Nb sample. The minimum

creep rate in tension was 1.50x10‘6 3'1 versus 1.13x10'6 8'1 in compression.

49

For the supertransus heat treated Ti-23Al-27Nb microstructures, with a gain size
of 177 um, three distinct n values were obtained over the stress range of 50-450 MPa at
650°C. The stress range of 50-172 MPa is characterized by a 11 value of 1.2 and a Qapp
value of 171 kJ/mol was obtained at 50 MPa over the temperature range of 705-760°C.
An 11 value of 2.0 was obtained from 172-318 MPa, and above 318 MPa a transition to a
11 value of 3.7 occurred.

The subtransus Ti-23Al-27Nb microstructures both displayed constant 11 values
over the stress range studied at 650°C and this suggests that one deformation mechanism
is dominant for these microstructures. The 3.8 um gain size microstructure produced a 11
value of 2.4 from 15-352 MPa and the 8.7 pm gain size microstructure produced a 11
value of 2.0 from 50-172 MPa. A Qapp value of 265 kJ/mol was obtained at 50 MPa for
both the subtransus microstructures over the temperature range of 650—764°C. Therefore,
decreasing the gain size in this material widened the stress range where gain boundary
sliding is suggested to be the dominant deformation mechanism, in addition to increasing
the measured minimum creep rate up to approximately two orders of magritude under the
same testing condition. The minimum creep rates at 172 MPa/650°C were 6.71x10'7 3",
9.93x10’8 3", and 4.27x10'9 5'1 for the 3.8 pm, 8.7 um, and 177 um gain size
microstructures, respectively. The creep behavior of the subtransus heat treated Ti-23Al-
27Nb alloy clearly portrays how the microstructure dominates the creep behavior for a
single composition.

Surface observations made from Ti-23Al-27Nb samples that were polished to a
metallogaphic finish, fiducially marked, and creep tested in vacuum revealed that gain

boundary sliding of the equiaxed gains was sigrificantly contributing to the macroscopic

50

creep strain for stresses within the n=2 regime. Local measurement of offset
displacements yielded strains as high as 6% for samples that were deformed to a total
creep strain of 9.8%. Surface gain boundary sliding was accommodated by gain
boundary cracking, and these cracks preferentially occurred at equiaxed O-equiaxed O
boundaries. Cracking within the bulk of the sample was also examined and occurred to a
lesser extent when compared to surface cracking.

TEM investigations of Ti-23Al-27Nb samples deformed at stresses below 318
MPa in the n=2 regime showed little evidence of dislocation activity within equiaxed 0
gain interiors. In addition, surface slip traces within equiaxed O gains were sparse,
even after being deformed to geater than 6% creep strain. Ti-23Al-27Nb samples
deformed in the high stress n 2 3.5 regime had a much higher dislocation density than
those deformed in the low-to-interrnediate stress regime. Dislocations piled up
preferentially at O-phase laths located near prior-BCC boundaries.

The Ti-25Al-24Nb and Ti-25Al-23Nb alloys displayed a similar trend as the Ti-
23Al-27Nb alloy with respect to gain size, in that the finer grained material exhibited
much larger minimum creep rates under the same testing conditions. Although 11 values
were not determined, a Qapp value of 311 kJ/mol was obtained for the Ti-25Al-23Nb
alloy at 50 MPa over the temperature range of 650-760°C.

The Ti-l2Al-38Nb alloy was creep tested in the stress range of 50-172 MPa and
the temperature range of 650-705°C. The minimum creep rate was found to decrease
with increasing gain size up to 138 um. Samples with a 337 um gain size displayed
similar minimum creep rates as samples with a 138 um gain size, which implies that

above 138 pm, the prior-BCC gain size is not influencing the minimum creep rate. All

51

the samples contained O-phase volume fiactions of approximately 30 percent. Creep
stress exponents fell in the range of 1.6-2.0 for the stress range of 50 to 100-135 MPa and
activation energies between 127-178 kJ/mol were obtained at 50 MPa fi'om 650-705°C.
Above 100-135 MPa at 650°C, 11 values geater than or equal to 3.5 were obtained and a
Qapp value of 256 kJ/mol was determined at 123 MPa from 650-760°C. So, based on n
and Qapp values, the Ti-l2Al-38Nb alloy was suggested to be deforming by gain
boundary sliding in the low stress regime and dislocation climb in the high stress regime.
At a stress of 50 MPa, a low Qapp value coupled with a stress exponent between 1 and 2
indicated Coble creep characteristics.

TEM investigations of the finest gained Ti-12Al-38Nb microstructure creep
tested at 50 MPa and 650°C revealed very few creep dislocations. A geater dislocation
density was observed within samples tested in the high stress regime. Dislocations were
observed to accumulate more often in the BCC phase than the O-phase, and pile ups of
dislocations were frequently observed at O-phase platelets.

F ully-O microstructures have displayed superior creep properties [Nandy et al.
(1993), Nandy et al. (1995)], so it follows that increasing the O-phase volume fraction in
Ti-Al-Nb alloys should lead to an increase in creep resistance. Increasing O-phase
volume fiaction does not always increase creep resistance, though. Supertransus Ti-
l2Al-38Nb microstructures displayed similar creep resistance to the Ti-23Al-27Nb, Ti-
25Al-24Nb, and Ti-25Al-23Nb microstructures even though they possessed only 30
volume percent O-phase. Irrespective of alloy chemistry and phase volume fraction, the
minimum creep rate was proportional to the inverse squared of the average gain size, for

gain sizes up to 177 um. Supertransus solutionized, large gained microstructures

52

displayed the best creep resistance over the wide compositional range studied,
irrespective of phase structure and phase volume fiaction. For intermediate stresses and
temperatures the equiaxed gain size of Ti-Al-Nb alloys appears to be the dominant
microstructural feature governing the creep behavior of these multiphase alloys.

Table IV summarizes the relevant literature providing 11 and Qapp values for Ti-
Al-Nb alloys, at the time this study commenced. Three stress regimes are clear, which
correspond to n=1, n=2, and n 2 3.5 regimes. In both the low and high stress regimes the
O alloys have the highest activation energies. This increase in apparent activation energy
for creep has been attributed to the ordered intermetallic constituent phases present. In
particular, the geater Nb concentrations present for O alloys are believed to geatly
decrease both lattice and gain boundary diffusion kinetics and hence increase the
activation energy required for diffusion.

Studying the creep deformation behavior of Ti-Al-Nb alloys was the main focus
of this thesis, therefore several objective were set forth to accomplish. The first goal was
to measure creep strain rates as a function of stress, temperature, and rrricrostructure over
the stress range of 10-400 MPa and 650-710°C. After this data was obtained creep stress
exponent values and apparent activation energies based on the measured creep strain rates
were calculated. These values were then compared to those obtained in a similar matter
for other Ti-Al-Nb alloys found in the literature. Accomplishing these objectives allowed
performance evaluation and the suggestion of dominant secondary-creep deformation
mechanisms. The next objective was to identify the dominant secondary creep
mechanisms as a function of stress, temperature, and strain rate using SEM deformation

observations and in-situ SEM experiments. Direct microstructural evidence of the

53

evolution of creep deformation on the sample surface as a function of displacement was
acquired through carefirlly selected in-situ creep experiments. A particular emphasis was
placed on observing gain boundary sliding deformation. Microstructure-creep
relationships were then constructed by completing the three aforementioned objectives.
The effect of boron additions on the creep behavior was then evaluated. Minimum creep
rates were determined as a function of stress at 650°C for the Ti-5Al-45Nb alloy in order

to determine the creep behavior of the BCC-phase for modeling purposes.

54

Table IV. Creep Parameters for or-Ti, org-based, and O-based Alloys

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Test Conditions Qapp,
Alloy Composition Heat Treatment c(MPa)/l‘emp_(°g n kJ/mol
Low-Stress Regime
Ti or
[Malakondaiah and Rao (1981)] various 1-2/550-865 1 104-121
Ti-24AI-11Nb a;
[Mishra and Banerjee (1990)] various 50-100/575-725, air 1 107-120
Ti—21Al-22Nb
[Smith et al. (1993)] an 30-90/650-760, air 1.4 na
Ti-22Al—23Nb
[Woodard and Pollock (1997)] na 69-110/650, air+vacuum 1.3 187
TizAle lO90°C/0.5h/WQ
[Boehlert and Miracle (1999)] +650°C/112h/WQ 50-172/650-760, air 1.2 171
Ti-12Al-38Nb
[Boehlert and Miraclell999)] various 50—135/650-705, air 1.6-1.9 127-178
Intermediate-Stress [Legime
Ti-22Al-23Nb 69-172/650-760,
[Hayes (1996)] 996°C/1hr/AQ air+aggon 2.8 327
Ti-25Al-23Nb
[Rowe and Larsen (1996)] 815°C/1hr 175-310/650, argon 2.8 na
TizAle 50-352/650-760,
[Boehlert and Miracle (1999)] various air+vacuum 1.8-2.3 265
Ti-12Al-38Nb 1200°C/5h/WQ 50-135/650-705
[Boehlert and Miracle (1999)] + 650°C/53h/WQ air+vacuum 1.9 256
Ti-26Al-27Nb
[Boehlert and Bmfljzoom IFZM Processed 172-317/650-760 2.3 na
High—Stress Iggime
Ti 01
[Malakondaiah and Rao (1981)] various 1-2/550-865 4.3 241
TI3AI a;
[Mendiratta and Lipsitt (1980)] lOOO°Cl4h/FC 138-312/650-800 4.3 206
Ti-24Al-11Nb a;
[Mishra and Banerjee Q9901] various 100-400/575-725, air 5 260
Ti-27Al-21Nb ll70°C/OQ 240-500/650-750, air
[Nandy et a1. (1993)] + 900°C/AQ 5-6 340
Ti-24A1-16Nb l 150°C/2.5°C/s
[Nandy et al. (1995)] + 750°C/AQ 150-540/700-750, air 4.2-4.3 37]
Ti-27Al-16Nb l l70°C/2.5°C/s
[Nandy et a1. (1995)] + 750°C/AQ 240-660/700-750, air 4.2-4.3 376
Ti-22Al-27Nb
[Rowe and Larsen (1996)] 815°C/ 1hr 310-380/650, argon 5.3 na
TizAle
[Boehlert and Miracle (1999)] various 317-442/650-760, air 3.7-5.1 346
Ti-12Al-38Nb
JBoehler-t and Miracle (1999)]] various 135-172/650-705, air 3.5-7.2 na
Ti-26Al-27Nb
jBoehlert and Bilgert (2001)] IFZM Processed 172-317/650-760 5.1 346

 

AQ: air quench, 0Q: oil quench, F C: furnace cooled, Ar: cooled in static argon gas; na: not available;
IFZM: Induction Float Zone Melting

55

 

5. Tensile Deformation Behavior

A wide range of tensile properties are attainable by varying the microstructure of
Ti-Al-Nb alloys through heat treatment and changes in composition. Examples of the
range in tensile properties determined at RT for various Ti-Al-Nb alloys are given in
Table V [Boehlert (2001)] and Table VI [Boehlert (2001)]. The tensile properties of Ti-
Al-Nb alloys are highly dependent upon phase composition, gain size, and phase volume
fraction.

For the BCC phase in particular, the elastic modulus, strength, and ductility are
highly dependent on chemistry. Increasing the Al content in the BCC phase favors the
BZ crystal structure, and this results in increased values of strength and elastic modulus,
with a simultaneous decrease in elongation-to-failure. This is evidenced by comparing
the tensile properties of the fully B2 Ti-25Al-24Nb alloy to those of the fully B Ti-12Al-
38Nb alloy (See Table VI [Boehlert (2001)]). The single phase B2 microstructures
exhibited mixed transganular and interganular fracture. Wavy slip, ductile dirnpling,
and cup-and-cone fracture were not observed for the B2 microstructures, but were readily
observed in the single phase B Ti-12Al-28Nb microstructures [Boehlert (2001)].

For the Ti-12Al-38Nb alloy, increasing the gain size from 33 pm to 138 um
resulted in a decrease in yield strength of 18% with elongation-to-failure relatively
unaffected. Interestingly, the largest gained (337 um) Ti-12Al-3 8Nb rrricrostructure had
the highest yield strength and lowest elongation-to-failure, and this is suggested to be a
result of inhomogeneous slip due to the large gain size [Banerjee (1976)]. Therefore, the
advantage of increased elongation-to-failure with decreased Al content in the B phase is

compromised at large gain sizes [Boehlert (2001)].

56

Single phase Ti-25Al-24Nb orthorhombic microstructures displayed RT
elongation-to-failure values 5 1% and interganular fiacture, however, planar slip traces
were observed on the sample surface after deformation [Boehlert (2001)].

Ti-25Al-24Nb microstructures containing both the BCC and 0 phases exhibited
the best combination of RT tensile properties. The increase in strength over those of the
individual constituent phase microstructures is due to the change in resulting
microstructure. The two phase O+BZ microstructures had a siglificantly finer gain size
which resulted in their geater strengths. The elongation-to-failure of the two-phase

microstructures was also geater, when the microstructure contained more than 20% B2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

phase.
Table V. RT Tensile Properties for Various Ti-Al-Nb Alloys
0.2% YS,
Alloy Composition Heat Treatment, °C MPa UTS, MPa at, %
Ti-l 5Al-45Nb
[Austin et al. (1993)] 1050°C/4h + 800°C/24h 865 924 15.1
Ti-22Al-23Nb 1050°C/2h +
[Smith et al. (1995)]] 815°C/8h/FC 836 1111 14.8
Ti-22Al-24Nb
[Rowe (1993)] 815°C/4h 1257 1350 3.6
Ti-22Al-25Nb 1000°C/ l h/Ar +
Bowe et al. (1991)] 815°C/2h/Ar 1245 1415 4.6
Ti-22Al-25Nb 1125°C/1h/BC +
[Rowe er al. (1991 )1 815°C/2h/Ar 1134 1175 0.9
Ti-22Al-27Nb
[Rowe et al. @931 815°C/lh/Ar 1294 1415 3.6
Ti-23Al-16Nb 1050°C/lh/WQ +
[Boehlert et alfl997c)] 850°C/2h/FC 691 906 14
Ti-23Al-23Nb
[Dary and Pollock (1996)] 760°C/100h 472 638 4
Ti-25Al-21Nb 1050°C/1h/Ar +
[Rowe et al. (1991)] 815°C/2h/Ar 847 881 0.4

 

WQ: water quench, F C: firrnace cool,

Ar: cooling performed in static argon gas, and BC: brick cooled (l.5°C/s)

57

 

32:8 5908

5:23 .265 .03 .38 0853 .0m .38 EBB -00 633;: .o: -<Z .9562; 8 e238 2: 25332 3% Eom
bee mEEw nag—Go 25352 Em is 625.qu 0.95 a Bow “.2:an Sun a 5.5 oowA 82:3 combs Exam—BS e:
was an

  

cZ-~<-F 259$ com magnum 35026932 98 momtaocm 26:3. PM ._> 2an

58

Incorporating the B2 phase into the microstructure decreases the amount of
adjacent O/O gain boundaries, which reduces stress concentration effects, in addition to
the ability of the BZ phase to blunt cracks which initiate and propagate along adjacent
O/O gain boundaries. Slip has been shown to be transmittable between adjacent O and
B2 gains, which reduces stress concentration effects that induce cracking [Majumdar et
al. 1995]. Planar slip is observed in the O and 012 phases and wavy slip is observed in the
B2 phase [Banerjee (1995), Majumdar et al. (1995), Boehlert et al. (1997c), Boehlert
(2001)]. The B2 phase and the 0 phase are based on the composition TizAle.
Therefore, the B2 phase typically has a lower Al content in the two-phase microstructure
as opposed to a fully-32 single phase microstructure (See Table VI). This leads to wavy
slip and ductile dirnpling within the BZ phase in the two-phase microstructure, which as
mentioned previously was not observed in the fully-B2 microstructures. In order for Ti-
Al-Nb alloys to possess adequate RT ductility, the BCC phase is an essential component.
In addition, Boehlert’s work has demonstrated that a minimum volume fiaction of 20
volume percent BCC phase is required for elongation-to-failure values >1% [Boehlert
(2001)].

The objective of the tensile behavior portion of this thesis was to determine the
tensile properties of these alloys at RT and elevated-temperature (650°C), in order to
determine microstructure-tensile property relationships, which are absent in the literature
for Ti-Al-Nb alloys with Ti content fixed at nominally 50 at% with A1 contents less than
22 at%. To the author’s knowledge, tensile properties have only been determined for one

alloy within this compositional range, a Ti-12Al-3 8Nb alloy [Boehlert (1999)].

59

6. Fatigue Deformation Behavior

A fatigue failure is defined as one that occurs due to application of a repetitive or
fluctuating stress that is much lower than the stress needed to cause failure on a single
application of load [Dieter (1986)]. For an alloy to be used as a biomaterial, resistance to
fatigue failure is the main structural requirement, as the implant will experience millions
of stress cycles within its useful lifetime.

Due to their excellent specific strength and electrochemical corrosion resistance in
addition to exhibiting exceptional biocompatibility characteristics (i.e. benigr biological
responses) among metallic biomaterials, pure Ti and or/B type Ti alloys are widely used
as structural biomaterials for the replacement of hard tissues in devices such as artificial
hip joints and dental implants. In particular, the specific alloy most widely used for
implants is Ti-6Al-4V (wt%) extra-low interstitial (ELI) because of its excellent
biocompatibility and its combination of high specific strength, fracture toughness, fatigue
and corrosion resistance, low density, ductility, and elastic modulus, oxidation resistance,
and conventional processability. However, vanadium (V) is potentially toxic in
elemental form [McKay et al. (1996)] and its elastic modulus is siglificantly geater than
that for bone; therefore, other alloying elements are currently being examined.

The recent trend in research and development of titanium alloys for biomedical
applications is to develop low rigidity B-type titanium alloys composed of non-toxic and
non-allergic elements with excellent mechanical properties, especially fatigue resistance,
and workability [ASTM designation draft #3, (2000), ASTM desigration draft #6 (2000),
Davidson and Georgette (1987), Niinorrri (2002)]. New titanium alloys for biomedical

applications are now being included in American Society for Testing and Materials

60

(ASTM) standards. For example, B Ti-15Mo (wt%) [ASTM desigration F2066-01
(2001)], Ti-35Nb-7Zr-5Ta (wt%) [ASTM designation draft #3 (2000)], and Ti-3Al-2.5V
(wt%) [ASTM desigration draft #6 (2000)] have been registered or are in the process of
being registered in ASTM standards. Beta-type titanium alloys have been developed in
order to obtain lower rigidity, which is considered effective for promoting bone healing
and remodeling. Although the rigidity of a/B type titanium alloys is less than that of Co-
Cr type alloys and stainless steels used for biomedical applications, it is still considerably
geater than that of the cortical bone [Davidson and Georgette (1987)].

Several investigations have included newer alloys based on the Ti-Al-Nb system,
such as Ti-6Al-7Nb (wt%) [Ti-10.5Al-3.6Nb (at%)] where Nb replaces V [Lopez et al.
(2003), Metikos-Hukovic et al. (2003), Cai et al. (2003), Iijima et al. (2003), Khan et al.
(1999)], Papakyriacou et al. (2000)], Semlitsch et al. (1992)], Watanabe et al. (2004),
Akahori et al. (2000b)]. Substitution of Nb for V is attractive as depending on the
concentration of Nb this does not result in degadation of several mechanical properties.

In this thesis Ti-Al-Nb alloys containing lower Ti concentrations were evaluated
to peruse their attractiveness for biomedical implant applications. Fatigue strength is
needed in order to assess the reliability of a material in medical implant applications as an
implant has to withstand not only one-time peak stresses, but also the several million load
cycles which it usually experiences during its lifetime. In this thesis the fatigue behavior
of the Ti-15Al-33Nb and Ti-21Al-29Nb alloys was evaluated and compared to those for

other biomedical Ti alloys.

61

B. Boron Modification of Titanium Alloys

Recent work has shown that the addition of trace amounts of boron (B) to
conventional titanium alloys, such as Ti-6Al-4V (wt%), decreases the as-cast gain size
by approximately an order of magritude [Tamirisakandala et al. (2005)]. This drastic
reduction in the as-cast gain size leads to siglificant benefits including increasing yield
strength while reducing or avoiding time spent on expensive and energy intensive
thermomechanical processing. The addition of B also leads to the presence of titanium
boride (TiB) whiskers within the conventional a+B microstructure [Tamirisakandala et
a1. (2005), Gorsse S. and Miracle DB. (2006)]. The presence of the TiB phase within
these microstructures leads to substantial increases in RT strength and elastic modulus,
while maintaining adequate elongation-to-failure values [Gorsse S. and Miracle DB.
(2006)]. To date, relatively few studies have focused on investigating the effects of
boron additions on the microstructure and mechanical properties of Ti-Al-Nb alloys
[Tang et al. (2001), Hagiwara et al. 2003a), Yang et al. (2003), Hagiwara et al. (2003b),
Emura et al. (2004)]. However, the initial results show boron additions to a Ti-22Al-
27Nb alloy to be beneficial for both RT tensile and RT high-cycle—fatigue properties
[Emura et al. (2004)].

The effect of boron additions on the microstructure and the impact of boron
additions on the creep and tensile deformation behavior of Ti-Al-Nb alloys were
evaluated. This was performed in hopes of further enhancing the creep resistance of
alloys that have shown the potential to exceed the elevated temperature capabilities of

conventional a+B Ti-alloys [Cowen and Boehlert (2006a), Cowen and Boehlert (2006b)].

62

CHAPTER 3

EXPERIMENTAL PROCEDURES

The aim of this Chapter is to describe the general experimental methods used to
accomplish the specific aims outlined at the end of Chapter 1. This Chapter discusses
alloy processing, the techniques used for microstructural characterization, and the
methodology and equipment used for creep, tensile, fatigue, and in-situ mechanical

testing.

A. Alloy Processing
1. Hot-Forged and Hot-Rolled Materials

1.1 Ti-15Al-33Nb and Ti-21AI—29Nb Materials

Deformation processing was performed on double vacuum-arc-remelted (V ARed)
Ti-15Al-33Nb and Ti-21Al-29Nb ingots, which were melted and processed by RTI
Titanium Corporation (Niles, Ohio). All forging and hot rolling was performed at
subtransus temperatures for both alloy compositions. The Ti-15Al-33Nb and Ti-21Al-
29Nb ingots were heated at 899°C (1650°F) and 982°C (1850°F), respectively, and
unidirectionally hot forged. An image of the forged ingots is shown in Figure 3.1 (a).
The ingots exhibited exterior cracks and therefore initially planned attempts for filrther
forging were not followed. Instead, attempts at subtransus hot rolling were made to
further breakdown the cast structure. Hot rolling operations for each alloy were
performed at the same temperatures as hot forging. Slabs (127 mm x 19.05 mm x 25.4

mm) were cut from the forged ingots and isotherrnally soaked at their respective

63

temperatures prior to hot rolling. Unidirectional multipass rolling was carried out
followed by interpass reheating for 5 minutes. The reduction per pass ranged between 5-
10 percent and the total reduction was approximately 90 percent. After rolling was
completed, the sheets were air-cooled to RT. The total effective true shear strain which
the original cast ingots underwent was calculated to be on the order of 2. The final sheet
thickness was approximately 1.5 mm. The as-processed (AP) sheet is illustrated in
Figure 3.1 (b). The surface cracking that occurred during hot forging was completely
eliminated. Overall, the alloys were quite amenable to subtransus processing as the as-
cast microstructure was successfully broken down as well as homogenized. The resulting
AP microstructures exhibited a fine equiaxed gain size of 3 pm. Thus, these alloys can
be successfully thermomechanically processed at temperatures below the BCC-transus.
This enables the possibility of obtaining a wide range of microstructures containing
varying phase volume fractions, gain sizes, and morphologies through post processing
heat treatment similar to other Ti-Al-Nb alloys [Boehlert et al. (1999)].

Since the Ti-15Al-33Nb and Ti-21Al-29Nb ingots were slow-cooled from the
melt and processed at subtransus temperatures, the 012 phase was present in the AP
microstructures. Two-phase O+BCC microstructures were targeted. The presence of the
012 phase is not desired as it generally leads to low RT elongation-to-failure values [Gogia

et al. (1998), Akkurt et al. (1991), Gogia et al. (1992), Boehlert et al. (19971»).

64

1.2 Ti-15Al-33Nb-0.5B Material

The Ti-15Al-33Nb-0.5B material was VARed and cast into a 0.250 kg button.
The button was then hot-rolled into sheet at 975°C, which is still below the BCC-transus

temperature.

1.3 Ti-SAl-45Nb Material

The Ti-5A1-45Nb material was triple VARed and cast into a cigar melt 18.2 mm
thick. The cigar melt was heated to 800°C for 30 minutes then unidirectionally hot rolled
with a 10 percent reduction per pass to a final thickness of 2.7 mm. All hot rolling was
performed at 800°C, with a three minute reheat between each pass. After rolling was
completed, the sheet was air-cooled to RT. The total effective true shear strain which the

original cigar melt underwent was calculated to be on the order of 0.8.

2. Hot Isostatically Pressed Material

2.1 Ti-15Al-33Nb-5B Material

The Ti-15Al-33Nb-5B material was induction-skull melted at Flowserve (Dayton,
Ohio). The elements were mixed in the correct stoichiometric amounts and then melted
and cast into an ingot of 70 mm diameter and 500 mm length. The Ti-15Al-33Nb-5B
ingot was then hot isostatically pressed (HIPed) for four hours at a pressure of 172 MPa

and temperature of 900°C. No subsequent thermomechanical processing was performed

on this alloy.

65

 

(b)

Figure 3.1. (a) The as-forged Ti-15Al-33Nb and Ti-21Al-29Nb ingots. The Ti-lSAl-
33Nb ingot is labeled 005 is and the Ti-21Al-29Nb ingot is labeled 006. Note that the Ti-
21A1-29Nb ingot exhibited more extensive cracking than the Ti-15Al-33Nb ingot. (b) An
as-processed Ti-21Al-29Nb sheet. Note the elimination of the cracking that occurred
during forging.

2.2 Ti-22Al-26Nb and Ti-22Al-26Nb-5B Materials

In order to examine a fabrication route alternative to ingot metallurgy, powder
metallurgy processing was performed on the Ti-22Al-26Nb and Ti-22Al-26Nb materials.
This latter approach has the potential to produce sigrificant cost reduction compared to
some alloy and metal matrix composite (MMC) processing methods [Smith et al. (2000)].
A limited amount of work has been done to understand the effects of powder
microstructures on the mechanical performance of O-based alloy MMCs. The Ti-22AI-
26Nb and Ti-22Al-26Nb-5B materials were produced in spherical powder form using
argon gas atomization at Crucible Research (Pittsburgh, PA). The powders were size
fractioned to 35 mesh, which corresponds to an average diameter of 500 um. Each
powder composition was placed into commercially pure Ti cans and degassed at 300°C
for 24 hours in preparation for HIPing at AF RL. The inner diameter of the cans was 22.9
mm and the outer diameter of the can was 25.4 mm and two cans were made for each
composition. The Ti-22Al-26Nb cans were filled with 0.135 kg and 0.065 kg of powder
and the Ti-22Al-26Nb-5B cans were filled with 0.149 kg and 0.76 kg of powder. The
cans were subsequently HIPed at 207 MPa and 1027°C for four hours. The initial ramp
up to temperature in the HIP vessel was such that it took 40 minutes to reach 700°C and
28 MPa. Thereafter it took another 20 minutes to reach 1027°C and 207 MPa. During
the ramp down it took 30 minutes to decrease the temperature to 300°C and 0 MPa. The
temperature was then decreased to RT rapidly. Both the temperature and pressure during
these ramps was approximately linear. It is noted that the materials remained below the

BCC-transus temperature throughout the HIP process. No subsequent thermomechanical

67

processing of these materials was performed after HIPing. The material was removed

from the cans using electron discharge machining (EDM).

B. Microstructural Evaluation

1. Chemical Composition Analysis

The bulk chemical composition of the Ti-Al-Nb-xB alloys was determined by
Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) and Inert Gas
Fluorescence (IGF). The bulk chenrical analyses for each alloy were performed by the
companies that processed each alloy.

The chemical composition of the constituent phases within the microstructure was
determined for selected samples through Electron Microprobe Analysis (EMPA). This
analysis was performed at Alfred University using a JEOL JXA-8200 Electron
Microprobe. High purity Ti, Al, and Nb standards were used in the EMPA analysis and

the spectral resolution of the instrument was 10 eV.

2. Heat Treatments

Selected Ti-15Al-33Nb and Ti-21Al-29Nb samples were wrapped in tantalum
(Ta) foils and heat treated in air for three hours at 855°C, 910°C, 960°C, 990°C, 1005°C,
1050°C, 1075°C, and 1105°C followed by water quenching. Heat treatments were also
performed for 200 hours at 855°C, 910°C, and 960°C on samples that were wrapped in
Ta foils and vacuum encapsulated in fused silica. The latter heat treatments were
performed in order to characterize the microstructural features present after chemical and

compositional equilibrium had been attained.

68

Microstructures for creep, tensile, and fatigue testing were produced for the Ti-
15Al-33Nb and Ti-21Al-29Nb alloys by solutionizing at either 960°C, 1005°C, or
1105°C for three hours followed by fumace cooling to 855°C and aging for 8 hours,
control cooling to 650°C and dwelling for 10 minutes, then furnace cooling to RT.
Furnace cooling was performed at a rate of 10°C/min and control cooling was performed
at a rate of 1°C/min. Two additional aged microstructures were also produced for the Ti-
15Al-33Nb alloy by solutionizing at 650°C for 100-250 hours followed by furnace
cooling. Henceforth, all aged microstructures will be referred to by their solutionizing
temperature: HT:650/100h/FC, HT:650/250h/FC, HT:960, HT:1005, and HT:1105. All
of the aging heat treatments were performed in flowing argon gas on samples wrapped in
Ta foils. A schematic of the aging heat treatments using HT:1005 as an example is

shown in Figure 3.2.

69

 

1200 . r .

 

 

 

 

...................... T 1721.417391‘1939911‘999931194.1139
1000 —_-- - --- 39.5.79 rsIi:15Al:3.3_N_b39911311531321.3939
10°C/min
g) 855°C 8hrs A
45‘ 800 - 1°Clmin ~
h 0
a 5° C Imin 650°C 10min
2 600 _ 4
Q)
:1.
o INC/min
E 400 —— -
0
H
200 — —
0 l l L
0 5 I0 15 20

Time, hrs

Figure 3.2. Schematic depicting the heat treatment schedule for the HT:1005 heat

treatment.

70

3. Sectioning, Cleaning, and Metallography

Samples were cut from the heat treated material from the face, transverse , and
longitudinal sheet orientations (see Figure 3.3) using either an abrasive saw with a silicon
carbide (SiC) cutting blade or a low-speed saw using a diamond blade. After sectioning
the samples were cleaned with acetone and ethanol. The cleaned samples were then
mounted in Konductometn‘ in preparation for ginding and polishing. Grinding was
accomplished using successively finer gits of SiC paper, with a typical schedule of 60
git, 240 git, then 600 git. After ginding, the samples were polished to a finish of 1 pm
with diamond paste suspensions. Colloidal silica with an average particle size of 60 nm
was used for the final polish, with polishing times ranging between 20 minutes to geater

than 2 hours per sample.

4. Microscopy

Optical Microsc0py (OM), Scanning Electron Microscopy (SEM), and
Transmission Electron Microscopy (TEM) were employed to evaluate the AP and heat
treated microstructures. High contrast, digitized, backseattered electron (BSE) images,
were obtained from a CamScan 44FE field emission SEM. The BSE images were used
to calculate gain size, calculate phase volume fiactions, and to examine phase
distributions and their morphologies. Brightfield (BF) images and selected area electron
diffraction patterns (SADPs) were obtained from a Hitachi H-800 TEM operated at 200
kV. TEM foils were prepared by ginding samples to a thickness of approximately 150

um after which 3 mm diameter disks were punched from the foils.

71

 

Rolling _ é

. . Q

Direction $
é

 

Transverse

 

 

 

Figure 3.3. A schematic depicting the three different sheet orientations from which
samples were taken for microstructural characterization.

72

The disks were twin-jet electropolished using a Struer’s tenupol 5 electropolishing unit

operated at 30-70 V and -30°C in a solution of 6 vol% H2SO4/94 vol% methanol.

5. Phase Volume Fractions

Since the contrast in a BSE image is directly proportional to the atomic number of
the material the image is captured from, different phases appear with different contrast in
a BSE image. Phases enriched in lighter elements (i.e. Al for the alloys in this thesis)
will appear darker, because the nucleus of lighter elements contains less protons and
neutrons, and therefore the backseattered electron yield decreases. For the BSE
photomicrogaphs present throughout this thesis, the 012 phase will appear as the darkest
contrast (black phase), the O-phase will appear with medium contrast (gey phase), and
the BCC phase will appear with the highest contrast (white phase). Approximately 20
BSE images were taken at different magrifications and from different areas of the face,
transverse, and longitudinal orientations of each sample. These images were then
analyzed using NIH Image and/or ImageJ image analysis software. This software allows
area-fiaction analysis of digitized images. This is performed by allowing the user to fill
in pixels of different contrast, which desigrate different phases in BSE images, and the
software counts the number of pixels in the filled region and divides this number by the
total number of pixels in the digitized image. Performing this analysis on several images
from different sheet orientations allows calculation of the average volume fiaction of
each phase. The phase volume fractions reported in this work should be considered
accurate within i5%, as this value was the maximum standard deviation obtained for

these measurements.

73

6. Grain Size Analysis

Average equiaxed gain size was determined using the Mean Line Intercept
Method [Hilliard (1964)] for subtransus heat treated microstructures and ASTM Standard
E112 [ASTM Designation E112-96e3] for selected Ti-15Al-33Nb and Ti-21Al-29Nb
supertransus heat treated microstructures. Selected samples were etched using a 75mL
H2O-20mL HF- 5mL HN03 etchant, which is a standard titanium etchant [Boyer (1985)],
to reveal gain boundaries. Several photomicrogaphs representing different
magrifications and different sheet orientations were analyzed for the gain size
calculations. Lath gain boundaries were not considered in the gain size analysis
performed in this study. Thus, for subtransus heat-treated microstructures the reported
gain size represents the average of all the equiaxed phases present in the microstructure
and for supertransus heat-treated microstructures the reported gain size is the average

prior-BCC gain size.

7. Differential Thermal Analysis

Differential thermal analysis (DTA) was performed to obtain an estimate of the
BCC-trarlsus temperature for each alloy. The Ti-15Al-33Nb and Ti-21Al-29Nb alloys
were gound into powder for the DTA analysis. Approximately 15 mg of the alloy
powders were used, along with an equivalent amount of high purity aluminum oxide as
the reference standard. The alloy powders and standard powders were heated in separate
platinum crucibles. The thermal schedule included an equilibration at 600°C after which
the samples were heated to 1200°C at a rate of 10 or 15°C/min while temperature and

temperature difference between the alloy and the reference standard were recorded. The

74

phase transformations were revealed by noticeable exotherms or endotherms in the

temperature difference versus temperature plots.

8. X-ray Diffraction

X-ray Diffraction (XRD) analysis was performed on the solution-treated and
quenched samples using Siemens Diffractometers with Cu Kc radiation sources operating
at 40 kV and 30 mA. Each scan was performed using a step size of 004° 20 and a count
time of 5 seconds. In addition, long count time scans were performed on selected

samples using a step size of 002° 20 and a count time of 30 seconds to determine the

ordering of the BCC phase.

C. Creep Testing

Flat, dogbone shaped samples (140 mm long x 25 mm wide with a 12.7 mm gage
length) were machined from the rolled sheets using either a nrill or an EDM. The dog
bones were then gound with SiC papers to at least a 240 git finish to remove the oxide
scale left from processing and any surface layers that may have been damaged and/or re-
cast during the EDM process.

Constant load, tensile-creep experiments were performed using four Applied Test
Systems, Inc. (ATS) lever arm creep testing flames. The stress was assumed to be
constant during the constant-load experiments because the total creep strain accumulated
by each sample was typically less than 3.5%. The 2370 series frames (two test frames)
had a 5:1 lever arm ratio and the 2410 and 2710 series test frames had a 20:1 lever-arm

ratio. The 20:1 lever-arm ratio facilitates creep testing at loads up to 10,000 lbs. The

75

loading direction was parallel to the rolling direction for all the creep and tensile
experiments. The load was measured with a resolution of 0.01 lbs using WSM 100 1b
(Model ABIOO-T), WSM 500 lb (Model AB), and ATS 2000 lb (Model 1210-AF and
Model 1210-AF-2K-B) load cells, on the 2710, 2410, and 2370 series flames,
respectively. Creep strain was monitored during the tests using linear variable
differential transformers (LVDTs) connected to 25.4 mm gage length ATS high-
temperature extensometers. The extensometer was attached directly to the gage length
section of each sample. The resolution of the strain measuring capability of the LVDT
was 0.0001. The strain rate resolution was estimated to be 2.8x10'9 8'1 according to the
resolution of the strain measurement technique. The creep data presented in this thesis is
based on the engineering stress and the plastic creep strain.

ATS single-zone shell furnaces (3320 series) equipped with MoSi2 heating
elements were used for the creep experiments conducted on all the ATS creep flames.
The temperature was monitored with three thermocouples for each test; one
thermocouple was spot-welded directly to the gage section of the dogbone, while the
other two were placed near the sample gage section but were not attached to the sample.
For each test, the temperature was kept within i 5°C of the targeted test temperature. All
creep tests were conducted in air at stresses between 10—400 MPa and temperatures
between 650-710°C. Each sample was soaked at the testing temperature for at least 60
rrrinutes before loading. The minimum creep rate was determined for each sample at
each of the specific stresses and temperatures. In order to determine creep stress
exponents and activation energies, constant temperature/load jump tests and constant

load/temperature jump tests were performed. This was accomplished by either changing

76

the load or temperature after the minimum creep rate had been achieved. The load was
increased by adding weights to the lever arm and temperature was controlled by
increasing or decreasing the voltage output to the furnace. The lever arms were

maintained level during the creep experiments in order to maintain a constant load.

D. Tensile Testing

Samples for tensile testing were prepared in the same manner as described for the
samples used for creep testing.

Tensile tests were performed at RT and 650°C using an Instron 8562 testing
machine equipped with and Instron 8500 controller, an Instron 4206 testing machine, and
a MT S 810 tensile testing machine, all equipped with 100 kN load cells. Strain was
measured using 25.4 mm gage length Instron or Epsilon extensometers attached to the
gage section of each dogbone and all tests were conducted at a strain rate of 1.3x10'3 3".
For the 650°C experiments, temperature was monitored in the same fashion as described
for the creep tests, using the same types of furnaces, and each sample was soaked at
650°C for at least 15 minutes prior to tensile testing. The temperature was controlled
within :t 5°C of the targeted test temperature.

The 650°C tensile tests performed on the Ti-15Al-33Nb-xB materials were
performed using a servohydraulic thermomechanical testing machine [Hartman and Russ
(1989)]. Digital photogaphs of the servohydraulic thermomechanical testing machine
used in this thesis are given in Figure 3.4 (a)-(c). The temperature was maintained
within :1: 5°C of the target temperature during testing and samples were allowed to soak at

temperature for at least 15 rrrinutes prior to testing. Strain was measured using a dual

77

quartz-armed high-temperature extensometer attached directly to the gage length of the
sample.

Yield Stress (Y S), Ultimate Tensile Stress (UTS), and Elongation-at-Failure (8f)
values were determined for each tensile test flom the. engineering stress versus
engineering strain curves. Young’s Modulus (E) for each sample was determined flom
the slope of the elastic portion of the stress versus strain curves. The YS was determined
according to the 0.2% offset method.

Selected specimens were polished to a metallogaphic finish prior to tensile
testing in order to reveal surface deformation features. The sample surfaces were imaged
after the samples were strained to a specific level and the tensile test was discontinued or

the sample surface behind the flacture surface was imaged after failure occurred.

78

 

 

a . .

(C)
Figure 3.4. Variable environment thermomechanical servohydraulic test machine used in
this thesis. Image (c) depicts the inside of the test chamber including the fliction grips
and extensometer.

79

E. Fatigue Testing

Fatigue tests were carried out at two institutions using two different test apparati.
In each case the fatigue testing was performed at a flequency of 10 Hz with a stress ratio
of R = 0.1 under a tension-tension stress mode in air at 295 K. A schematic of the fatigue
specimen geometry used by both institutions is available in Figure 3.5 (a).

The fatigue experiments performed at Toyohashi University of Technology
(Toyohashi, Japan) used an electro-servohydraulic fatigue testing machine [Niinomi
(2003)]. The maximum applied stresses ranged between 820 and 950 MPa (2 0.86 UTS)
and 1x107 cycles were considered to be “nm out” after which the unflactured sample was
removed. Typically, more than one sample was tested in each condition, and such
samples were polished through 60 nm colloidal silica in order to identify surface
deformation through SEM observations of post-tested specimens.

The fatigue experiments performed at the Fraunhofer Institute (F reiburg,
Germany) used servohydraulic testing machines with a chain assembly and stainless steel
gips in order to avoid transverse stresses. One of the servohydraulic testing machines is
depicted in Figure 3.5 (b). For samples tested at the Frannhofer Institute, the maximum
stresses ranged between 350 and 750 MPa (0.35—0.75 UTS). Due to the sigrificant
fatigue data scatter, up to five specimens were tested for each stress level chosen. For
such testing, 2x106 cycles was considered to be run out, and all the samples which
experienced run out were removed florn the machine and subsequently examined in RT
tensile experiments, identical to those described previously, in order to characterize

strength loss due to fatigue.

80

 

 

0,01

 

(b)

Figure 3.5. (a) A schematic of the fatigue specimen geometry used. All values are given
in mm. (b) Image of the fatigue set-up at the F raunhofer Institute. A pulsating sinusoidal
load (tension—tension mode) was used for fatigue testing performed.

81

F. In-situ Tensile and Creep Testing

For the in-situ tensile and in-situ tensile-creep experiments, flat dog-bone-shaped
samples, with gage section dimensions of 3 mm wide by 2.5 mm thick by 10 mm long,
were EDM cut. A slight reduction in width of 0.6 mm at the middle of the sample gage
section was introduced in selected samples to ensure that failure occurred at this location.
The specimens were glued to a metallic platen and polished using an automatic polishing
machine (See Figure 3.6).

These tests were performed using a screw-driven tensile stage built by Ernest F.
Fullam, Incorporated (Lantham, NY), placed inside the chamber of the CamScan 44FE
SEM (see Figure 3.7). The load-displacement-time relationship was obtained live during
the experiments using MTESTW version F 8.8e data acquisition and control software
(Adrnet, Inc., Norwood, MA) that allowed for either load-control or displacement-control
testing. The displacement data acquired during the experiments comprised that of the
sample as well as the gipping fixtures, and the displacement values reported in this study
do not represent the sole displacement of the sample.

Since the pressure within the SEM chamber never exceeded 10'6 torr, oxidation
did not affect the in-situ BSE imaging characterization. In fact, no evidence of oxidation
was observed even after 162 hours of creep testing at 650°C. Digital images of post
creep tested samples are shown in Figure 3.8 (a). Figure 3.8 (a) clearly shows that all the
deformation that occurred during the creep experiments occurred within the section of the
sample’s gage length located directly above the tungsten-based heating element and that
the integrity of the sample’s surface was maintained throughout the duration of the

experiment.

82

The in-situ stage allows for tensile or tensile-creep testing at temperatures ranging
flom RT to 1000°C and loads up to 4.4 kN (1000 lbs). A 1000 lb load cell was attached
to the end of the displacement assembly to monitor load. The tensile and creep
experiments were performed in load-control mode and the samples were loaded to the
creep stress at a rate of 50 lbs/min. The creep experiments were performed in the stress
range of 335-400 MPa at 650°C. For the elevated temperature tensile experiments and
the creep experiments, the samples were heated using a 6 mm diameter tungsten-based
heating element placed just below the gage section of the sample. The heating element is
surrounded by an insulating ceramic sheath, and only one face of the heating element is
exposed. A DC power supply operated at a constant voltage was used to produce the
current needed to resistively heat the tlmgsten-based heating element. The temperature
was monitored by a thermocouple spot-welded directed to the sample’s gage section (see
Figure 3.8 (b). An additional thermocouple was placed below the heating element and
was also monitored throughout the experiments. The temperature was maintained within
i 5°C of the target temperature. For the 650°C tensile and creep experiments the samples
were held at temperature for a minimum of 30 minutes prior to addition of load. An open
bath chiller circulated 20°C distilled water through copper lining which continuously
cooled the heater during the experiments.

The following list denotes the procedures involved in performing the in-situ creep
experiments:

1. The vacuum chamber door of the SEM was vented.
2. The SEM porthole cover was replaced with a specially made port hole cover which

contained feed-throughs for the quick-disconnect cooling lines, two thermocouple wires,

83

the load and displacement control connections, and the DC power supply connects for the
heater.

3. The tensile stage was then secured to the sample stage with set screws.

4. The cables/wires from the tensile stage were then connected to their mating fixtures
located on the inside half of the specially made porthole cover. The mating fixtures for
the cables/wires on the outside half of the specially made porthole cover were then
connected to the chiller, data acquisition computer, DC power supply, and the
temperature readout.

5. The water chiller, temperature readout, heater DC power supply, and ADMET stage
controller were then turned on. The MTESTW data acquisition and control software was
then opened on the computer. Once the software boots, the load and displacement values
are automatically set to zero. The desired load-time profile, the desired data acquisition
rate, and sample dimensions were then input into the MTESTW data acquisition and
control software.

6. A thermocouple was then spotted welded directly to the gage section of the sample.
The sample was then carefully placed within the fliction gipping fixtures, and secured by
tightening four screws (2 within each fixture) located within the fixtures. Care was taken
when tightening the gips not to apply a load to the sample. If a small load was applied
either in tension or compression, the load was removed by careflllly adjusting the
displacement of the stage with the ADMET controller.

7. The vacuum chamber door of the SEM was then closed and the SEM was evacuated.

8. The stage was then positioned in the proper position for imaging the gage section of

the sample. Due to the size constraints of the stage, the typical working distance used for

84

BSE imaging was 15-21 mm. Pretest BSE images were then acquired from the desired
areas of the sample before heating or applying load.

9. The voltage output to the heater was then increased in slow increments until the
desired temperature was reached. Due to the thermal strains that arise upon heating due
to the constraint of the sample in the gipping fixtures, a compensating load was
maintained by the ADMET displacement controller. The stress resolved on the sample
was maintained at 0 i 5 MPa during heating to compensate for the thermal strains.

10. After the temperature remained stable at the desired test temperature for at least 30
minutes, the MTESTW software control was activated. This resulted in loading the
sample at a rate of 50 lbs/min to the desired stress while simultaneously displaying a real-
time displacement versus time plot.

1 1. The sample was then imaged at desired times/displacements throughout the duration
of the experiment, and voltage was only applied to the SEM filament when imaging was
performed. This necessitated a full beam alignment each time imaging was performed.
12. The water level in the chiller was continuously monitored during the experiments
and refilled as necessary. The cold trap of the SEM was typically refilled with liquid
nitrogen every 8 hours, with a maximum time of 15 hours in between refills. Constantly
maintaining an appropriate level of liquid nitrogen in the cold trap was essential for

sustaining the integity of the vacuum system during the creep experiments.

85

 

Figure 3.6. Tensile and creep samples (approximately 38 mm long) glued to a metallic
platen for metallogaphic polishing prior to evaluation of the in-situ deformation.

86

  
    

‘ 1 , $1.”.

rlrv'mllIII'I" n
_ 2 b l -J

 

Figure 3.7. A digital image of the in-situ tensile stage. The sample is gripped between
the two fixtures shown in the middle of the tensile stage. The 6 mm diameter tungsten-
based heating element is visible directly beneath the gage section of the sample and is

surrounded by the rectangular ceramic sheath.

87

 

 

 

Figure 3.8. Samples creep tested for up to 162 hours within the SEM using the tensile
stage are shown in (a). Note the preservation of the sample surfaces and the localization
of the deformation in the gage section to the portion of the sample gage section that sat
directly above the 6 mm diameter tungsten-based heating element The typical placement
of the thermocouple attached directly to the sample gage section is illustrated in the BSE
image in (b). The white circle below the specimen in (b) is the tlmgsten-based heating
element.

88

CHAPTER 4

RESULTS

This Chapter presents the results obtained flom the microstructure
characterization, creep experiments, tensile experiments, and fatigue experiments
performed on the Ti-Al-Nb alloys studied in this thesis. Within each subsection of this
Chapter, i.e. microstructure, creep properties, tensile properties, and fatigue properties,
the results obtained for the monolithic alloys are described first, followed by the results

obtained for the boron-modified alloys.

A. Microstructure Results

1. Bulk Chemical Compositions

The bulk chemical compositions determined flom ICP-OES and [GP analyses are
given in Table VII. The measured alloy compositions matched the targeted nominal alloy
compositions well and the oxygen content was kept at acceptable levels for each alloy
studied in this thesis. The oxygen content of the Ti-5Al-45Nb alloy was exceptionally
low. This alloy was triple VARed at the AF RL foundry using a process that consistently
produces low-oxygen containing Ti alloys. The oxygen content is important because
oxygen occupies interstitial lattice sites and can be detrimental to mechanical properties,
in particular elongation-to-failure. The addition of oxygen to Ti-Al-Nb alloys raises the
BCC-transus temperature and stabilizes the 012 phase, which is not desired in this alloy
system as geater 012 volume flactions lead to lower at values [Gogia et al. (1998), Akkurt

et al. ( 1991), Gogia et al. (1992), Boehlert et al. (l997b)].

89

Table VII. Bulk Chemical Compositions of the Ti-Al-Nb Alloys

 

 

 

 

 

 

AlloyComposition Ti, at% Al, at% Nb, at% B, at% 0, ppm
Ti-5Al-45Nb 49.5 5.14 45.36 0 336
Ti-15A1-33Nb 51.4 15.3 33.3 0 1100
Ti-2lAl-29Nb 50.7 20.6 28.7 0 790
Ti-22Al-26Nb 50.7 21.4 26.8 <0.15 1 127

 

 

 

 

 

 

 

2. Differential Thermal Analysis

Figures 4.1, 4.2, 4.3, and 4.4 show differential thermal analysis curves for as-
processed Ti-SAl-45Nb, Ti-lSAl-33Nb, Ti-21Al-29Nb, and Ti-22Al-26Nb samples,
respectively.

From Figure 4.1 it can be seen that the Ti-5Al-45Nb alloy showed no noticeable
exothermic or endothermic reactions over the temperature range of 600-1200°C, which
indicates that no phase transformations occurred in this alloy.

From Figure 4.2 it can qualitatively be stated that the BCC-transus temperature
lies between 943-1059°C for the Ti-15A1-33Nb alloy. For the Ti-2lAl-29Nb alloy, the
BCC-transus occurs between 1009-1110°C, as shown in Figure 4.3. The temperature at
which each alloy was thermomechanically processed is indicated on each DTA plot,
which is clearly below the BCC-transus temperature for each alloy.

The BCC-transus temperature was 1130°C for the Ti-22Al-26Nb alloy as
estimated flom the DTA analysis presented in Figure 4.4. This value is slightly higher
than that measured for powder metallurgy-processed Ti-22Al-26Nb alloys which
estimated the BCC-transus to be between 1120-1125°C [Smith et al. (1999), Smith et al.
(2000)]. The slightly higher temperature value is believed to be caused by the increased
oxygen content of 1127 ppm for the Ti-22Al-26Nb alloy in this study versus 860-1080

ppm for the Ti-22Al-26Nb alloys produced in a similar manner by Smith and coworkers

90

0.6
0.5
0.4
0.3
0.2
0.1

 

  

processing temperature

 

Temperature Difference, °C

s'=
l—r

 

_0.2 l l l 1 l
600 700 800 900 1000 1100 1200

Temperature, °C

 

Figure 4.1. Differential thermal analysis curve for a Ti-5Al-45Nb specimen. The sample
was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min.

91

 

11059I°C T

Si -0.6 -
2 -0.8 _.
é“ 853°C 943°C
9 -1 -
2 1
5
g -1.2 — processing temperature -
3.
E -1.4 - -
1-3

-1.6

 

 

 

600 700 800 900 1000 1100 1200

Temperature, °C

Figure 4.2. Differential thermal analysis curve for a Ti-15Al-33Nb powder specimen.
. The sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The
temperatures indicated on the curves represent the peak value of the exotherm/endotherm.

92

-002 1 1 T 1

 

 

 

11110°C

Si
6'): -0.4 — -
r.
g 0 6
is ' ' . . 1
G I 1009 C
2 .
a -008 J
E
8. 837°C 925°
5 -1— -
0
H

processing temperature

 

 

 

-1.
too 700 800 900 1000 1100 1200

Temperature, °C

Figure 4.3. Differential thermal analysis curve for a Ti-21Al-29Nb powder specimen.
. The sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The
temperatures indicated on the curves represent the peak value of the exotherm/endotherm.

93

 

1.8

T

  

 

1.6

   

processing temperature
(1027°C)
1.4

a

1.2

Temperature Difference, °C

 

 

 

1500 700 800 900 1000 1100 1200

Temperature, °C

Figure 4.4. Differential thermal analysis curve for a Ti-22Al-26Nb specimen. The
sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The
temperatures indicated on the curves represent the peak value of the exotherm/endotherm.

94

3. As-Processed/Solution-Treated and Water Quenched/Solution-
Treated and Aged Microstructures

The AP microstructures of the Ti-15Al-33Nb and Ti-21Al-29Nb alloys are shown
in Figure 4.5. Through hot forging and hot rolling at subtransus temperatures, the
microstructure of the cast ingots was successflllly refined and homogenized. The AP
microstructures contained only the 012 and BCC phases and exhibited a fine equiaxed
gain size of approximately 3 um.

The microstructures resulting from solutionizing at 855°C, 910°C, 960°C, 990°C,
1005°C, 1050°C, 1075°C, and 1105°C for 3 hours followed by water quenching are
shown for the Ti-15Al-33Nb and Ti-21Al-29Nb alloys in Figures 4.6-4.9 and Figures
4.10-4.13, respectively. In each BSE photomicrogaph, the BCC phase is the lightest
contrast phase (white), the 012 phase is the darkest contrast phase (black), and the O-phase
is the medium contrast phase (gey).

The O-phase precipitates in a Widmanstatten lath morphology for the Ti-lSAl-
33Nb and Ti-21Al-29Nb alloys when solutionized at temperatures less than or equal to
910°C. At 960°C, the O-phase has disappeared from the Ti-15A1-33Nb alloy while it
still remains in the Ti-21Al-29Nb alloy but occurs in more of an equiaxed morphology.
This transition to more of an equiaxed morphology can also be seen in the Ti-21Al-29Nb
sample solutionized at quenched flom 910°C, but the Widmanstatten lath morphology is
still dominant at 910°C when compared to 960°C (see Figures 4.8 (a) and 4.8 (b)). The
globularization of the O-phase appears to be the result of the coarsening of O-phase laths
with increasing temperature. The morphological transition observed in this thesis agees
‘ well with the results previously obtained by Boehlert et al. for Ti-25Al-25Nb and Ti-

23Al-27Nb alloys [Boehlert et al. (1999)]. The author’s found that solutionizing above

95

875°C produced microstructures containing equiaxed gains of the BCC and 0 phases.
Solutionizing below 875°C resulted in Widmanstatten precipitation of O-phase plates
within the BZ gains. Therefore, the O-phase morphology over the compositional range
of 21-25 at% Al is dependent upon initial solutionizing temperature within the O+BCC
phase field, and higher solutionizing temperatures favor a more equiaxed morphology.

At 960°C, only the 012 and BCC phases are present in the Ti-15Al-33Nb alloy, and
this provides evidence of a two phase 012+BCC phase field existing flom 960-990°C.
Above 990°C the Ti-15Al-33Nb alloy is fully-BCC, and BCC gain gowth occurs with
increasing solutionizing temperature within the BCC phase field (See Figures 4.5-4.7).
Since fully-BCC microstructures equilibrated at and above 990°C, this temperature
provides the upper limit for the BCC-transus temperature for the Ti-15Al-33Nb alloy.

From 990-1005°C, only the 012 and BCC phases are present in the Ti-21Al-29Nb
microstructures. A very small volume of O-phase may have been present at 990°C, but
the volume flaction of O-phase measured was less than the standard deviation of its
measured value. As with the Ti-15Al-33Nb alloy, sigrificant BCC gain gowth occurred
with increasing solutionizing temperature in the BCC phase field. For solution treatment
temperatures 2 1050°C the Ti-21Al-29Nb alloy is fully-BCC, therefore 1050°C provides
the upper limit for the BCC-transus temperature of the Ti-21Al-29Nb alloy.

Since the 012 phase was present in both the AP Ti-15Al-33Nb and AP Ti-21Al-
29Nb alloys, it remained within the all microstructures which were solutionized for three
hours. Microstructures flee of the 012 phase were only produced by solutionizing above
the BCC-transus. For the temperatures examined, the diffusion kinetics were either too

‘ sluggish below the BCC-transus temperature and/or the samples were not solutionized

96

long enough to remove the 012 phase flom the microstructure. This finding is sigrificant
because the 012 phase is only an equilibrium phase within the 012+BCC phase field that
exists just below the BCC-transus for the Ti-15Al-33Nb and Ti-21Al-29Nb alloys.

Backscattered electron images of the solution-treated and aged microstructures
developed through the heat treatment schedules described in Chapter 2 are shown in
Figures 4.14-4.16. These heat treatments were devised in order to produce thermally
stable microstructures that contained varying amounts of the O and BCC phases, gain
sizes, phase compositions, and phase morphologies. These microstructures were used for
the mechanical property studies.

The subtransus HT:960 heat treatment succeeded in precipitating the O-phase into
the microstructure of both alloys while still maintaining a fine equaixed gain size. Only
Widmanstatten O-laths were present in the Ti-15Al-33Nb microstructure (see Figure 4.14
(a)), while both equiaxed and lath O-phase were present in the Ti—21Al-29Nb
nricrostructure (see Figure 4.14 (b)).

The HT:1005 heat treatment produced a large-gained (120 um) supertransus Ti-
15Al-33Nb microstructure, in which equiaxed prior-BCC gain boundaries were evident.
The same heat treatment produced a fine gained (12 um) subtransus Ti-21Al-29Nb
microstructure. The HT: 1005 heat treatment accommodated the precipitation of a similar
amount of O-phase within each alloy’s microstructure as the HT:960 heat treatment, but
produced larger equiaxed-gained microstructures than the HT:960 heat treatment.

The HT:1105 heat treatment produced a larger prior-BCC gained supertransus

microstructure than the HT :1005 heat treatment for the Ti-15Al-33Nb alloy. This heat

97

treatment was devised in order to evaluate the effect of increasing the prior-BCC gain
size on the mechanical properties of the Ti-15Al-33Nb alloy.

Compared with all the heat treatments performed, the HT:650/100-250h/FC heat
treatments stabilized the largest amount of O-phase in the Ti-15Al-33Nb alloy. In
addition, these microstructures had the finest distribution of O-phase laths and maintained
an equiaxed gain size of 4 pm. Figure 4.17 compares BSE photomicrogaphs of the AP
and HT:650/100h/FC Ti-15Al-33Nb microstructures. Figure 4.17 provides evidence that
the AP Ti-15A1-33Nb microstructure was devoid of the O-phase. Directly aging at
650°C produced the Ti-15A1-33Nb microstructure with the largest volume flaction of 0-
phase laths. The O-phase laths had nanometer scale average width dimensions after the

650°C heat treatment.

98

 

(b
Figure 4.5. Backscattered electron images of the AP microstructures: (a) Ti-15Al-33Nb
and (b) Ti-21Al-29Nb. The (12 phase is the dark equiaxed phase and the BCC phase is the
continuous matrix phase in each alloy.

99

 

(b)

. Figure 4.6. Backscattered electron images of Ti-15Al-33Nb solutionized and water
quenched microstructmes: (a) 855°C and (b) 910°C.

100

 

(b)
Figure 4.7. Backscattered electron images of Ti-15Al-33Nb solutionized and water
quenched microstructures: (a) 960°C and (b) 990°C. The pitting present in (b) occurred
due to over-etching the sample.

101

 

(b)

Figure 4.8. Backscattered electron images of Ti-15Al-33Nb solutionized and water
quenched microstructures: (a) 1005°C and (b) 1050°C.

102

 

(b)

Figure 4.9. Backscattered electron images of Ti-15A1-33Nb solutionized and water
quenched nricrostructures: (a) 1075°C and (h) 1105°C.

103

 

(b)

Figure 4.10. Backscattered electron images of Ti-21Al-29Nb solutionized and water
quenched microstructures: (a) 855°C, and (b) 910°C.

104

 

(b)

Figure 4.11. Backscattered electron images of Ti-21Al-29Nb solutionized and water
quenched microstructures: (a) 960°C and (b) 990°C.

105

 

(b)

Figure 4.12. Backscattered electron images of Ti-21Al-29Nb solutionized and water
quenched microstructures: (a) 1005°C and (b) 1050°C.

106

 

(b)

Figure 4.13. Backscattered electron images of Ti-21A1-29Nb solutionized and water
quenched microstructures: (a) 1075°C, and (b) 1105°C.

107

 

ISAI-

Figure 4.14. Backscattered electron images of HT:960 microstructures: (a) Ti-
-29Nb.

33Nb and (b) Ti-21Al

108

:2
-.t;
L'd L‘s-
..;~ >.

 

Figure 4.15. Backscattered electron images of HT:1005 microstructures: (a) Ti-15Al-
33Nb and (b) Ti-21Al-29Nb. The arrows in (a) point to some of the prior-BCC gain
boundary triple-points.

109

 

(b)

Figure 4.16. Backscattered electron images of Ti-15Al-33Nb (a) HT:650°C/100h/FC and
(b) HT:1105 microstructures. Note the extremely fine O-phase laths present in (a) and
the arrows in (b) indicate prior-BCC gain boundary triple-points.

110

 

Figure 4.17. Comparison of microstructures of AP Ti-15AI-33Nb in (a) and
HT:650°C/100h/FC in (b). Note the presence of a large volume of O-phase laths in (b)
with no O-phase present in (a).

111

Figure 4.18 (a) shows an image of the AP Ti-5A1-45Nb sheet surface that was
obtained flom a stereomicroscope. The average equiaxed gain size for the Ti-5Al-45Nb
alloy was 4500 um. This alloy was flrlly-BCC, even after heat treatment at 650°C for 66-
80 hours, and the disordered BCC structure was identified by TEM analysis as will be
described later within this Chapter.

A BSE image of the hot isostatic pressed (HIPed) Ti-22Al-26Nb alloy is shown in
Figure 4.18 (b). The Ti-22Al-26Nb alloy contained 6% 012 phase, 72% O-phase, and
22% BCC phase by volume with an average equiaxed gain size of 59 um. No porosity
or unconsolidated gas-atomized-powder precursors were evident within the
microstructure of the HIPed compact. Therefore, HIPing under the pressure and
temperature conditions outlined within Chapter 2 successfully produced a fully sintered
alloy flee of porosity. No subsequent thermomechanical processing or heat treatments

were performed on this alloy.

112

 

(b)

Figure 4.18. Stereomicroscope image of the surface of the AP Ti-5Al-45Nb rolled sheet
is shown in (a) and a BSE image of the As-HIPed Ti-22Al-26Nb alloy is shown in (b).

113

4. Phase Volume Fractions

The phase volume flactions for the heat treated microstructures are given in
Tables VIII and IX for the Ti-15Al-33Nb and Ti-21Al-29Nb alloys, respectively. For the
solution-treated and water quenched samples a gaph of phase volume flaction versus
solutionizing temperature is presented in Figure 4.19. For a given temperature within the
O+BCC phase field, the Ti-21Al-29Nb alloy contained a larger O-phase volume flaction
than that for the Ti-15Al-33Nb alloy. This is due to its higher Al content in the Ti-21Al-
29Nb alloy and the fact that the stoichiometric O-phase composition is Ti2Ale.

Table VIII. Measured Phase Volume Fractions for Ti-l 5Al-33Nb Microstructures

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Heat Treatment 012 Vp O Vp BCC Vp
AP 10 0 90
855°C /3h/WQ 5 34 61
910°C /3h/W Q 5 17 78
960°C /3h/WQ 5 0 95
990°C /3h/WQ 0 0 100
1005°C /3h/WQ O O 100
1050°C /3h/WQ 0 0 100
1075°C /3h/WQ O 0 100
1105°C /3h/WQ 0 0 100
HT:650°C/100h 1 1 73 16
HT:96O 5 37 58
HT: 1005 1 44 55
HT:1105 1 63 36
Table IX. Measured Phase Volume Fractions for Ti-21Al-29Nb Microstructures
Heat Treatment 012 Vp O Vp BCC Vp
AP 4 0 96
855°C/3h/WQ 4 83 13
910°C/3h/WQ 3 50 47
960°C/3h/WQ 1 38 61
990°C/3h/WQ 6 0 94
1005°C/3h/WQ 5 0 95
1050°C/3h/WQ O 0 100
1075°C/3h/WQ 0 0 100
1105°C/3h/WQ 0 O 100
HT:96O 0 72 28
HT:1005 5 78 I7

 

 

 

 

 

 

 

114

100 .

 

7" "Ti-15Al-33Nb O-Phase Vp
E 80 : ' * Ti-21AI-29Nb O-Phase Vp q
0
CB
E
o 60 — -
E .
5 I
3 .
> 40 - _ .
3 a.» ”
a ,
.:1
“r 20 - -
O "-3

 

 

 

 

(g . X .1. . . 1.
50 900 50 1000 1050 1100 1150

Solutionizing Temperature, °C

. Figure 4.19. O-phase volume flaction versus solutionizing temperature for the Ti-15Al-
33Nb and Ti-21Al-29Nb alloys.

115

5. Grain Size

The average equiaxed gain size for each Ti-15A1-33Nb and Ti-21Al-29Nb

microstructure is given in Table X and a plot of gain size versus solutionizing

temperature is shown in Figure 4.20. There is minimal equiaxed gain gowth apparent

below 990°C and 1050°C for the Ti-15Al-33Nb and the Ti-21Al-29Nb alloys,

respectively.

Table X. Measured Equiaxed Grain Size for Ti-15Al-33Nb and Ti-21Al-29Nb

 

 

 

 

 

 

 

 

 

 

 

 

 

Microstructures
Heat Treatment Ti-lSAI-33Nb Grain Size, [1m Ti-21Al-29Nb Grain Size, um
855°C/3h/WQ 4 5
910°C/3h/WQ 5 5
960°C/3h/WQ 6 8
990°C/3h/WQ 100 7
1005°C/3h/WQ 120 12
1050°C/3h/WQ 144 135
1075°C/3h/WQ 157 150
1105°C/3h/WQ 173 196
650°C/100h/FC 4 N/A
HT :960 6 8
HT: 1005 120 12
HT: 1 105 173 N/A

 

 

 

 

 

N/A indicates not available because the indicated heat treatment was not performed on
the Ti-21Al-29Nb alloy.

116

200

 

~ . "Ti-15Al-33Nb ,'
- Ti-21AI-29Nb
150 _ -
E II I
:1. ,
51 100 _ . -
=
a
t-
e
50 1. ..

 

 

/’ VI /
5:7. :1 ‘ ir/l’ . ‘ ' 7 -
1

91507“ 900 950 10100 1050 1100 1150

Solutionizing Temperature, °C

 

Figure 4.20. Measured equiaxed gain size versus solutionizing temperature for the Ti-
15Al-33Nb and Ti-21A129Nb alloys.

117

6. BCC Phase Ordering

The BCC phase in Ti-Al-Nb alloys can either be ordered B2 or disordered B. This
structural transition is dependent upon alloy composition and heat treatment [Kestner-
Weykamp et al. (1989), Bendersky et al. (1991), Rhodes et al. (1993)].

The presence of {100} and {210} superlattice reflections in the Intensity versus
20 XRD pattern or in SADPs occur when the BCC phase is ordered. The (210)
superlattice reflection will occur with a weak intensity at approximately 64° 20, based on
the lattice parameter of 0.323 nm. In electron diffraction patterns obtained from TEM,
superlattice reflections of the <100> type will be present in SADPs of a {001} type zone
axis.

Long count time (0.02° 20 step size and 30 second count time) XRD scans were
performed on fully-BCC Ti-lSAl-33Nb and Ti-21Al-29Nb microstructures that
underwent the following heat treatment: 1075°C/3h/WQ. Figure 4.21 depicts the long
count time scans performed on the Ti-15A1-33Nb and Ti-21Al-29Nb fully-BCC
microstructures. Figure 4.22 (a) displays a [001] zone axis SADP in which the <100>
superlattice reflections are present and the associated BF image is available in Figure
4.22 (b). This SADP and BF image were taken from a Ti-21Al-29Nb sample that
underwent the following heat treatment: 910°C/3h/WQ. A SADP taken from a Ti-SAl-
45Nb sample that was heat treated at 650°C for 66-80 hours followed by furnace cooling
is presented in Figure 4.23. The absence of <100> superlattice reflections in the [001]
SADP confirms that the crystal structure of the Ti-SAl-45Nb alloy is disordered BCC.

The presence of the (210) superlattice peak in Figure 4.21 (c) and the <100> super

lattice reflections present in Figure 4.22 (a) confirm B2 ordering in the Ti-21Al-29Nb

118

sample and a lack of ordering in the Ti-lSAl—33Nb sample. From this it can be
concluded that nominal Al additions of up to 15 at% to Ti-Al-Nb alloys with a Ti content
of 50 at% do not favor a predominantly ordered B2 crystal structure and that the nominal
Al alloy content needed to favor ordering of the BCC phase lies in the range of 17-20

at%.

119

7550 I . ,7

 

 

 

 

 

 

a 700 ,
fl
:5
3 650 s
3“»
'5 600 , z
E
a
E 550 ‘ r
500606T ”62* 63* 64*”65
29,
(a)
650 2 ,
(210)
600 ,
a
5 550
8
32500 ,
.5 r
g 450, r
E
r' 400 ,
3 5 0 L ~~~~i~~—Lw 7 7477 7," #777, ”71,727 ,,i a,
60 61 62 63 64 65
29,0
(b)

Figure 4.21. Long count time scans between 60-65° 20 for (a) the Ti-15Al-33Nb alloy,
and (b) the Ti-21A1-29Nb alloy. Note the presence of the superlattice (210) reflection in

(b)-

120

.1
(020)

(’“0 (010) C
. (110)

,
"~””’ (401))

(100) .

«4.10) (0-10) .
'(1-0)

(300')

 

Figure 4.22. (a) [001] zone axis SADP and (b) its associated BF image obtained from a
Ti-21Al-29Nb 910°C/3h/WQ sample.

121

(020) (110) (200)

(-110) (1-10)

(-200) (—l—10) (0-20)

 

Figure 4.23. A [001] zone axis SAD pattern obtained from a Ti-SAl-45Nb sample. The
lack of <100> superlattice reflections confirm that this alloy has the disordered BCC
structure.

122

7. Equilibrium Heat Treated Microstructures

In order to determine the volume fraction and chemistry of the phases present at
equilibrium, selected Ti-15Al-33Nb and Ti-21Al-29Nb samples were vacuum
encapsulated and heat treated for 200 hours at 855°C, 910°V, and 960°C followed by
water quenching (it is assumed that chemical and compositional equilibrium is reached
after solutionizing for 200 hours). Backscattered electron images of these
microstructures are available in Figures 4.24426. The chemistry of the constituent
phases of these microstructures was quantified through EMPA analysis. These values
along with the measured phase volume fractions are presented in Table XI. EMPA
analysis was also performed on HT2960 and HT:1005 Ti-lSAl-33Nb samples and these
constituent phase chemistries are also given in Table XI.

Several differences in microstructure are evident upon comparing the
microstructures of the samples solutionized for three hours in air with the samples
vacuum encapsulated and solutionized for 200 hours in air.

With respect to O—phase morphology, it can readily be seen from Figures 4.24-
4.26 that an equiaxed morphology is preferred over a Widmanstatten morphology for
longer solution treatment times. This phenomenon is attributed to the O-phase laths
coarsening with increased dwell time. The samples solutionized for three hours exhibited
similar 0 and BCC phase volume fiactions compared to those solutionized for 200 hours
(Compare Tables VIII, IX, and XI). This suggests that within the first three hours of
solutionizing, the volume fraction of the phases approaches the equilibrium value, but
increased dwell time coarsens the Widmanstatten O-phase into more of an equiaxed

morphology.

123

The on; phase was nearly eliminated fiom all the Ti-21Al-29Nb microstructures
when solutionized for 200 hours. In each of these microstructures, the standard deviation
in the a2 phase volume fraction always exceeded the mean value, and in these
microstructures the mean value was always less than one volume percent. The
disappearance of the a; phase can be accounted for by the fact that it is not an
equilibrium phase at these temperatures. The significantly increased dwell time fiom 3 to
200 hours allowed for the on; phase to dissolve through diffusion of the Ti, Al, and Nb
atoms into the 0 and/or BCC phases. Since the a; phase was present in the AP material,
its removal fi'om the microstructure is due to increased dwell time at the solutionizing
temperature.

The effect of nominal alloy composition on phase chemistry is directly reflected
in the measured chemical compositions of the constituent phases of each of the alloys. In
the a; and 0 crystal structures, the Ti, Al, and Nb atoms sit on specific sites and this is
reflected by the measured compositions which indicate that each phase maintains its
stoichiometric chemistry relatively well. The composition of the BCC phase is
significantly different that that for the O and 0.; phases In particular, the BCC phase in
the Ti-21Al-29Nb alloy contained 5.1 at% more Al on average than the BCC phase in the
Ti-15A1-33Nb alloy. The higher Al content in the BCC phase of the Ti-21Al-29Nb alloy
is due to the higher nominal Al content of the Ti-21Al-29Nb alloy compared to the lower

nominal Al content of the Ti-15A1-33Nb alloy.

124

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Q3 5: wow Nu ".mm m.mN me mm «B Sc 8: o 052.5383 32%-? H Ni.
3% M: 3m mm m.m~ 5mm w. fl m Q. «E S: «E o 0320858 9.23.16, _ mi.
QR wdm m. _ m 3 man 5.3 :m mm 8: E: «E o Ogoofimmw 32%-? _ Ni.
m6». of «.96 mm on w. 8 NS 3 «.2 9mm mdo fl moofihm pmeA<mT§
msm WE cdv mm NS NR 9% pm v.2 mam mam m 89.5» anE<mEH
wém WE flow co 8: «E E: o 5.2 mdm $3 3 OBEooroa 922-3% TE.
93 NE New 3 «\a 8: E: m 5.3 w.mm Woo o goofioa 9.23-28 T?
m.mm 2 n. _ m me «E S: E: mm E: 3E «E o OBEoonmw pmeA<m TE.
$3 $3 $3 n> $3 $3 $3 m> $3 $3 $3 9» 06.30839; 5589.50
...2 ._< .a. £2 .2. .z. ...z .2 .F .8: 3?
23.5 00m 02:3 oEEeieatc 23.5 so

 

 

 

 

 

mogogmouomz £4??? USS—om com moEmEono 893 can maouofim oEBO> 3.3m 3.3302 .Un 23B

125

 

Figure 4.24. Backscattered electron images of 855°C/200h/WQ microstructures: (a) Ti-
15A1-33Nb and (b) Ti-21Al-29Nb.

126

4?:

f ,,-‘
'1 > ' .

ta .
, '1.

9"
.4165
:51.

l

.5 A

v .4

' ’4 2"

6 "J
W, I. .»

1

. 1".
A, -,

 

Figure 4.25. Backscattered electron images of 910°C/200h/WQ microstructures: (a) Ti-

15Al-33Nb and (b) Ti-21Al-29Nb.

127

 

Figure 4.26. Backscattered electron images of 960°C/200h/WQ microstructures: (a) Ti-
15Al-33Nb and (b) Ti-21Al-29Nb.

128

8. The Effect of Boron Modification on Microstructure

Table XII lists the measured chemical compositions and Table XIII lists the
measured phase volume fractions and grain sizes for the Ti-15Al-33Nb-xB and Ti-22Al-
26Nb-XE alloys. The targeted compositions were maintained adequately well in each
alloy, and it is particularly noteworthy that the oxygen content was maintained at a low
level in the Ti-lSAl—33Nb—5B alloy. The Ti-22Al-26Nb—xB powder metallurgy
processed materials exhibited significantly higher oxygen contents than their ingot
metallurgy counterparts. However, the oxygen contents were within the ranges measured
for other Ti-22Al-26Nb powder metallurgy processed alloys [Smith et al. (2000), Smith
et al. (1999), Rhodes et al. (2000)]. Figure 4.27 presents data from a DTA analysis
performed on the Ti-22Al-26Nb—5B material. The BCC-transus temperature was not
affected by the addition of 5 at% boron to the Ti-22Al-26Nb alloy as a value of 1130°C
was determined from DTA analysis for both the Ti-22Al-26Nb alloy and the Ti-22A1-
26Nb-5B alloy. The average composition of the boride needles, as determined by EMPA
in the Ti-15A1-33Nb-5B alloy, was 46.8B-29.0Nb-23.0Ti-1.2Al (at%). The average
composition of the boride needles determined by EMPA in the Ti-22Al-26Nb-5B alloy
was 46.3B-31.5Nb—22.2Ti (at%). The boride phase compositions reported are the
average of measurements taken from 20 different boride needles within each
microstructure. The measured compositions suggest a phase chemistry of BzTiNb.

Since the Ti-22Al-26Nb and Ti-22Al-26Nb-5B alloys were produced by HIPing
gas-atomized powders, the morphology of the powder precursors were characterized
using SEM and BSE images are available in Figure 4.28. Photomicrographs of the Ti-

22Al-26Nb-5B alloy are given in Figure 4.29. Larger borides, up to 158 um long and 22

129

um wide, were present in the Ti-22Al-26Nb-5B atomized powders and the HIPed
compacts when compared to the Ti-15Al-33Nb-5B alloy. Based on the BSE images, the
borides exhibit what appears to be faulting or coring from the center of the boride to the
boride-matrix interface, see Figure 4.30. It is noted that although no porosity was evident
in the HIPed Ti-22Al-26Nb-xB materials, the microstructure of the Ti-22Al-26Nb-5B
material was not fully homogenized as some of the gas-atomized powders were not
completely sintered. Thus, a portion of this material’s microstructure contained
incompletely sintered spheres as the original powder particles maintained their circular
boundaries as is readily evident in Figure 4.31. This is expected to have played a role in

the low elongation-to-failure values obtained that will be discussed in subsequent

portions of this thesis.

Table XII. Measured Bulk Chemical Compositions of the Ti-Al-Nb-xB Alloys

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Nominal Composition Ti, at% Al, at% Nb, at% B, at% 0, ppm
Ti-15Al-33Nb 51.4 15.3 33.3 0 1100
Ti-15Al-33Nb-0.5B 52.4 14.3 32.9 0.4 1450
Ti-15Al-33Nb-5B 45.3 15.9 33.4 5.3 130
Ti-22Al-26Nb 50.7 21.4 26.8 <0.15 1 127
Ti-22Al-26Nb-5B 47.4 21.4 25.6 4.5 1653
Table XIII. Measured Phase Volume Fractions and Grain Sizes of the Ti-Al-Nb-xB
Alloys
(1; 0 Bee Boride
Alloy Composition Condition Vp Vp Vp Vp Grain Size*, pm
Ti-15Al-33Nb AP 10 0 90 0 3
Ti-15Al-33Nb HT:1005 1 44 55 O 120 +/- 32
Ti-15Al-33Nb-0.5B AP 22 0 78 0 11 +/- 4
Ti-15Al-3 3Nb-0.5B HT: 1 005 0 69 31 0 130 +/- 29
Ti-15Al-33Nb-5B AP 1 69 21 9 48 +/- 12
Ti-15Al-33Nb-5B HT:1005 4 65 26 5 63 +/- 13
Ti-22Al-26Nb AP 6 72 22 0 59 +/-3
Ti-22Al-26Nb—5B AP 20 66 12 2 23

 

 

 

 

 

 

 

 

* Grain Size represents prior BCC grain size, except for Ti-22Al-26Nb-5B where it
represents all the equaixed phases present in the microstructure

130

 

 

 

 

 

0.5 —
processing temperature (1027° C)

0 w

-0.5 —

Temperature Difference, °C

 

 

_1 l b l l l
600 700 800 900 1000 1100 1200
Temperature, °C

Figure 4.27. Differential thermal analysis curve for a Ti-22Al-26Nb-5B specimen. The
sample was equilibrated at 600°C and heated to 1200°C at a rate of 15°C/min. The

' temperatures indicated on the curves represent the peak value of the exotherm/endotherm.

131

 

22Al-26Nb-xB gas atomized powders

i-

fT

unages o

Backscattered electron

4.28.
(a) Ti-22A1-26Nb (b) Ti-22Al-26Nb—SB.

igure

F

132

 

tinctures at (a)

mICI'OS

i-22Al-26Nb-5B

imagesofT

magnification.

Backscattered electron

' cation and (b) high

.29.

4

Figure

low

133

3pm

 

_, .,
...—.3

(b)
Figure 4.30. High-magnification BSE SEM images of the borides in (a) Ti-22Al-26Nb-
SB and (b) Ti-15Al-33Nb-5B. Coring or faulting appears to exist across the boride cross-
section. The cracking within the borides in is due to in—situ tensile testing, and is not an
artifact of processing.

134

.. ., ‘9 r

- viii-":1: ~

“-11: 3.1.. 1W?”
x "AV W t

4 'n,’
‘ ’[n 1
1‘ .
"3."- .
,

1.8%): '«i‘.

.. .,

$613,626 ‘
a: _i 9'93). ; "r,

reggx .. b ‘ ' .
a A J“

. V

on

‘56? '1 «“9

J" talk: '*
we??? . ‘ - (It.

 

Figure 4.31. Backscattered electron images illustrating incompletely homogenized gas
atomized powders evident in the microstructure of the Ti-22Al-26Nb-SB alloy afier
HIPing.

135

Backscattered electron images of the Ti-15Al-33Nb-0.5B microstructures are
presented in Figure 4.32. Figure 4.33 displays BSE images of the Ti-15Al-33Nb-SB
microstructures. Examples of Electron Backscatter Diffraction Patterns (EBSPs)
obtained fiom the boride phase in the Ti-15Al-33Nb-5B ally are presented in Figure 4.34.
They have been indexed according to the orthorhombic B27 TiB phase structure [Kooi et
al. (2003), Lu et al. (2001)] using the estimated lattice constants (a=0.6184 nm,
b=0.31262 nm, c=0.46833 nm) predicted by Trinkle through density functional theory
calculations [Trinkle (2006)], and are based on the average compositions measured
through EMPA analysis of the boride needles. Using Trinkle’s lattice parameters,
SADPs obtained from the boride phase present in the Ti-15Al-33Nb-5B alloy have been
indexed and are available in Figure 4.35. Therefore, the boride phase within these
microstructures is assumed to have the B27 orthorhombic crystal structure where half the
Ti (c) Wyckofi‘ positions (0.177, 0.25, 0.123) are occupied by Ti atoms and the other half
are occupied by Nb atoms. TEM bright field images revealed the faceted boride
structure with apparent faulting for the Ti-15A1-33Nb-SB material, see Figure 4.35 (c).
Convergent Beam Electron Diffraction (CBED) patterns obtained from low index zones
from the boride phase are needed to definitively prove the crystal structure of the boride

phase in these alloys, and this is a topic for future research.

136

 

(b)

Figure 4.32. Backscattered electron images of the Ti-

(a) AP and (b) HT:1005.

33Nb-0.53 microstructures:

15A]

137

 

        
 
  

9 :7 ' A, J ‘ v

mg: Rife»- :/~x=.v1‘.

t ,- »-:~«v r- 3.: w/ ;~
I “'4 '9‘ fl‘cfir‘ ”w (‘t
’ ‘44‘75‘5 13‘: i ‘»’ 3;.

ka- ‘7uz'v5‘_.‘ l .

, (11‘4". 2:75:75; 4 ’yoi‘ué

ZKZ' .s‘JlI} :é'l . ,Nr‘;".7;~_.;+
I" “91“ .- !y-L 1./, ' ‘1 ’5‘];

 

"‘5" \

. fir] ' \:n’
* We;
NW9 I 4n;-
v (5??!

5.:’°”".‘.., fl” - ”‘91.? ’9 ~W”

  

(b
Figure 4.33. Backscattered electron images of the Ti-lSAl-33Nb-SB microstructm‘es: (a)
As-HIPed and (b) HT :1005.

138

.1. 'I 1‘. .. ‘ ' ,‘i. h.“ . .“‘.‘_ a l
‘ 1-885' Fri-1‘53“ ' 2.797. . 1-1243‘ PH-SSB'

' O

 

(c) (d)

Figure 4.34. EBSPs obtained from the boride phase in Ti-15A1-33Nb-5B. The patterns
in (c) and (d) indicate the same patterns in (a) and (b) indexed according to the
orthorhombic 827 TiB phase structure.

139

 

(C)

Figure 4.35. SADPs obtained from a boride needle within a Ti-15Al-33Nb-SB sample:
(a) [130] zone axis and (b) [121] zone axis. The SADPs in (a) and (b) are from the same
boride, along the same kikuchi band, but are not from the boride in the upper left-hand
comer of (c), which is a BF image of a boride within a Ti-lSAl-33Nb-5B HT:1005
sample. Note what appears to be coring or faulting within the boride.

140

B. Creep Behavior Results

1. Minimum Creep Rates, Creep Stress Exponents, and Apparent
Activation Energies

The creep behavior of the Ti-5Al-45Nb, Ti-15Al-33Nb, and Ti-21Al-29Nb alloys
resembled that for most pure metals and alloys, exhibiting the primary, secondary, and
tertiary stages of creep [Evans and Wilshire (1985), Hertzberg (1996)]. In order to
determine secondary creep stress exponents and apparent activation energies, constant
temperature/load jump experiments and constant load/temperature jump experiments
were performed. This was accomplished by either changing the load or temperature after
a minimum creep rate had been achieved. The creep strain rates were deemed the
minimum creep rate when the strain rate remained constant with respect to time for at
least 50 hours under each testing condition at 650°C. For the lowest applied stresses
evaluated at 650°C, the minimum creep rate was typically achieved after ISO-200 hours.
At higher temperatures, the minimum creep rate was achieved in less than 50 hours under
certain creep testing conditions. Experimental data obtained from a typical load-jump
experiment conducted at 650°C is shown in Figure 4.36 and experimental data obtained
from a typical temperature-jump experiment conducted at 150 MPa is shown in Figure
4.37 . The minimum creep rates determined from the load-j ump experiments were
verified to equal those obtained from samples tested at a single stress and temperature.

The measured minimum creep rates and average equiaxed grain sizes of the creep
tested samples are given in Table XIV, Table XV, and Table XVI for the Ti-SAl-45Nb,
Ti-21Al-29Nb, and Ti-15A1-33Nb microstructures, respectively. Table XVII presents the
creep exponents and apparent activation energies calculated based on the minimum creep

rates measured for the Ti-5A1-45Nb, Ti-15Al-33Nb, and Ti-21Al-29Nb microstructures.

141

 

 
 
   
    
 

 

2.5 _ T=650°C
0:250 MPa

.\° 2 ” ”
a" 0:225 MPa
2 1.5 _ -
V; 0:200 MPa
8 1 .
I-
u

0.5 _

0

 

 

0 510 160 130 260 230 360 350

Time, hrs

Figure 4.36. Creep strain versus time data obtained from a Ti-15A1-33Nb HT:1105
sample during a load-jump experiment.

142

 

1.5

j I I l

6=150 MPa

  
    
     

T=710°C

.\° 1_ -
.5

fl

1:

m

G1

0

2 0.5 — -
U T=650°C

 

 

 

o 510 160 130 260 250 360 350

Time, hrs

Figure 4.37. Creep strain versus time data obtained from a Ti-lSAl-33Nb HT :1105
sample during a temperature-jump experiment.

143

The stress dependence of the minimum creep rate at 650°C is depicted for Ti-
15Al-33Nb, Ti—21Al-29Nb, and Ti-SAl-45Nb in Figures 4.38, 4.39, and 4.40,
respectively. All three plots are combined in Figure 4.41 in order to compare the relative
secondary-stage creep resistances of all the microstructures. From Figure 4.41 it can
clearly be seen that the Ti-lSAl-33Nb HT:1005 and HT:1105 microstructures were the
most creep resistant microstructures, while the Ti-SAl-45Nb alloy displayed the worst
creep resistance. Apparent activation energies determined from the temperature
dependence of the minimum creep rate are presented in Figures 4.42-4.45 for the Ti-SAl-
45Nb, Ti-15Al-33Nb, and Ti-21Al-29Nb alloys.

The multistep aging heat-treatments produced thermally stable microstructures.
This was confirmed by analyzing phase volume fractions from samples taken fiom both
the grip and gage length sections of samples after being subjected to creep for 350-800
hours. Differences in O-phase volume fraction were never greater than 2%, which were
within the estimated error (i 5%) of the phase volume fraction measurement technique.

For the Ti-21Al-29Nb alloy, the minimum creep rates decreased with increasing
equiaxed grain size from 8 to 12 pm. For the Ti-15A1-33Nb alloy, the minimum creep
rates decreased with increasing equiaxed grain size from 4-120 um. However, for the
120 and 173 pm grained microstructures, the minimum creep rates were similar, and
therefore the prior-BCC grain size did not appear to influence the minimum creep rates
for grain sizes greater than 120 um. Figure 4.46 is a plot of creep strain versus time for
three Ti-15Al-33Nb microstructures creep tested under conditions of 172 MPa/650°C.
Figure 4.46 shows that under identical testing conditions, the greatest primary creep

resistance was displayed by the supertransus HT: 1005 microstructure. Intermediate

144

primary creep resistance was displayed by the supertransus HT:1105 microstructure. The
worst primary creep resistance was displayed by the subtransus heat-treated, fine-grained

HT:960 microstructure.

Table XIV. Measured Minimum Creep Rates and Grain Size for the Ti-SAl-45Nb Alloy

 

 

 

 

 

 

 

 

 

 

 

 

 

Heat Treatment, °C GIT, MPal°C Minimum Creep Rate, 8'1 Grain Size, pm
650°C/66-80h/FC 10/650 1.41x104r 4500
650°C/66-80h/FC 23/650 2.38x10”“ 4500
650°C/66-80h/FC 34/650 3.40x10”8 4500
650°C/66-80h/FC 34/670 4.90x10‘Ir 4500
650°C/66-80h/FC 34/690 8.50x10T 4500
650°C/66-80h/FC 34/710 1.39x10'T 4500
650°C/66-80h/FC 40/650 5.04x10’8 4500
650°C/66-80h/FC 48/650 1.18x10'7 4500
650°C/66-80h/FC 63/650 1.90x10‘T 4500
650°C/66-80h/FC 75/650 2.53x10'7 4500
650°C/66-80h/FC 75/670 4.61x10’7 4500

 

 

 

 

Table XV. Measured Minimum Creep Rates and Grain Size for Heat-Treated Ti-21Al-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

29Nb Microstructures
Heat Treatment, °C arr, MPal°C Minimum Creep Rate, s" Grain Size, pm
HT:960 48/650 4.57x10'9 8
HT:960 48/710 1.01x10'8 8
HT:960 73/650 7.02x10‘9 8
HT:960 77/710 1.49x10'8 8
HT:960 99/650 7.64x10’9 8
HT:960 122/650 1.16x10‘8 8
HT:960 148/650 1.50x10'8 8
HT:960 170/650 1.98x10‘“ 8
HT:1005 48/650 363,110”10 12
HT:1005 73/650 6.31x10’m 12
HT:1005 100/650 3.18x10‘9 12
HT:1005 126/650 3.74x10'9 12
HT:1005 148/650 6.54x10'9 12
HT:1005 148/670 8.93x10'9 12
HT:1005 148/690 1.65x10‘“ 12
HT:1005 148/710 3.60x10‘8 12
HT:1005 172/650 1.10x10'8 12
HT:1005 225/650 3.80x10“ 12
HT:1005 250/650 5.93x10'8 12

 

 

 

 

 

145

 

 

Table XVI. Measured Minimum Creep Rates and Grain Sizes for Heat-Treated Ti-lSAl-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

33Nb Microstructures
Heat Treatment, °C o/T, MPaPC Minimum Creep Rate, 11'1 Grain Size, um

HT:650/250h/FC 49/650 1.07x10‘8 4

HT:650/250h/FC 75/650 1.43x10”“ 4

HT:650/250h/FC 99/650 3.1 1x10“8 4

HT:650/250h/FC 126/650 119,1107 4

HT:650/250h/FC 149/650 2.07x10‘7 4

HT:650/250h/FC 170/650 4.46x10'7 4
HT:960 49/650 1.04x10'9 6
HT:960 75/650 1.71x10'9 6
HT:960 99/650 2.48x10’9 6
HT:960 125/650 6.75x10‘9 6
HT:960 150/650 9.34x10‘9 6
HT:96O 171/650 2.22x10”r 6
HT:1005 148/650 5.78x10'T 120
HT:1005 148/670 8.36x10’9 120
HT:1005 148/690 1.51:110‘8 120
HT:1005 150/710 2.01x10’8 120
HT:1005 171/650 7.60x10'9 120
HT:1005 201/650 107,1108 120
HT:1005 226/650 1.53x10'8 120
HT:1005 251/650 2.08x10’8 120
HT:1005 273/650 5.19x10‘K 120
HT:1005 273/670 1.06x10'7 120
HT:1005 273/690 2.47x10‘T 120
HT:1105 151/650 4.81x10"’ 173
HT:1105 151/670 1.08x10‘8 173
HT:1105 151/690 124,1108 173
HT:1105 151/710 2.26x10’8 173
HT:1 105 172/650 8.65x10jf 173
HT:1 105 200/650 1.32x10" 173
HTzl 105 225/650 1.70x10'8 173
HT:1 105 249/650 241,1108 173
HT:1 105 249/670 7.18x10‘8 173
HTzl 105 275/650 3.39x10'“ 173
HT:1105 275/670 7.04xle 173
HT:1105 275/683 1.44x10'7 173
HTzl 105 275/690 1.82x10'7 173

 

146

 

Table XVII. Measured Creep Exponents and Apparent Activation Energies for the Ti-
5Al-45NB, Ti-15Al-33Nb, and Ti-21Al-29Nb Heat-Treated Microstructures

 

Alloy and Heat Treatment, °C I (III, MPal°C I n [ o/T, MPaI°C l Qapp, kJ/mol

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ti-SAl-45Nb
650/66-88h/FC 10-34/650 0.7 34/650-710 184
34-75/650 2.6 75/650-670 217
Ti-15Al-33Nb
HT:650/250h/FC 49-99/650 l .5
99-170/650 4.7
HT:96O 49-99/650 1 .3
99-171/650 3.8
HT: 1005 148-226/650 2.5 150/6SO-710 163
226-273/650 6 273/650-690 288
HT:1105 151-275/650 3.1 151/650-710 181
275/650-690 317
Ti-21Al-29Nb
HT:96O 48-170/650 1.1 48/650-710 100
HT: 1005 48-250/650 3.2 148/650-710 215

 

 

 

 

 

 

147

D

Ti-lSAl-33Nb HT:650/250h/FC
Ti-15Al-33Nb HT:960

O

 

 

 

 

10" .— - Ti-lSAl-33Nb HT:1005
t - Ti-15Al-33Nb HT:1105
"rm C p
.2? 4
E 10-7 T 1
1:. i
8 .
l-
U _
E 10”8 1 . I; 1- .
E : n = 3.8 r J n = 2-5 ‘
'2
10'” .“5 ."i 1 . T =85???
10 100 1000
Stress, MPa

Figure 4.38. Plot of log 8...... versus log a for Ti-l 5Al-33Nb microstructures creep tested
at T = 650°C.

148

' Ti-21Al-29Nb HT:960

 

 

 

 

10‘7 E—— . Ti-21Al-29Nb HT:1005 ....
'7
.2“
a 10.8 g ?
e. E i
Q j
i: : ~
U 1
E -9
.5 1
2 ,,

10.10 . 4 - .....1 . T=650°C

10 100 1000
Stress, MPa

Figure 4.39. Plot of log 8m versus log 6 for Ti-21Al-29Nb microstructures creep tested
at T = 650°C.

149

_6 Ti-5Al-45Nb HT:650°C/66h-80h/F C

 

 

 

 

 

10 -4 ff
.9.“
3
5 10'7 _ 1
E
5
E
.E
2
e ‘ T = 650°C
10 10 ‘ L L 100
Stress, MPa

Figure 4.40. Plot of log 8...... versus log 0 for the Ti-5Al-45Nb alloy creep tested at
T = 650°C.

150

-l

Creep Rate, 8

immum

M

 

 

L o//

10" :

1- I
1-

 

10-10 ”T = 6508C,

10 L100

 

1
Ti-SAl-45Nb HT:650/66-88Il/FC
Ti-lSAl—33Nb HT:650/250h/FC
Ti-lSAl-33Nb HT:960
Ti-lSAl-33Nb HT:1005
Ti-lSAI-33Nh HT:1105

Ti—Z l ..\|—2‘).\'h 11129610
Ti—ZlAl-29Nb HTleOS

T ‘ 1000

Stress, MPa

Figure 4.41. Plot of log 2...... versus 10g 6 for the Ti-5A1-45Nb, Ti-lSAl-33Nb, and Ti-
21Al-29Nb alloys creep tested in this thesis.

151

‘ Ti-5Al-45Nb a = 34 MPa
' Ti-SAl-45N b o = 75 MPa

 

'14 1 1 1 1 1 1 1
710°C 690°C 670°C 650°C
"-m -14.5 _ . ~
.2? Qapp = 217 kJ/mol ‘
g -15 — ..
he
3 -15.5 — 4
L1
0
E -16 — —
E
g -16.5 _ -
5 -17 ~ .

 

 

 

 

-17. l l L I l l 1
10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9
10,000/T, 1/K

Figure 4.42. Activation energies determined at o = 34 MPa and o = 75 MP8 for the Ti-
5Al-45Nb alloy.

152

’ Ti—15Al-33Nb HT:1005 0' = 150 MPa
' Ti-lSAl-33Nb HT:1105 0' =151MI’a
-17

 

710°C '690'°C ‘ 670°C' 650°C

-18.5 —

 

In Minimum Creep Rate, s'1
.L.
\D

 

 

 

—19. l I l l l l 1
10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9
10,000/T, l/K

Figure 4.43. Activation energies determined at a = 150 MPa for the Ti-15Al-33Nb
HT:1005 and HT:1105 heat-treated microstructures.

153

’ Ti-15Al-33Nb HT:1005 0' = 273 MPa
" Ti-15Al-33Nb HT:1105 0' = 275 MPa

-14.5

 

710°C 690°C ' 670°C 650°C

-1

1"
UI
1

.q

..'..

y-

u:
1

1L

9‘

m
1

 

_Qapp = 317 kJ/mol

In Minimum Creep Rate, 8
.'..
ax

\\\\
'.-

 

 

 

..17. 1 1 l l I
10.3 10.4 10.5 10.6 10.7 10.8 10.9
10,000/T, l/K

Figure 4.44. Activation energies determined at a = 275 MPa for the Ti-15Al-33Nb
HT:1005 and HT:1105 heat-treated microstructures.

154

’ Ti-21Al-29Nb HT:960 0' = 48 MPa
' Ti-21A1-29Nl) HT:1005 0' =148 Ml’a

 

 
 

-1605 I I T I I I I
710°C 690°C 670°C 650°C

7m, -17 _ -
Q I

*5 _

m -17.5L -
8 ~~ e

0

5 -18 % . » Qapp = 215 kJ/moh
E 1‘

g -18.5 L .‘ —
Is

2 -19 I —
E: Qapp = 100 kJ/mol

 

 

 

-19. I 1 I I | l I
10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9
10,000/T, 1/K

Figure 4.45. Activation energies determined at o = 48 MPa and 0' = 148 MP3 for the Ti-
21Al-29Nb HT2960 and HT:1005 heat-treated microstructures.

155

Creep Strain, %

Figure 4.46. Creep strain versus time plot for heat-treated Ti-15Al—33Nb microstructures

1.4—

1.2

' Ti-15Al-33Nb HT: 960°C

' Ti-15Al-33Nb HT: 1005°C

' Ti-15Al-33Nb HT. 1105°C

 

 

o=172 MPaT=650°C

d=6um
L
Awnlnflnlfln ld=12101lm

.__.

 

 

Time, hrs

creep tested at 172 MPa/650°C.

156

2. Proposed Deformation Mechanisms

The two supertransus heat-treated Ti-15Al-33Nb microstructures displayed the
greatest creep resistance of all the microstructures tested. Over the stress range of 148-
226 MPa for the HT:1005 microstructure and 151-275 MPa for the HT:1105
microstructure, constant 11 values were obtained which implies that a single deformation
mechanism was dominant over this stress range. The 11 value was 2.5 for HT:1005 and
3.1 for HT:1105. The Qapp values determined at 150 MPa were 163 kJ/mol and 181
kJ/mol for HT:1005 and HT:1105, respectively. At 275 MPa the Qapp values were 288
kJ/mol and 317 kJ/mol for HT:1005 and HT:1105, respectively. For the stress range of
148-226 MPa for the HT:1005 microstructure, grain boundary sliding is the suggested
deformation mechanism. For stresses above 226 MPa dislocation climb is the suggested
deformation mechanism. The singular 11 value for the HT:1105 microstructure suggests
that one deformation mechanism is operating from 151-275 MPa. However, the different
Qapp values acquired at 150 MPa and 275 MPa suggest different mechanisms are active.
Due to the similar creep strain versus life behavior of both microstructures, the
conclusion was drawn that the same deformation mechanisms are operative for the
HT:1105 and HT:1005 microstructures.

The Ti-15A1-33Nb subtransus heat-treated microstructures exhibited worse creep
resistance than the Ti-15Al-33Nb supertransus heat-treated microstructures. The Ti-
15Al-33Nb HT:960 microstructure had an 11 value of 1.3 from 49 to 99 MPa and an 11
value of 3.8 from 99 MPa to 171 MPa. These 11 values suggest Coble creep or grain
boundary sliding deformation in the low-stress regime and a transition to dislocation

climb creep deformation at stresses above 99 MPa.

157.

The Ti-15Al-33Nb HT:650/250h/FC microstructure had an 11 value of 1.5 fi'om
49-99 MPa, which suggests either a diffusional mechanism or grain boundary sliding to
be active. A transition to dislocation-climb controlled creep is suggested to have
occurred from 99-170 MPa (11 = 4.7 for this stress range).

The creep parameters calculated for the Ti-21Al-29Nb microstructures suggest
that a different mechanism is active for each of the microstructures examined at 650°C.
Indications of a Coble creep mechanism are shown for the HT:960 microstructure, which
had a creep stress exponent value of 1.1 over the total applied stress range of 48-170
MPa. An apparent activation energy of 100 kJ/mol was acquired at 48 MPa for this
microstructure. The HT:1005 microstructure shows characteristics of dislocation glide or
dislocation climb controlled creep with an 11 value of 3.2 over the applied stress range of
48-250 MPa and a Qapp value of 215 kJ/mol determined at a stress of 148 MPa.

The single phase Ti-5Al-45Nb alloy displayed an 11 value of 0.7 over the applied
stress range of 10-34 MPa, with a Qapp value of 184 kJ/mol acquired at a stress of 34
MPa. Over the stress range of 34-75 MPa the creep stress exponent transitioned to a
value of 2.6. and a Qapp value of 217 kJ/mol was obtained over the temperature range of
650—670°C at 75 MPa. This alloy shows indications of Coble creep deformation in the
low-stress regime and grain boundary sliding in the high stress regime.

The deformation mechanisms suggested in this section are based on the calculated

n and Qapp values and will be further discussed in Chapter 5.

158

3. Creep Deformation Behavior

Creep stress exponent values and Qapp values allow one to suggest the dominant
secondary creep deformation mechanism. However, they provide no direct
microstructural evidence of the deformation mechanism. To obtain direct microstructural
evidence of the potentially dominant deformation mechanisms, an in-situ creep testing
methodology was developed using the Ernest Fullam tensile stage described in Chapter 3.

Creep experiments were performed on metallographically prepared samples,
using the in-situ stage placed within the vacuum chamber of the CamScan 44 FE SEM.
Performing creep experiments in this manner offered three key advantages in observing
creep deformation. First, the integrity of the sample surface was maintained for the
duration of the creep experiments because they were performed in the vacuum
environment of the SEM chamber. This allowed the detrimental effects of surface
oxidation during creep to be avoided. Second, the surface deformation was chronicled as
a function of displacement and time, as multiple BSE images were acquired throughout
each experiment. The creep displacement and time data were continuously acquired as
well. Third, the creep deformation was observed to operate while the sample was
deforming by creep. The third advantage provides one with direct evidence of which
microstructural feature controls the creep deformation behavior.

The n and Qapp values for many Ti-Al-Nb alloys [Banerjee (2006), Boehlert and
Miracle (1999), Hayes (1996), Rowe and Larson (1996), Boehlert and Bingert (2001),
Cowen and Boehlert (2006a)] have suggested grain boundary sliding to be the dominant
mechanism in the low-to-intermediate stress regime. To the author’s knowledge only one

study has shown through fiducial line experiments that grain boundary sliding and grain

159

boundary cracking contribute significant amounts to the overall creep strain [Boehlert
and Miracle (1999)]. Therefore, the main objective of the in-situ creep experiments was
to observe and characterize grain boundary sliding and cracking in the Ti-l 5Al-33Nb
alloy.

Before the capability to perform in-situ SEM creep experiments existed, one of
the main methods to show experimental evidence of grain boundary sliding was to
fiducially mark a metallographically prepared sample, and image the sample surface
before and after creep testing in a vacuum environment. While fiducial mark offsets can
be imaged after creep deformation and local strains can be calculated from these offsets,
there is no capability to actually witness and chronicle the grain boundary sliding
deformation phenomenon occurring while the sample is continuously experiencing creep
deformation. Using some of the previous methods, the creep test must be discontinued in
order to image and record the deformation at a given strain level. Therefore, the creep
test must be stopped and restarted from RT and zero load every time imaging is
performed. This requires the sample to experience a new stage of primary creep
deformation each time the sample is removed and then reloaded. The advantage of this
methodology is that once the creep test begins it runs continuously. The sample
experiences the same creep deformation it would experience in a conventional constant-
load creep test.

The Ti-15Al-33Nb HT:1005 microstructure was chosen for in-sim creep testing
because its 11 and Qapp values suggested grain boundary sliding to be the dominant
deformation mechanism over the stress range of 148-226 MPa at 650°C. Two

experiments were carefully chosen in order to directly observe the manner in which the

160

microstructure deformed during creep. One experiment was performed in the n=2.5
stress regime. The other experiment was performed in the n=6.0 stress regime, with the
main goal of directly observing grain boundary sliding in the n=2.5 stress regime. Figure
4.47 is a plot of displacement versus time for the Ti-15Al-33Nb HT:1005 samples creep
tested at 650°C. One sample was tested at a stress of 225 MPa and the other sample was
tested at a stress of 300 MPa.

The evolution of grain boundary cracking deformation as a fimction of creep
displacement in a Ti-15Al-33Nb HT:1005 sample tested at 225 MPa/650°C is presented
in the BSE images in Figure 4.48 (a)- 4.48 (11) (Figure 4.48 spans pages 164-170 and the
loading direction is horizontal in each image). Figure 4.49 presents BSE images of the
most extensive grain boundary cracking/sliding/upheaval observed in the TH 5Al-33Nb
HT:1005 sample tested at 225 MPa/650°C. This experiment verifies, through surface
observation, that cracking and crack accommodated sliding of prior-BCC grain
boundaries is a dominant creep deformation mechanism operative at 225 MPa/650°. The
following sequence of events was observed. Cracks initiated within prior—BCC grain
boundaries first. Afier enough cracking occurred to accommodate the grain boundary
sliding/upheaval process, all subsequent deformation was observed to occur
predominantly by cracks opening to allow grain boundary sliding/upheaval. Sparse
microcrack nucleation within some of the O-phase laths located within prior-BCC grains
was observed, but was not as extensive as the deformation experienced by the
cracking/sliding/upheaval of prior-BCC grain boundaries. The first microcracks within
the O-laths initiated at a creep displacement of 166 um, and were measured to propagate

to a final length of 4 pm at a final creep displacement of 412 pm. This experiment not

161

only confirmed that grain boundary cracking followed by sliding/upheaval was the
dominant creep deformation mechanism, but also showed that BCC phase in these alloys
experiences the majority of the creep deformation. The prior-BCC grain boundaries are
the dominant microstructural feature controlling the creep deformation under the given
testing conditions.

The evolution of creep deformation as a function of creep displacement in a Ti-
15Al-33Nb HT:1005 sample tested at 300 MPa/650°C is presented in the BSE images in
Figure 4.50 (a)- 4.50 (h) (Figure 4.50 spans pages 172-175 and the loading direction is
horizontal in each image). The prior-BCC grain boundaries were also the locus of
damage under these testing conditions. Cracking, followed by sliding/upheaval of the
prior-BCC grain boundaries also occurred at 300 MPa/650°C in this sample, but differed
from the sample creep tested at 225 MPa/650°C in that sliding/upheaval was not
observed to occur during the secondary stage of creep at 300 MPa/650°C. Grain
boundary sliding/upheaval did not occur in this sample until displacement levels in the
tertiary stage of creep were reached.

The cracking and accommodated sliding/upheaval of prior-BCC grain boundaries
dominated the creep deformation observed for the supertransus Ti-15A1-33Nb HT: 1005

samples creep tested in-situ.

162

' Ti—15Al-33Nb HT:1005 Creep Tested at 225 MPa
' Ti-15Al-33Nb HT:1005 Creep Tested at 300 MPa

Samples Were Not Taken To Failure

 

 

 

 

 

 

1.2
1 _ _
a
E ..I
E
0
E _
8
a
a: -
a
0'2 é _ =2.371r10”"t~;l ‘
mm T=650°C
0 I I I
0 50 100 150 200

Time, hrs

Figure 4.47. Displacement versus time data collected during insitu creep testing of the
Ti-15A1-33Nb HT:1005 samples. The black curve represents a stress of 225 MPa and the
red curve represents a stress of 300 MPa.

163

1| - hf. . ' «\"I ‘ ' V/r’ //\‘WV“2:‘W:{: 7. "’t " ' 1, t-
I 1‘ J07," V fix "LA - 513V " ‘7‘- H
" .11 . " ‘ N§§§V§V~ unwrap
.II' ' ii. .1 I'D-k ‘1“! (u‘r‘ {‘4' th/
"I "”w » . 3“ ' K “1,2 “\fl» «Tu L, '2'; .
' ' , W1%}i = We emigre-z .

‘ V 4‘ ./ ” ”‘7' " V‘ 4'. .1 c

I
$9
6

    

   

A;-

:14
IF
‘\
PUT
\
VA"
‘ 2:, I
‘7.
he
77,—.
r
,1 -

A" -.
1‘91 .
. '1'

r!“
13%:
21'!
\‘
S
{If
3;
bl_ 1 _
\‘fitvitb 1.4.54
, '1 :7)‘«S_ ' 1’
A ’.' ,‘z‘éik .
fr:
011
" :r
i ‘71
fl I
1.5
‘1IK
231‘
1,.
JR»
"1
.11
44" <
‘4'

.
L5

1

:2? .1

. F

153
$0
- N
1.77"

-;;£‘.’::-. ( W53,“
.3 8:91:5174‘.

.'
‘\

L.»—

.I

1111;.
7'“? ’
r\ 7‘E‘t‘7
(‘4‘). 1 «I

,5 1
7 .1,
:42?

‘v 5
$8.21:
. IKE?”

9’2.

“ ”’2;
L ‘

JR;

\

a?“
LN
‘1‘".

. a
v»...
é-

u!

x' 1‘» ;\\ii\‘, xii",

1"; " ‘ V‘I ‘,_ .

- Isl ‘ ., _ \ .. A»); . ., ,1 . g
{5’1 \ "’ ~54’ $1<§5$$é§$7 95;“ {29$}? .5181: A? . 1:71-99"
.r— \ ‘ l/z ,5‘.$‘j,\\~ A0.\\ j /'4.;:.';(/ ,, {A [14‘1“ , ..

‘ 30 m \\§>.<. ‘ N: “a," .‘J *-"e.’\‘v‘.‘.\:.v-ta'.‘. - Pretest
Ilfltlml‘wl\ ”\1‘1 «4"». 7.7“ " ‘1 . \7‘.~ fibwafi .\, ' Y’s“: .

I—‘d...

\B\

’1’ /

   

 

"5" * r .- ' < a": +1 L; agate

\~
1:? .ri’sfi'y:

2: " , .- r s “glee/1.1.42.

“fifevzz ' / ”52?"-'“-l‘\ “311‘

8388:2111 ,gfl,;’1.ri\flli0.
Six/A « /. 1.

, .. ,. V3": ~' «am 17:41 V's-.5371-
‘~ . 1‘1» - ~ W /‘-5“ wilr \\ ’ -. ‘1

- ‘ (71",; " . K, /A’V’; A\._‘.‘.$ ,ir/J‘ Yme .p

_ _, 1 ‘ - .. '3 - .7 .. v/.i®1\~fi‘gm%t’;afgff 9r ,5 -15

" "7 "~2‘::':?:€ 0 r... ,1

(b)
Figure 4.48. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-15Al-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand comer of micrographs (a)-
(n). The loading direction is horizontal in each image.

164

r 4 fixxvm M4 . g‘..§..—J.A., yl
... 44.2.24 M4. Cher/... 347/ .4. 1. A4 e
.444. .4”. Jam. 4443.11411/ ‘(IHV/ m ,4:
1.1.7 an.” .2» flat/... :4 “3‘54... 1:...
A

44%.? ......a. PWuW .;
.4 ...;n ...mmx 8.5.11.9 . EV .
p.4..4.....,......,.../ 64...: \ee.

.44.... ..4. .14. «.34.. .. \ .
4......,....,,. 24.3.4476 1J4ufiowu.\4m\ 4.4 ......A
......44444? .1... 14%.
$1401.44...” 2W... R54 444...».74 :\

......tt... Jam” .\_... ..44“...

flag» “‘1 0. LI.“ §..W\\\

5.... ..- 9‘. 4,. ...
3. Ling.”

0‘5
yr—“

:43

       

m
m
I.
3
0

'—."ll

:37

  

     
      

«ilk-.1” 4
fix ..3. m...
4%.... “1.04444 “543V §W440m
7...... ...... .84.... ..

.../Ii 7 44.1! ,

4.4.4....
. . fidnmvhmflmx
U, ”43% .\\.I Adv/e.

(d)

Backscattered electron images illustrating the evolution of creep

deformation in a Ti-15Al-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand comer of micrographs (a)-

(n). The loading direction is horizontal in each image.

165

Figure 4.48.

    

. 1... . ,
... .. ..2... ..14.V.1,../............

  

 

.1/.. .14... i. :4, , (4...? . 5...”... .....k... 4 ..., , . .....:.._. 4 . a}
\ 4..N:r‘>.wrww.‘l~y47.4U((1\v4!14‘r\”.417u1‘~ all“. . I... 4 .. ..44meb‘Jll, r... .W( _V11{ ”PM: ..
Aywk 44.... ..V, I. .‘ x ..s 11/ ., . 1. 3"... ’/ 3
\‘ V 4 uufiflvnl’lx 4% —I.\a‘..z.z. 141$.ru174 A m 1.. . 1 1. ...414‘ ...! «JUN, (A —‘u.\..£\s/.d. Fin! .Vv m
£311.91; 44.4.1.6..Qi) .v... ..m.....w.sw..1...4r m .,.. _ . . 4....~1\H../ $41.. Wu. m
<\»\1V..r ......54. 41’“. ”I.Q4;.!. \1.1«4.. ..1 fl .IJW,\V4\14V..J .,.... .,...\I. F 1...!» 9.44wa v.\ \wkfi
,...mwu... 441...».441. La...“ ”.19.... ...... a ,. .\\. 11.41.41. .11.... 4A»? 4...: L :3»... \
KEV 4.].w4..v 4 ...l, 4?. ...fl “...—W? . \41. .m4v1...“_.,k.\..& 0 .wmup\.’(“.,44..4 t 1.4.7.0...» Ilka... VX1114“-
1¢IF44 R4442]! (4.01%.? ”4W1/: .. .4. «A4144! (”fix ...rr.._..v.,\‘r\.1-H/A.Vas....h .' ”$471144 {avg/.0 01%»). v..KN££7.VIW\W.4,M.n4.I.l/\n 1t... .
.... ._ €441... V...»1 ..1....€.x\..,.. 3.. ..44 «@1ng ..aks1....,......u1...., u. ,
. k.\\.. 4.49 4.4/«..44». .1\4../..éf1./$ V ‘9 blll..$ww1 4W4.41\.1.V1: JV /4. 3.1/v1.4 4
FL5HU V53... 1./»..) 4....» ., .147. . 1.1 \ ... ,. . I .....l.w1/....\r,...d .1.! ..4 5.x \.
11.7,‘y/1x420i 4
/

llama? ..VV awn

   

     
    

     
  

  
         

. ,q r! 4 .\. .
/ 1A4. N 4.4444914... 441.544.. ., MM: .4..,.....,.. .3 , WNWJ...» 1-xm\...my
7 .1. . .. . .. 1A.. . ., .11.. . . \.\._|. T l 41.4 .... ,. ..I...
“WM“. . ...: .. ... .,.. . .1.... .... ...wflr . .11.... \ ....

4

(f)

Backscattered electron images illustrating the evolution of creep

deformation in a Ti-lSAl-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-

(n). The loading direction is horizontal in each image.

    

“PM“ ,4 1...... Mw1g444311. ...le. \Mk‘
\Ha N {Skis 0.x.»
. 4 (P4114? 51.1

\dr , M «Emu. JIWMWV .9.

kw...\\.l1.\\\|\\..lr;uw /.u flwmxfii ”1.04
..1... W .

   

 

4.

 

. .,., V. . .
44.1.3444. a m a
mew..44m4afi w 4
. p‘\11; ”my ‘4! W

....

166

. \

   
 

  

‘55..

  
 

   

r
'1 In,

 

A [void ‘1" l x c»? 3' \‘"*x~*:_ . v w '
995% ‘.~"\"M‘-\ \.~ t, L71 ”/4" J WA nil...
-1 QVVCQ¢ “"’
5 ‘ / 1 ,

\ 3
V " Efl
‘ «A:

”74.

\
,
1 A
1’
x

:17,

l:
/

,1. :
l/Zuy

M4
In.
‘ P”

Q."

1
as
I.
-
0‘,
"He“

"\
”PE"
.1574!
47""

v
t

V‘
. ,a‘ ,4-
v (any, “
.,, .
4 .- ., 4.‘ v -
‘yv 4 1;. "-
I ‘\ 7’2/42‘
:’ ..
A.\~4¢h
_._

        
    
  

  
             
   

flk‘,¢ ‘- -\J
\. _ , ”31‘,

III ‘\\(~ x ‘,_ ' . . I! a,“ .7» ~ ‘ . ‘4‘: .I '7 _ ‘ .
W‘Al \ “\i r7? 2% ~ 15‘? xv IL!
- 1¥\@;§\&nul.‘i Nbszéjg‘? Jy'flill.

g; #3: it}??? V‘fififii‘, gag» alga, ‘1 A}! . 1.51915

\ »\‘ .443!” ? \me 37. ~ Yin/$41;

4335~ fiNt/Q"}WL
9159”".3m3g} '5
r- , flaw: tag-3e- » »

    

Figure 4.48. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-15A1-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-
(n). The loading direction is horizontal in each image.

167

                  

v .,. ,.s‘i 3‘ « x. in. _, .. .. .. ...:Uhfl. ., .
,4 xix wits/£4 m
.....4 4 . )2 . .
,mevmkifm .....éyfiig} m
«hair! . 3.11. .wh..3.
.7; #5:“ 1:”. .V i. ,..... .

Ni «x..- w.
.....RM .5112 n . ,.VuMCch 4:3,...
..r

_ J, .01 FW. r A ..
.vu\\ f 7..“ . .... . .r ,..»: VWNW.<\....
Nilflg.» Fa 1.w..¥..m...,.. .... (it? :menhrnceow.) x
finifi . 5. .. . damage 3
, . . 1‘ , I : a) g d
. A

.r._.2flvuwmym«wwm \. .. (...... 2
ix v \ , ,... u... .. ., .

     

t ..
a x .3
M3 _. A a V

gawwxxx/ » 3M. .,mv ...
Mu. .,émsmx ... Kwflmww/j .,
“w“ 2%» . ,

in? w
_ . y,
yfi. \wa ,

. 7; . «1.11.11: ,

               

... . , z ,
_ r I . V fiwb
m; </ i t i ,
iftvflfmxmw pm mo‘hisz. . ..
,i ,.. ,\ if? dfi/ Arno \ «Hy/(k.
. /Aflmhfivmw.wlgimxoi 4 5

FHA 4V . ~.. a . ‘ a“: , .vi 5.1}. 4N». M
\. 4. H f“? , fig:

.

\ n a W
I 4 lg ; . ,.
.. Iflxmmmwwoww..\v NxfifiwWWCflh . . .. .. . f .. J... .

i
t x ... 5.
fl - . ME. as
1%,, i... ,. £17

”4,.
/a

t “x. m ...» \t i
a m; _

/

    

I in

Dr

WJbQY It

3,: .-

“\_\ ~.. ‘\m“finu' .. i , .

_ wfiJfl.‘ ..F\m. ......” .4»... ,.\\.. .
:3 iii 1.6a «.... .,

...r.“ ..w...’ Q 5‘“... by]. m. $34,”... .

 
   

 

(i)

Backscattered electron images illustrating the evolution of creep

deformation in a Ti-15Al-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-

(n). The loading direction is horizontal in each image.

168

Figure 4.48.

         

 

-.....15 amt“... “IVA... :. l: . . arfiiNr 4.44:. {7...‘11111 .M .....at?

. l 4 .4; l Av $5.. .4. \’./.,w “.“A xV/n/ l m
#19:». .4...H‘Mi.a\ ._ “.11....” m .13.! (4‘5): 4..\ 1.4.. 411.3???) fin u. \.VL _ \va.” m n
7. .Ir..\\. .. .. . .1.. m _ . ud¢\ a fit... .7. 1344...... mu
...k... #41,... .,.. 1 . 2.x .. . :br.fi.....a\\$..: u
\\\.)!44Isv I. 4f . 55 ...l §ldvI ‘1va.. \LFFLIEWWM‘ 0A.....W..l4. \ \ M. -
Ammkmx4 34. W/WI: 0“ Pb 4.1... \w. ”VJ-W. 0..! on

14...,41144v.\w.4. W ...... .. ,. kl I.) DIV/4‘ ....«.. gym” ALA . ......PVV. <

adv...) ,.. .4 “a W449 A» a. . ,../,...

4_ t. :7 , a . xvi)”; 5\/ 7 >74. 6;. .4
4/ /.(-. .Mwuifl WNW-”~44: .1”... ~ W K ..r .\\.\1-\ \wxv’WNWV/ my \st§‘ ‘1. “off 7(4JV W] ”was.

. ..v {\J ~

           

.. .V. .. is. M.
\> 4 \v.

A: ...? . .. 7 *9...
. .. fim . F 44.4mm»... ix

 

W1;

 

        

Il‘l
I‘\

     

...
«M4
as».
.\ \‘v .
u. _/ .
m .,4.
u.“ .\ x
. ...
‘3‘; t... L x

(1)

Backscattered electron images illustrating the evolution of creep

deformation in a Ti-15Al-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand comer of micrographs (a)-

(n). The loading direction is horizontal in each image.

Figure 4.48.

169

,7W:V¢ ul rm Hp. Trill . » Ix 1;: . 3‘2) Vl.‘ “7...qu 4.:
rill; km a : .1 E/JNJ ,wwl ..«AFHTQG
4 ) .- ..(‘fi‘ . ...-K“

 

fleiflwrb A: 35‘. m
“1' (A .lllfli v“. r}. . .\..‘..._\.a .. ‘1', .A.w.v\.7m:k., . .IKVL: ..

V .A ... a, «wxy/iwiur 0.
fire 1 v leg «newsman. ,.
a m __,. _ kafilwilfilwg/ 1.: .<.. :x ,2
.,A .,. mm i @an ‘, £5 w... ..2, 2.2... .
rfl , x x. , ..VJ. , . , ,. .
f4 (3‘ )1! «.1. . . thc/Ju‘nwAP [I .‘\...\e W
«$35: Mimi. _. {,.. ,3 s.
..b. , , «Nu,....&un_x..am... .,.» , A? ‘
.aéflmirrltuf? % Swivihxlx .

; Y7A7A2~Hf

 
  

.P
on.
w

     

     

             

, RLYMJ 1&3 ..
V ar/f.

     

     

30 pm

.‘I\\ N 3" n.

or \M a

.a
\.. st
$1.? _. x} it

 

(n)

Backscattered electron images illustmting the evolution of creep

deformation in a Ti-lSAl-33Nb HT:1005 sample creep tested at 225 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of microgmphs (a)-

(n). The loading direction is horizontal in each image.

170

Figure 4.48.

,. as

a A . : be?

F/b ‘

~ Jl.
.. his"! .‘ , . .
. .w, w”. . H.
\ /, r \ .

5,

.
”Hume
. 4].! .

 

“f . A ,l--.

(b)

Figure 4.49. Backscattered electron images showing

right-hand
ntal in each image.

indicated in the lower

on is ho '

the most extensive grain boundary
1005 sample creep tested at

' observed in the Ti-15Al-33Nb HT

inmmis

8

hs (a)—(h). Thel

225 MPa/6563?}. The total displacement

micrograp

U

 

corner of

171

«—

(b)
Figure 4.50. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-15A1-33Nb HT :1005 sample creep tested at 300 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-
(h). The loading direction is horizontal in each image.

 

172

0.61 mm

 

Figure 4.50. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-lSAl-33Nb HT:1005 sample creep tested at 300 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-
(h). The loading direction is horizontal in each image.

173

‘~ v \, ._ . .
_ , A .,.: a.» .,.; w 1’ , .
' .4» ..e: w”! Irv/xix;- .- i «a “a, . ’
7'71 ‘4” v‘fi‘kmfkv» -.
1" V.” falfl‘fl ‘iv ’d"_(‘\ $13555" g fjaxz' "’
u or". \f‘uvlif.’ I, ,, §.\\ 1;»;- .. ..‘T . ,
/'M I A} \ '

~f " .153" “\M‘
. V/n / V’ ‘ ”V
,1 a a“ ‘ J
v: I & Q": ' “ \.\‘" _¢.
(‘75: *l ' ./ l2! 3‘ “
, v :3" 41 "j “‘“f ,7 {at
I .. '\ .. 7 ‘ , \7'. ‘ f i?“\
.1"— , } a A, "

 

(D
Figure 4.50. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-15A1-33Nb HT :1005 sample creep tested at 300 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-
(h). The loading direction is horizontal in each image.

174

..

J‘- . ‘2\:F..,1_:f—-" ftp/IQ“ '“V-i ', '7"-
it a? "A" ’ /’ fl

     

Nil; ~ - . g —
x -‘-r - r
t t " H-
,Y ~ ~ g , ‘

    
       

   

  

    

 
  
  

        

 
 

-“" ./_ x .. . r-«HI' .A‘
‘ 9 q ‘ l “J. ‘. -’, ' .
-- .‘ 2.25333 »,» g. 121,»
. 4V... “3 . _‘ _“ ',;.’/__ ’v
I .->._‘ If“ p \w ‘ r 2“ ' ,g-x-z .
x it." as. c. \J’,» g; )
, . l--/.x'- ..\- ivy \ 11> \.
t \ , 4:13., :11". ,'~ . x .
r fixv.‘ 5;; a."
.x ‘1" ",A(i .1 ._ ~‘,' ..
‘ ‘ L " ‘ (is
.

       
  

,"1 t1 "(7 ‘.
“ta
“‘7"

’4

 
   
 

   
  

 

a

_ fix"?

 
 
  

   
  

(b)
Figure 4.50. Backscattered electron images illustrating the evolution of creep
deformation in a Ti-15A1-33Nb HT:1005 sample creep tested at 300 MPa/650°C. The
total displacement in mm is indicated in the lower right-hand corner of micrographs (a)-
(h). The loading direction is horizontal in each image.

175

The in-situ methodology developed allowed the evolution of surface deformation
to be chronicled during creep deformation. However, sub-surface observations were
needed in order to correlate the surface creep deformation observed to the creep
deformation observed within the bulk. A Ti-15Al-33Nb HT:1005 sample was
conventionally creep tested in air under test conditions of 172 MPa/650°C. At 650°C, a
stress of 172 MPa lies within the n=2.5 regime for this microstructure. This sample
failed at a creep strain of 1.7% after 820 hours. Severe edge cracking is evident in Figure
4.51 (a), which was believed to be a result of environmentally assisted oxygen
embrittlement. Figures 4.51 (b) through 4.53 show BSE images taken from the interior
of the post-mortem sample. Internal cracking is evident and was localized to the prior-
BCC grain boundaries, which is in agreement with the in-situ deformation observations
obtained in the same stress regime for this microstructure. The amount of grain boundary
cracking observed on the sample surface was more extensive than that within the bulk.
The degree of constraint on the sample surface is much less than that within the bulk,
therefore a greater extent of cracking and sliding/upheaval would be expected to occur at
the sample surface when compared to the interior. Edge cracking did not occur
preferentially over prior-BCC grain boundary cracking for the sample tested in vacuum.
The edge cracks were also observed to preferentially propagate along the prior-BCC
grain boundaries into the sample interior. Therefore, the environmental stability of Ti-

Al-Nb alloys at projected application temperatures raises a strong concern.

176

 

(b)
Figure 4.51. Backscattered electron images of (a) Severe environment-assisted edge
cracking and (b) internal cracking at prior-BCC grain boundaries observed within the
bulk of a Ti-15A1-33Nb HT :1005 sample which failed at a creep strain of 1.7% after 820
hours of creep deformation under conditions of 172 MPa/650°C.

177

  
 

   
 
        
  

    

/; . .,-/e...“ x s -25...
A /l‘\.\\‘ 3 ’»5

(b)
Figure 4.52. Backscattered electron images of internal cracking at prior-BCC grain
boundaries observed within the bulk of a Ti-lSAl-33Nb HT:1005 sample which failed at
a creep strain of 1.7% alter 820 hours of creep deformation under conditions of 172
MPa/650°C.

   

/’ .
Arm-{47$ -

178

,.. ”.../iii}, .1. ”PM ..I
“/I /fii...\. \D.4‘.,Zxa a r'4\>Wa: lull.“ ”17
. _ . . . lfind..fl6¢2.

II

-. ../ v a “Ni V [.../v .
. ,. u~ .*# 51" v. \z s; \ «Hr/‘rWfl/(Ata)/.
. a: a. _. ,, a I..w..t,...v¥$w

A

 

  

      

 

 

...71...» 1'
,, p a ... . r . J.A
. Y t r
..a ' njfly ‘3 '/
kt?!» I...“ \ ...”...

     

      

.-...wwae..r
(ml.W\\.\ \\\.««:,\,
Av\\. \s t .\ WW1.» «V a. l t

xx... \wflsvwv \ $ .«x v . _
\V. ‘1. 4‘ AVxfi/Zwasr affirm 4/ ,
ewes... , . ....\> .4.

.f.&%.. , /,.,.
\anfl\\\..}, .V;.0' /\ A, a .
. ~.ea.../...........M.¢cn.“ M/ll/ \L '_/ \V; ..I

\
“WA

 

 

Figure 4 53. Backscattered electron images of internal cracking at prior-BCC grain
boundaries observed within the bulk of a Ti-lSAl-33Nb HT:1005 sample which failed at

a creep strain of 1.7% after 820 hours of creep deformation under conditions of 172

MPa/650°C.

179

4. The Effect of Boron on Creep Behavior

4.1 Minimum Creep Rates, Creep Stress Exponents, and Apparent
Activation Energies for the Boron-Modified Alloys

Since the Ti-15Al-33Nb HT:1005 microstructure displayed the best creep
resistance out of all the Ti-15Al-33Nb microstructures produced, the boron-modified
alloys were heat treated in the same manner and will be referred to in the same manner,
i.e. Ti-15Al-33Nb-O.SB HT:1005 and Ti-lSAl-33Nb-SB HT:1005. The creep behavior of
the boron-modified alloys also resembled that for most pure metals and alloys, exhibiting
the primary, secondary, and tertiary stages of creep [Evans and Wilshire (1985),
Hertzberg (1996)]. Experimental data obtained from a typical load-jump experiment
conducted at 650°C is shown in Figure 4.54. The data obtained from a typical
temperature-jump experiment conducted at 150 MPa is shown in Figure 4.55. Creep
stress exponents and apparent activation energies were determined in the same manner
for the boron-modified alloys as the monolithic alloys.

The stress dependence of the minimum creep rate for each Ti-15Al—33Nb-xB
composition is displayed in Figure 4.56. The temperature dependence of the minimum
creep rate is given in Figure 4.57, and the creep strain versus time behavior at 650°C is
given in Figure 4.58. From Figure 4.58 it is evident that the Ti-15Al-33Nb-SB alloy
displays greater primary creep resistance than both the monolithic and Ti-15Al-33Nb-
0.5B alloys. After 40 hours of creep testing under conditions of 275 MPa/650°C the Ti-
15Al-33Nb alloy accumulated a creep strain of 1.2% and the Ti-15Al-33Nb—SB alloy
accumulated a creep strain of 0.52%. The Ti-15Al-33Nb—O.5B alloy accumulated a creep
strain of 1.98% after 40 hours of creep testing a 250 MPa/650°C. Clearly, the addition of

0.5 at% boron to Ti-15Al-33Nb is deleterious for creep resistance, but the addition of 5

180

at% boron is beneficial for creep resistance. Table XVIII lists the minimum creep rates
determined for the Ti-Al-Nb-xB alloys.

Comparing the minimum creep rates for identical applied stresses and
temperatures, the Ti-15Al-33Nb-5B alloy exhibited the lowest creep rates while the Ti-
15Al-33Nb-O.SB alloy exhibited the highest creep rates. For example, at 172 MPa/650°C
the minimum creep rates were 7.6x10'9 5", 1.6x10’8 3", and 2.3x10'9 8'1 for the Ti-15Al-
33Nb, Ti-15Al-33Nb-0.5B, and Ti-15Al-33Nb—SB alloys, respectively. Thus, the Ti-
15A1-33Nb-5B alloy exhibited almost an order of magnitude lower minimum creep rate
than the Ti-15Al-33Nb-O.SB alloy. Table XIX lists the calculated n and Qapp values for
the stress and temperature ranges examined for the Ti-15A1-33Nb-xB materials.

The effect of boron addition on the displacement versus time behavior for the Ti-
22Al-26Nb alloy is shown in Figure 4.59. Under creep testing conditions of 400
MPa/650°C, the minimum creep rate for the Ti-22Al-26Nb alloy was 5.60x10'8 8'1
compared to 4.20x10'7 s" for the Ti-22Al-26Nb—5B alloy. The Ti-22Al-26Nb-5B alloy
exhibited a creep rate almost an order of magnitude higher than the monolithic Ti-22Al-
26Nb alloy. Figure 4.58 clearly illustrates that the Ti-22Al-26Nb-SB alloy exhibited
poorer primary and secondary creep resistance than the monolithic alloy. Included in
Figure 4.58 is creep displacement versus time data obtained for the Ti-15Al-33Nb-5B
alloy under the same testing conditions. The Ti-15Al-33Nb-5B alloy is more creep
resistant than the Ti-22Al-26Nb-5B alloy, and interestingly displayed almost the same
minimum creep rate as the Ti-22Al-26Nb alloy (see Table XIX), but exhibited a larger

primary creep strain.

181

Table XVIII. Measured Minimum Creep Rates and Grain Sizes for the Ti-Al-Nb-xB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Microstructures
Alloy and Heat Treatment, GIT, Minimum Creep Rate, Grain Size,
°C MPaI°C s" pm
Ti-15Al-33Nb-O.5B HT:1005 125/650 8.01x10" 130
Ti-lSAl-33Nb—O.SB HT:1005 150/650 9.73x10'9 130
Ti-lSAl-33Nb-0.SB HT:1005 150/670 1.24x10'8 130
Ti-15A1-33Nb-0.5B HT:1005 150/690 2.33x10‘8 130
Ti-15A1-33Nb-0.SB HT:1005 150/710 6.45x10‘8 130
Ti-15Al-33Nb-0.5B HT:1005 172/650 1.60x10‘r 130
Ti-15Al-33Nb-O.SB HT:1005 200/650 2.19x10‘8 130
Ti-15Al-33Nb-0.SB HT:1005 225/650 4.04x10‘8 130
Ti-lSAl-33Nb—0.SB HT:1005 250/650 7.55x10'8 130
Ti-15A1-33Nb-0.SB HT:1005 250/670 1.67x10'7 130
Ti-15A1-33Nb-0.SB HT:1005 250/690 4.27x10'7 130
Ti-15Al-33Nb-SB HT:1005 172/650 2.30x103’ 63
Ti-15Al-33Nb-SB HT:1005 200/650 3.29m? 63
Ti-15Al-33Nb-SB HT:1005 225/650 4.69x10'9 63
Ti-15Al-33Nb-SB HT:1005 250/650 5.81x10'9 63
Ti-15Al-33Nb-SB HT:1005 275/650 8.49x10'9 63
Ti-15A1-33Nb-SB HT:1005 275/670 1.58x10‘8 63
Ti-15A1-33Nb-SB HT:1005 275/690 4.61x10‘8 63
Ti-lSAl-33Nb-SB HT:1005 275/710 9.25x10if 63
Ti-15A1-33Nb-SB HT:1005 300/650 1.22x10‘8 63
Ti-15A1-33Nb—SB HT:1005 325/650 1.54x10‘“ 63
Ti-15A1-33Nb-SB HT:1005 340/650 1.91x10’8 63
Ti-15Al-33Nb-SB HT:1005* 400/650 5.21x10" 63
Ti-22A1-26Nb* 400/650 5.60x104r 59
Ti-22Al-26Nb-SB* 375/650 3.16x10'7 23
Ti-22Al-26Nb—5B* 400/650 4.52x10'7 23

 

 

 

 

 

 

* indicates this test was performed using the Ernest Fullam Tensile Stage

Table XIX. Measured Creep Exponents and Apparent Activation Energies for the
HT: 1005 Heat-Treated Ti-l 5Al-33Nb-xB Materials

 

 

 

 

 

 

 

Alloy Composition GIT, MPaPC II (III, MPaI°C Qapp, kJ/mol
Ti-15Al-33Nb 148-226/650 2.5 150/650-710 163
226-273/650 6.0 273/650-690 288
Ti-15Al-33Nb-O.SB 125—200/650 2.2 150/650-710 236
200-250/650 5.5 250/650—690 320
Ti-15Al-33Nb-5B 172-340/650 3.2 275/650-710 310

 

 

 

 

182

 

2.5 .

 

1.5 r

   

172 MPa

Creep Strain, %

—t

T = 650°C
100 150 200 250 300
Time, hrs

 

 

 

 

0 510

Figure 4.54. Creep strain versus time data obtained from a Ti-lSAl-33Nb-0.SB HT:1005
sample during a load-jump experiment.

183

 

 

 

 

 

5 . _
c\° 4 _
.5
E
or 3~
Gt
0
2 2 _
u

1 t

0 0' = 250 MPa

0 110 2‘0 310 4'0 5‘0 60
Time, hrs

Figure 4.55. Creep strain versus time data obtained from a Ti-15Al-33Nb-O.SB HT:1005
sample during a temperature-jump experiment.

184

' Ti-15A1-33Nb HT:1005
' Ti-15Al-33Nb-0.SB HT:1005
‘ Ti-lSAl-33Nb-SB HT:1005

—.

10'7 :—

i—i
6
do

 

Minimum Creep Rate, s'1

 

 

- . . . .T.=6.59°.C
100 1000
Stress, MPa

 

Figure 4.56. Plot of log 2.... versus log a for the Ti-5Al-45Nb alloy creep tested at T =
650°C.

185

’ Ti—15Al-33Nb-0.5B 150 MPa
" Ti-15AI-33Nb- 0. 5B 250 MPa
' Ti-lSAl-I33Nb- 5B 275 MPa

     
 

 

 

 

Qapp= 320 kJ/mol
q; -15 L ‘ 3
‘5 0.
a: “n t
% -16 L _
Q o .
U . '
E 17 ‘ \9 Q —
:1 - e app — 310 kJ/mol—
.2 \
.E “
2 -18 r —
E: Qapp 236 kJ/mol .

-19
10.1 10. 210. 310. 410. 510. 610. 710. 810. 9

10,000/T, l/K

Figure 4.57. Activation energies determined at 0 =150 MPa, 0 = 250 MPa, and o = 275
MPa for the Ti-15Al—33Nb—0.5B and Ti-15Al-33Nb-5B heat-treated microstructures.

186

' Ti-lSAl-33Nb HT:1005
' Ti-15AI-33Nb-0.5B HT:1005
‘ Ti-15AI-33Nb-5B HT:1005

 

 

 

7r='650°'c ' ' ' ' I
2.5 _ -
e\° 2 p r
g II...
a I...
‘v'i 1-5 r .-' o=250 MPa“
9. .-'
Q I. o ..
2 l — I. . . O . . . '4
U .' . . - ' o=275MPa
0'5 1 . . ‘AAAAAA“““““‘A“‘
“Hum“ o=275 MPa
0 l l I A l l I
0 5 10 15 20 25 30 35 40

Time, hrs

Figure 4.58. Creep strain versus time behavior under test conditions of 250-
275MPafT=650°C in air for the Ti-15Al-33Nb-xB materials.

187

' Ti-15AI-33Nb-5B
' Ti-22Al—26Nb
‘ Ti—22Al-26Nb-5B

 

0' = 400 MPa T = 650°C

 

 

Displacement, mm

 

 

0 1‘0 210 30 4'0 5'0 60 70
Time, hrs

Figure 4.59. Displacement versus time behavior for Ti-Al-Nb-xB materials under creep
testing conditions of 400 MPa/650°C in a vacuum atmosphere acquired using the in-situ
tensile stage.

188

4.2 Proposed Deformation Mechanisms for the Boron-Modified Alloys

Based on the calculated n and Qapp values, at least two dominant creep
mechanisms are suggested to be active as a function of stress for the Ti-15Al-33Nb-xB
alloys. Grain boundary sliding is suggested in the low-stress (o < ZOO-250 MPa) regime
and dislocation climb is suggested in the high-stress (o >200-250 MPa) regime for both
the monolithic and Ti-15Al-33Nb-0.SB alloys.

The Ti-15Al-33Nb-SB alloy displayed a constant 11 value of 3.2 over the entire
stress range studied with a Qapp value of 310 kJ/mol at 6 =275 MPa. Therefore,
dislocation glide or climb is suggested as the dominant deformation mechanism over the
stress range studied. Out of all the Ti-15Al-33Nb-xB alloys creep tested in this thesis,
this alloy produced the lowest primary creep strains and lowest minimum creep rates.

Creep deformation mechanisms were not suggested for the Ti-22Al-26Nb—xB
alloys because these alloys were tested using the in-situ stage. Since an extensometer
was not used, the minimum creep rates determined from the total displacement were not

used to calculate creep stress exponents and apparent activation energies.

4.3 Creep Deformation Behavior of the Boron-Modified Alloys

The same in-situ creep testing methodology described previously was also applied
to the boron-modified alloys. Figure 4.60 (3)-(h) shows the creep deformation evolution
on the surface of a Ti-22Al-26Nb-5B sample during an in-situ creep experiment
conducted at 375 MPa/T=650°C (Figure 4.60 spans pages 191-194 and the loading
direction is horizontal in each image). To gain a perspective of the amount of cracking

endured by this sample before failure occurred, low magnification BSE images are

189

presented in Figure 4.61 . The sequence of deformation events observed in the Ti-22Al-
26Nb-5B alloy included boride cracking at a total machine displacement of 240 um
(immediately upon reaching the creep stress) followed by more extensive boride cracking
with increased displacement. Cracks nucleated and grew preferentially within and
around the boride phase. Crack nucleation also occurred within adjacent a; grains,
however this occurred secondary to crack nucleation and growth from the boride phase.
Therefore, the inhomogeneously distributed large borides within the microstructure acted
as stress concentration sites for crack nucleation and growth, as opposed to providing
strengthening.

Since the creep tests performed on the Ti-22Al-26Nb alloy were performed using
the in-situ stage, unlike that for the Ti-15Al-33Nb-xB alloys, the observed deformation
behavior for the Ti-22Al-26Nb alloy is discussed in this section. Figure 4.62 depicts
BSE images taken from a Ti-22Al-26Nb sample creep tested at 400 MPa/650°C. The
deformation was localized to the prior-BCC grain boundaries in this alloy, and cracks
nucleated at and propagated predominantly through and along the prior-BCC grain
boundaries.

To gain insight into the deformation that occurred within the Ti-15Al-33Nb-SB
alloy, post-mortem BSE images were taken from the sample surface located behind the
fracture surface available in Figure 4.63. Contrary to expectation, the majority of the
boride laths appeared undamaged (uncracked) after creep failure, but evidence exists of
severe cracking along prior-BCC grain boundaries in addition to evidence of prior-BCC
grain boundary sliding/upheaval. This suggests that in the boron-modified alloys, the

BCC phase is experiencing the majority of the creep deformation.

190

 

(b)
Figure 4.60. Backscattered electron images illustrating the evolution of creep
deformation in at Ti-22Al-26Nb-5B sample creep tested at 375 MPa/650°C. The total
displacement in mm is indicated in the lower right-hand corner of micrographs (a)-(h).
The loading direction is horizontal in each image.

191

u

‘1

- -' A.» ,t »,.'. . , , 0.28mm
he rim-2,, - . -

3 3 f
3* s ‘

3 -'3
’r” 4

t
e: - -_

- Wren.

 

” ($3; 5-..; .1
(d)
Figure 4.60. Backscattered electron images illustrating the evolution of creep
deformation in at Ti-22Al-26Nb-5B sample creep tested at 375 MPa/650°C. The total
displacement in mm is indicated in the lower right-hand comer of micrographs (a)-(h).
The loading direction is horizontal in each image.

192

 

Figure 4.60. Backscattered electron images illustrating the evolution of creep
deformation in at Ti-22Al-26Nb-5B sample creep tested at 375 MPa/650°C. The total
displacement in mm is indicated in the lower right-hand corner of micrographs (a)-(h).
The loading direction is horizontal in each image.

193

 

(h)

Backscattered electron images illustrating the evolution of creep
deformation in at Ti-22Al-26Nb-5B sample creep tested at 375 MPa/650°C. The total

Figure 4.60.

displacement in mm is indicated in the lower right-hand comer of micrographs (a)-(h).

The loading direction is horizontal in each image.

194

 

g
g

urface crac
The 10

1V6 S

         

  

,6»

..n.
. . 1

ustrating extens

              

 

images

   

        

.. .. I15. _ .
1 . . . Her} «13‘
.. . . L .. .,TVrJaw.‘ I... .
. I». ..1. via” ... ... .
v44 . ,.t: a . . . . _ . . ,
a: . v... .343 .....w} . gin...
. ,. ...y... ._ _ .... , ..
. .géugxfi t9. .

   

  

              

.
y)..- .

”61.4,: _

    
           

.. . . .. .. ._ x , n

. . . T .. ._ . bi. Vtfir
H it . . .. , . .1. ..
.. ,r it‘dfiwxfinfigr . . V .. .. , c7... ,.

  

ill
26Nb-SB sample creep tested at 375 MPa/650°C

   
            

     
  

 

             

9.15 ..
A..+..,...:.,,%.. . .. , .. M
, . 4, . ..................~.:..7€...
...: .trv ; . ., .. .Wta dew [1... 2
. . . ., 2

i-

Backscattered electron
ntal in each image.

. ,... . a . .
.__. ...... 1.. . .,
...u. ......fl... ....Ak..&m+f.

.. fieaagxinasa... ,.

.14.“ is

as... $33,199. ...,ag, . ..
J :1. _ r411...“ . _ 9.: ,..mwflwfiki
_. u ...?......a., £5... .

.. . . b V
. . . . . r . .. 1.. ...l. .. .,
.1»... .... .. . .. attuaapmv

 

    

. a...

...,

on is ho '

100 u
4.61

'bited by a T

 

igure

F
e

195

7.1:?

\
V K\ v
"\

 

L)
1

(b)
Figure 4.62. Backscattered electron images taken during a 650°C creep experiment at
400 MPa for a Ti-22Al-26Nb sample. Both images were acquired at a creep
displacement of 0.41 mm. The white arrows indicate some of the numerous prior-BCC
grain boundary cracks. The loading direction is horizontal in each image.

196

 

(b)
Figure 4.63. Backscattered electron images of a post-mortem Ti-15Al-33Nb-SB sample
creep tested at 400 MPa/650°C. Note that the majority of the boride needles remain
uncracked while the cracking and sliding/upheaval of prior-BCC grain boundaries
accounted for the majority of the deformation.

197

C. Tensile Behavior Results
1. Room-Temperature Tensile Behavior

1.1 Stress versus Strain Behavior at Room-Temperature

Figure 4.64 depicts the stress versus strain curves for the Ti-lSAl-33Nb and Ti-
21Al-29Nb alloys tested at RT. Table XX displays RT tensile properties.

The AP Ti-lSAl-33Nb microstructure displayed the best balance of RT tensile
properties. The stress-strain behavior of this microstructure resembled that of elastic-

perfectly plastic materials, with little-to-no work hardening. The high strength and 8f

values are attributed to the fine equaixed grain size of 3 um and the high volume fraction
(0.90) of BCC phase. The heat treatments, which precipitated the O-phase into the
microstructure and changed the average equiaxed grain size, drastically altered the RT
tensile behavior of the Ti-15Al-33Nb alloy. The solution-treated and aged
microstructures, which contained O—phase volume fractions between 0.37-0.63, exhibited
strain hardening behavior and UTS values between 836-867 MPa while maintaining 8f
values greater than 2%. The Ti-15A1-33Nb HT:96O microstructure maintained an 8f
value of 4.4%, and it is noted that the BCC phase in this microstructure contained 1.4 at%
less Al than the BCC phase in the Ti-lSAl-33Nb HT :1005 microstructure. The loss in
strength displayed by the supertransus O+BCC Ti-15Al-33Nb microstructures when
compared to the AP microstructure can be accounted for by the increase in prior-BCC
grain size. The decrease in elongation-to—failure for the O+BCC microstructures in
comparison to the AP microstructure can be accounted for by the increase in O-phase
volume fraction. Previous work has shown the O-phase to posses an insuflicient number

of active slip systems at RT [Popille and Douin (1996)]. Increasing O-phase volume

198

fraction at the expense of BCC phase volume fraction tended to increase RT strength at
the expense of elongation-to-failure. This was experienced by the HT:650/100h/FC
microstructure which contained 73 volume percent O-phase and exhibited the highest
strength and the lowest elongation-to-failure. This data suggests that in order to maintain
RT ductility in this alloy, the O—phase volume fraction must be kept under 0.70, which
agrees with previous findings [Boehlert (2001)].

The AP Ti-21Al-29Nb microstructure illustrates the importance of BCC phase
structure and Al content of the BCC phase on RT tensile behavior of Ti-Al-Nb alloys
[Boehlert (2001)]. Previous work has shown that fully B2 microstructures exhibit brittle
behavior at RT. Increasing the A] content in the BZ phase has also been shown to
generate lower 8f values at RT [Boehlert (2001)]. The Al content of the BCC phase in the
AP microstructures was not determined by EMPA, but Table XI clearly shows that the
BCC phase in the Ti-21Al-29Nb alloy always contained more A] than the BCC phase in
the Ti-15A1-33Nb alloy. The AP Ti-21Al-29Nb microstructure (3 um grain size, 0.96 B2
phase, 0.04 a; phase) displayed an elongation-to-failure value of 1.7%, in comparison to
16.8% for the AP Ti-15A1-33Nb microstructure (3 um grain size, 0.90 [3 phase, 0.10 or;
phase). Both microstructures contained the same grain size and relatively the same phase
volume fractions, but differ in that the BCC phase is ordered in the Ti-21Al-29Nb alloy
and is disordered in the Ti-lSAl-33Nb alloy. The ordered nature of the B2 phase in the
AP Ti-21Al-29Nb alloy is believed to account for the low elongation-to-failure value
obtained.

The single phase fully-B Ti-SAl-45Nb alloy exhibited the lowest strength and

highest elongation-to-failure of all the alloys studied in this thesis, while simultaneously

199

having the largest prior-BCC grain size (4500 um). If the grain size of the Ti-5Al-45Nb
alloy was reduced through a different thermomechanical processing route, its RT tensile
strength would only be expected to increase. This could have been accomplished through
further hot working. The large elongation—to-failure values exhibited by the fully-B Ti-
5Al-45Nb alloy provide fimher evidence of the brittle behavior exhibited by
predominantly B2 Ti-Al-Nb microstructures with a high BCC phase Al content; this type

of brittle behavior was illustrated by the AP Ti-21Al-29Nb alloy.

200

0 Ti—SAl-45Nb H'l‘:650/66—8011/19‘C
I Ti-lSAI-33Nb AP
I Ti-15Al-33Nb HT: 65(_)/100h/1’(,‘
" * Ti-lSAI—SJNI) HT: 960
A '1‘i-15Al-33Nh HT: 1005
1200 —-O - Ti-15A1-33Nb HT: 1105
"I i--2 I .\i—2".5\31) AP
'I‘i-Zl \I-Z‘Wh ”'1‘: mm.)
L Ti-21Al-29Nb HT: 1005
A * Ti-22Al-26Nb As-Hll’l’cd —

1000

..;N»

00
G
c
‘1

 

 

 

N
91
E i}
m" 600 ‘1' -
8 :b‘O-ooo oo-Looo-ooooO"O"°“°"-Oo_
3:, t "i
m 400 i “‘3
i
200;; *
0; L . 1 ‘
0 5 10 15 20 25

Strain, "/o

Figure 4.64. Room-temperature tensile behavior for the Ti-5A1-45Nb, Ti-lSAl-33Nb, Ti-
21Al-29Nb, and Ti-22Al-26Nb microstructures.

201

Table XX. Room-Temperature Tensile Properties of the Ti-SAl-45Nb, Ti—15A1-33Nb Ti-
21Al-29Nb, and Ti-22Al-26Nb Microstructures

 

 

 

 

 

 

 

 

 

 

 

 

E, 3!,
Alloy Heat Treatment, °C GPa 0.2% YS2 MPa UTS, MPa %
Ti-SAl-45Nb 650/66-80h/FC 86 545 559 24.9
Ti-l 5Al-33Nb AP 94 876 916 16.8
Ti-15Al-33Nb HT:650/100h/FC 1 12 n/a 920 0.8
Ti-15Al-33Nb HT:960 103 778 867 4.4
Ti-15A1-33Nb HT: 1005 1 1 1 799 852 2.1
Ti-15Al-33Nb HT: 1 105 113 786 836 2.7
Ti-21Al-29Nb AP 103 972 1010 1.7
Ti-21Al-29Nb HT: 960 1 12 n/a 830 0.8
Ti-21Al-29Nb HT: 1005 115 803 868 1.6
Ti-22Al-26Nb As-HIPPed 131 1008 1059 1.2

 

 

 

 

 

 

 

n/a indicates 0.2% requirement was not met and the fracture stress was taken as the UTS

The heat treated Ti-21Al-29Nb microstructures displayed comparable strength
values to those of the heat-treated Ti-l 5Al-33Nb microstructures, however their
elongation-to-failure values were always less than 2%. This behavior can be explained
by the fact that the O-phase volume fiaction for the heat-treated Ti-21Al-29Nb
microstructures was greater than or equal to 0.72. If supertransus Ti-21Al-29Nb
microstructures had been developed, it is believed that their elongation-to-failure values
would have been even lower due to the large prior-BCC grains that would have been
present, in addition to the large volume fraction of O-phase.

The Ti-22Al-26Nb alloy exhibited high strength combined with low elongation-
to-failure, similar to the RT tensile behavior of the Ti-21Al-29Nb alloy. This is due to
the Ti-22Al-26Nb microstructure containing an O-phase volume fraction of 0.72 and its

intermediate prior-BCC grain size of 59 um.

202

1.2 Room-Temperature Tensile Deformation Behavior

Secondary electron (SE) images of the fracture surfaces of the Ti-15A1-33Nb and
Ti-21Al-29Nb microstructures tested at RT are shown in Figures 4654.68. The fracture
surface morphologies of the Ti-15Al-33Nb microstructures display a greater fraction of
ductile failure than those for the Ti-21Al-29Nb microstructures in all conditions. The Ti-
21Al-29Nb fracture surfaces exhibited more of a flat, cleaved morphology. The TH 5Al-
33Nb fractures surfaces appeared more fibrous and torn, and contained much larger areas
of ductile dimpling. These observations correlate well with the higher 8f values displayed
by the Ti-15A1-33Nb alloy compared to the Ti-21Al-29Nb alloy. As the O-phase volume
fraction increased and the BCC volume fraction decreased in the Ti-lSAl-33Nb
microstructures, the fracture surface morphology contained a larger fiaction of
intergranular cleavage-type features. Mixed mode fi'acture surface morphologies were
typical of the Ti-lSAl-33Nb O+BCC microstructures.

Figure 4.69 shows SE images taken fi'om the surface of a fully-BCC Ti-5Al-45Nb
sample which was metallographically prepared prior to failing at a strain of 24.9% at RT.
The ductile characteristics of the disordered B phase are evident upon examination of
Figure 4.69. Wavy and planar slip traces, slip bands, evidence of cross-slip, and
numerous surface slip steps are present and characteristic of a BCC alloy deformed to a
high strain level.

Figure 4.70 shows surface slip traces and cracking that occurred on the surface for
an AP Ti-15Al-33Nb sample that failed at a strain of 16.8% in tension at RT. BCC phase
wavy slip bands are evident in Figure 4.67 (b), in addition to slip compatibility between

the BCC and «2 phases.

203

 

(b)
Figure 4.65. Secondary electron images of RT tensile fracture surfaces: (a) Ti-15Al-
33Nb AP and (b) Ti-21Al-29Nb AP.

204

 

Figure 4.66. Secondary electron images of RT tensile fiacture surfaces: (a) Ti-15Al-33Nb
HT2960, (b) Ti-21Al-29Nb HT:960.

205

 

Figure 4.67. Secondary electron images of RT tensile fi‘acture surfaces: (a) Ti-15Al-
33Nb HT:1005 and (b) Ti-21Al-29Nb HT:1005.

206

 

Figure 4.68. Secondary electron images of RT tensile fracture surfaces: (a) Ti-lSAl-

33Nb HT:650/100h/FC and (b) Ti-15Al-33Nb HT:1105.

207

 

Figure 4.69. Secondary electron images of surface slip observations taken from a frilly-
BCC Ti-5Al-45Nb sample tensile tested to failure at RT. The sample failed at a strain of
24.9% at RT.

208

 

Figure 4.70. Secondary electron images of surface slip and cracking exhibited by a Ti-
15Al-33Nb sample tensile tested to failtn'e at RT. The sample failed at a strain of 16.8%
at RT.

209

1.3 The Effect of Boron Modification on Room-Temperature Tensile
Behavior

Representative RT stress versus strain curves for the Ti-lSAl-33Nb-xB materials
are shown in Figure 4.71, while Figure 4.72 presents the RT stress versus strain behavior
for the Ti-22Al-26Nb-xB materials. The average RT tensile properties of the Ti-Al-Nb-
xB alloys are given in Table XXI.

The addition of 0.5 at% boron to the Ti-lSAl-33Nb alloy caused almost no
difference in the RT tensile behavior in both the AP and HT: 1005 conditions.

The addition of 5 at% boron to the Ti-15Al-33Nb alloy produced significant
strengthening at the expense of elongation-to-failure. The heat-treated Ti-15Al-33Nb—5B
alloy experienced an increase in elastic modulus of 11 percent, an increase in Y8 of 22
percent, an increase in UTS of 21 percent, and a decrease in elongation-to-failure of 52
percent. This strengthening can be accounted for by both the changes in microstructural
features and by the presence of the boride needles within the microstructure. The Ti-
lSAl-33Nb-SB alloy contained 66 volume percent O-phase compared to 44 volume
percent O-phase in the monolithic alloy. Increasing O-phase volume fiaction has been
shown to increase the strength of Ti-Al-Nb alloys [Cowen and Boehlert (2006a), Boehlert
(2001), Boehlert (1999)]. The prior-BCC grain size of the Ti-15Al-33Nb-5B alloy was
approximately half that of the monolithic alloy, which leads to the possibility of some
strengthening due solely to the reduction in grain size. Strengthening is provided by the
addition of 5 volume percent reinforcing boride needles. It is proposed that the boride
needles carry more of the load within the microstructure than the other constituent
phases, until a critical amount of cracking within the borides occurs upon which the load

is then transferred to the matrix.

210

The Ti-22Al-26Nb—5B alloy displayed an inferior tensile performance when
compared to the Ti-22Al-26Nb monolithic alloy. The Ti-22Al-26Nb-SB samples
produced an UTS 39 percent lower than the monolithic alloy. The presence of the boride
needles, in addition to a finer grain size, would be expected to strengthen the
microstructure in a similar fashion as suggested for the Ti-15Al-33Nb-5B alloy. The
detriment in strength is believed to be caused by the addition of the large boride needles
which initiated fracture and exhibited a brittle response. Hypereutectic boron
compositions have resulted in large borides and low elongation-to-failure values for
conventional Ti alloys [Gorsse and Miracle (2006), Yolton and M011 (1996)]. In
addition, incompletely sintered spherical compacts were postulated to be preferred sites
for cracking and decohesion under tensile stress and would be expected to decrease the
elongation-to-failure values. It is also noted that the Ti-22Al-26Nb-SB material exhibited

a greater volume fraction of the a; phase than the Ti-22Al-26Nb monolithic alloy, which

would lead to more cracking and lower elongation-to-failure values.

Table XXI. RT Tensile anerties of The Ti-Al-Nb—xB Alloys

 

 

 

 

 

 

 

 

 

Heat Treatment, E, 0.2% YS, UTS, 8;,

Alloy °C GPa MPa MPa %
Ti-lSAl-33Nb AP 94 876 916 16.8
Ti-l 5Al-33Nb HT:1005 l 1 1 799 852 2.7
Ti-15Al-33Nb-0.SB AP 95 842 932 13.9
Ti-15A1-33Nb-0.5B HT:1005 99 757 888 4.6
Ti-15A1-33Nb-5B HT:1005 124 1030 1077 1.3
Ti-22Al-26Nb AP 131 1008 1059 1.2
Ti-22Al-26Nb-SB AP 144 na 643 0.5

 

 

 

 

 

 

 

211

+Ti—15Al-33Nb AP
' Ti-15A1-33Nb HT:1005
‘ ‘ Ti-15A1-33Nb-0.SB AP
1200 _ ' Tl-15Al-33N1)-0.513 HT:1005 _

 

 

 

 

 

" iii—i ”$111k:) lii‘:iil:..;§

1000 - -

a 800 -
9-1 .
E .

m» 600 :2 —
a .9
2: i

m 400: -
't
3

200‘; _

00 3 11) 1‘5 20

Strain, %

Figure 4.71. Ti-15Al-33Nb-xB RT stress versus strain behavior.

212

1200 l T I l l l l

 

 

 

1000 —
.1 800L
5
m. 600 —
m 400 .
E --°--Ti-22Al-26Nb
200 1r Ti-22Al—26Nb-SB ‘

 

 

00 0:2 014 0:6 ois i 112 1:4 1.6

Strain, %

Figure 4.72. Ti-22Al-26Nb-xB RT stress versus strain behavior.

213

2. Elevated-Temperature Tensile Behavior

2.1 Stress versus Strain Behavior at 650°C

Figure 4.73 shows the stress versus strain curves for the Ti-15Al-33Nb and Ti-
21Al-29Nb alloys tensile tested at 650°C. Table XXII displays the corresponding 650°C
tensile properties. Table XXIII shows the percentage change in tensile properties for
each microstructure when tested at 650°C compared to RT.

At 650°C, all alloys had lower E, YS, and UTS values and increased elongation-
to-failure values. These results follow expected trends. The increase in elongation-to-
failure values for all the microstructures can be attributed to the thermally assisted
activation of a larger number of slip systems at 650°C. Previous studies on O-phase
alloys have shown pyramidal dislocations to become active at elevated temperatures
along with increasing activity of the basal and prismatic slip systems with increasing
temperature up to 800°C [Popille and Douin (1996), Banerjee (1995), Banerjee (1997)].
The microstructures that exhibited the most dramatic increase in elongation-to-failure
were the heat treated microstructures, which contained larger O-phase volume fiactions.
The most striking example of this was the Ti-15Al-33Nb HT :650/100h/FC
microstructure. This microstructure displayed only a 6 percent decrease in UTS
compared to RT and a dramatic increase in 8f. This is attributable to its large volume
fiaction of 0 phase (0.73) and fine grain size of 4 pm. Heat treated samples that did not
exhibit a 0.2% YS at room-temperature, due to the limited number of slip systems
available, did at 650°C. All microstructures strain hardened at 650°C. The elastic

modulus of all the microstructures decreased by approximately 30 percent at 650°C.

214

 

Table XXII. 650°C Tensile Properties of the Ti-15Al-33Nb and Ti-21Al-29Nb
Microstructures

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E, 0.2% YS, UTS,

Alloy Heat Treatment, °C GPa MPa MPa at, %
Ti-15Al-33Nb AP 66 592 601 12.8
Ti-15Al-33Nb HT:650/100h/FC 77 770 865 7.9
Ti-15Al-33Nb HT:960 66 522 561 8
Ti-15Al-33Nb HT:1005 76 546 588 5.9
Ti-15A1-33Nb HT: 1 105 94 531 609 9.5
Ti-21Al-29Nb AP 71 810 825 2
Ti-21Al-29Nb HT: 960 82 656 664 1.6
Ti-21Al-29Nb HT: 1 005 77 537 657 5.2

Table XXIII. Difference In Tensile Properties at 650°C Compared to Room Temperature
% % %
Heat Decrease Decrease Decrease %
Treatment, in E-mod, in 0.2% in UTS, Increase

Alloy °C GPa YS, MPa MPa in at, %
Ti-15Al-33Nb AP 30 32 34 n/a
Ti-15A1-33Nb HT:650 31 n/a 6 969
Ti-l 5Al—33Nb HT:960 36 33 35 184
Ti-15A1-33Nb HT: 1 005 32 32 31 283
Ti-lSAl-33Nb HT: 1105 17 32 27 353
Ti-21Al-29Nb AP 31 17 18 15
Ti-21Al-29Nb HT: 960 27 n/a 20 196
Ti-21Al-29Nb HT:1005 33 33 24 320

 

 

 

 

 

 

 

n/a indicates not applicable
2.2 650°C Tensile Deformation Behavior

Secondary electron images of the fracture surfaces of the microstructures tested at
650°C are shown in Figures 4.74-4.77. These fracture surface observations correlate well
with the increased elongation-to-failure values obtained. With the exception of the AP
Ti-21Al-29Nb alloy, all the fracture surfaces display a greater area of ductile fracture
characteristics at 650°C when compared to RT. The AP Ti-21Al-29Nb displayed brittle

behavior at RT and 650°C as displayed by its low 8f values.

215

—®— Ti-15Al-33Nb AP
0 Ti-lSAl-33Nb HT:650/100h/FC
A Ti-lSAI-33Nl) llT:960
‘l i-l.5 \l—.i.i\l> ll 1‘: 1005
’44} r Ti-lSAl-33Nb HT:1105
1000 ‘ fi— Ti-21AI-29Nb AP
G” Ti-21Al-29Nb HTz960
. ,-- Ti-21A1-29Nb HT: 1005

800

600

Stress, MPa

 

 

 

T=650°C
o 2 4 6 8 10 12 14

Strain, %

>—

Figure 4.73. 650°C tensile behavior of the Ti-15A1-33Nb and Ti-21Al-29Nb alloys.

216

 

Figure 4.74. Secondary electron images of 650°C tensile fi'acture surfaces: (a) Ti-15Al-
33Nb AP and (b) Ti-21Al-29Nb AP.

217

 

Figure 4.75. Secondary electron images of 650°C tensile fractme surfaces: (a) Ti-lSAl-
33Nb HT:960 and (b) Ti-21Al-29Nb HT:960.

218

 

Figure 4.76. Secondary electron images of 650°C tensile fracture surfaces: (a) Ti-lSAl-
33Nb HT :1005 and (b) Ti-21Al-29Nb HT:1005.

219

 

Figure 4.77. Secondary electron images of 650°C tensile fracture surfaces: (a) Ti-15Al-
33Nb HT:650/100h/FC and (b) Ti-15A1-33Nb HT:1105.

220

2.3 The Effect of Boron Modification on 650°C Tensile Behavior

Figure 4.78 illustrates the stress versus strain behavior at 650°C for the Ti-lSAl-
33Nb-XE materials. Figure 4.79 illustrates the stress versus displacement behavior at
650°C for the Ti-22Al-26Nb-xB materials, respectively. The average 650°C tensile
properties are listed in Tables XXIV.

At 650°C, the heat-treated Ti-15Al-33Nb-0.5B alloy experienced a decrease in
elastic modulus of 18 percent, a decrease in Y8 of 45 percent, a decrease in UTS of 15
percent, and an increase in 8f of 61 percent when compared to the monolithic alloy tested
at 650°C. As with the RT behavior for this alloy, these results remain unexplained from
a microstructural standpoint.

At 650°C, the heat-treated Ti-lSAl-33Nb—5B alloy experienced an increase in
elastic modulus of 17 percent and increase in Y8 of 16 percent when compared to the
monolithic alloy tested at 650°C. This strengthening can be accounted for by the same
reasoning as the strengthening exhibited at RT, namely the reinforcement from boride
needles, a decrease in prior-BCC grain size, and an increase in O-phase volume fi'action.

The 650°C tensile data for the Ti-22Al-26Nb-xB materials was measured using
the Ernest Fullam, Inc. tensile stage. As stated in Chapter 3, the in-situ experiments did
not provide an accurate measure of the actual tensile strain experienced across the gage
section of the sample. Therefore, the values for E, 0.2% YS, and a; could not be
calculated from the curves presented in Figure 4.79. However, the UTS values were
measured from the stress versus displacement curves. It is evident that a significantly
greater displacement was achieved for the Ti-22Al-26Nb monolithic alloy, and therefore

a significantly greater elongation-to—failure value is estimated for this alloy compared to

221

the Ti-22Al-26Nb-SB alloy. At 650°C an inferior tensile response was exhibited by the
Ti22Al-26Nb-5B alloy. This occurred due to its inhomogeneous microstructure which
contained incompletely sintered powder compacts and large boride needles in which
cracks initiated.

Figure 4.80 shows BSE images acquired from a Ti-15Al-33Nb-5B HT:1005
sample that was metallographically prepared prior to tensile testing to a strain of 1.7% at
650°C in vacuum. An extensive amount of cracking is observed to occur within the
borides, with no evidence of slip occurring within the 0+ BCC matrix. No cracking
within the boride needles in the AP or HT:1005 Ti-15Al-33Nb-5B alloy was evident
before tensile testing.

Surface slip damage observations from Ti-22Al-26Nb and Ti-22Al-26Nb-5B
samples tensile tested at 650°C are presented in Figure 4.81 and Figure 4.82,
respectively. Wavy slip was evident in the BCC phase of the Ti-22Al-26Nb alloy, along
with slip compatibility between the BCC and or; phase and the BCC and O-phase.
Preferential cracking at adjacent or; boundaries was evident for the Ti-22Al-26Nb alloy.
Boride cracking without evidence of matrix slip is shown for the Ti-22Al-26Nb-SB alloy

in agreement with observations made for the Ti-15Al-33Nb-5B alloy.

Table XXIV. 650°C Tensile Properties of the Ti-Al-Nb-xB Alloys

 

 

 

 

 

 

 

 

 

 

 

 

 

E, 0.2% YS, UTS, 8;,

Alloy Heat Treatment, °C GPa MPa MPa %
Ti-15Al-33Nb HT:1005 76 546 588 5.9
Ti-15A1-33Nb-0.5B HT: 1005 62 300 498 15
Ti-lSAl-33Nb-5B* HT: 1005 91 650 >784 >1.7
Ti-22Al-26Nb AP n/a n/a 934 n/a
Ti-22Al-26Nb-SB AP n/a n/a 625 n/a

 

na: not applicable; * this sample was not taken to failure and it was tested in a vacuum
environment (10'7 torr) to reveal surface defamation events

222

+ Ti—15Al-33Nb HT:1005
- —~ Ti-15A1-33Nb-0.SB HT:1005
...-...— Ti-15Al-33Nb-5B HT:1005

800 _ q I |
f Sample Did Not Break
700 .. g

Stress, MPa

 

 

 

l l

5 10 15

Strain, %

Figure 4.78. Ti-15A1-33Nb-xB 650°C stress versus strain behavior.

223

+ Ti-22Al-26Nb
“ ‘1" Ti-22Al-26NB—SB

 

 

1000 ——4— . . —.————4
800 — .
a? 600
2 _
g 400 — -
m
200 _ _
T = 650°C

 

 

00 0:2 0:4 016 0.18 1 112 1.4
Displacement, mm

Figure 4.79. Ti-22Al-26Nb-xB stress versus strain behavior at 650°C. These
experiments were performed using the tensile stage inside the SEM chamber and accurate
strain measurements were not available, though it is clear that the Ti-22Al-26Nb sample
underwent significantly more displacement and strain than the Ti-22Al-26Nb-5B sample.

224

 

Figure 4.80. Backscattered electron images of boride cracking at different locations on
the surface of a Ti-15Al-33Nb-SB HT tensile sample tested to a strain to a strain of 1.7%
at 650°C in vacuum.

225

 

Figure 4.81. Backscattered electron images of surface damage observations made behind

the fracture surface afier tensile testing a Ti-22Al-26Nb sample to failure at 650°C.

226

 

Figure 4.82. Backscattered electron images of surface damage observations made behind
the fracture surface after tensile testing a Ti-22Al-26Nb-SB sample to failure at 650°C.
Part of the fracture surface is visible in the upper right-hand comer of (a).

227

D. Fatigue Results

1. S-N Behavior

The S-N curves for the AP Ti-15A1-33Nb and Ti-21Al-29Nb alloys tested by the
Fraunhofer Institute are illustrated in Figure 4.83. Considering the 95 percent confidence
band for the fatigue live values, determined according to the appropriate ASTM Standard
[ASTM Standard E 739-80 (1988)], one alloy was not clearly superior to another. Run-
out samples were noted for both alloys at maximum stresses of 350-450 MPa. Although a
fatigue limit was not observed based on the data (i.e. there were failures even at the
lowest maximum stress of 350 MPa), it is expected to be close to 350 MPa as the
majority of the samples tested at this maximum stress level exhibited run out.

The S-N curves of all the AP alloy samples tested in air at RT are provided in
Figure 4.84. The two institutions tested the specimens in different stress regimes. The
F raunhofer Institute data was obtained at o S 750 MPa while the Toyohashi University of
Technology data was obtained at o 2 820 MPa. It is possible that the different test
apparati may have led to a discrepancy in the fatigue life data. Another explanation for
the occurrence of relatively early failures and run-outs at many of the stress levels could
be the potential existence of two flaw populations. One of the flaw populations (e.g.
surface notches) leads to early failures, but was not present on all the tested specimens. In
order to compare the current alloy results with other Ti alloys, the data obtained by
Toyohashi University of Technology were used as all the Ti alloys compared were tested
using an identical testing apparatus and similar maximum applied cyclic stresses
[Akahori et al. (2000b), Niinomi (2003), Okazaki and Hizume (1994), Akahori et al.

(2000a)]. Figure 4.85 compares the S-N behavior for the AP and HT:1005 heat-treated

228

Ti-15Al-33Nb and Ti-21Al-29Nb samples. The AP samples maintained higher fatigue
lives on average than the heat-treated samples, and this may have been related to the
increased RT tensile strength, which is particularly important for high-cycle fatigue and
is in agreement with recent correlations between tensile strength and high-cycle fatigue

behavior for a Ti-22Al-27Nb alloy [Hagiwara et al. (2004)].

229

 

700 I

 

 

 

a - Ti-21Al-29Nb AP
5 650 . -- - Ti-15A1-33Nb AP
.r 600 — .
E
a 550i r
O
'7': 500 __ ~
>:
“g 450 — - - - ~
5
E 400 _ —
“a
E 350 r I O I a
300 L- L
104 105 106 107

Number of Cycles to Failure

Figure 4.83. Maximum cyclic stress versus fatigue life for the AP alloys tested at the
Fraunhofer Institute.

230

Maximum Cyclic Stress, MPa

 

 

 

 

O Ti-lSAI-33Nb AP Toyohashi
I Ti-lSAl-33Nb AP Fraunhofer
O Ti-21Al-29Nb AP Toyohashi
A Ti-21Al-29Nb AP Fraunhofer
V Ti-15A1-33Nb HT:1005 Toyohashi
I l l": l ,‘&l"::!\ 11 HT: 1 11“.; Tu) ulmxlll
1000 44 4 4 4 4 4 4 4
. k .
L . 9
900 L k b V ‘ C Q _
~ V
- v
800 — —
i 1
700 — —
: AA
600 L —
E :
500g- 5
C A A A
400 1 4
: A A A j
300 ' 4 i l 6 7
1000 10 10 10 10

Number of Cycles to Failure

Figure 4.84. Maximum cyclic stress versus fatigue life for the all alloys tested.

231

 

 

 

0 Ti-15A1-33Nb AP
I Ti-ZlAl-29Nb AP
9 Ti-15Al-33Nb HT:1005
A Ti-21Al-29Nb HT:1005
960 4 . 4 . 4 ,
.,., F - :
E 940 f A j
r- . 4
m" 920 E A I —
§ : I I 4
a 900 F A o I o I j
o g e ‘
'5 330 L . . . 4
6} t 0. . O :
E 860 L .. i
5 i -
g 840 L . —_
a : 1
E 820 _— . .

 

 

 

800 “ L L 4 L
1000 104 105 106 107
Number of Cycles to Failure

Figure 4.85. Maximum cyclic stress versus fatigue life for the AP and HT:1005 alloys
tested at the Toyohashi University of Technology.

232

2. Fatigue Deformation Behavior

Figure 4.86 presents SE images of the entire fracture surface of both an AP Ti-
lSAl-33Nb sample and an AP Ti-21Al-29Nb sample. Secondary electron images
acquired from fatigue crack initiation, propagation, and overload regions are presented in
Figures 4.87-4.92. Surface slip traces were evident on all the AP fatigue samples, see
Figures 4.93 and 4.94.

The run-out samples were subsequently examined in RT tension, as described in
the experimental section, and the associated data is listed in Table XIX. These samples
did not exhibit lower strength values (on average) than those which were not fatigue
tested. Thus no clear evidence, in terms of strength loss, was apparent that fatigue testing

reduced tensile strength.

Table XXV. RT Tensile UTS Values of the AP Samples that Exhibited Run-Out

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Alloy Number of cycles Maximum Cyclic Stress, MPa UTS, MPa
Ti-15Al-33Nb 2,000,000 350 887
Ti-15Al-33Nb 2,000,000 350 1002
Ti-15Al-33Nb 2,000,000 350 907
Ti-15Al-33Nb 2,000,000 450 972
Ti-15Al-33Nb 2,000,000 550 928

Average 939
Ti-21Al-29Nb 2,000,000 350 995
Ti-21Al-29Nb 2,000,000 350 1048
Ti-21Al-29Nb 2,000,000 350 1021
Ti-21Al-29Nb 2,000,000 350 1023
Ti-21Al-29Nb 2,000,000 400 897
Ti-21Al-29Nb 2,000,000 450 1106
Ti-21Al-29Nb 2,000,000 450 1062

Average 1022

 

233

 

 

(b)
Figure 4.86. Secondary electron images of the fatigue fracture surfaces of (a) an AP Ti-
15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles and (b) an AP Ti-21Al-29Nb sample tested at a maximum cyclic stress of
930 MPa which failed afier 19,087 cycles.

234

 

Figure 4.87. Secondary electron images of a fatigue crack initiation site in an AP Ti-
15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed alter
37,972 cycles.

235

 

Figure 4.88. Secondary electron images of a fatigue crack propagation region in an AP
Ti-15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles.

236

 

Figure 4.89. Secondary electron images of a fatigue crack overload region in an AP Ti-
15Al-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles.

237

 

Figure 4.90. Secondary electron images a fatigue crack initiation site in an AP Ti-21Al-
29Nb sample tested at a maximum cyclic stress of 930 MPa which failed afier 19,087
cycles.

238

 

Figure 4.91. Secondary electron images of a fatigue crack propagation region in an AP
Ti-21Al-29Nb sample tested at a maximum cyclic stress of 930 MPa which failed after
19,087 cycles.

239

‘

..i '3" ‘11

 

Figure 4.92. Secondary electron images of a fatigue crack overload region in an AP Ti-
21Al-29Nb sample tested at a maximum cyclic stress of 930 MPa which failed afier
19,087 cycles.

 

Figure 4.93. Backscattered electron images of surface slip traces exhibited by an AP Ti-
lSAl-33Nb sample tested at a maximum cyclic stress of 865 MPa which failed after
37,972 cycles.

241

 

Figure 4.94. Backscattered electron images of surface slip traces exhibited by an AP Ti-
21Al-29Nb sample tested at a maximum cyclic stress of 930 MPa which failed after
19,087 cycles.

242

CHAPTERS

DISCUSSION

This Chapter provides a discussion and analysis of the results obtained in Chapter
4, and each subsection is organized based upon the specific aims outlined at the end of
Chapter 1. The phase field ranges determined for the alloys in this thesis are discussed
first, followed by the phase transformation behavior, and the effect of boron additions on
the microstructure. Microstructure-creep relationships are constructed and the in-situ
grain boundary cracking and sliding/upheaval results are discussed. Creep deformation
mechanisms are suggested based on pure metal theory and supported with the
microstructural deformation observations made. An empirical model based on
constituent phase data was developed in an attempt to predict the minimum creep rates of
two-phase O+BCC microstructures. The effect of boron additions on creep deformation
concludes the creep discussion. Microstructure-tensile property relationships, and the
effect of boron on these relationships, are discussed. The fatigue behavior observed for
the Ti-Al-Nb alloys in this thesis is compared to the fatigue behavior for other Ti alloys

being considered for biomedical applications.

A. Microstructure Discussion

1. Phase Equilibria and Phase Field Ranges

One of the motives in alloy selection for this work was to firrther develop the
understanding of phase equilibria spanning the compositional range of Ti-23Al-27Nb to

Ti-12Al-38Nb, so alloys with nominal compositions of Ti-17Al-33Nb and Ti-22Al-28Nb

243

were selected. In particular, the Ti-l 7Al-33Nb composition was chosen with the intent of
being able to produce a microstructure containing approximately equal volume fractions
of the O and BCC phases. It was postulated that equal combinations of the constituent
phases would impart creep resistance, while simultaneously maintaining room-
temperature elongation-to-failure. While the resulting alloy compositions were in
actuality Ti-15Al-33Nb and Ti-21Al-29Nb, the results of the phase evolution study were
still beneficial.

The pseudobinary phase diagram based on Ti = 50 at%, first constructed by
Bendersky and coworkers [Bendersky et al. (1991)] and later modified to incorporate the
results of Boehlert and coworkers [Boehlert et al. (1999)] is depicted in Figure 5.1. It is
it evident that phase equilibria data was lacking between the compositions of Ti-23Al-
27Nb and Ti-12Al-38Nb. From the current results, this pseudobinary diagram has been
readjusted, see Figure 5.2. Figure 5.2 displays the isopleth with TiAl and TiNb bounding
compositions that incorporates the results obtained from this thesis [Cowen and Boehlert
(2006a), Cowen and Boehlert (2006b)] with the previously acquired data [Bendersky et
al. (1991), Boehlert et al. (1999)].

An isopleth is a vertical section taken through a ternary phase diagram, so there
are important differences which distinguish isopleths from true binary phase diagrams.
In an isopleth the reaction horizontal that would be present in a true binary phase diagram
is replaced by a three-phase region that has no fixed form [Rhines (1956)]. If straight or
horizontal boundaries exist in an isopleth, they exist by coincidence, because there is no

requirement for a definite number of bounding curves [Rhines (1956)].

244

In Figure 5.2, the transformation temperatures are accurate, but the compositions
of the individual phases cannot be read off the diagram, except in the single-phase fields.
Therefore, the lever rule cannot be applied in the two phase fields to calculate the
equilibrirun volume fraction and composition of the individual phases in the two-phase
fields, as it could be applied for a true binary phase diagram.

Figure 5.3 displays a 900°C ternary slice of the Ti-Al-Nb system [Rowe et al.
(1993)]. The colored data points on the ternary slice are data points taken from EMPA
data collected from the Ti-15Al-33Nb and Ti-21Al-29Nb samples heat treated for 200
hours at 900°C followed by water quenching. The compositions measured for the BCC
phase in each alloy lie on the Ti = 50 at% tie lie between the BCC and 0 phase fields.

The chemical compositions of the constituent BCC phase measured by EMPA
from the Ti-15Al-33Nb and Ti-21Al-29Nb samples solutionized at 910°C for 200 hours
followed by water quenching were in excellent agreement with the chemical
compositions predicted by the 900°C ternary slice. The measured phase volume fractions
were 85% and 9% for the BCC phase and O-phase, respectively, in the Ti-15Al-33Nb
sample solutionized for 200 hours at 910°C followed by water quenching. These values
are in excellent agreement with the volume fractions predicted by the lever rule, which

are 90% and 10% for the BCC phase and O-phase, respectively.

245

  
    

 

 

 

\\
1400 4 —-L°9°"d m [3 _
Dgz - 2 4‘ ~ 2
‘3‘”— 0mg iii.) -
08+ 2 \
9 s r \
if "002+B2 E E E \ E E B
11m _ . O+a2+BZ = \ C] _
r:- a2+B2fi a B2
E O+a22+B25~
E if
E 900 — —
.— A
700 r 0 0+5 -
0
TIM 712mm: TINb

 

COMPOSITION

Figure 5.1. Pseudobinary diagram based on TiAl and TiNb, for Ti = 50 at%, with
differing ratios of Al:Nb, first developed by Bendersky and coworkers [Bendersky et al.
(1991)] and later modified by Boehlert and coworkers [Boehlert et al. (1999)].

246

 

 

 

 

 

 

 

2 § §§ a
Legend g5. $5 5 5. g?
DB ‘\ flag @333 U,
-B2 \ gggr 33;: ;
o0+ \ gigs ”fig :
3+ 2 \r‘ngmé ééms is
A
'B2 + a2 ‘
1400 I [3 + (12 I\\ 0 —l
A0 + B2 + on
{0.) 130 2 \‘ —
2' 1200 32 ‘ B -
a \
Q1100 _._: “mg 0 _
:- B2 4114’ : I D
5100 O+Bz+azz 1g;r B
[— k
900 11’ ‘
80
700
TiAl TizAle 3 TiNb
Composition

Figure 5.2. An isopleth based on Ti = 50 at%, with differing ratios of Al:Nb. This
diagram was constructed from the phase equilibria results of this thesis added to the work
of [Bendersky et al. (1991)] and [Boehlert et al. (1999)].

247

 

 

 

 

 

 

Figure 5.3. A 900°C ternary slice of the Ti-Al-Nb system [Rowe et al. (1993)]. The
black and white data points represent EMPA measurements from Boehlert et al. (1999).
EMPA measurements obtained from the BCC phase in Ti-15Al-33Nb and Ti-21Al-29Nb
samples solutionized at 910°C for 200 hours followed by water quenching are shown on
this diagram. The red point is the nominal composition of the Ti-15Al-33Nb alloy and is
also the composition of the BCC phase in this alloy at 910°C. The blue point represents
the nominal composition of the Ti-21Al-29Nb alloy and the orange point is the
composition of the BCC phase in the Ti-21Al—29Nb alloy at 910°C.

248

1.1 Ordering of the BCC Phase

The BCC phase in Ti-Al-Nb alloys can either be ordered 82 or disordered B. This
structural transition is dependent upon alloy composition and heat treatment [Kestner-
Weykamp et al. (1989), Bendersky et al. (1991), Rhodes et al. (1993)].

Superlattice reflections occur when the lattice points are occupied by specific
types of atoms in the B2 structure, which makes the effective scattering power of each
crystallographic plane unequal to the effective scattering power of the average of the
individual atoms [Kittel (1996)]. When the crystal structure is disordered B, the effective
scattering power of each crystallographic plane is equal to the effective scattering power
of the average of the individual atoms, and the BCC structure factor is obeyed (i.e. for
given Miller Indicies (th), h+k+l = even for an allowed reflection). Since the basis for
the BZ structure in Ti-Al-Nb alloys is Al or Nb atoms at the 000 position and Ti atoms at
the '/2‘/2‘/2 position, the space lattice is simple cubic, and all the reflections allowed for
the simple cubic lattice will occur, though the additional peaks will have a lower
intensity.

The XRD and TEM analyses provided proof that the BCC phase is disordered in
the Ti-5A1-45Nb and Ti-lSAl-33Nb alloys due to the lack of superlattice reflections in
the SADPs and XRD 29 scans. The EMPA data obtained from the Ti-15A1-33Nb alloy
(see Table X1 in Chapter 4) showed that the Al content measured in the BCC phase was
12.5-15 at%. The presence of superlattice reflections in the SADPs and XRD 29 scans
obtained from the Ti-21Al-29Nb alloy proved that the BCC phase in this alloy is ordered
BZ. However, the degree of order has not been quantified; this is important because there

could be a mix of ordered and disordered domains within the BCC phase, and only the

249

ordered domains would contribute the superlattice reflections. Since higher nominal Al
contents favor ordering (see Figure 5.2), it is fair to assume that the degree of order
within the BZ phase increases within increasing A] content. The EMPA data obtained
from the Ti-21Al-29Nb alloy (see Table XI) showed that the Al content measured in the
BCC phase was between 18-20.8 at% compared to 12.5-15 at% in the Ti-15A1-33Nb
alloy. These values were dependent on the heat treatment performed. Thus, when the Al
content is 2 18 at% in the BCC phase, the BCC phase within the alloy is suggested to be
predominantly ordered 32. The BZ phase field border therefore extends to at least 18
at% Al and this is reflected in the isopleth presented in Figure 5.2. Therefore, a
predominantly ordered BZ structure is not expected for nominal Ti-Al-Nb alloy

compositions containing less than 18 at% A1.

1.2 The BCC Phase Field

The Ti-SAl-45Nb alloy was thermomechanically processed at 800°C. Differential
thermal analysis, SEM, and TEM analyses of samples taken fiom the AP sheet revealed a
single-phase BCC microstructure. In addition, samples aged at 650°C for 66-80 hours
followed by furnace cooling also contained single-phase BCC microstructures. Upon
examination of Figure 5.2 it can readily be seen that the temperature range of 650-800°C
lies within O+BCC phase field for the compositional range of Ti-12Al-38Nb to Ti-25Al-
25Nb. Since the BCC phase was the only phase detected in the Ti-SAl-45Nb alloy over
the temperature range of 650-800°C, it is assumed that this composition remains single-
phase BCC up to the melting point, and therefore a BCC transus temperature would not

exist.

250

When solution-treated and water quenched from 990°C and above, the Ti-l 5Al-
33Nb alloy exhibited only the BCC phase. A similar result occurred for the Ti-21Al-
29Nb alloy when solution-treated and water quenched from 1050°C and above. The
BCC-transus temperature was estimated to be 980°C for the Ti-15A1-33Nb alloy and
1040°C for the Ti-21Al-29Nb alloy. These temperature values were based on the
aforementioned microstructural observations in addition to the results from the DTA,
XRD, and grain size analyses.

A solution treatment study was not performed on the Ti-22Al-26Nb alloy because
previous research has shown the BCC-transus for this alloy to be 1125°C [Smith et al.
2000)]. From the DTA analysis performed, the BCC-transus for the Ti-22Al-26Nb alloy
in this thesis was estimated to be 1130°C. The slight increase in BCC-transus
temperature is believed to be due to the higher oxygen content, 1127 ppm oxygen, in the
Ti-22Al-26Nb alloy of this thesis versus 860-1080 ppm for the Ti-22Al-26Nb alloy
examined by Smith et al. (2000). Oxygen stabilizes the a phase in conventional titanium
alloys, and previous studies have shown increased oxygen content to raise the BCC-
transus temperature in Ti-Al-Nb alloys [Szaruga et al. (1992), Rhodes et al. (1993),
Takag and Ouchi (1997)]. Figure 5.2 shows that decreasing the Alle ratio fi'om 1 to
0.32 decreases the BCC-transus temperature for Ti-Al-Nb alloys with a Ti content of 50
at%. At an Alle ratio of 0.11, a BCC-transus temperature no longer exists. Therefore,
the minimum Al content needed for the O-phase to be an equilibrium phase in Ti-Al-Nb
alloys with a nominal Ti content of 50 at% lies between nominal Al contents of 5-12 at%.

To understand the effect of increasing the Al/Nb ratio on the kinetics of BCC

phase grain growth, the activation energies for BCC grain growth were calculated for the

251

Ti-15A1-33Nb and Ti-21Al-29Nb alloys. The following empirical equation describes
normal grain growth for single-phase materials under isothermal heat treatment

conditions [Hu and Rath (1970), Seetharaman and Semiatin (1997)]:
n n _
d ‘ do - kt (1)

where n is the grain growth exponent, d is the heat-treated BCC grain size, d0 is the pre-

heat treated grain size and in this case the BCC grain size of the AP sheet, and t is the
annealing time. The variable k is estimated by the Arrhenius equation:

.. :Qil’fi
k — k0 exp( RT (2)

where k0 is a kinetic constant, Qapp is the activation energy for grain growth, R is the gas
constant, and T is the absolute temperature. For most single-phase metals, the value of n
ranges between 2-10 due to the drag force exerted by solute atoms on grain boundaries
[Higgins et al. (1992)]. Taking n=3, the corresponding values of k are represented on the
ln k versus (l/T) plot depicted in Figure 5.4 for samples solutionized at different
temperatures above the BCC-transus temperature followed by water quenching. The
activation energy for BCC grain growth in the Ti-lSAl-33Nb alloy was determined to be
194 kJ/mol over the temperature range of 990-1105°C and the activation energy for BCC
grain growth in the Ti-21Al-29Nb alloy was determined to be 311 kJ/mol over the
temperature range of 1050-1105°C.

Diffusion data on the TiNb system predicts activation energies of 180-188 kJ/mol
for a Ti-29Nb alloy and 198-210 kJ/mol for a Ti-33Nb alloy [Shewmon (1963)]. The

activation energy determined for BCC grain growth in the Ti—15Al-33Nb alloy correlates

252

well with that for the Ti-33Nb alloy (within 6 percent), but the activation energy for the
Ti-21Al-29Nb alloy does not. This suggests that the addition of up to 15 at% Al in TiNb
does not hinder grain growth diffusion kinetics, but the addition of 21 at% Al in TiNb
significantly increases the activation energy required for grain growth. The previous
statement corresponds well with the results that show the BCC phase is disordered in the
Ti-lSAl-33Nb alloy and is ordered in the Ti-21Al-29Nb alloy. As suggested in a creep
study performed on a Ti-26Al-15Nb alloy, increasing the degree of order within a given
ordered crystal structure would impose the requirement of greater correlation atom jumps
in diffusion [Nandy et al. (1995)]. A greater amount of thermal energy would have to be
added to the system in order for the Al atoms to perform diffusional jumps to the nearest
neighboring Al sites in the 32 lattice. Hence, this could account for the higher activation
energy required for grain growth in the B2 phase in the Ti-21Al-29Nb alloy compared

with the B phase in the Ti-15A1-33Nb alloy.

253

 

   

 

 

 

6'5 * . Ti-21Al-29Nb *
' Ti-15Al-33Nb
6 _ Qapp= 311 kJ/mol_
x Qapp= 194 kJ/mol
:
5.5 ~ —
5 _ a
4.5 1 m 1 1 1 i 1!
7.2 7.3 7.4 7.5 7.6 7.7 7.8 7. 8
10,000/T,1/K

Figure 5.4. BCC-phase grain growth kinetics for the Ti-lSAl-33Nb and Ti-21Al-29Nb
samples heat-treated in the BCC phase field for three hours followed by water
quenching.

254

 

y .

1.3 The BCC+012 Phase Field

The B+otz phase field has been confirmed to exist below the BCC-transus
temperature for the Ti-15Al-33Nb alloy. This phase field encompasses a narrow
temperature range spanning approximately 960-980°C. It is noted that no solution
treatments were performed between 910-959°C. The [Hog phase field could therefore
begin at a temperature between 910-959°C, but this phase field ends at 980°C. Equaixed
grains of only the 112 and [3 phases were present in the Ti-15A1-33Nb samples
solutionized at 960°C for 3-200 hours followed by water quenching. The 012 volume
fraction should be at its maximum at the lowest temperature within the B+a2 phase field.
The maximum volume fi'action of a; phase present in all of the heat treated Ti-15Al-
33Nb microstructures was 10 volume percent, and this occurred for the sample
solutionized for 200 hours at 960°C followed by water quenching. This result reinforces
the suggestion that 960°C is an appropriate lower bound for this phase field.

The chemical composition of the [3 phase was 50.7Ti-14.5Al-34.8Nb and the
chemical composition of the on phase was 59.4Ti-20.9Al-19.7Nb in samples solutionized
for 200 hours at 960°C. To the author’s knowledge, this is the largest amount of Nb
observed in the a; phase in Ti-Al-Nb alloys. For a Ti-26Al-17Nb alloy, the composition
of the a; phase was measured by EMPA to be Ti-27Al-10Nb [Banerjee (1995)]. For a
Ti-23Al-27Nb alloy, the composition of the a; phase was measured by EMPA to be Ti-
25Al-16Nb [Boehlert et al. (1999)]. Therefore, it is suggested that 19-20 at% Nb is the
maximum amount of Nb solubility in the 012 phase in the Ti-Al-Nb alloy system, based on

the 59.4Ti-20.9Al-19.7Nb composition measured in this thesis.

255

In the a; phase the Al concentration remains relatively constant because the Al
atoms occupy specific sites, while the Ti and Nb contents vary because they both occupy
the same lattice sites randomly (See Figures 2.1 and 2.2). The current work suggests that
the Al content in the 012 phase can be as low as 21 at%. The Nb content of the a; phase
has important implications in the area of the mechanical properties of the a; phase. In
both stoichiometric and non-stoichiometric Ti3Al, extensive basal slip of <a> dislocations
is only observed when the stress axis is oriented at 45° to the basal plane and the Schmid
Factor is at its maximum [Inui et al. (1994), Minonishi (1991)]. It has been suggested
that increasing the Nb content of the or; phase lowers the difference in magnitude
between the critical resolved shear stress for prismatic slip and the critical resolved shear
stress for basal slip [Banerjee (1995)]. The increased Nb content in the a; phase in the
Ti-15Al-33Nb alloy could account for one of the reasons for the increased RT
elongation-to-failure values exhibited by the Ti-lSAl-33Nb alloy compared to other a;
phase containing alloys.

The Ti-lSAl-33Nb alloy is the lowest Al containing Ti-Al-Nb alloy, based on a
nominal Ti content of 50 at%, in which the a; phase has been observed. Heat treatments
performed in the range of 650—1200°C proved that the Ti-12Al-38Nb alloy studied by
Boehlert was devoid of the (12 phase [Boehlert (1999)]. Bulk chemical analysis of the Ti-
12Al-38Nb alloy revealed a composition closer to Ti-l3Al-39Nb. Therefore, when the
nominal Ti-Al-Nb alloy composition contains 14-25 at% Al, the a; phase will be formed
in the microstructure.

The temperature range for the 82+a2 phase field was approximately 980—1005°C

for the Ti-21Al-29Nb alloy. Ti-21Al-29Nb samples that contained only equaixed a2 and

256

equiaxed B2 grains were produced when solutionized for three hours at both 990°C and
1005°C, respectively. The B2+a2 phase field was found to exist over the temperature
range of 1000-1070°C for a Ti—23Al-27Nb alloy [Boehlert et al. (1999)]. It is noted that
no solution treatments were performed between 1005-1049°C, therefore the BZ+a2 phase
field could extend past 1005°C and exist up to 1049°C for the Ti-21Al-29Nb alloy.

For Ti-Al-Nb alloys, the two phase BCC+a2 phase field narrows with increasing

nominal Nb content over the compositional range of Ti-25Al-25Nb to Ti-15Al-33Nb.

1.4 The O+BCC-tor; Phase Field

The temperature range for the B2+a2+0 phase field for a Ti-23Al-27Nb alloy was
found to be 975-1000°C [Boehlert et al. (1999)]. The Ti-lSAl-33Nb and Ti-21Al-29Nb
samples solutionized at temperatures lying within the O+B2+012 phase field for the Ti-
23Al-27Nb alloy produced a BZ+a2 microstructure for the Ti-21Al-29Nb alloy and a
fully B microstructure for the Ti-lSAl-33Nb alloy. The existence of the three-phase field
has been included in Figure 5.2 but its existence for nominal alloy compositions
containing less than 23 at% A] can only be speculated. In this thesis all the samples
solution treated at subtransus temperatures for 3 hours followed by water-quenching
possessed some volume fi'action of a; phase in the microstructure; this is due to
subtransus processing and will be discussed shortly.

Since Figure 5.2 is an isopleth, the narrowing of the proposed three phase field
with increasing Nb content in addition to the downward curvature of the three-phase field
with decreasing temperature is philosophically valid. Though not necessary, because it is

an isopleth, the shape of the three phase field can be validated through the application of

257

Gibb’s Phase rule [Reed-Hill and Abbaschian (1994)]. Gibb’s phase rule is given by the
following equation:

P + F = C + 2 (3)
where P is the number of phases, F is the number of degrees of freedom, and C is the
number of components. In this phase field P equals 3 (O, BCC, and 012 phases) and C
equals 3 (Ti, Al, and Nb are the alloy components). Solving for F then yields a value of
2, so it is valid for both temperature and composition to vary over the indicated range as
long as pressure remains fixed.

By examination of Figure 5.2 it can be seen that the a; phase is only an
equilibrium phase at temperatures within in the BCC+a2 phase field. One of the goals of
this thesis was to produce O+BCC microstructures devoid of the (12 phase in order to
enhance RT ductility. Microstructures fiee of the a; phase were produced by
solutionizing above the BCC-transus or by solutionizing Ti-21Al-29Nb samples in the
O+BCC phase field for 200 hours followed by water quenching.

For both the Ti-15Al-33Nb and Ti-21Al-29Nb alloys, the a; phase was removed
through solution treatment above the BCC-transus followed by water quenching. This
method is easy to accomplish, but incurs the debit of retaining large prior-BCC grains,
which are typically deleterious to RT ductility, and are not easily removed from the
microstructure without subtransus temperature hot working. For creep driven structural
applications supertransus heat treated microstructures may be acceptable, but for RT
structural applications large prior-BCC grains are undesirable.

In the Ti-21Al-29Nb alloy the a; phase was retained in the microstructure afier

solutionizing for three hours in the O+BCC phase field, but was eliminated by increasing

258

'J.‘ in

   
  
  
    
 
  
   

~12» 1111 u

,_s}liklb

" «11 min“. ‘I
t-l blan-

., '1» at! I.

' )ln’hw V '

‘1 1".1 It: [
~‘lr 111%.. T
:' xtmfl 1 ..
‘ 1,7,1le '._-

¢
'v'C-tAI r.
-1

the solutionizing time to 200 hours in the O+BCC phase field. This is attributed to the
sluggish diffusion kinetics of the a; phase at lower temperatures, since it is only a stable
phase at temperatures just below the BCC-transus where diffusion is more rapid. The 0-
phase was observed to form as a “rim” around the a; phase, which produces an
additional diffusion barrier and further traps it within the microstructure. It is assumed
that 200 hours at the solutionizing temperature allowed enough time for the stabilization
of equilibrium microstructures. While the removal of the a; phase is desirable, the
necessity of heat treating for long periods of time (up to 200 hours) for its removal is
impractical in an industrial environment.

The presence of the a; phase in every subtransus Ti-lSAl-33Nb and Ti-21Al-
29Nb microstructure, with the exception of the Ti-21Al-29Nb samples solutionized for
200 hours, is a result of the alloy processing and in particular the temperatures at which
the processing was performed. When the master alloys were melted and cast into ingot
form, the ingot contained every constituent phase because the ingot was slow cooled fi'om
the melt through every phase field. Since both alloys were hot-worked at subtransus
temperatures, the a; phase present in the ingot still remained after hot rolling into sheet
form. In addition, the Ti-21Al-29Nb ingot and sheets were processed at 982°C, which
lies in the BZ+0t2 phase field for this composition.

If the O+B2+a2 and O+B+012 phase fields do exist for compositions between Ti-

23Al-27Nb and Ti-15A1-33Nb, their respective temperature ranges are extremely narrow.

259

1.5 The O+BCC Phase Field

The two phase O+BCC phase field existed below 910°C for the Ti-15Al-33Nb
alloy and below 960°C for the Ti-21Al-29Nb alloy. Ti-15Al-33Nb samples heat treated
at 650°C for 100-250 hours resulted in the formation of the highest volume fraction of 0-
phase. Increasing the solutionizing temperature within the O+BCC phase field resulted
in a larger BCC phase volume fraction at the expense of O-phase volume fraction. The
size of the O-phase laths also increased when both the solutionizing temperature and
solutionizing time were increased. Heat treating for 200 hours within this phase field
resulted in the O-phase morphology transitioning fiom a lath morphology to an equiaxed
morphology due to O-lath coarsening. This morphological transition occurred in both the
Ti-15Al-33Nb and Ti-21Al-29Nb alloys, see Figures 4.244.26. Therefore, O-phase
morphology is not only dictated by the temperature the heat treatment is performed at,
but also the dwell time of the heat treatment. Identical heat-treatments within the
O+BCC phase field always resulted in greater O-phase volume fractions for the Ti-21Al-
29Nb alloy compared to the Ti-15Al-33Nb alloy. The amount of O-phase that can be
formed within a given Ti-Al-Nb microstructure is therefore strongly dependent upon the
nominal Al content because the O-phase stoichiometry is TizAle. Increasing the
nominal Al content in Ti-Al-Nb alloys increases the amount of O-phase attainable.

In Figure 5.2, the horizontal border between the O+B2 and O+B phase fields was
initially suggested by Bendersky and coworkers [Bendersky et al. (1991)]. Since the
phase evolution study conducted in this thesis provided no further evidence to refute or
support the temperature at which this border is suggested to exist at, it has not been

modified. Long term aging of the Ti-21Al-29Nb alloy at various temperatures between

260

650-800°C followed by EMPA and TEM analysis of the BCC phase would provide

insight into where the BZ/B border exists within the O+BCC phase field.

2. Phase Transformation Mechanisms

Three possible phase transformation mechanisms have been found to occur for Ti-
Al-Nb alloys but only one phase transformation mechanism was observed in this thesis
[Boehlert et al. 1999), Muraleedharan et al. (1992b), Bendersky et al. (1991), Rowe and
Hall (1991), Rowe (1991b), Rowe and Larsen (1996), Boehlert et al. (1997a), Rowe and ‘
Gigliotti (1990)]. The phase transformation mechanism observed was conventional
Widmanstatten precipitation of the O-phase from the BCC phase. The AP material for
both the Ti-15Al-33Nb and Ti-21Al-29Nb alloys contained only the BCC and a2 phases.
Therefore, the transformation of the parent BCC phase to Widmanstatten O-phase
occurred for all the aging heat treatments performed on the sheet material.
Widmanstatten precipitation occurs when plate, or even needle-shaped precipitates grow
in such a manner that they are aligned along specific crystallographic planes or directions
of the matrix crystals [Reed-Hill and Abbaschian (1994)].

Below 875°C in the O+BCC phase field, grain boundary diffusion and grain
growth kinetics are more sluggish due to lower temperatures, and Widmanstatten
precipitation of the O-phase has been shown to be the dominant phase transformation
mechanism [Boehlert et al. (1999)]. The growth of the O-phase laths was observed to
first initiate at the prior-BCC grain boundaries. Growth of the Widmanstatten O-laths
then proceeded into the interior of the prior-BCC grains. This phase transformation

mechanism requires lattice reconstruction along with thermally activated diffusion of

261

atoms across the O and BCC phase boundaries. High concentration gradients and large
lattice mismatches occur at the tips of the growing O-laths [Boehlert et al. (1999)].
These features increase the chemical fi’ee energy difference between the O and BCC
phases and assist in driving the phase transformation of BCC to 0. Chemical and
compositional equilibrium is reached when the chemical potential of each of the
components is equal within both the BCC and 0 phases.

Orientation relationships (OR) arise between the constituent phases when phase

transformations occur. The OR in which the a; phase grows from the parent B2 phase is

the same OR found between the a and B phases in pure zirconium described by Burgers
[Burgers (1934)]. The classic Burger’s OR is [l l-20]a//[l-II]B; (0001)01//(110)B. The
OR between the O and B2 phases is [1-11]B2//[110]O; (110)B2//(001)O [Muraleedharan
(1992b)]. In both orientation relationships the closest-packed directions and closest-
packed planes of each phase are parallel with one another.

For the Ti-15Al-33Nb solution treated and water quenched microstructures, the
O-phase always precipitated first as laths with the Widmanstatten morphology. The Ti-
15Al-33Nb samples that were solutionized at 855°C, 910°C, and 960°C for 200 hours
were the only microstructures for this alloy composition in which the O-phase adopted an
equiaxed morphology. This morphological transition is attributed to O-phase coarsening
with increased dwell time. For Ti-lSAl-33Nb samples solutionized at 855°C, 910°C, and
960°C for 3 hours the O-phase was present only in lath form. The size of the
Widmanstatten O-phase laths was a function of initial solutionizing temperature. For the

HT:650°C/100h/FC microstructure, the O-phase lath width was estimated to be on the

262

order of 50-100 nm. For heat treatments performed at 855°C and above, O-phase lath
width is typically in the range of about 1-5 um.

For the Ti-21Al-29Nb solution treated and water quenched microstructures, the
O-phase occurred in both lath and equiaxed morphologies. For solutionizing
temperatures between 855-960°C the O-phase occurred in both morphologies, regardless
of the solutionizing dwell time. For the 960°C/200h/WQ microstructure, the O-phase
occurred in a fully-equaixed morphology. Lower solutionizing temperatures favored a
Widmanstatten morphology, and higher solutionizing temperatures favored an equiaxed
morphology. This observation is in agreement with previous results obtained for Ti-
25Al-25Nb and Ti-23Al-27Nb alloys [Boehlert et al. (1999)].

As mentioned previously, Widmanstatten precipitation of the O-phase was the
only phase transformation mechanism observed in the solution treated and aged Ti-lSAl-
33Nb and Ti—21Al-29Nb microstructures. The reason Widmanstatten precipitation was
the only phase transformation mechanism observed is because rapid quenching from the
BCC phase field to RT was not involved in the aging heat treatments performed. Rapid
quenching leads to a high degree of supersaturation of the BCC phase. This provides
more of a chemical driving force for the phase transformations not observed in this thesis,
which were reviewed in Chapter 2.

The composition-invariant and cellular precipitation phase transformation
mechanisms reviewed in Chapter 2 were avoided by performing the multi-step aging heat
treatments without water-quenching. Avoiding these phase transformation mechanisms
was advantageous in that these mechanisms have been shown to produce thermally

unstable microstructures. The discontinuous cellular phase transformation did not reach

263

completion even after 304 hours of aging at 650°C in a Ti-23Al—27Nb alloy that was first
solutionized and quenched from 950°C [Boehlert et al. (1999)]. The projected in-service
operating temperature range for orthorhombic titanium-aluminides is 550—800°C. Clearly
the cellular precipitation phase transformation must be avoided in order to maintain
microstructural stability in creep-driven applications. The disadvantage to performing
the aging heat treatments through furnace and control cooling was that the desired two-

phase O+BCC microstructures contained 1-5 volume percent a; phase afler aging.

3. The Effect of Boron Modification on the Microstructure of Ti-Al-Nb
Alloys

The addition of 0.5 at% boron to the Ti-15Al-33Nb alloy did not produce a
significantly different grain size, and no boride needles were visible within the
microstructure in either the AP or HT:1005 conditions. This suggests that 0.5 at% boron
is soluble within one or more phases within the microstructure. It is noted that no EMPA
analysis was performed on the phases in this alloy. Afler the heat treatment, both the Ti-
15Al-33Nb and Ti-15Al-33Nb-0.5B microstructures contained prior-BCC grain sizes
between 120-130 um, suggesting that 0.5 at% B did not significantly affect the BCC-
transus temperature or reduce the grain size. The increased volume fraction of the 012
phase within the AP microstructure of the Ti-15Al-33Nb-0.5B alloy is an effect of
processing at 975°C, and further confirms that the two phase a2+B phase field exists
below the [3 phase field for the Ti-15Al-33Nb alloy. Electron microprobe analysis of the

constituent phases in the Ti-15Al-33Nb-0.5B alloy is required to determine the boron

264

  
    
 
  
    
  
  
    

1 .(1 141; an

' ’ 11.1.1100
‘ ' ..1' ,‘H '; ' v ~

r

11‘ 1MB“

4.
Atheist. 7

. ‘ .
,1
“an?!“ M'
3 my)“. a»; .

I 1"

'U 101110 «- ‘ _
1) l..1{.d’m ‘.
1'1 JiJfiib J:

' "Hill? (1.;

. . ' 1:11 Jflhni -1 .-
b

- 1 , manning _
.-‘ Vii-1T but;

n
, . . .. fl a. .
W... 'A'rI‘M ..... _ _ .. . .. .... -_ 1 ”a: 0. ‘ .
' '1" 1 .1_1 .I

‘1». 04: U
'.

-; .- a ., . ._ g , u , 1 ‘1‘ Wife; .

:;l'."]"<1u ._
_l

Weary..."

l

content present within the BCC phase, 0 phase, and in particular the a; phase in order to
help elucidate the reason for the large increase in (12 volume fiactions.

The addition of 5 at% boron significantly altered the microstructure of both the
Ti-15Al-33Nb and Ti-22Al-26Nb alloys. Boride needles were evident within the Ti-
15Al-33Nb-5B and Ti-22Al-26Nb-5B microstructures ranging in length fi'om 3-158 um
and up to 22 um in width. The maximum length of boride needles observed in the Ti-
15Al-33Nb—SB alloy was approximately 50 um, and this length corresponds to the
approximate minimum length of boride needles observed in the Ti-22Al-26Nb—5B alloy.
This is significant because the Ti-15Al-33Nb-5B alloy displayed superior tensile strength
and creep resistance to the Ti-15Al-33Nb alloy and the tensile strength and creep
resistance of the Ti-22Al-26Nb-5B alloy were inferior to the Ti-22Al-26Nb alloy.
Therefore, the mechanical properties of this alloy system are sensitive to the size scale
and distribution of the boride phase.

The prior-BCC grain size in the Ti-lSAl-33Nb—5B and Ti-22Al-26Nb—SB samples
was significantly reduced when compared to the Ti-l 5Al-33Nb, Ti-lSAl-33Nb-0.5B, and
Ti-22Al-26Nb alloys. In fact, the Ti-15Al-33Nb-5B alloy exhibited a grain size half that
or the Ti-lSAl-33Nb and Ti-lSAl-33Nb—0.SB alloys after an identical supertransus
1005°C heat treatment was performed on all three alloys. The reduction in grain size in
the Ti-15Al-33Nb-5B is attributed to the homogenous distribution of fine boride needles
pinning prior-BCC grains from growing. Interestingly, the boride needles occurred on
the same length scale as the prior-BCC grain size (48-63 um) in the Ti-15A1-33Nb-5B
alloy. The a; phase volume fiaction was similar in the Ti-15Al-33Nb HT:1005 and the

Ti-15A1-33Nb-5B HT:1005 alloys. The Ti-15Al-33Nb—5B HT:1005 alloy contained 25

265

volume percent more O-phase than the monolithic HT:1005 alloy. The increase in 0-
phase volume fiaction is thought to be due to the different processing routes employed
for each alloy. The boron-modified alloy was cast and HIPed (172 MPa/900°C/4h) while
the monolithic alloy was hot forged and hot rolled at 899°C. The O-phase was present in
the as-HIPed Ti-15Al-33Nb-5B microstructure before the HT:1005 heat treatment was
performed, while it was absent in the as-rolled microstructure. The HT:1005 heat
treatment restored the O-phase in the Ti-lSAl-33Nb rolled sheet material. Since the 0-
phase transforms from the parent BCC phase upon cooling in supertransus heat treated
Ti-Al-Nb alloys (as discussed in the phase transformation subsection of this chapter) the
increased volume fiaction of O-phase does not decrease the prior-BCC grain size.
Therefore, the 50% reduction in prior-BCC grain size in the Ti-15Al-33Nb-5B alloy must
be due to the homogeneous distribution of boride needles obstructing BCC grain growth.

The boride needles occupied only 2% of the volume of the Ti-22Al-26Nb-5B
microstructure. These needles were not homogeneously distributed. In this case, the
reduction in grain size of the Ti-22Al-26Nb-5B alloy in comparison to the Ti-22Al-26Nb
alloy is suggested to be primarily a result of the increased volume fiaction of a; phase,
which is believed to have inhibited BCC grain growth. The Ti-22Al-26Nb-5B alloy
contained 20% or; phase by volume in comparison to 6% or; phase measured in the Ti-
22Al-26Nb alloy. The boride needles are not believed to be the primary cause for the
reduction in grain size due to the their smaller volume fiaction, their non-homogeneous
distribution, and because they were present on a length scale approximately 2-7 times
larger than the measured equiaxed grain size of 23 um. In the Ti-22Al-26Nb-5B alloy,

microstructural features that would be expected to increase strength (smaller grain size,

266

higher 012 volume fraction, and boride phase needles) and creep resistance (higher a2
volume fiaction and boride phase needles) when compared to the monolithic alloy did not
result in enhanced mechanical behavior.

The BCC-transus of Ti-22Al-26Nb alloys produced through powder metallurgy is
between 1120-1125°C [Smith et al. (1999), Smith et al. (2000)]. DTA experiments
indicated that the Ti-22Al-26Nb—5B BCC-transus was 1130°C. Thus the addition of 5
at% boron did not significantly affect the BCC-transus temperature for the Ti-22Al-26Nb
alloy. A DTA analysis was not performed on the Ti-15Al-33Nb-5B alloy.

The most significant microstructural finding was the chemical composition of the
boride phase. The EMPA data from the boride phase in both the Ti-15Al-33Nb-5B and
Ti-22Al-26Nb-5B alloys suggests a phase composition of BzTiNb. This is significant
due to the fact that boron additions to Ti-6Al—4V (wt%) alloys have produced boride
whiskers with a composition of TiB or TiB2 [Tamirisakandala et al. (2005), Gorsse and
Miracle (2006)]. The SADPs displayed in Figure 4.35, indexed according to the lattice
parameters determined through density functional theory calculations [T rinkle (2006)],
provide strong evidence that the crystal structure of the boride phase is orthorhombic
B27. This phase appeared with lighter contrast in BSE images when compared to the TiB
phase [Tarrririsakandala et al. (2005), Gorsse and Miracle (2006)] due to the fact that Nb
is substituting for Ti on half of the titanium (c) Wyckoff sites (0.177, 0.25, 0.123) in the
B27 lattice. Convergent beam electron diffi'action (CBED) patterns of low index zone

axes obtained fi'om a single boride are needed to confirm the crystal structure.

267

 

B. Creep Discussion

1. Microstructure-Creep Relationships

For the Ti-15Al-33Nb microstructures the minimum creep rate decreased with
increased O-phase volume fiaction, with two exceptions. The minimum creep rates of
the Ti-15Al-33Nb HT:1105 microstructure (V f O-phase = 0.63) exceeded those of the Ti-
15Al-33Nb HT:1005 microstructure (V f O-phase = 0.44) for 0' 2 225 MPa. The second
exception was the Ti-15Al-33Nb HT:650°C/250h/FC microstructure, which displayed the
worst creep resistance of all the Ti-15Al-33Nb microstructures. This was unexpected
because this microstructure contained the largest volume fiaction of O-phase (V f O-phase
= 0.73) of all the Ti-15Al-33Nb nricrostructures. The fine grain size, 4 pm, of this
microstructure is expected to have contributed to its poor creep resistance. This
microstructure not only contained the most O-phase by volume, but the O-phase also
occurred on the finest size scale in this microstructure. The average O-lath width was
measured to be on'the order of about 50-100 nm, compared to about 1-5 um for the
HT:960, HT:1005, and HT:1105 microstructures. A decrease in O-lath spacing would be
expected to increase the creep resistance of these alloys as a decrease in O-lath spacing
within the BCC grains implies a lower mean-flee slip path within the BCC grains.
Decreasing the mean-free slip path should promote strain hardening due to pile up of
BCC dislocations at O-phase grain boundaries. The pile up of BCC phase dislocations at
O-phase laths is a feature that has been observed in post creep deformed Ti-Al-Nb alloys
[Boehlert and Miracle (1999), Boehlert (1999)]. Steady state creep deformation is
achieved during dislocation climb controlled creep when strain hardening is balanced by

recovery processes. Increasing the rate of strain hardening would decrease the minimum

268

creep rate for dislocation-climb controlled deformation, which is suggested to occur for
this microstructure at 0' 2 99 MPa. This sample still exhibited the poorest creep
resistance for O+BCC microstructures at stresses less than 99 MPa where dislocation
climb creep was not expected. Based on n and Qapp values small grain sizes would be
expected to exhibit worse secondary creep resistance if dislocation climb was not the rate
controlling secondary creep mechanism. Therefore, the inferior creep resistance
displayed by the Ti-15Al-33Nb HT:650°C/250h/FC microstructure cannot be due to
volume fiaction effects and is due to its fine equiaxed grain size of 4 pm.

The AP Ti-15Al-33Nb and Ti-21Al-29Nb alloys were not creep tested, but they
would have had poor creep resistance. The BCC phase would have been transforming
into the O-phase during creep deformation at 650°C, therefore the AP Ti-15Al-33Nb and
Ti-21Al-29Nb nricrostructures would also have incurred additional creep strain due to
microstructural instability.

The subtransus Ti-21Al-29Nb HT:960 and HT: 1005 microstructures contained
larger O-phase volume fractions than the identically heat treated Ti-15Al-33Nb
microstructures. Both Ti-21Al-29Nb microstructures performed comparable to the
subtransus 960°C heat treated Ti-15Al-33Nb rrricrostructure and out-performed the
650°C heat-treated Ti-15Al-33Nb microstructure. However, both the supertransus heat
treated Ti-15Al-33Nb microstructures displayed superior creep resistance to the
subtransus Ti-21Al-29Nb microstructures. Thus higher nominal Al contents are not
necessary in order to increase the creep resistance of Ti-Al-Nb alloys. Supertransus heat
treatments were not performed on the Ti-21Al-29Nb alloy because the corresponding

increase in prior-BCC grain size would have further decreased RT 8f values. The

269

previous statement is supported by the fact that the supertransus heat treated Ti-15Al-
33Nb microstructures displayed RT 8f values approximately half those of the subtransus
heat treated Ti-15Al-33Nb microstructures.

The microstructural features of the single phase Ti-5Al-45Nb alloy provided
helpful insight into the microstructure-creep relationships of multiphase Ti-Al-Nb alloys.
The Ti-5Al-45Nb alloy produced a similar minimum creep rate (1.41x10’8 8") under test
conditions of 10 MPa/650°C as the Ti-15Al-33Nb HT :1005 microstructure produced
(1.53x10'8 5") under test conditions of 226 MPa/650°C. The Ti-5Al-45Nb alloy had an
equiaxed grain size 37.5 times larger than the Ti-15A1-33Nb HT:1005 microstructure,
which was the most creep resistant microstructure produced. This is significant because
increasing the equiaxed grain size resulted in greater creep resistance for all the Ti-15Al-
33Nb and Ti-21Al-29Nb microstructures studied, with the exception of the trade-off
observed above a 120 um grain size for the Ti-15Al-33Nb alloy. The fact that the BCC
phase in Ti-5Al-45Nb is disordered could also have contributed to the decrease in creep
resistance. However, the lack of O-phase in this microstructure was expected to have
been the primary reason for the poor creep resistance of this alloy compared to the other
Ti-Al-Nb alloys studied. This is supported by the fact that the subtransus Ti-15Al-33Nb
HT:960 microstructure (V f disordered [3 phase = 0.37) displayed greater creep resistance
than the subtransus Ti-21Al-29Nb HT :960 (V f = 0.28 ordered B2 phase) microstructure
and comparable creep resistance to the Ti-21Al-29Nb HT:1005 (V f = 0.17 ordered B2
phase) microstructure. The poor creep resistance of the Ti-5Al-45Nb alloy showed that a
minimum volume fraction of O-phase is required for acceptable creep resistance at the

stress levels that would be experienced in gas turbine engine environments.

270

Upon review of the literature for O-based and 012-based Ti-Al-Nb alloys, it is clear
that supertransus microstructures exhibit superior creep resistance to subtransus
microstructures [Mishra and Banerjee (1990), Smith et al. (1993), Woodard and Pollock
(1997), Boehlert and Miracle (1999), Boehlert (1999), Hayes (1996), Rowe and Larsen
(1996), Boehlert and Bingert (2001), Nandy et al. (1993), Nandy et al. (1995), Cowen
and Boehlert (2006a), Cowen and Boehlert (2006b)]. The common theme between all
Ti-Al-Nb alloys which contain the O, BCC, and a2 phases is that supertransus
microstructures contain larger equiaxed grain sizes than subtransus microstructures. In
addition to containing larger equiaxed grain sizes, the microstructures that were
supertransus solutionized and aged typically contained more O-phase by volume than
subtransus solutionized and aged microstructures. Therefore, higher temperature
solutionizing heat treatments followed by appropriate aging heat treatments impart
geater creep resistance due to both an increase in equaixed gain size and an increase in
O-phase volume fraction. The solution-treated and aged Ti-Al-Nb alloys studied in this
thesis were all aged at the same temperature, 855°C. These microstructures obeyed the
trend of O—phase volume fiaction and average equiaxed grain size simultaneously
increasing with an increase in solutionizing temperature.

While O—phase volume fraction was believed to be the dominant microstructural
feature in imparting creep resistance to Ti-Al-Nb alloys [Mishra and Banerjee (1990),
Smith et a1. (1993), Smith et al. (1995),Woodard and Pollock (1997), Hayes (1996),
Rowe (1993), Rowe et al. (1993), Rowe and Larsen (1996)] the results of this thesis and
previous works [Boehlert and Miracle (1999), Cowen and Boehlert (20063), Cowen and

Boehlert (2006b)] suggest that average equiaxed gain size is more dominant than 0-

271

phase volume fiaction. The large-gained supertransus Ti-15Al-33Nb microstructures
outperformed the fine-gained subtransus Ti-21Al-29Nb microstructures, despite the Ti-
21Al-29Nb microstructures containing a larger volume fiaction of the O-phase. The
creep behavior of the Ti-5Al-45Nb alloy indicates that a minimum volume fraction of 0-
phase is required in order for the microstructure to posses adequate creep resistance; the
equaixed gain size of this microstructure was 4500 um and it possessed the worst creep
resistance of all the alloys in this thesis. For a Ti-12Al-38Nb alloy, the minimum creep
rates were found to decrease with increasing gain size from 33 to 138 um [Boehlert
(1999)]. The Ti-12Al-38Nb samples with a 337 um gain size displayed similar
minimum creep rates as the samples with a 138 um gain size, which implies that above
138 um, the prior-BCC gain size does not affect the minimum creep rate. A similar
result was obtained in this thesis for the Ti-15Al-33Nb alloy, as increasing the prior-BCC
gain size above 120 um did not further decrease the minimum creep rate. All the Ti-
12Al-3 8Nb microstructures contained O-phase volume fiactions of approximately 30
volume percent and produced minimum creep rates lower than or comparable to the Ti-
21A1-29Nb microstructures which contained 72-78 volume percent O-phase and a 8-12
um equiaxed gain size. The minimum volume fraction of O-phase required for adequate
creep resistance is therefore suggested to be 30 volume percent. This statement is
supported by the fact that the minimum volume fraction of O-phase present in the
solution-treated and aged Ti-lSAl-33Nb microstructures was 37 volume percent. For the
rrricrostructures that contained more than 30 volume percent O-phase, the average
equiaxed gain size dominated the creep resistance for the stresses and temperatures

examined.

272

 

2. Prior-BCC Grain Boundary Cracking and Sliding/Upheaval

The in-situ creep experiments provided evidence that the equiaxed prior-BCC
gain boundaries were the locus of damage during creep deformation of the supertransus
heat treated microstructures. Prior-BCC gain boundary cracking during creep
deformation was observed in the Ti-15Al-33Nb, Ti-15Al-33Nb-5B, and Ti-22Al-26Nb
alloys. Internal cracking was evident within the bulk of conventionally creep tested Ti-
15Al-33Nb HT: 1005 samples tested in air. This cracking was localized to the prior—BCC
gain boundaries, which is in ageement with the in-situ deformation observations
obtained in the same stress regime for this microstructure. However, the amount of gain
boundary cracking observed on the sample surface was more extensive than that within
the bulk. The degee of constraint on the sample surface is much less than that within the
bulk, therefore a geater extent of cracking and sliding/upheaval would be expected to
occur at the sample surface.

The environmental stability of Ti-Al-Nb alloys at projected application
temperatures raises a strong concern due to the large degee of edge cracking observed
for samples creep tested in air. As would be expected, environmentally-assisted edge
cracking did not occur preferentially over prior-BCC gain boundary cracking for the
samples tested in vacuum.

ln-situ creep tests performed on the Ti-15Al-33Nb HT:1005 samples indicated
that the cracking and sliding/upheaval of prior-BCC gain boundaries was the dominant
creep deformation mechanism under test conditions of 225 MPa/650°C. Figure 5.5
shows the prior-BCC gain boundary crack opening displacements measured for the three

prior-BCC gain boundaries labeled GB 1, GB 11, and GB III in Figure 5.6. Figure 5.6 is

273

a higher magrification BSE image of the same prior-BCC gain boundaries displayed in
the damage evolution images shown in Figure 4.48 (a)—Figure 4.48 (n). Figure 5.7
displays BSE images of the two prior-BCC gain boundaries that exhibited the largest
degee of deformation of all the prior-BCC gain boundaries located within the gage

section of the Ti-15Al-33Nb HT: 1005 sample tested at 225 MPa/650°C.

274

 

 

 

 

9 Maximum Displacement of GB 1, pm

I Maximum Displacement of GB II, um
E A Maximum Displacement of GB 111, um
1 V Maximum Displacement of the Most Deformed GB, pm

a 8 Maximum Displacement of the 2nd Most Deformed GB, um
*" 30 ——-
: F 1 1 1 1
0
E .
O _ _
a 25
_
at
.2 v
a: 20 — - -
a
g 15 1. V L _.
5
g .
l—~ ' —1
g 10
o—
g e
U 5 T e . . I T
a ' . . t
e R X ‘

5 e - = = I I I ' I
E 0 . I .1 1 1 m
a 0 100 200 300 400 500
2 Total Creep Displacement, um

Figure 5.5. Maximum gain boundary displacements measured for the gain boundaries
labeled 1, II, and III in Figure 5.5 as a function of the measured total creep displacement.
The maximum gain boundary cracking displacement for the gain boundaries shown in
Figure 5 .6. The gain boundaries shown in Figure 5.6 displayed the largest extent of
gain boundary cracking/sliding out of all the gain boundaries contained within the gage
section of the Ti-15Al-33Nb HT: 1005 sample creep tested under conditions of 225
MPa/650°C.

275

   
    

   

.’
/

”I b a, T.

’7.‘ [A ‘1‘.

‘3 m1; 111
10 um “u _/

II' I\‘ “\

   
 

Figure 5.6. Backscattered electron images of gain boundaries 1, II, and III at creep
displacements of (a) 0 mm and (b) 0.41 mm. This figure shows the gain boundaries
from which the displacement measurements in Figure 5.6 were measured. The loading
direction is horizontal in each image.

 

* Is a.‘ ‘x.
«51815.

J]

..
'1

, a a

‘1‘.

. ...I.

;
\t;
1»

.‘II,
\o .

p

F.
\

... r

a.
.r..

.

 

.
n
u
M
v

(b)

Figure 5.7. Backscattered electron images obtained at a creep displacement of 0.41 mm

for (a) the most extensive amount of gain boundary on" ‘- '
the second most extensive amount of gain boundary - .,.. 1'

loading direction is horizontal in each image.

'and (b)

l ‘I'j‘

 

The

a ...-“my r

 

277

The total creep strain is the sum of the strain accumulated through intraganular
deformation and the strain accumulated through gain boundary sliding. The importance
of gain boundary sliding as a deformation mechanism during creep depends on its
contribution to the total creep strain and the interrelationship between sliding and
intragranular deformation [Garofalo (1965)]. If gain boundary sliding is dependent upon
prior intraganular deformation, it serves primarily as an accommodation process and not
as a rate controlling process [Garofalo (1965)]. Upon examination of Figures 4.48(a)-(n)
and Figures 5.7 and 5.8 it is clear that the prior-BCC gain interiors remained essentially
undeforrned in comparison to the gain boundaries. In particular, Figure 5.6 (b) shows
that intraganular deformation occurred only in areas adjacent to the prior-BCC gain
boundaries; the interiors of the prior-BCC gains appear exactly the same as they did
before creep deformation. The distances between O-phase laths located at the interiors of
the prior-BCC gains were measured as a function of total creep displacement in order to
provide a quantitative estimate of the amount of displacement that occurred within the
prior-BCC gain interiors. A maximum displacement of 0.1 pm was obtained afier
measuring the distances between approximately 20 different O-laths located within prior-
BCC gain interiors before and after 412 pm of total creep displacement. The in-situ
creep experiments demonstrated that the prior-BCC gain boundary deformation was not
caused as accommodation to intraganular deformation. Therefore, it is fair to state that
the gain boundary cracking and sliding/upheaval observed were the strain rate
controlling processes, and not an accommodation mechanism to intraganular

deformation.

278

Grain boundary sliding occurs on boundaries randonrly oriented to the specimen
surface and the tensile axis, therefore computing the longitudinal strain contributed by
gain boundary sliding is difficult [Garofalo ( 1965)]. Measurements were performed in
order to quantify the amount of local deformation occurring within the microstructure and
provide a quantitative estimate of the amount of the total displacement that was
contributed by gain boundary cracking/sliding/upheaval. Grain boundary crack opening
displacements and gain boundary sliding/upheaval displacements were measured from
the BSE images acquired at specific total creep displacements. At a total creep
displacement of 412 um, maximum displacements of 7 um, 5 um, and 5 pm were
measrued for GB I, GB II, and GB 111, respectively; maximum displacements of 26 um
and 20 um were measured for the most deformed GB and the 2lid most deformed GB,
respectively. Summing the measured displacements yields a value of 63 pm, which
represents 15.3% of the total creep displacement measured. As previously stated, the
displacement measured during the in-situ tests also includes the displacement of the
gipping fixtures. This implies that the displacement value measured from cracking and
sliding/upheaval represents a higher percentage of the total creep displacement than
15.3%. It is also sigrificant that 15.3% of the total creep displacement value could be
constituted by the cracking and sliding/upheaval of 5 prior-BCC gain boundaries. The
combined cracking and sliding/upheaval of the two most deformed gain boundaries
accounted for 11.2% of the total creep displacement. Figure 5.8 displays low
magrification images which illustrate the variation in gain boundary deformation
observed throughout the sample tested at 225 MPa/650°C. The gain boundaries in

Figure 5.6 (GB 1, GB 11, and GB 111) are present in the upper-middle region of Figure 5.8

279

(a) and Figure 5.8 (b). Comparing Figure 5.6 and Figure 5.8, the deformation
experienced by GB 1, GB 11, and GB 111 represents a smaller degee of deformation than
the average deformation experienced by all the prior-BCC gain boundaries within the
sample gage length. While it would be impractical to measure the crack opening
displacement of every prior-BCC gain within the gage section of the sample, it is
reasonable to state that deformation due to gain boundary cracking and sliding/upheaval

accounts for the majority of the total creep displacement measured.

280

 

)

(b

: 1 005
lacement of

-15Al-33Nb HT

i

images obtained fi'om the T
(a) Pretest and (b) at a total creep d

Backscattered electron

sample creep tested at 225 MPa/650°C

8

5

Figure

13p
in area

illustrates

The boxed-

ntal in each image.

on is ho

ading d' ’

The 10
the same area shown in

412 pm.

6

5

Figure

281

3. Creep Deformation Mechanisms

3.1 The Apparent Activation Energy for Creep

Constituent O-phase and BCC phase lattice self-diffusion data are unavailable in
the literature. Self—diffusion data is available for the a; phase, and was presented in
Table II in Chapter 2. The lattice self-diffusion activation energy data for Ti3Al was
obtained by performing radioactive tracer (“T i) experiments applying the Boltzrnann-
Matano method and Darken’s equation [J . Rusing and C. Herzig (1996)]. The activation
energies for Ti, Al, and Ti3Al diffusion in Ti3Al are 288, 374, and 312 kJ/mol,
respectively [J . Rusing and C. Herzig (1996)]. Activations energies obtained from creep
experiments performed on Ti3Al, Ti3Al+5Nb, and Ti-24Al-11Nb 01; alloys have fallen
within the range of 107-120 kJ/mol at low stresses and 206-285 kJ/mol at high stresses
[Mishra and Banerjee (1990), Mendiratta and Lipsitt (1980), Malakondaiah and Rao
(1981)]. The values obtained at high stresses correspond well with the activation energy
for lattice self-diffusion determined by Rusing and Herzig. The activation energies
determined at low stresses are approximately half the value of those determined at high
stresses. The activation energies determined at low stresses are therefore suggested to be
representative of gain boundary diffusion, based on the fact that the activation energy for
gain boundary diffusion in pure metals and alloys is typically half that of lattice self-
diffusion [Evans and Wilshire (1985), Hertzberg (1996)].

In the a; phase present within Ti-Al-Nb alloys, the Ti and Nb atoms share the
same lattice sites while the Al atoms occupy fixed lattice sites. Microprobe data from the
a; phase correlates well with the specific site occupancies as the Al content is uniformly

close to 24 at% and the Ti and Nb contents change with variations in nominal alloy

282

content (See Table XI and [Boehlert et al. (1999)]). The fiequency factor and the
activation energy for Nb diffusion in Ti3Al have been determined to be 3.15x10‘1 mzs'l
and 339 kJ/mol, respectively [Breuer et al. (1999)]. The value of 339 kJ/mol is 51 kJ/mol
higher than the activation energy for Ti self-diffusion in Ti3Al [J . Rusing and C. Herzig
(1996)]. The authors attribute the difference in activation energies to be due to an
increase in the migation energy as well as the formation energy of vacancies in the
neighborhood of Nb atoms [Breuer et a1. (1999)]. It is therefore fair to conclude that if
increasing the Nb content increases the activation energy for lattice self-diffusion in
Ti3Al, then 0 alloys will posses higher activation energies for lattice self-diffusion than
(12 alloys due to the fact that O alloys contain higher nominal Nb contents than a; alloys.
The activation energies determined fi'om creep experiments performed on Ti-Al-Nb
alloys have shown that O alloys possess higher apparent activation energies for creep
when compared to 012-based alloys and conventional Ti alloys (see Table IV in Chapter
2). The increase in activation energy for creep is one of the main explanations given for
the increased creep resistance of O-based alloys over a; alloys and conventional Ti alloys
[Boehlert and Miracle (1999), Peters et al. (2003)].

Until diffusion data for the O-phase and for O alloys is obtained, the activation
energies obtained for O alloys must be compared to those known for Ti3Al and 012-based

alloys.

283

1 film

, 11“ d“.

  
  
  
  
   
     

. Il'JVl --
" ""0 1
1.1.11;

'1

HO.- o' ‘
1 I [all M
_. 75100:! d

' lliw

‘ ~11 no} -

,’ 112d!
I‘Jilu‘i‘in I s .
-..'tz me i J

432-1157! 3
l
.

,u

t ”4113131)": In

7

' ‘ ' » .. .. . . - rut "5"" . _ . J
1 ' ' " ‘-
.H ‘

. '. .-

'~ . ..m C so!*

3.2 Creep Deformation Mechanisms Based on Pure Metal Theory

Pure metal creep theory has been used extensively to suggest the dominant
secondary creep mechanisms in the O alloy literature that has been cited in this thesis.
Therefore, in order to rank and compare the alloys studied in this thesis with those of
other works, pure metal creep theory was used to suggest the dominant secondary creep
deformation mechanisms based on n and Qapp values calculated from the power-law
equation for creep strain rate. In addition, the in-situ SEM observations were used to
identify the dominant secondary creep mechanisms as described in the previous section
of this Chapter. The alloys studied in this thesis are not pure metals or single-phase
disordered alloys. They are multiphase alloys that contain both disordered solid-solution
phases and intermetallic phases. However, the gain boundary cracking and
sliding/upheaval observed during in-situ creep experiments correlates well with the creep
stress exponent obtained fiom pure metal theory. The BCC phase in the Ti-15Al-33Nb
alloy is a disordered solid solution of Ti, Al, and Nb atoms and the majority of the
deformation observed in-situ occurred at prior-BCC gain boundaries. The disordered
BCC phase microstructure of the Ti-5Al-45Nb alloy displayed the worst creep resistance,
despite having the largest gain size of 4500 um. This illustrated the need for a minimum
volume fraction of O-phase for creep resistance. In the multiphase alloys that contained a
significant volume fiaction of the O-phase, the microstructural feature that dominated the
creep behavior was the equaixed gain size.

The measured creep stress exponents and apparent activation energies suggest
three different deformation mechanisms are active for the stress and temperature range

examined.

284

The low-stress regime corresponded to n values ranging 0.7 to 1.3 and Qapp
values ranging 100-184 kJ/mol. Coble creep deformation is suggested to be the dominant
deformation mechanism in this regime. Harper-Dom creep was not considered because
the average equiaxed gain size was found to dominate the creep behaviOr of the alloys
studied in this thesis. One of the requirements for Harper-Dom creep is that
microstructures with different gain sizes produce identical creep strain rates when tested
under identical conditions [Owen and Langdon (1996)]. Nabarro-Herring creep
deformation occurs by diffusion of atoms and vacancies through the bulk of the material.
Nabarro-Herring creep deformation was disregarded because the activation energies
calculated were much lower than known values for Ti3Al and values for O and (12 alloys
determined from creep experiments.

The intermediate-stress regime corresponded to n values ranging fiom 1.5 to 3.2
and Qapp values ranging from 163-217 kJ/mol. Therefore, gain boundary sliding is
suggested to be the dominant deformation mechanism for intermediate stresses at 650°C.
The in-situ creep tests showed that the majority of the creep deformation occurred
through gain boundary cracking and sliding/upheaval. In addition, the intraganular
deformation that was observed only occurred adjacent to the prior-BCC gain boundaries.
Measurements of crack opening and gain boundary sliding/upheaval displacements
demonstrated that the macroscopic creep displacement measured could be accounted for
by the opening of gain boundary cracks and sliding/upheaval of prior-BCC gains.
Grain boundary sliding models predict the strain rate due to gain boundary sliding to be
directly proportional to oz, i.e. n=2, and inversely proportional to the gain size [Langdon

(1970), Crossman and Ashby (1975), Evans and Wilshire (1985)]. Figure 5.9 is a plot of

285

minimum creep rate versus oz/d for the all the alloys studied in this thesis with an n value
in the range of 1.5-2.8, with the exception of the Ti-5Al-45Nb alloy. The minimum creep
rate versus 02/d for the Ti-5Al—45Nb alloy is presented separately in Figure 5.10; the
4500 um gain size of the Ti-5Al-45Nb alloy makes the linear relationship
indistinguishable due to the scale of the x-axis in Figure 5.9. When the creep stress
exponent lies in the range of 2.5-3.1 a linear relationship exists between the minimum
creep rate and GZ/d. The Ti-21Al-29Nb HT:1005 microstructure displayed a creep stress
exponent value of 3.2 over the stress range of 48-250 MPa. However, the creep stress
exponent value over the stress range of 48—172 MPa is 2.8. Therefore, gain boundary
cracking and/or sliding/upheaval could be contributing a sigrificant amount of the total
creep strain to the Ti-21Al-29Nb HT:1005 microstructure for stresses less than or equal
to 172 MPa. In-situ creep testing of the Ti-21Al-29Nb alloy would enable qualification

of the previous statement.

286

1', ft;

' d1 «have

. 1141' “I

 

 

  

 

 

 

' Ti-15Al-33N b HT:650°C/250h/FC
' Ti-lSAI-33Nb HT:1005
‘ Ti-lSAl-33Nb HT:1105
3 5 10,, v Ti-21Al-29Nb HT:1005
.7 3 10'”- 3
g 2 5 10'? -
m d=173p.m d=4umx
8: 2 10'8— —
5
.3
51510— , -
E 1 10“- .- -
-- +d=120 um d=1211m
2 5 10'9_ - / / ' 4
,/1/ '
0 r W/ 1 1 1
0 5 106 1 107 1.5 107 2 107 2.5 107
ozld

Figure 5.9. The measured minimum creep rate versus oz/d for the Ti-15Al-33Nb and Ti-
21Al-29Nb microstructures creep tested within the 11 value regime of 1.5-3.2 at 650°C.

287

310"7 1 1 1 re 1 1

 

 

 

 

 

v-t -7
M 2.510 - —
45‘
‘5
m 210-7- a
Q.
8
1- 1510-7- —~
U
E
g 110-7r —
IE
2 510'“— —
0 l l l l l l
6 103 8 11)3 1 104 1.2 1041.4 1041.6 1041.8 10“ 2 10‘

ozld, sz5

Figure 5.10. The measured minimum creep rate versus 0'2/d for the Ti-5Al-45Nb alloy
creep tested in the n=2.6 regime at 650°C.

288

u—‘fi “fir ‘

'11.

1""
a.
f

'5

 

The high—stress regime corresponded to n values ranging fiom 3.8-6.0 and Qapp
values ranging 288-317 kJ/mol. Dislocation climb would be suggested to be the
dominant deformation mechanism in the high-stress regime based on pure metal theory.
Combining the calculated n and Qapp values with the in-situ deformation observations of
gain boundary cracking in the Ti-15Al-33Nb HT:1005 supertransus microstructure, the
rate controlling mechanism in the high stress regime could also be interganular crack
gowth that is dependent on lattice self-diffirsion or gain boundary diffusion.

Within the high stress regime, the Ti-15Al-33Nb supertransus microstructures
displayed superior creep resistance to all the subtransus microstructures. The Ti-lSAl-
33Nb supertransus microstructures consisted of large prior-BCC gains containing a more
uniform and tightly spaced dispersion of Widmanstatten O-phase laths when compared to
the HT:960 subtransus microstructure. The glide and climb of dislocations through the
BCC phase would therefore be more difficult in the supertransus microstructures due to
the decrease in mean-free slip path in the BCC phase. BCC phase dislocations have been
observed to pile up at the O/BCC interfaces in creep deformed specimens [Boehlert and
Miracle (1999)], so by incorporating more O/BCC interfaces through supertransus heat
treatment, the resistance to dislocation climb controlled creep should increase. The
Weertrnan model of dislocation creep states that the climb of the dislocations at the head
of a dislocation pile up is rate controlling when dislocation climb is the dominant
deformation mechanism [Weertrnan (1968)]. If the Weertrnan model of dislocation creep
is assumed to be rate controlling for supertransus Ti-15Al-33Nb microstructures,
increasing the number of O/BCC interfaces would promote strain hardening and decrease

the minimum creep rate.

289

The results of this thesis suggest that the average equiaxed gain size affects the
stress at which the onset of dislocation controlled creep is suggested to occur in the Ti-
15A1-33Nb alloy, if the Weertrnan model is assumed to be rate-controlling. This is
sigrificant because gain size theoretically should not to have an effect when dislocation
climb controlled creep is the dominant mechanism. The TH 5Al-33Nb HT:650/250h/FC
microstructure contained an O-phase volume fraction of 0.73 and an average equaixed
gain size of 4 um. The Ti-153Al-33Nb HT:1005 nricrostructure contained an O-phase
volume fraction of 0.55 and a prior-BCC gain size of 120 um. The separation of 0-
phase laths was on the order of 50-100 nm in the 650°C heat treated microstructure as
opposed to 1-5 pm in the 1005°C heat treated microstructure. Therefore, if the
Weertrnan model is assumed, the 650°C nricrostructure should have out performed the
1005°C microstructure in the n 2 3.8 regime. In actuality, the 650°C microstructure
performed the worst out of all the Ti-lSAl-33Nb microstructures. The prior-BCC gain
boundaries were observed to be the locus of damage in terms of cracking and
sliding/upheaval in-situ in the Ti-15Al-33Nb HT :1005 microstructure. Decreasing the
gain size 116 um could account for the decrease in creep resistance based on the
vulnerability of the BCC gain boundaries observed in the Ti-15Al-33Nb, Ti-15Al-33Nb-
5B, and Ti-22Al-26Nb alloys. The increase in minimum creep rates displayed by the Ti-
15Al-33Nb 650°C/250h/FC microstructure is suggested to be due to increased gain

boundary area available for cracking and sliding/upheaval in both stress regimes.

290

3.3. A Composite Creep Model Based on Constituent Phase Data

When deciding what type of model was applicable to the creep data obtained for
multiphase Ti-Al-Nb microstructures, important considerations were made.

First, the validity of the assumptions that the models are based on must be
considered. Some models assume both phases are ductile, other models assume one
phase is ductile and the second phase only experiences elastic deformation, and other
models assume one phase is ductile and the second phase is rigid and non-deforming.
Creep models typically assume that only one phase in the multiphase microstructure
experiences creep deformation. The literature available for models in which both phases
are creeping in a two-phase system are sparse [Chan (2002), Mileiko (1970), Kelly and
Street (1972), Bullock et al. (1977), Henshall and Strum (1996), Henshall et al. (1997),
Bartholomeusz and Wert (1994), Bartholomeusz and Wert (1995)]. However, this
assumption would be the most applicable for the Ti-Al-Nb alloy system. From the
microstructural deformation observations made in Chapter 4, it was concluded that the
BCC phase experiences the geatest minimum creep rates, and the worse overall creep
resistance of the constituent phases. Planar surface slip was observed to occur in the 0-
phase, but never to an extent approaching that of the wavy slip observed in the BCC
phase. The fiacture surfaces revealed ductile dimpling within the BCC phase and
transganular cleavage fracture across 0 gains. The in-situ creep experiments revealed
that the prior-BCC gain boundaries experience the majority of the deformation, but the
O-phase does experience some deformation.

The microstructure in which the constituent single-phase data for the model

comes from must be taken into account. The gain size of the single-phase

291

microstructure that the constituent phase creep strain rate comes from will be significant.
In this work, the gain size of Ti-Al-Nb alloys has shown to be an important
microstructural parameter that affects the creep deformation behavior over a wide range
of stresses for a number of different nominal alloy compositions.

The chemical composition of the phases can influence the creep properties. This
was illustrated in this work by the difference in BCC phase Al content in both the Ti-
15Al-33Nb and Ti-21Al-29Nb alloys. For example, when both alloys are single-phase
BCC, the BCC phase composition is the nominal alloy composition. In O+BCC
microstructures, the Al is transferred from the BCC phase when the O-phase precipitates.
For example, the chemical composition of the B2 phase was Ti-18Al-31Nb in a Ti-21Al-
29Nb 910°C/200h/WQ two-phase microstructure (V f O-phase=0.72 Vf B2 phase=0.28).
The chemical composition of the [3 phase was Ti-12.5Al-37.9Nb in the Ti-15Al-33Nb
HT:960 two-phase microstructure(V f O-phase=0.37 Vf B phase=0.58).

The morphology of the phases may change when present in the dual-phase
microstructure. While single-phase microstructures typically consist of equiaxed gains,
dual-phase microstructures may exist in a wide range of morphologies. Some possible
two-phase morphologies include: second phase equiaxed gains, a fine dispersion of
second phase particles, a coarse dispersion of second phase particles, second phase
particles segegated at gain boundaries, as a second phase gain boundary film, or as
second phase laths, plates, or needles such as Widmanstatten laths. The microstructures
creep tested in this thesis typically contained equiaxed BCC gains, equiaxed a2 gains,
and Widmanstatten O-laths which gew from the prior-BCC gain boundaries into the

gains.

292

Constituent phase data collected from single-phase nricrostructures with the same
equiaxed gain size, the same chemical composition in the two-phase microstructure, and
the same morphology in the two-phase microstructure are needed in order to be able to
neglect the effect of change in nricrostructure on the model. Clearly, it is difficult if not
impossible to meet the previous three requirements for the Ti—Al-Nb alloy system. These
three conditions could possibly be met in the limiting case of a two-phase microstructure
consisting of equiaxed gains of each phase of nominally the same size. This type of
microstructure was not achieved through the heat treatments performed. A
nricrostructure based on equiaxed gains of both the O and BCC phases would be invalid
because the O-phase was shown to exist predominantly in lath form at 650°C. The model
to be applied to the Ti-15Al-33Nb alloy would ideally be based on equiaxed gains which
contain second-phase needles or laths. It is the author’s belief that Finite Element
modeling of a microstructure containing equiaxed prior-BCC gains which contain
Widmanstatten O-phase laths would be the most viable method of determining the creep
strain rate of the individual 0 and BCC phases during creep deformation. From single
phase data, it is apparent that the constituent phases in Ti-Al-Nb alloys creep at different
rates in multiphase microstructures. An empirical model will now be described in an
attempt to model the creep strain rate based on constituent phase data.

Figure 5.11 portrays the stress dependence of the minimum creep rate at 650°C
for fully-BCC, O+BCC, and fully-O Ti-Al-Nb alloys. Upon examination of Figure 5.11,
it appears that the fully-BCC and fully-O microstructures provide the lower and upper
bounds, respectively, for the creep resistance displayed by Ti-Al-Nb alloys. Therefore,

an empirical model for the minimum creep rate of O+BCC microstructures was

293

developed based on the constituent phase strain rates of the fully-BCC Ti-5Al-45Nb alloy
and the fully-O Ti-26Al-27Nb alloy. The Ti-5Al-45Nb alloy was used for BCC phase
properties as it was the first Ti-Al-Nb alloy produced to remain fully-BCC at 650°C. The
Ti-26Al-27Nb alloy was chosen because its fully-O microstructure consisted solely of O-
laths. Most fully-O alloys, such as the Ti-27Al-21Nb alloy shown in Figure 5.12, consist
of equaixed O gains. Since the O+BCC microstructures creep tested in this thesis
contained predominantly Widmanstatten O-laths, the O-phase morphology of the Ti-
26Al-27Nb alloy was felt to be more representative of that found in the Ti-15A1-33Nb
and Ti-21Al-29Nb microstructures. The strain rate of the BCC phase is assumed to
behave according to the power-law stress relationship obtained for the Ti-5Al-45Nb alloy
for stresses 2 34 MPa given by:

EBCC = 3.83x10'12 0'2'6 (4)

where o is the applied creep stress. The strain rate of the O-phase is assumed to behave
according to the power-law stress relationship obtained for the Ti-26Al-27Nb alloy for

stresses S 317 MPa given by:

$0 = 1.47x10’150'2'3 (5)

where o is the applied creep stress. Therefore, the total creep strain rate on a volume

fraction weighted basis is given by:

5W =3.83x10“20'2'6V/BCC +1.47x10'”az‘3Vfl, (e)

The creep rates predicted by this model for the Ti-15Al-33Nb and Ti-21Al-29Nb alloys

are presented in Figure 5.12. For microstructures with an equiaxed gain size S 6 pm, the

294

J) '

   
   
   

; ’1' (h,
-; .1163
. 11(1))”.
Hurry]

-: .1an

creep rates predicted by the model are within approximately one order of magritude of
the experimentally measure creep rates. For equaixed gain sizes 2 8 urn the values
predicted by the model over predict the strain rate by as much as three orders of

magnitude, but are typically within two orders of magritude.

295

Ti—SAl-45Nb Fully Equiaxed BCC
Ti- 12 \I—3 8Nb 29% Lath 0+ 71% Equiaxed BCC [Boehlert (I999)
TI-15Al-33N b 44% Lath O + 55% Equiaxed BCC

Ti- 26Al- 27Nb Fully Lath O [Boehlert and Bingert (2001)]
Ii- 27 \l—21\hl11111.1\t(l()[\antlycral.(I‘Nill

itVebl.

10'7 ,— °, 4

.81
10 .—

Creep Rate, s-l

 

rnlmum
ll
.1»)
IV)
7

10‘9 :— I x H. t . ,../y n: 5.1 i:

M
e,\
3

 

 

10"”T=6.50‘.’C......1 . -
10 100 1000
Stress, MPa

 

Figure 5.11. The stress dependence of the minimum creep rate of fully-BCC, O+BCC,
and fully-O Ti-Al-Nb alloys.

296

The strain rates predicted by the model deviate from the experimentally measured
values due to several reasons. First, this model relies on the assumption that both phases
exhibit power-law creep behavior. Second, the creep stress exponents and the power-law
creep constant A were assumed to remain constant with applied stress. For the Ti-5Al-
45Nb alloy the creep stress exponent is assumed to maintain a value of 2.6 for all applied
stresses 2 34 MPa. For the Ti-26Al-27Nb alloy the creep stress exponent is assumed to
maintain a value of 2.3 for all applied stresses 3 317 MPa. Based on the transitions in
creep exponents observed in this thesis, the assumption of constant 11 and A values is
suggested to have more validity for the Ti-26Al-27Nb alloy than the Ti-5Al-45Nb alloy.
Since the Ti—5Al-45Nb alloy is a disordered solid solution, a transition in creep stress
exponent to a value of 4 or higher would be expected at some applied stress above 75
MPa.

The nricrostructures of the single-phase alloys that the constituent phase data is
based upon are not the most accurate representation of the O+BCC microstructures creep
tested. However, they are the most accurate that are available.

In the case of the Ti-5Al-45Nb alloy, its gain size of 4500 um is 26-1125 times
larger than the equiaxed gain size of all the two-phase alloys creep tested in this thesis.
The limited quantity of this alloy that was produced for this thesis was the only quantity
of this alloy ever produced. Since it was processed at 800°C, and no BCC-transus was
found for this alloy, reducing the gain size would only have been possible by melting,
casting, and hot-rolling more of this material at lower temperatures. The crystal structure
of the Ti-5Al-45Nb alloy was determined to be disordered BCC. The crystal structure of

the BCC phase in the Ti-15Al-33Nb alloy was also determined to be disordered BCC, but

297

the minimum Al content was measured to be 12.5 at % Al in the BCC phase in this alloy.
The crystal structure of the Ti-21Al-29Nb alloy was determined to be ordered B2, in
addition to containing a minimum Al content of 18 at%. Therefore, the large difference
in Al concentration of the BCC phase is expected to be another cause of the discrepancy
in creep rates between the estimated/modeled creep rates and those measured
experimentally.

The Ti-26Al-27Nb alloy is the most creep resistant O alloy produced, to date. As
previously mentioned, its microstructure consisted completely of O-laths, and no
equiaxed O-phase. Therefore, this type of microstructure was thought to be more
representative of the O-phase found within the O+BCC microstructures in this thesis.

The Ti-27Al-21Nb alloy examined by Nandy et al. (1993) was not used for the
constituent O-phase data as it had an equiaxed microstructure, was creep tested in
compression, and exhibited an n value of 6.2 over the stress range examined. Tension-
compression effects on the creep behavior of Ti-Al-Nb alloys have shown to be
negligible do to similar minimum creep rates occurring during both tension and
compression creep [Boehlert and Miracle (1999)]. However, all the alloys in this thesis
were creep tested in tension, therefore the Ti-26Al-27Nb alloy (which was tensile-creep
tested) had the added benefit of consistency in testing method. The HT:650, HT:960, and
HT: 1005 Ti-15Al-33Nb microstructures displayed a transition in 11 value below 317 MPa.
Therefore, basing the creep rate of the O-phase on an 11 value of 6.2 for stresses 5 317

was thought to be less valid that using the n value of 2.3 over this stress range.

298

— Ti-lSAI-JJNb HT:650°C/250II/FC Model

0 Ti-lSAI—JJNb HT:650°C/250h/FC Experiment
_ Ti-lSAl-JJNb HT:960 Model

0 'l‘i-ISAI-JJNb IIT:960 Experiment
— ’I'i-I5 \I-33\h II I :IIIIIS “odd

I 'I'i-l5.\I-33\h II T: I005 Ilrpcrimcnt
'I'i—IS \I~33\h HT:1 105 “(1ch

A l'i~l5.\|-33\h II I':I l05 Experiment
_'li-2I \l-Z"\h II 1:01.11 \lndrl

 

V li-ZI \l-Z‘)\h II INMI I\p\-r1meut ——v——.—7—7——r—Y—,—d
— I1—2I \l-Z'Hh II lzltm< \IIItIt‘I
L Ii-ZI\Il"\I>|||:IIIII5I\pt-1imt‘nl

1.

-l

10*5 -

 

Creep Rate, s
H
9.

4
4
'0‘

inrmum
07

M

 

 

Stress, MPa

Figure 5.12. The minimum creep rates determined experimentally and by the model
given in equation (6) versus applied stress. The solid lines represent the model
calculation and the data points are from experiment.

299

The only microstructural feature this empirical model is based on is the volume
fraction of the constituent phases. For the O+BCC microstructures this model was
applied to, the average equiaxed gain size was shown to have more of an effect on the
minimum creep rate than the O-phase volume fraction. The average equiaxed gain size
was not incorporated into the empirical model developed. Figure 5.13 displays the
minimum creep rates determined by the model, after being normalized by the average
equiaxed gain size (in pm) of each microstructure. Taking the Ti-15Al-33Nb HT:1005
nricrostructure as an example, the minimum creep rates predicted by the model were
divided by 120. Normalizing the model strain rates by the average equaixed gain size
can be rationalized due to the fact that theory predicts diffusional creep mechanisms
[Nabarro (1948), Herring (1950), Coble (1963)] and gain boundary sliding deformation
[Langdon (1970)] to be inversely proportional to the gain size. Normalizing by the
average equaixed gain size produced model minimum creep rates within the same order
of magritude as the experimentally measured creep rates, with the exception of the Ti-
15Al-33Nb HT:960 and Ti-21Al-29Nb HT:1005 microstructures. The maximum
deviation observed for the Ti-15Al-33Nb HT:960 and Ti-21Al-29Nb HT:1005
microstructures was one order magritude. Table XXVI lists the minimum creep rates
determined by the normalized and unnormalized empirical model as well as the
experimentally measured minimum creep rates.

The results of the empirical model are interesting, but due to the lack of a physical
basis are unconvincing. Further work is needed in order to model the minimum creep

rate of O+BCC microstructures and Finite Element modeling is recommended.

300

— Ti-l5Al-33Nb HT:650°0250h/FC Model
0 TI—lSAl-33Nb HT:650°C/250IIIFC Expcrlmellt
Ti-lSAI-JJNb HT:960 Model
0 l'i-lSAl-JJVI) lsz960 Experiment
— 'l‘i-IS \I<33\h ll'l':l005 \lndel
I Ii-IS.\I«53\b ll |:I005 Experiment
— Iii~l5 \I»33\h HT:1105 Model
A 'l'i—ISM-JJVh ll l:l I05 F.\peri1nent
—- Ii»lI \IVZ‘)\h II Iz‘IMI \Itnlcl
V Ii—ZI \l-I‘I\Ir H L960 IApt‘r‘imcnt , f x , , , Y ,?

 

 

— Ii—ZI \I 2"\|1III:II|0.< \lmIt-I
L Iirll\I-Z‘I\hIII:I00.<l\pcr11nt-nt

10‘7

vrvrrt

LML LIJII

Creep Rate, s-l

 

10'9 5

inrmum
1 1v1I

10'10‘ . . . , n 1265.0“;
10 100 1000

Stress, MPa

M

 

 

 

Figure 5.13. The minimum creep rates determined experimentally and by the model
given in equation (6) versus applied stress. The model values obtained from equation (6)
were then normalized by the average equiaxed gain size. The solid lines represent the
model calculation and the data points are from experiment.

30]

Table XXVI. Minimum CreeLRates Determined by Experiment and Empirical Model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Applied Creep Stress, MPa I Model 1 Experiment LModel Normalized
Ti-15Al-33Nb HT:650
49 5.51x10'8 s" 1.07x10'8 s" 1.38x10'8 s"
75 1.67x10'7 s“ 1.43x10'8 s" 4.16x10'3 s“
99 3.43x10W s" 3.11x10'8 s" 8.57x10'8 s"
126 6.42x10'7 s" 1.19x10'7 s;r 1.6Ox10'7 s"
149 9.92x10'7 s" 2.07x10'Ts°T 2.48x10'7 s‘I
170 1.40x10'6 s" 4.46xlO’Ts'I 3.49x10'7 s‘I
Ti-lSAl-33Nb HT:960
49 5.51x10“ s" 1.04x10'9 s'I 9.18x10’9 s"
75 1.67x10'7 s" 1.71x10'9 s" 2.78x10‘8 s"
99 3.43x10'7 s;T 2.48x10'9 s" 5.71x10"‘s‘I
125 6.28x10'7 s'I 6.75x10'Ts'I 1.05x10'7 s°‘
150 1.01x10'6 s'I 9.34x10'9 s:I 1.68x10'7 s"I
171 1.42x10'6 ST 2.22x10‘8 s'I 2.37x10'7 s'I
Ti-ISAl-33Nb HT:1005
148 7.40x10'7 s" 5.78x10'9 s" 6.16x10'vs'1
171 1.08x10‘ s" 7.60x10'9 s;I 8.97x10“’ s'I
201 1.64x10‘6 s'I 1.07x10“s" 1.37x10‘8 s"
226 2.22x10'6 s;I 1.53x10“‘?~;I 1.85x10'8 s'1
251 2.92x10'6 s'I 2.08::104’};I 2.43x10’8 s"
273 3.63x10'6 s" 5.19x10'8 s" 3.03x10'8 3"
Ti-lSAI-33Nb HT:1105
151 6.38x10’7 s;T 4.81x10'9 9;1 3.69x10'9 s"
172 8.94x10'7 s‘I 8.65x10'“ s" 5.17x10‘9 s’1
200 1.32x10'6 s‘1 1.32x10’8 s" 7.65x10'9 s"
225 1.8Ox10'6 s" 1.70xle s“ 1.04x10'8 s'I
249 2.34x10'° s" 2.41x104‘ s" 1.35x10’8 9;1
275 3.03x10'6 s" 3.39x10'8 s'I 1.75x10'8 s'I
Ti-21AI-29Nb [1sz
48 2.52x10'8 s" 4.57x10‘9 s'I 3.15x10‘9 s'1
73 7.49x10'Ts'I 7.02x10’9 s'I 9.37x10'9 s;T
99 1.65x10'7 s" 7.64x10'9 s;I 2.07x10‘8 s"
122 2.85x10‘7 s" 1.16x10‘K s" 3.56x10'8 s"
148 4.71x10‘79;I 1.50x10’8 s" 5.88x10’8 s'1
170 6.75x10'7 s" 1.98x10“‘s‘I 8.44x10" s"
Ti-21AI-29Nb HT:1005
48 1.53x10‘8 s" 3.63x10'T0 s" 1.28x10‘9 s“
73 4.55x10'8 s" 6.31x10’m s" 3.79x10'9 s'1
100 1.03x10‘7 s" 3.18x10‘vs'f 8.6Ox10’9 s'I
126 1.88x107 s" 3.74x10'VsT 1.57x104‘ s"
148 2.86x10’7 s" 6.54x10'9 s'I 2.38x10‘8 s"
172 4.22x10'7 s'I 1.10x10'8 s" 3.52x10'8 s'1
225 8.49x10'7 s" 3.8Ox10'8 s" 7.08x10‘8 s'1
250 1.12x10'6 3* 5.93x10'8 s" 9.31x10’8 s'I

 

 

 

 

302

 

4. The Effect of Boron Modification on Creep Deformation

The addition of 0.5 at% boron increased the minimum creep rates while the
addition of 5 at% boron decreased the minimum creep rates for the Ti-lSAl-33Nb alloy.

Based on the calculated n and Qapp values, at least two dominant creep
mechanisms are suggested to be active as a function of stress. Grain boundary sliding is
suggested in the low-stress (o < ZOO-250 MPa) regime and dislocation climb is suggested
in the high-stress (o > ZOO-ZSOMPa) regime for both the Ti-lSAl-33Nb and Ti-lSAl-
33Nb-0.53 alloys. In addition, lower activation energy values were obtained in the low-
stress (163-236 kJ/mol) regime than in the high-stress regime (288-320 kJ/mol) for both
the Ti-lSAl-33Nb and Ti-lSAl-33Nb-O.SB alloys. The addition of 0.5 at% boron to the
Ti-15Al-33Nb alloy increased the minimum creep rates over the examined stress and
temperature range, but the transition stress between the two apparent deformation
mechanisms was relatively unaffected by the boron addition. The onset of dislocation
climb controlled creep is suggested to occur at 200 MPa in the Ti-lSAl-33Nb-0.SB alloy,
while it is not suggested to occur below 226 MPa in the monolithic Ti-15A1-33Nb alloy.

The Ti-lSAl-33Nb-5B alloy displayed a constant n value of 3.2 over the entire
stress range studied with a Qapp value of 310 kJ/mol at a = 275 MPa. Therefore,
dislocation glide or climb is suggested as the dominant deformation mechanism over the
stress range studied. The combination of 66 volume percent O-phase, intermediate prior-
BCC grain size of 63 pm, and fine and homogeneous dispersion of 5 volume percent
boride needles throughout the microstructure account for the superior creep resistance
exhibited by this alloy. The boride needles were postulated to impede prior-BCC grains

frdm sliding, pose a substantial obstacle for blocking the climb of dislocations, and carry

303

more of the load within the microstructure than the rest of the constituent phases. At
concentrations above the suggested minimum O—phase volume fiaction of 0.30,
increasing the prior-BCC grain size up to 120 um always resulted in lower minimum
creep rates for the Ti-15Al-33Nb alloy. The prior-BCC grain size of the Ti-15A1-33Nb-
5B alloy was half that of the monolithic alloy, yet it exhibited superior creep resistance.
Therefore the increase in creep resistance was postulated to be due to both the presence
of the boride needles and the increased volume fi‘action of O-phase.

Microscopy investigations of the sample surface of the Ti-lSAl-33Nb-SB alloy
after failure under creep conditions of 400 MPa/650°C revealed that the prior-BCC grain
boundaries were the locus of damage. In fact, the majority of the boride needles located
directly behind the fracture surface and throughout the gage section remained uncracked
after creep failure (See Figure 4.63). The damage accumulation at the prior-BCC grain
boundaries in the Ti-15A1-33Nb-SB alloy appeared exactly the same as was observed
during in-situ creep testing of the monolithic Ti-lSAl-33Nb alloy. Cracking and
sliding/upheaval occurred almost exclusively at the prior-BCC grain boundaries. Twice
as many prior-BCC grain boundaries were available for cracking and sliding/upheaval
deformation in the Ti-lSAl-33Nb—5B alloy, yet it possessed greater creep resistance than
the monolithic alloy. These observations provide further evidence that the increase in
creep resistance of the Ti-15Al-33Nb-5B alloy was due to the boride phase, and reinforce
the suggestion that the borides carry the majority of the load. Dislocation pile ups have
been observed to fiequently occur at O/BCC interfaces in creep deformed samples
[Boehlert and Miracle (1999)], and pile ups of dislocations from both phases are also

suggested to occur at the boride interface. Observations of dislocation interactions

304

between the constituent phases and the boride needles are needed in order to further
explain the enhancement in creep resistance.

The Ti-22Al-26Nb-5B alloy exhibited a minimum creep rate almost an order of
magnitude higher than that of the monolithic Ti-22Al-26Nb alloy. The poorer creep
resistance of the Ti-22Al-26Nb—5B alloy is expected to be a combined effect of boride
cracking and decohesion of the incompletely sintered spherical compacts. Hagiwara and
Emura [Hagiwara and Emura (2005)] observed improved creep strength with increased
TiB volumes for Ti-6Al-ZSn-4Zr-2Mo (wt%)l 5-20 mass% TiB alloys produced through
blended elemental powder metallurgy via HIPing. To the author’s knowledge Hagiwara
and Emura’s work is the only study documenting the effects of the addition of a boride
phase on the creep behavior of conventional Ti alloys. In the Ti-6Al-ZSn-4Zr-2Mo
(wt%)/ 5-20 mass% TiB alloy and the Ti-15A1-33Nb-SB alloy the boride phases were
significantly finer than those for the Ti-22Al-26Nb-SB material. The chemical
composition of the boride phase was measured to be nearly the same in both alloys.
Therefore a degradation in mechanical properties due to a compositional difference could
not have accounted for the inferior response of the Ti-22Al-26Nb-5 B alloy. Thus the size
and morphology of the boride needles appeared to dictate the creep response, with
benefits gained when they were present on a fine scale and homogeneously distributed.
Once the borides crack (which was observed to occur immediately upon reaching the
creep stresses examined), the load is expected to be shifted to the constituent phases and

the initiated cracks are suggested to act as stress concentrators increasing the creep rate.

305

C. Tensile Behavior Discussion

1. Monolithic Alloy Room-Temperature Tensile Discussion

The volume fraction, structure, and chemistry of the BCC phase control the RT
elongation-to-failure of Ti-Al-Nb alloys. Elongation-to-failure values greater than 2%
were maintained as long as the BCC volume fraction exceeded 30 percent. Previous
research has indicated that a minimum volume fraction of 0.20 BCC phase is required to
maintain RT ductility [Boehlert (2000)]. For B dominated microstructures and fully-B
microstructures, 8f values 2 16.8% were obtained in this thesis.

The AP Ti-21Al-29Nb microstructure illustrated the importance of BCC phase
structure on RT tensile behavior of Ti-Al-Nb alloys. Previous work has shown that fully-
B2 microstructures exhibit brittle behavior at RT [Boehlert (2001)]. The AP Ti-21Al-
29Nb microstructure (3 pm grain size, 0.96 B2 phase, 0.04 a; phase) displayed an
elongation-to-failure value of 1.7%, in comparison to 16.8% for the AP Ti-lSAl-33Nb
microstructure (3 pm grain size, 0.90 [3 phase, 0.10 a; phase). Both microstructures
contained the same grain size and relatively the same phase volume fi‘actions, but differed
in that the BCC phase is ordered in the Ti-21Al-29Nb alloy and is disordered in the Ti-
15Al-33Nb alloy. In fact, the AP TI-21Al-29Nb microstructure contained more BCC
phase by volume that the AP Ti-15Al-33Nb microstructure. The high Al content and
ordering of the BZ phase in the AP Ti-21Al-29Nb alloy is believed to account for the low
8f value obtained.

Increasing the Al content in the 32 phase has been shown to generate lower (if
values at RT [Boehlert (2001)]. The A] content of the BCC phase in the AP

microstructures was not determined by EMPA, but Table XI clearly shows that the BCC

306

phase in the Ti-21Al-29Nb alloy always contained more Al than the BCC phase in the Ti-
15Al-33Nb alloy. The Ti-15A1-33Nb HT:960 microstructure exhibited an elongation-to-
failure value twice that of the Ti-15A1-33Nb HT:1005 microstructure. The increase in
elongation-to-failure can partially be attributed to the decrease in Al content of the BCC
phase. The BCC phase in the Ti-15A1-33Nb alloy contained 12.5 at% Al after the
HT:960 heat treatment and it contained 13.9 at% Al after the HT:1005 heat treatment.

So, in addition to decreasing the 8f value due to an increase in O-phase volume fraction

and grain size, the increase in BCC phase Al content could also account for the lower 8f
value exhibited by the Ti-15A1-33Nb HT:1005 microstructure.

Figure 5.14 displays a RT stress versus strain curves for a Ti-12Al-38Nb alloy
[Boehlert (1999)] and the Ti-SAl-45Nb alloy. The yield strength values were 545 and
553 MPa for the Ti-SAl-45Nb and Ti-12Al-38Nb alloys, respectively. Both alloys
experienced a load-drop afier yielding. This stress strain behavior suggests that the Al
content does not affect the RT tensile strength over the wide compositional range of 5-12
at% Al in this alloy system. Since the BCC phase in both the alloys is a disordered solid
solution of Ti, Al, and Nb atoms, increasing the Al content does not have a strong
influence on strength and ductility. Once the A] content is high enough to order the BCC
phase, which was shown to require a nominal alloy composition of at least 17 at% A1, a
strong correlation between increased A] content in the B2 phase and increased RT
strength at the expense of ductility has been observed [Boehlert (2001)].

Interstitial oxygen content has been shown to have an important influence on the
tensile behavior of Ti-Al-Nb alloys [Gogia et al. (1998), Akkurt et a1. (1991), Gogia et al.

(1992), Boehlert et al. (l997b)]. The oxygen contents were measured to be 1100 ppm

307

and 790 ppm for the Ti-15A1-33Nb and Ti-21Al-29Nb alloys, respectively. Interstitial
oxygen contents > 890 ppm produced 8f values < 1% for both supertransus and
subtransus Ti-25Al-24Nb, Ti-25Al-23Nb, and Ti-23Al-27Nb microstructures [Boehlert
(2001 )]. The previous comparison suggests that the maximum RT 8f value attainable was
displayed by the Ti-21Al-29Nb alloy, yet there may still be room for further
improvement of the RT elongation-to-failure of the Ti-15A1-33Nb alloy by reducing the

oxygen content below 1100 ppm.

308

 

- Ti-5A1—45Nb Fully B d = 4500 pm
+Ti—12Al-38Nb Fully [5 d = 138 pm

600 T l I l .

 

 

Stress, MPa
ca
3
G

 

 

 

200 ~
100 4
0
30

Strain, %

Figure 5.14. Room-temperature tensile stress versus strain curves for fully-[3 Ti-5Al-
45Nb and Ti-12Al-38Nb [Boehlert (1999)] alloys. The Ti-12A1-38Nb alloy did not fail.
It was unloaded after reaching 27% strain.

309

2. Boron-Modified Alloy Room-Temperature Tensile Discussion

The addition of 0.5 at% B to the Ti-15A1-33Nb alloy caused almost no difference
in the RT tensile behavior in both the AP and HT:1005 condition due to the 0.5 at% B
addition not producing significant changes in the microstructure. The AP Ti-15Al-33Nb-
0.5B alloy contained 12 volume percent more a; phase than the AP Ti-15Al-33Nb alloy,
but this was attributed to processing at 975°C within the (n+0 phase field, and not the
addition of boron.

The addition of 5 at% boron to the Ti-15Al-33Nb alloy produced significant
strengthening at the expense of 8f. The heat treated Ti-15A1-33Nb-5B alloy experienced
an increase in elastic modulus of 11 percent, an increase in Y8 of 22 percent, an increase
in UTS of 21 percent, and a decrease in 8f of 52 percent. This strengthening can be
accounted for by both changes in microstructural features and by the presence of the
boride needles within the microstructure. The Ti-15A1-33Nb-5B alloy contained 66
volume percent O-phase compared to 44 volume percent in the monolithic alloy and
increasing O-phase volume fraction has been shown to increase the strength of Ti-Al-Nb
alloys [Cowen and Boehlert (2006), Boehlert (2001), Rowe et al. (1991), Boehlert et al.
(1997c), Smith et al. (1995)]. The prior-BCC grain size of the Ti-lSAl-33Nb-5B alloy
was half that of the monolithic alloy, which leads to the possibility of strengthening due
to the reduction in grain size. It is proposed that the boride needles carry more of the
load than the constituent phases in the microstructure, until a critical amount of cracking
within the boride needles occurs upon which the load is then transferred to the matrix.
These events have been shown to occur previously through surface observations made

during in-situ tensile deformation in a Ti-6Al-4V-1B (wt%) alloy [Boehlert et al. (2006)].

310

The Ti-22Al-26Nb—5B alloy displayed an inferior RT tensile performance when
compared to the monolithic alloy. The Ti-22Al-26Nb—5B samples produced an UTS 39
percent lower than the monolithic alloy at RT. The presence of the boride needles, in
addition to a finer grain size, would be expected to strengthen the microstructure in a
similar fashion as that found for the Ti—lSAl-33Nb-SB alloy and a Ti-6Al-4V-1B (wt%)
alloy [Boehlert et al. (2006)]. The detriment in strength was believed to be caused by the
addition of the large boride needles which initiated fracture and exhibited a brittle
response. Hypereutectic boron compositions have resulted in large borides and low 8f
values for conventional Ti alloys [Gorsse and Miracle (2006), Yolton and M011 (1996)].
In addition, incompletely sintered spherical compacts were postulated to be preferred
sites for cracking and decohesion under tensile stress and would be expected to decrease
the 8f values. It is also noted that the Ti-22Al-26Nb—SB material exhibited a greater
volume fraction of a; phase than the Ti-22Al-26Nb monolithic alloy. This would be
expected to result in lower 8f values due to the greater likelihood of crack formation at
adjacent (12 grains. Preferential cracking at adjacent or; boundaries is a typical
deformation feature observed within Ti-Al-Nb alloys [Boehlert (2001), Gogia et al.
(1992), Boehlert et al. (l997b)].

Ti-22Al-26Nb has proven to be a ductile alloy if care is taken to
thermomechanically process it into the appropriate microstructure and maintain
acceptable oxygen contents [Smith et al. (1999), Smith et al. (2000), Rhodes et al.
(2000), Rowe (1993), Rowe and Larsen (1996)]. The average RT 8f value for the Ti-
22Al-26Nb monolithic alloy in the current study, 1.2%, was almost identical to that

measured for a powder metallurgy HIPed Ti-22Al-26Nb alloy [Smith et al. (2000)].

311

 

 

However, they noted that their worked foil of this composition exhibited an 8f value of
approximately 5%. Thus with additional hot work of the HIPed material, the
microstructure of the incompletely sintered HIPed compact may be more completely
homogenized and expected to exhibit greater 8f values. Improved tensile strength due to
boron addition in a Ti-22Al-27Nb (at%)/6.5 mass% TiB material has been observed
[Emura et al. (2004)]. The RT tensile properties from their work are provided in Table
XXVII. The powder metallurgy material produced in their study was also processed
through HIPing gas atomized powders, but differs from the alloy in this study in that the
HIPed compact was then hot rolled. The hot rolling resulted in a finer distribution of
borides than that for the Ti-22Al-26Nb—5B alloy in the current study. Thus, it is believed
that hot rolling of the Ti—22Al-26Nb-SB compact would possibly have eliminated the
detrimental defects shown in Figure 4.31, while at the same time refining the size of the
boride needles and homogenizing their distribution throughout the microstructure.

Table XXV II. Room-Temperature Tensile Properties for a Ti-22Al-27Nb Alloy With
and Without 6.5mass% TiB

 

 

 

 

 

 

YS, UTS,
Alloy and Processing Route MPa MPa 8;, %
Powder Metallurgy Ti-22Al-27Nb 701 848 6.8
Powder Metallurgy Ti-22Al-27Nb/ 6.5 mass% TiB 1068 1260 2.3
Ingot Metallurgy Ti-22A1—27Nb 818 992 4

 

 

 

 

 

3. Monolithic Alloy 650°C Tensile Discussion

BCC metals are characterized by a strongly temperature dependent component of
the flow stress and the yield stress increases rapidly with decreasing temperature [Reed-
Hill and Abbaschian (1994)]. The microstructures in this study exhibited a decrease in

yield stress with an increase temperature, in agreement with other works performed on

312

O+BCC alloys [Dary and Pollock (1996), Boehlert (2000)]. In addition, the
microstructures that did not exhibit a 0.2% yield stress at RT due to low fracture strains
did exhibit a yield stress at 650°C. Figure 5.15 shows the relationship between yield
strength and temperature for the Ti-15Al-33Nb alloy, the Ti-21Al-29Nb alloy, and other
Ti-Al-Nb O-based alloys [Boehlert (1999), Boehlert (2001), Dary and Pollock (1996)].
The decrease in strength and increase in elongation-to-failure is attributed to thermal
activation of a larger number of slip systems in the O-phase. Previous studies on O-phase
alloys have shown pyramidal dislocations to become active at elevated temperatures
along with increasing activity of the basal and prismatic slip systems with increasing

temperature up to 800°C [Popille and Douin (1996), Banerjee (1995), Banerjee (1997)].

313

 

0 — Ti-25A1-17Nb [Dary and Pollock(1996)]
I Ti-23Al-27Nb [Boehlert (2001)]
—— -—+——— Ti-22Al-23Nb [Dary and Pollock (1996)]
+ Ti-21Al-29Nb HT:1005
v 1145 \l-33\h HT:1005
1200 —— u Ti-12Al-38Nb [Boehlert (1999)] ——1»

1000

l
1

co
9
€

600

Yield Strength, MPa

 

400

 

 

200

 

0 160 260 360 460 560 600 760 800

Temperature, °C

 

Figure 5.15. Yield strength versus temperature for selected O+BCC Ti-Al-Nb
microstructures.

314

4. Boron-Modified Alloy 650°C Tensile Discussion

Upon increasing the temperature from RT to 650°C, the heat treated Ti—15Al-
33Nb-0.5B alloy experienced a decrease in elastic modulus of 18 percent, a decrease in
Y8 of 45 percent, a decrease in UTS of 15 percent, and an increase in sf of 61 percent in
comparison to the Ti-15Al-33Nb HT:1005 alloy. This behavior is contrary to what is
expected with the addition of boron to Ti alloys [Gorsse and Miracle (2006), Boehlert et
al. (2006)], and so far this effect remains unexplained. The 650°C yield strength of the
Ti-15Al-33Nb-0.5B HT:1005 alloy decreased by 45 percent when compared to the Ti-
lSAl-33Nb HT:1005 alloy.

Figure 4.76 shows a BSE images acquired flour a Ti-15Al-33Nb-5B HT:1005
sample that was metallographically prepared prior to tensile testing to a strain of 1.7% at
650°C in vacuum. An extensive amount of cracking is observed to occur within the
borides, with no evidence of slip occurring within the O+BCC matrix. This is contrary to
what was observed in a Ti-6Al-4V-1B (wt%) alloy [Boehlert et al. (2006)], in that afier
extensive boride cracking occurred, slip was observed to emanate from the boride/matrix

interface. The propagation of slip through the (1+0 matrix was then observed to occur

extensively, which was believed to account for the large 8f values obtained.

The Ti-22Al-26Nb-5B alloy produced an UTS 33 percent lower than the
monolithic alloy at 650°C. The inferior tensile behavior at elevated temperature is
suggested to be due to the same reasons proposed for the decrease observed at RT. The
addition of the large boride needles initiated fracture and exhibited a brittle response.
That is, the cracking borides created stress intensity for crack growth in the matrix, rather

than slip. The incompletely sintered spherical compacts appeared to be preferred sites for

315

I; ban--En‘asxr ~

L.

cracking and decohesion under tensile stress and would be expected to decrease strength

and 3f values at 650°C.

D. Fatigue Discussion

The ranges of fatigue strength of conventional and biomedical Ti alloys, including
Ti-6Al-4V (wt%) ELI and Ti-6Al-7Nb (wt%) alloys obtained from the literature
[Akahori et al. (2000b), Niinomi (2003), Okazaki and Hizume (1994), Akahori et al.
(2000a)], are shown in Figure 5.16 in comparison to the Ti-15Al-33Nb and Ti-21Al-
29Nb alloys. The current alloys exhibited greater fatigue strength than the Ti-6Al-7Nb
(wt%) alloy and the other [3 Ti alloys, while the fatigue lives were comparable to those
for Ti-6Al-4V (wt%) at a given maximum applied stress. The compositions of the Ti-Al-
Nb alloys studied in this thesis in weight percent were Ti-9.8Al-47.0Nb (wt%) [Ti-21Al-
29Nb] and Ti-6.9Al-51.7Nb (wt%) [Ti-15Al-33Nb]. Thus the additional Al and Nb
concentration at the expense of Ti improved the fatigue life as evident in the comparison
with Ti—6Al—7Nb (wt%). This is considered to be related to the greater RT tensile
strength exhibited by the Ti-21Al-29Nb and Ti-15Al-33Nb alloys compared with the
values for the Ti-6Al-7Nb (wt%) alloy (E=110 GPa, YS=773 MPa, UTS: 893 MPa, and
at: 5.5%) [Watanabe et al. (2004)].

The good fatigue strength of the Ti-15A1-33Nb and Ti-21Al-29Nb alloys
compared to other Ti alloys is considered to be due to the balance of strength and
ductility brought by the O and BCC phases. Two-phase O+BCC microstructures are
proposed to increase resistance to fatigue crack initiation and slow fatigue crack

propagation when compared to single-phase microstructures. It has been reported that,

316

for Ti alloys with relatively fine microstructures, a large portion of the fatigue life is
occupied by the small fatigue crack initiation and propagation life [Hagiwara (1998)].
This may be one reason why the fine grained AP microstructures exhibited good fatigue
strength compared to the other biomedical Ti alloys.

The fatigue cracks initiated at the specimen surface and propagated parabolically
towards the specimen interior. Relatively wide striations were observed in the stable
crack growth region, and equiaxed dimples were observed in the overload or fast fracture
region. Such fracture surface morphologies are generally observed for ductile metallic
materials. It is noted that a smaller area of dimpled regions was observed for the Ti-
21Al-29Nb alloy than for the Ti-15Al-33Nb alloy, which may have been a result of the
lower elongation-to-failure values and the B2 crystal structure. In addition, the fracture
surface in the crack initiation and stable crack growth regions was relatively flat for Ti-
21Al-29Nb compared with the more tortuous Ti-15A1-33Nb surfaces. Considerable facets
were observed near the crack initiation sites in both the low cycle and high cycle fatigue
life regions for Ti-21Al-29Nb. In the fast fracture area, a mixed-mode fiacture surface
composed of intergranular fiactures, facets, and dimples was observed. Therefore, the
fatigue crack initiation and propagation characteristics of the Ti-21Al-29Nb alloy

resembled more of a brittle fiacture than that for the Ti-15A1-33Nb alloy.

317

 

 

O Ti-l5AI-33Nb AP
I Ti-ZlAl-29Nb AP
9 Ti-lSAI-33Nb HT:1005
A Tl-ZlAl-29Nb HT:1005
V Ti-29Nb-I3Ta-4.6Zr ST+673K |Niin0mi (2003)|
k Ti-h \l—4\' \\ idmanstzlttcn \lglhu I ()lsu/uki :lnll Hizume (1994). Akahori 1'! ul. (20002111
0 Ti-6Al-4V Equiaxed Alpha | ()kazaki and Hizume (I994), Akahori at al. (20003)]
9 ll t. \I "\ll 15”.. l‘: ill:.ll'\ \ijllld I \iullléll'll’ldll.(2531173551
“ _
E 1100 r 9. j
“ - -
m _ -
8 1000 f 0 j
I: : . -' ° . :
m L. -
a 900 :— A A A». .z 0 k I j
= P It” 1‘ Q‘.. . i
g: b o .8“ O . I
- . .
U 800 f ’ . t' j
,_ L.‘ k v . .1
E t ’ 0 J
a r- k V 4
E 700 j V j
m i 3
g L- K .1
i- l l l l ‘1

 

 

 

1—1
a
no

500
1000 10“ 1o5 106 107
Number of Cycles to Failure

Figure 5.16. Maximum cyclic stress versus fatigue life for the Ti-15A1-33Nb and Ti-
21A1-29Nb alloys in comparison to other Ti alloys to be used for biomedical applications.

318

 

   

Surface slip traces were evident on all the AP fatigue samples. Wavy slip
occurred within the ductile BCC phase and slip transfer compatibility between the BCC
and (:2 phases was observed. The ductile behavior is expected to have had a beneficial
effect especially within the low cycle fatigue regions where crack propagation is expected
to dominate fatigue lives. Surface slip was more evident within the BCC phase than the
(12 phase. Thus the BCC phase played a significant role in promoting low cycle fatigue
resistance at RT for the Ti-Al-Nb alloys, as low cycle fatigue is dominated by crack
propagation for Ti alloys [Wagner (1996)].

The run-out samples were subsequently examined in RT tension and these
samples did not exhibit lower strength values (on average) than those which were not
fatigue tested. Thus no clear evidence, in temis of strength loss, was apparent that fatigue
testing reduced tensile strength. The corresponding fracture surfaces exhibited similar
features to those for virgin samples which were not fatigue tested. No significant
differences with respect to the number of crack initiation sites from those specimens
which only underwent RT tensile fracture were evident. Thus, the run-out samples did
not show deleterious effects fiom the fatigue cycling whatsoever. It is suggested that any
microcracks which may have developed during fatigue in such specimens were not
detrimental to the tensile strength. Therefore, it is believed that few large cracks initiated

and grew during fatigue for the specimens that exhibited run out.

319

CHAPTER 6

SUMMARY AND CONCLUSIONS

A. Summary

This thesis constituted a study of several of the aspects of the physical and
mechanical metallurgy of advanced O+BCC Ti-Al-Nb alloys targeted by the Air Force
for structural applications. Seven different Ti-Al-Nb-xB alloy compositions were
evaluated in this thesis, and only the Ti-22Al-26Nb alloy composition had been
previously studied. Publication in peer-reviewed material science journals of the work
performed in this thesis has made data available in the scientific literature that was
previously non-existent. Microstructure characterization and creep deformation behavior
were the main focus of this thesis, although the tensile and fatigue behavior were also
evaluated. The most innovative results of this thesis were produced from the in-situ
creep testing methodology developed.

The subtransus processing of the Ti-lSAl-33Nb and Ti-21Al-29Nb alloys enabled
a wide variety of microstructures to be produced via post-processing heat treatment. In
particular, subtransus processing provided the ability to control grain size, and produce
both fme-grained and course-grained microstructures. The Ti-15Al-33Nb alloy showed
increased workability at subtransus temperatures due to the lower Al content and lower
flow stress when compared to the Ti-21Al-29Nb alloy. Subtransus processing is crucial
for Ti-Al-Nb alloys because controlling the microstructure was shown to have a bigger

influence on alloy performance than controlling the alloy composition. Taking this

320

aspect into account makes O-based alloys much more favorable than y-TiAl based alloys
for similar end-use applications.

Powder metallurgy processing of the Ti-22Al-26Nb alloy proved to be a
successful alternative processing route to conventional ingot metallurgy processing.
HIPing gas atomized powders resulted in a Ti-22Al-26Nb microstructure that exhibited
excellent strength and creep resistance. The Ti-22Al-26Nb-5B alloy possessed inferior
mechanical properties due to the presence of large boride needles and unsintered powder
precursors within the microstructure. Further hot working of the HIPed compacts
produced was suggested in order to homogenize the microstructure of the Ti-22Al-26Nb—
5B alloy.

The knowledge gap for Ti-Al-Nb phase equilibria between the compositional
range of Ti-23Al-27Nb to Ti-12Al-38Nb that existed before this work began was
successfully filled. The BCC-transus temperatures were estimated along with the
temperature ranges of the BCC+a2, O+BCC+0L2, and O+BCC phase fields. The Ti-15Al-
33Nb alloy is the lowest Al containing Ti-Al-Nb alloy, based on a nominal Ti content of
50 at%, in which the (12 phase has been observed. Since the Al content measured by
EMPA in the a; phase was 21-24 at%, a nominal alloy Al content of at 15 at% is needed
for the a; phase to be a constituent phase. Therefore, when the nominal alloy
composition contains 14-25 at% Al, the a; phase will be formed in the microstructure.
Due to the thermal schedule of the aging heat treatments performed, conventional
Widmanstatten precipitation of the O-phase from the BCC phase was the only phase
transformation mechanism observed. The addition of 0.5 at% boron to the Ti-15Al-33Nb

alloy did not produce a significantly different grain size, and no boride needles were

321

visible within the microstructure in either the AP or HT:1005 conditions. This suggests
that 0.5 at% boron is soluble within one or more phases within the microstructure. The
addition of 5 at% boron significantly altered the microstructrue of both the Ti-15A1-33Nb
and Ti-22Al-26Nb alloys. Boride needles were evident within the Ti-15A1-33Nb-5B and
Ti-22Al-26Nb-5B microstructures ranging in length from 3-158 um and up to 22 pm in
width. The maximum length of boride needles observed in the Ti-15Al-33Nb-5B alloy
was approximately 50 um, and this length corresponded to the approximate minimum
length of borides observed in the Ti—22Al-26Nb-5B alloy. This is significant because the
Ti-15Al-33Nb-5B displayed superior tensile strength and creep resistance to the Ti-15Al-
33Nb alloy and the tensile strength and creep resistance of the Ti-22Al-26Nb-5B alloy
were inferior to the Ti-22Al-26Nb alloy. Therefore, the mechanical properties of this
alloy system are sensitive to the size scale and distribution of the boride phase. The most
significant discovery that resulted due to boron modification of the microstructure was
determining the chemistry and suggesting the structure of the boride phase. SADPs
indexed according to the lattice parameters determined through density functional theory
calculations [Trinkle (2006)], provide strong evidence that the crystal structure of the
boride phase is orthorhombic B27. In the boride phase within Ti-Al-Nb alloys, Nb is
suggested to substitute for Ti on half of the titanium (c) Wyckoff sites (0.177, 0.25,
0.123) in the B27 lattice.

Based on creep stress exponents and apparent activation energies three separate

deformation mechanisms were suggested for the stress and temperature range examined.

322

The low-stress regime corresponded to n values ranging 0.7 to 1.3 and Qapp
values ranging 100-184 kJ/mol. Coble creep deformation was suggested to be the
dominant deformation mechanism in this regime.

The intermediate-stress regime corresponded to n values ranging 1.5 to 3.2 and
Qapp values ranging 163-217 kJ/mol. Grain boundary sliding was suggested to be the
dominant deformation mechanism for intermediate stresses at 650°C. The in-situ creep
testing showed that the majority of the creep deformation occurred at the equiaxed grain
boundaries. In addition, the sparse amount of intragranular deformation that was
observed occurred adjacent to the grain boundaries, while grain interiors remained
essentially undeforrned. Examination of the interiors of samples conventionally creep
tested in air in the same stress regime revealed internal cracking localized to the prior-
BCC grain boundaries. However, the cracking and sliding/upheaval observed on the
sample surface occurred to a greater extent than within the bulk of the alloy.
Measurements of crack opening and grain boundary sliding/upheaval displacements
demonstrated that the macroscopic creep displacement measured could be accounted for
by the opening of grain boundary cracks and sliding/upheaval of prior-BCC grains. In
the suggested grain boundary sliding regime, the minimum creep rates determined from
experiment were directly proportional to the square of the applied creep stress normalized
by the average equiaxed grain size.

The high-stress regime corresponded to n values ranging from 3.8-6.0 and Qapp
values ranging 288—317 kJ/mol. Dislocation climb was suggested to be the dominant
deformation mechanism in the high stress regime based on pure metal theory.

Combining the n and Qapp values with the fact that the in-situ observations revealed

323

 

 

 

    
 
 
  
  

-. mun .
""mmd ... l '
,‘ {11.50045
'1: ljll’tq ‘ --'_‘
' ’, lug . A.
.rnlgm ..“
rulfl'wJ (1%}- 3V

V :11

grain boundary cracking as the dominant deformation feature, the rate controlling
mechanism in the high stress regime is most likely intergranular crack growth that is
dependent upon lattice self—diffusion or grain boundary diffusion. Within the high stress
regime, the Ti-15Al-33Nb supertransus microstructures produced lower strain rates than
all the subtransus microstructures. The onset stress for the suggested dislocation climb
regime was a function of microstructure and initiated at lower stress levels for the
subtransus microstructures.

A rule—of-mixtures empirical model based on constituent phase volume fractions
and strain rates was developed in order to predict the minimum creep rate of two-phase
O+BCC microstructures. When the strain rates calculated by the model were normalized
by the averaged equiaxed grain size of the microstructure, the deviation between the
experimentally measured strain rate and the strain rate predicted by the model did not
exceed one order of magnitude.

For the O+BCC microstructures, the average equiaxed grain size had more of an
impact on the minimum creep rate than the O-phase volume fraction for the stress and
temperature range studied. However, the results from the Ti-5Al-45Nb alloy suggested
that a minimum volume fraction of O-phase was required for acceptable creep resistance
at the stress levels that would be experienced in gas turbine engine environments. The
common theme between all Ti-Al-Nb alloys which contain the O, BCC, and a2 phases is
that supertransus heat treated microstructures contained larger equiaxed grain sizes than
subtransus heat treated microstructures. In addition to containing larger equiaxed grain
sizes, the supertransus solutionized then aged microstructures typically contained more

Orphase by volume. Therefore, higher temperature solutionizing heat treatments

324

 

 

 

. 5

"P300

     
  
 

MIL-1m
.-.;l . d“ -

_.'l 11 J] l

“’i'i'J'J “I .: —

-‘ f...
i.‘ II .11], l.

imparted greater creep resistance due to both an increase in equaixed grain size and an
increase in O-phase volume fraction. While O-phase volume fraction was believed to be
the dominant microstructural feature in imparting creep resistance to Ti-Al-Nb alloys
[Mishra and Banerjee (1990), Smith et al. (1993), Smith et al. (1995),Woodard and
Pollock (1997), Hayes (1996), Rowe (1993), Rowe et al. (1993), Rowe and Larsen
(1996)], the results of this thesis and previous work [Boehlert and Miracle (1999), Cowen
and Boehlert (2006a), Cowen and Boehlert (2006b)] suggest that average equiaxed grain
size is more dominant than O-phase volume fraction. The large-grained supertransus Ti-
15A1-33Nb microstructures outperformed the fine-grained subtransus Ti-21Al-29Nb
microstructures, despite the Ti-21Al-29Nb microstructures containing a larger volume
fiaction of the O-phase. The creep behavior of the Ti-5Al-45Nb alloy indicated that a
minimum volume fraction of O-phase is required in order for the microstructure to posses
adequate creep resistance; the equaixed grain size of this microstructure was 4500 um
and it possessed the worst creep resistance of all the alloys in this thesis.

A combination of an O-phase volume fraction 2 30 volume percent and a
moderately large average equiaxed grain size produces the best creep resistance for two-
phase O+BCC microstructures. The O-phase is needed to strengthen the microstructure
by providing resistance to intragranular deformation and by carrying more of the load
than the BCC phase (in a similar manner that the boride needles were proposed to carry
more of the load than the constituent phases in the boron-modified alloys). Decreasing
the number of equiaxed grain boundaries decreases the area available for grain boundary
cracking and sliding/upheaval, which was the dominant deformation mode observed in-

situ. In addition, increasing the average equiaxed grain size inherently provides

325

resistance to Nabarro-Herring creep deformation, Coble creep deformation, and grain
boundary sliding creep deformation. The important point for the O-based Ti-Al-Nb
alloys is that the microstructural dependence of the creep behavior observed in this thesis
has been displayed over a wide range of nominal alloy compositions [Cowen and
Boehlert (2006a), Cowen and Boehlert (2006b), Boehlert and Miracle (1999), Boehlert
(1999)]. For solely creep-driven structural applications where RT ductility is not the
main requirement, Al-rich and Nb—lean Ti-Al-Nb alloys are desirable due to enhanced
oxidation resistance (more Al provides more A1203 as a protective oxide) and lower
density (less Nb). Orthorhombic-based Ti-Al-Nb alloys can posses adequate creep
resistance if the average equiaxed grain size and O-phase volume fiaction are controlled
through subtransus processing and appropriate post-processing heat treatments,
regardless of nominal alloy composition.

When 5 at% boron was added to the Ti-Al-Nb microstructures, a significant
increase in creep resistance was gained when the morphology and distribution of the
boride phase was controlled. Large borides with an uneven distribution throughout the
microstructure resulted in preferential sites for crack nucleation and growth at low creep
displacements. Fine borides that were homogenously dispersed throughout the
microstructure decreased the minimum creep rate by up to one order of magnitude. The
borides were postulated to carry more of the load than the constituent phases in the
microstructure until a critical amount of cracking occurred, and then the load was shed to
the matrix. Post-mortem surface investigations revealed the majority of the borides in the
Ti-15Al-33Nb-5B alloy had not cracked afier creep deformation. However, the prior-

BCC grain boundaries within the microstructure had experienced extensive cracking and

326

 

 

    
  
    
  
  
   

11,. 'm .
1“ II MWW‘

I “a” .
min-(m .
mm .33. .
by,"
9.18.

.1 rut-ll”

. O
r . ' . ,
' '4! l»
‘
u ,H ‘ . . . '-' J;-
' " 1"» lrfill!m.. . ‘
A L. ‘
’ .
3'1.) .. l a‘ - l ‘ ‘
' ' . w - willie on. . '
V 1’ r l .
1= 1'. .1 t- 1; fig“ ,I:‘
‘l: ' l

._.l : .4 » , I 7' .. r, ......r......,, ,.." -u" (I. .
1“".

 

sliding/upheaval. Therefore, within the boron-modified Ti-15Al-33Nb-5B alloy, the
prior-BCC grain boundaries were also the locus of creep damage.

The volume fraction, structure, and chemistry of the BCC phase Control the RT
elongation-to-failure of Ti-Al-Nb alloys. In O+BCC microstructures, the volume fiaction
of BCC phase is important as 8f values greater than 2% were maintained as long as the
BCC volume fraction exceeded 30 percent. Increasing the Al content of the BCC phase
above 17 at% favored ordering, and this resulted in decreased 8f values in both O+BCC-
and BCC-dominated microstructures. For the A] content range of 5-12 at%, no effect of
BCC phase Al content on RT tensile properties was observed in fully-BCC
microstructures.

The addition of 0.5 at% boron to the Ti-15Al-33Nb alloy caused almost no
difference in the RT tensile behavior in both the AP and HT :1005 condition. The
addition of 5 at% boron to the Ti-15Al-33Nb alloy produced significant strengthening at
the expense of 8f. The detriment in strength observed in the Ti-22Al-26Nb alloy was
believed to be caused by the addition of the large boride needles which initiated fracture
and exhibited a brittle response.

The Ti-15Al-33Nb and Ti-21Al-29Nb alloys exhibited greater fatigue strength
than the Ti-6Al-7Nb (wt%) alloy and the other [3 Ti alloys, while the fatigue lives were
comparable to those for Ti-6Al-4V (wt%) at a given maximum applied stress. These

alloys therefore show promise for further consideration as biomaterials.

327

B. Conclusions

1. Processing

1. Traditional hot forging and hot rolling performed at subtransus temperatures
successfully produced homogeneous Ti-15Al-33Nb and Ti-21Al-29Nb sheet material.
The fine grain size of the as-processed material would not have been attainable through
supertransus thermomechanical processing. Lower nominal Al contents decrease the
flow stress and hence increase the ease with which Ti-Al-Nb alloys can be processed at
subtransus temperatures.

2. The powder metallurgy processing route produced a Ti-22Al-26Nb alloy that exhibited
acceptable strength and creep resistance. Powder metallurgical processing did not
produce a homogenous microstructure for the Ti-22Al-26Nb-5B alloy examined. Hot
rolling of the Ti-22Al-26Nb-xB HIPed compacts was suggested to further homogenize
the microstructure of the Ti-22Al-26Nb—5B alloy and improve the mechanical properties

of both alloys.

2. Microstructure

1. A wide variety of microstructures with different grain sizes, phase volume fractions,
phase chenristries, and morphologies can be can produced through post-processing heat
treatment of subtransus processed Ti-Al-Nb alloys. The ability to evaluate the effect of a
range of grain sizes on mechanical behavior would not have been possible if all the alloys
had been supertransus processed.

2. The nominal Al content required to order the BCC phase in Ti-Al-Nb alloys lies in the

nominal alloy composition range of 17-20 at% A1.

328

3. The Ti-lSAl-33Nb alloy is the lowest Al containing Ti-Al-Nb alloy, based on a
nominal Ti content of 50 at%, in which the (12 phase has been observed. The highest Nb
content within the a; phase present in Ti-Al-Nb microstructure to date was measured to
be 19.7 at% Nb.

4. Widmanstatten precipitation of the O-phase from the parent BCC phase was the only
phase transformation mechanism observed in this study.

5. The EMPA data from the boride phase in both the Ti-lSAl-33Nb-5B and Ti-22Al-
26Nb-5B alloys suggests a phase composition close to BzTiNb. SADPs suggested that
the boride phase within the Ti-Al-Nb microstructures posses the B27 orthorhombic

crystal structure.

3. Creep Behavior

1. Supertransus heat treated O+BCC microstructures exhibit greater creep resistance that
subtransus heat treated O+BCC microstructures due to large average equiaxed grain sizes
and O-phase volume fractions greater than 30 volume percent.

2. The creep behavior of the single-phase BCC Ti-5Al-45Nb alloy demonstrated that a
minimum volume fi‘action of 0.30 O-phase is required in order for Ti-Al-Nb alloys to
posses adequate creep resistance.

3. Increasing the nominal alloy Nb content up to 33 at% at the expense of the nominal
alloy Al content is not deleterious to the creep resistance of O+BCC Ti-Al-Nb

microstructures.

329

4. The in-situ creep experiments demonstrated that the prior-BCC grain boundaries were
the locus of damage accumulation in the form of cracking and sliding/upheaval during
creep deformation.

5. An empirical model was developed for O+BCC microstructures based on the
constituent phase properties of fully-equaixed BCC phase and fully-lath O-phase Ti-Al-
Nb alloys. When the minimum creep rate predicted by the model was normalized by the
average equiaxed grain size in pm, the maximum difference between measured and
experimental minimum creep rates is within one order of magnitude. This implies that
microstructure is more significant than classical mechanisms.

6. When present on a fine scale and homogeneously distributed, boride phase needles can
decrease the minimum creep rate of Ti-Al-Nb alloys up to one order of magnitude by
carrying more of the load than the constituent phases. The 5 at% boron-modified Ti-
15Al-33Nb microstructure outperformed the monolithic Ti-15Al-33Nb microstructure,
despite the fact that the monolithic microstructure possessed an average equaixed grain

size twice that of the boron-modified microstructure.

4. Tensile Behavior

1. Room-temperature elongation-to-failure values > 2% are only achievable by Ti-Al-Nb
microstructures which contain 2 30 volume percent BCC phase.
2. Increasing the Al content of the BCC phase favors ordering, and B2 dominated

microstructures display elongation-to-failure values < 2% at room-temperature.

330

3. No effect of Al content on room-temperature tensile behavior was observed when the
stress versus strain behavior of fully-BCC Ti-5Al-45Nb alloy was compared that of a
fully-BCC Ti-12Al-3 8Nb alloy.

4. The addition of 5 at% boron to the Ti-15A1-33Nb provided significant strengthening at
the expense of Sr at both room-temperature and 650°C.

5. When the grain size is kept below 175 pm, the O-phase volume fiactions is S 70
volume percent, and the Al alloy contents is between 12-15 at%, room-temperature
elongation-to-failure values greater than 2% and elevated-temperature creep resistance

are both attainable.

5. Fatigue Behavior

1. The as-processed Ti-15Al-33Nb and Ti-21A1-29Nb alloys possessed fatigue lives
comparable to or greater than conventional Ti alloys that are currently utilized in
biomedical applications. Since strength and fatigue life are considered to be the main
mechanical criteria for structural biomaterials, the alloys in this thesis merit further

evaluation for biomedical applications.

331

C. Recommendations for Future Work

1. The lattice-self diffusion and grain boundary diffusion coefficients and activation
energies need to be determined for the O and BCC phases as a function of temperature
and phase composition.

2. In order to confirm that the boride crystal structure is orthorhombic B27, CBED
analysis must be performed.

3. Analysis of the dislocation structures of post creep deformed samples is required in
order to provide evidence for the deformation mechanisms suggested in this work.
Dislocation density measurements made from samples tested in different stress regimes
along with observations of interactions (i.e. pile ups, pinning, and arrangement of
dislocation networks) between constituent phase dislocations would provide additional
evidence to support the suggested mechanisms. The intragranular deformation that
occurred within all the microstructures that could not be resolved by SEM, needs to be
characterized by TEM. In particular, the dislocation structure in the Ti-lSAl-33Nb—5B
alloy needs to be analyzed in order to determine a microstructural basis for the superior
creep resistance displayed. At this point in time it is known that the boride needles have
improved the creep resistance, but an understanding of the dislocation interactions
involved is required.

4. A deeper assessment of thermally activated creep crack mechanisms is needed.

332

BIBLIOGRAPHY

Akahori T., Niinomi M., F ukunaga K., and Inagaki I., Metall. Mater. Trans. A, 31 1937-
1948 (2000a).

Akahori T., Niinomi M., Fukunaga K., and Inagaki 1., Metall. Mater. Trans. A, 31 1949-
58 (2000b).

Akkurt A.S., Liu 6., and Bond G.M., Mat. Res. Soc. Symp. Proc., 213 455-460 (1991).

Ardell AJ. and Lee S.S., Acta Metal]., 34 2411-2423 (1986).

ASTM Designation E112-96e3, Standard Test Methods for Determining Average Grain
Size, American Society for Testing and Materials, West Conshohocken, PA.

ASTM Designation E 73 9-80, Standard Practice for Statistical Analysis of Linear or
Linearized Stress-Life and Strain-Life Fatigue Data, American Society for Testing and
Materials, West Conshohocken, PA (1988).

ASTM Designation draft #3, Standard Specification for Wrought Titanium—35niobium—
7zirconium—5 tantalum Alloy for Surgical Implant Applications (UNS R5835 0), American
Society for Testing and Materials, Philadelphia, PA (2000).

ASTM Designation draft #6, Standard Specification for Wrought T itanium—3aluminum—
2. 5 vanadium Alloy Seamless Tubing for Surgical Implant Applications (UNS R5 6320),
American Society for Testing and Materials, Philadelphia, PA (2000).

ASTM Designation F2066-01, Standard Specification for Wrought T itanium—I 5
Molybdenum Alloy for Surgical Implant Applications, American Society for Testing and
Materials, Philadelphia, PA, (2001).

Austin C.M., Dobbs J.R., Fraser H.L., Konitzer D.G., Miller D.J., Parks M.J., Schaeffer
J .C., and Sears J .W., Technical Report WL-TR-93-4059, (1993).

333

Banerjee D., lntermetallic Compounds Principles and Practice, Edited by Westbrook H.J.
and Fleischer R.L., John Wiley and Sons, New York, NY, (1976).

Banerjee D., Gogia A.K., Nandy T.K., and Joshi V.A., Acta Metall, 36 871-882 (1988).

Banerjee 1)., Phil. Mag. A, 72 1559-1587 (1995).

Banerjee D., Prog. Mater. Sci, 42 135-158 (1997).

Banerjee D., Presentation given at The Minerals, Metals, and Materials Society Annual
Meeting, March 12-16, San Antonio, TX, (2006).

Bartholomeusz M.F. and Wert J .A., Metal]. Mater. Trans. A, 25 2161-2171, (1994).

Bartholomeusz M.F. and Wert J .A., Metal]. Mater. Trans. A, 26 3257-3264, (1995).

Bendersky L.A., Boettinger W.J., and Roytburd A., Acta Metall. Mater., 39 1959-1969
(1991).

Bendersky LA. and Boettinger W.J., J. Res. Natl. Inst. Stand T echnol., 98 585-606
(1993).

Bernstein I.M., Trans. Metal]. Soc. AIME, 239 1518 (1967).

Blackburn M.J. and Smith M.P., Technical Report AF ML-TR-78-l8, (1978).

Blackburn M.J. and Smith M.P., Technical Report WRDC-TR-89-4095 (1989).

Boehlert C.J., Majumdar B.S., Seetharaman V., Miracle DB, and Wheeler R. in:
Structural Interrnetallics, Edited by Nathal M.V., Darolia R., Liu C.T., Martin P.L.,
Miracle D.B., Wagner R., and Yamaguchi M., The Minerals, Metals, and Materials
Society, Warrendale, PA, 795-804 (1997a).

334

 

Boehlert C.J., Majumdar BS, and Eylon D., Key Eng. Mater., 127-131 843-50 (1997b).

Boehlert C.J., Majumdar B.S., Krishnamurthy S., and Miracle D.B., Metal]. Trans. A, 28
309-323 (1997c).

Boehlert C.J., Mater. Sci. Eng. A, 267 82-98 (1999).

Boehlert Cl. and Miracle D.B., Metal]. Mater. Trans. A., 30 2349-2367 (1999).

Boehlert C.J., Majumdar B.S., Seetharaman V., and Miracle D.B., Metal]. Trans. A, 30
2305-23 (1999).

Boehlert C.J., Mater. Sci. Eng. A, 279 118-129 (2000).

Boehlert C.J., Metal]. Mater. Trans. A, 32 1977-88 (2001).

Boehlert CJ. and Bingert J .F., J. Mater. Process. Tech, 117 401-409 (2001).

Boehlert C.J., Cowen C.J., Tamirisakandala S., McEldowney D.J., and Miracle D.B., Scr.
Mater., 55 465-468 (2006).

Boyer R.R., in: ASM Handbook, Metallography and Microstructures, Edited by Mills
K., Davis J .R., Destfani J .D., Dieterich D.A., Crankovic G.M., Frissell H.J., Jenkins
D.M., Cubberly W.H., and Stedfeld R.L, American Society of Metals International,
Materials Park, OH, 9 460, (1985).

Breuer J ., Wilger T., Friesel M., and Herzing C., lntermetallics 7 381-388 (1999).

Bullock E., McLean M., Miles D.E., Acta Metal], 25 333-344 (1977).
Burgers W.G., Physica, 1 561-586 (1934).

Cai Z., Shafer T., Watanabe I., Nunn M.E., and Okabe T., Biomaterials, 24 213-218
(2003)

335

 

 

Chan K.S., Mater. Sci. Eng. A., 337 59-66 (2002).
Coble R.L., J. App. Phys, 34 1679-1682 (1963).

Conrad H., Doner M., and DeMeester B., in: Titanium Science and Technology, Edited
by Jaffee RI. and Burte H.M., Plenum, New York, 969, (1973).

Cowen CJ. and Boehlert C.J., Phil. Mag. A, 86 99-124 (2006a).
Cowen OJ. and Boehlert C.J., lntermetallics, 14, 412-422 (2006b).
Crossman F.W. and Ashby M.F., Acta Metal], 23 425-440 (1975).
Dary RC. and Pollock T.M., Mater. Sci. Eng. A., 208 188 (1996).

Davidson J .A. and RS. Georgette F.S. in: Proceedings of the Implant Manufacturing and
Material Technology, Society of Manufacturing Engineers, Dearborn, MI, EM87-122-1—
EM87-122-26 (1987).

Dieter G.E., Mechanical Metallurgy, 3ml Edition, McGraw-Hill, Boston, MA, (1986).

Emura S., Liu J ., and Hagiwara M. in: Titanium 1995, Edited by Blenkinsop P.A., Evans
W.J., and Flowers H.M.. The University Press, Cambridge, UK, 1 404-410 (1996).

Emura S., Yang 8.1., and Hagiwara M., Metal]. Mater. Trans. A, 35 2971-2979 (2004).

Evans R.W. and Wilshire B., Creep of Metals and Alloys, The Institute of Metals, New
York, NY, (1985).

Eylon D., Hall J.A., Pierce C.M., and Ruckle D.L., Metall. Trans. A, 7 1817-1826 (1976).

Garofalo F., Fundamentals of Creep and Creep-Rupture in Metals, The Macmillan
Company, New York, NY, (1965).

336

 

Gogia A.K., Nandy T.K., Muraleedharan K., and Banerjee D., Mater. Sci. Eng. A, 159
73-86 (1992).

Gogia A.K., Nandy T.K., Banerjee D., Carisey T., Strudel J .L., and Franchet J .M.,
Intermetallics, 6 741-48 (1998).

Gorsse S. and Miracle D.B., Acta Mater., 51 2427-2442 (2006).

Hagiwara M., Mater. Japan, 37 35-38 (1998).

Hagiwara M., Emura S., and Araoka A., Mater. Sci. Forum, 426-32 1715-1720 (2003a).

Hagiwara M., Emura S., Araoka A., Kong B.-O., and Tang F., Met. Mater. Int., 9 265-72
(2003b).

Hagiwara M., Araoka A., Yang S.J., Emura S. and Nam S.W., Metal]. Mater. Trans. A,
35 2161-2170 (2004).

Hagiwara M. and Emura S., Presentation given at the Workshop on Titanium Alloys
Modified with Boron, Dayton, OH, 11-13 October, (2005).

Harper J. and Dom J .E., Acta Metallurgica, 5 654-665 (1957).

Hartman G.A. and Russ S.M., Metal Matrix Composites: Testing, Analysis and Failure
Modes, Edited by Johnson W.S., American Society for Testing and Materials,
Philadelphia, PA 43 (1989).

Hayes R.W., Scr. Metal], 34 1005-112 (1996).

Herring C., J. App. Phys, 21 437-45 (1950).

Henshall G.A. and Strum M.J., Acta Mater., 44 3249-3257 (1996).

337

 

Henshall G.A., Subramanian P.R., Strum M.J., Mendiratta M.G., Acta Mater., 45 3135-
3142 (1997).

Hertzberg R.W., Deformation and Fracture Mechanics of Engineering Materials, 4th
Edition, John Wiley and Sons, New York, NY, (1996).

Higgins G.T., Wiryolukito S., and Nash P., Grain Growth in Polycrystalline Materials,
Edited by Abbruzzese G. and Brozzo P. Trans Tech Publications, Aederrnannsdorf,
Switzerland, (1992).

Hilliard J .E., Met. Prog., 78 99-100 (1964).
Hu H. and Rath B.B., Metall. Trans, 1 3181-84 (1970).

Hull D. and Bacon D.J., Introduction to Dislocations, 4th Ed., Elsevier Butterworth-
Heinemann, Burlington, MA, (2004).

Iijima D., Yoneyama T., Doi H., Hamanaka H., and Kurosaki N., Biomaterials, 24 1519-
24 (2003).

Inui H., Toda Y., Shirai Y., and Yamaguchi M., Philos. Mag. A, 69 1161 (1994).
Langdon, T.G., Phil. Mag, 22 689-700 (1970).

Langdon T.G. and Yavari P., Acta Metal], 30 881-887 (1982).

Langdon T.G., Mater. Sci. Eng. A, 283 266-73 (2000).

Li D. and Boehlert C.J., Metal]. Mater. Trans. A., 36 2569-2584 (2005).

Lopez M.F., Jimenez J .A., and Gutierrez A., Electrochimica Acta, 48 1395-1401 (2003).

338

 

Lu W., Zhang D., Zhang X., Wu K, Sakata T., and Mori H., J. Alloys Compounds, 327
240-247 (2001).

Kelly A. and Street K.N., Proc. R. Soc. Land. A., 328 283-293 (1972).

Kestner-Weykamp H.T., Ward C.W., Broderick T.F., and Kaufman M.J., Scr. Metal], 23
1697-1702 (1989).

Khan M.A., R.L. Williams R.L., and DP. Williams D.F., Biomaterials, 20 631-637
(1999)

Kittel C., Introduction to Solid State Physics, 7th Ed., John Wiley and Sons, New York,
NY, (1 996).

Kooi B.J., Pei Y.T., and De Hosson J. Th. M. , Acta Mater., 51 831-845 (2003).

Majumdar B.S., Boehlert C.J., Rai A.K., and Miracle DB. in: High Temperature
Ordered lntermetallic Alloys- VI, Edited by Horton J ., Baker I., Hanada S., Noebe RD,
and Schwartz D.S., Materials Research Society, Pittsburgh, PA, 364 1259-1265 ( 1995).

Malakondaiah G. and Rao P.R., Acta Metal], 29 1263 (1981).

Marquardt B.J., Scarr G.K., Chesnutt J .C., Rhodes CG, and Fraser H.L., Technical
Report WRDC-TR-89-4133, (1989).

McKay G.C., Macnair R., MacDonald C., and Grant M.H., Biomaterials, 17 1339-44
(1996)

Mendiratta MG. and Lipsitt H.A., J. Mater. Sci, 15 2985 (1980).

Metikos-Hukovic M., Tkalcec E., Kwokal A., and Piljac J ., Surface & Coatings
Technology, 165 40-50 (2003).

Mileiko S.T., J. Mater. Sci, 5 254-261 (1970).

339

Miller W.H., Chen R.T., and Starke E.A., Metall. Trans. A, 18 1451-1468 (1987).

Minonishi Y., Philos. Mag. A, 63 1085 ( 1991).

Mishra RS. and Banerjee D., Mater. Sci. Eng, A, 130 151-164 (1990).

Mishra R.S., Banerjee D., and Mukherjee A.K., Mater. Sci. Eng, A, 192-193 756-762
(1995)

Mozer B., Bendersky L.A., Boettinger W.J., and Rowe R.G., Scr., Metal]. Mater., 24
2363-2368 (1990).

Mukherjee A.K., Bird J.E., and Dorn J.E., Trans. ASM, 62 155 (1969).

Muraleedharan K., Nagender S.V., and Banerjee D., Scr. Metal]. Mater., 24 27-32
(1990)

Muraleedharan K., Gogia A.K., Nandy T.K., Banerjee D., and Lele S., Metall. Trans. A,
23 401-415 (1992a).

Muraleedharan K., Nandy T.K., Banerjee D., and Lele S., Metall. Trans. A., 23 417-431
(1992b)

Muraleedharan K., Banerjee D., Banerjee S., and Lele S., Philos. Mag. A, 71 1011-1036
(1995a).

Muraleedharan K., Nandy T.K., Banerjee D., Intermetallics, 3 187-199 (1995b)

Nabarro F.R.N., Report in Conference on Strength of Solids, The Physical Society,
London, (1948).

Nandy T.K., Mishra RS, and Banerjee D., Scr. Metal]. Mater., 28 569-574 (1993).

340

Nandy T.K., Mishra R.S., Gogia A.K., and Banerjee D., Scr. Metal]. Mater., 32 851-856
(1995)

Nandy T.K. and Banerjee D., Intermetallics, 8 915-928 (2000).

Narayana Murty S.V.S. and Nageswara Rao B., J. Mater. Process. T ech., 104 103-109
(2000)

Niinomi M., Metal]. Mater. Trans. A, 33 477-486 (2002).

Niinomi M., Biomaterials, 24 2673-2683 (2003).

Owen D.M. and Langdon T.G., Mater. Sci. Eng. A, 216 20-29 (1996).

Okazaki M. and Hizume T., J. Soc. Mater. Sci. Japan, 43 1238-1244 (1994).

Papakyriacou M., Mayer H., Pypen C., Plenk H. Jr., and Stanzl-Tschegg S., International
Journal of Fatigue, 22 873-888 (2000).

Peng J ., Mao Y., Li S., and Sun X., Mater. Sci. Eng. A, 299 75-80 (2001).

Peters M., Hemptenmacher J ., Kumpfert J ., and Leyens C., Titanium and Titanium
Alloys, Edited by Leyens C. and Peters M., Wiley-VCH Verlag, Weinheirn, Germany,
2003.

Popille F. and Douin J., Phil. Mag. A., 73 1401-1418 (1996).

Szaruga A., Saqib M., Omlor R, and Lipsitt H.A., Scr. Metal], 26 757-790 (1992).

Rachinger W.A., J. Inst. Metals, 81 33-41 (1952-1953).

Reed-Hill RE. and Abbaschian R., Physical Metallurgy Principles, 3rd ed. PWS
Publishing Company, Boston, MA, (1994).

341

 

Rhines F.N., Phase Diagrams in Metallurgy, McGraw-Hill Book Company, Inc., New
York, NY, (1956).

Rhodes C.G., Graves J .A., Smith RR, and James M.R. in: Structural Intermetallics,
Edited by R. Darolia, J .J . Lewandowski, C.T. Liu, P.L. Martin, D.B. Miracle, and M.V.
Nathal, The Minerals, Metals, and Materials Society, Warrendale, PA, 45-52 (1993).

Rhodes C.G., Srrrith P.R., Hanusiak W.H., and Shephard M.J., Metal]. Mater. Trans. A,
31 2931-2941 (2000).

Rowe R.G. and Gigliotti M.F.X., Scr. Metal], 24 1206-1214 (1990).
Rowe R.G., US. Patent Number 5,032,357 (1991a).

Rowe R.G. and Hall E.L. in: High Temperature Ordered lntermetallic Alloys-IV,
Symposia Proceeding of the Materials Research Society, Edited by Johnson L.A., Pope
DP, and Stiegler J .O., Materials Research Society, Pittsburgh, PA, 231, 449-454
(1991b)

Rowe R.G., Konitzer D.G., Woodfield AP, and Chesnutt J .C., High Temperature
Ordered lntermetallic Alloys-IV, Vol. 231, Edited by Johnson L.A., Pope DP, and
Stiegler J .O., Materials Research Society, Pittsburgh, PA, 703-708, (1991).

Rowe R.G. in: Titanium 1992 Science and Technology, Edited by Froes F .H. and Caplan
I., The Minerals, Metals, and Materials Society, Warrendale, PA, I 343-350 (1993).

Rowe R.G, Banerjee D., Muraleedharan K., Larsen M., Hall E.L., Konitzer D.G., and
Woodfield AP. in: Titanium 1992 Science and Technology, Edited by Froes F .H. and
Caplan I., The Minerals, Metals, and Materials Society, Warrendale, PA, I 1259-1266
(1993)

Rowe R.G. and Larsen M. in: Titanium 1995, Edited by Blenkins0p P.A., Evans W.J.,
and Flowers H.M.. The University Press, Cambridge, UK, 1 364-371 (1996).

Ruano O.A., Wadsworth J., and Sherby 0.1)., Acta Metal], 36 1117-28 (1988).

342

Rusing J. and Herzig C., Intermetallics, 4 647 (1996).

Sagar P.K., Banerjee D., Muraleedharan K., and Prasad Y.V.R.K., Metal]. Mater. Trans.
A, 27 2593-2604 (1996).

Seetharaman V. and Semiatin S.L., Metall. Trans. A, 28 947-54 (1997).

Semlitsch M.F. , Weber H., Streicher R.M., and Schon R., Biomaterials, 13 781-88
(1992)

Shewmon P.G., Difiitsion in Solids, McGraw-Hill, New York, NY (1963).

Smith P.R., Khobaib M., and Graves J .A., Scr. Metal], 29 1313-1318 (1993).

Smith P.R., Porter W.J., Kralik W.J., and Graves J.A. in: Metal Matrix Composites,
Edited by Poursartip A. and Street K.N., Woodhead Publishing Ltd., Cambridge, UK, 2
731-738 (1995).

Smith P.R., Rosenberger A., and Shepard M.J., Scr. Mater., 41 221-228 (1999).

Smith P.R., Rosenberger A., Shepard M.J., and Wheeler R., J. Mater. Sci. 35 3169-3179
(2000)

Takeuchi S. and Argon A.S., J. Mater. Sci, 11 1542 (1976).

Tamirisakandala S., Bhat R.B., Tiley J .S., and Miracle D.B., Scr. Mater., 53 1421-6
(2005)

Takag S. and Ouchi C., Mater. Trans. JIM, 38 285-291 (1997).

Tang F., Nakazawa S., and Hagiwara M., Mater. Sci. Eng. A, 315 147-52 (2001).

343

 

Trinkle D.R., submitted to Scr. Meta]. Mater., (2006).

Wagner L., Fatigue life behavior, in: Fatigue and Fracture Properties of Titanium
Alloys, ASM Handbook, Fatigue and Fracture, Edited by Lampman S.R, American
Society of Metals International, Materials Park, OH, 19 829—845 (1996).

Watanabe 1., Tanaka Y., Watanabe E., and l-Iisatsune K., The Journal of Prosthetic
Dentistry, 92 278-82 (2004).

Weertrnan J ., Trans. Am. Soc. Met, 61 681-694 (1968).

Weertrnan J. and Blacic J ., Geophys. Res. Lett., 11 117-120 (1984).

Worth B.D., Jones W., and Allison J .E.: Metall. Trans. A, 26 2947-2959 (1995).

Woodard S.R., Pollock T.M., and Dary F.C., Final Report for Grant
No.F3360194MT488, Carnegie Mellon University, Pittsburgh, PA, (1996).

Woodard SR. and Pollock T.M., Technical Report WL-TR-97-4082, (1997).

Yang S.J., Emura S., Hagiwara M., and Nam S.W., Scr. Mater., 49 897-902 (2003).

Yolton CF. and Beckman J.P., Mater. Sci. Eng. A, 192/193 597-603 (1995).

Yolton CF. and Moll J .H. in: Titanium 1995, Edited by Blenkinsop P.A., Evans W.J.,
and Flowers H.M.. The University Press, Cambridge, UK, 1 2755 (1996).

Zhang J .W., Li S.Q., Zou D.X., Ma W.Q., and Zhong Z.Y., Intermetallics, 8 699-702
(2000).

344

 

   

1111111111111111111111111