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

 

I'll , Gd

3 U .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘M mu

 

THE EFFECT OF HEAT TREATMENTS ON MICROSTRUCTURES
AND PRIMARY CREEP DEFORMATION OF INVESTMENT
CAST TITANIUM ALUMINIDE ALLOYS AND
POLYSYNTHETICALLY TWINN ED (PST) CRYSTALS

By

Dong Yi Seo

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Materials Science and Mechanics

1998

 

 

ABSTRACT

THE EFFECT OF HEAT TREATMENTS ON MICROSTRUCTURES
AND PRIMARY CREEP DEFORMATION OF INVESTMENT
CAST TITANIUM ALUMINIDE ALLOYS AND
POLYSYNTHETICALLY TWINNED (PST) CRYSTALS

By

Dong Yi Seo

Several heat treatments were developed and applied to several investment cast duplex
(equiaxed+lamellar) TiAl alloys (Ti-45~47A1-2Nb—2Mn+0.8v%TiB2XDTM, Ti-47Al-
2Nb-2Mn+0.8TiBz XDTM with interstitial elements, Ti-47A1-2Nb-2Cr, and Ti-47Al-2Nb-
1Mn-0.5W-0.5Mo-0.28i) in an effort to enhance creep properties. Creep behavior in
polysynthetically twinned (PST) crystals with different orientations was also investigated.

Primary creep resistance of W-Mo-Si alloys can be improved by about 10 times
with heat treatment, and the XD alloys with additions of interstitials can be improved by
about 7 times, and 47-2-2 can be improved by about 3 times, and the XD alloys can be
improved slightly, or not at all when the Al level is lower. The variation in creep
resistance with heat treatment can be explained by differences observed in the
microstructures and textures produced by the various heat treatments. The XD alloys with
high 0 show large lamellar volume fraction (40%), since oxygen is an on stabilizer. The
addition of carbon in the XD alloys results in grain size refinement and improvement of
primary creep resistance due to the formation of fine precipitates. These XD alloys show
no additional benefit from carbon concentrations higher than 0.065 wt% C. Precipitates in
the W-Mo-Si alloys nucleated and grew at a faster rate in the deformed part of the

specimen, as compared to the grip section, and the precipitates were generally smaller

and more homogeneously dispersed. These observations indicate that strain assisted
nucleation and/or growth of precipitates accounts for much of the excellent creep
resistance of the W-Mo-Si alloy.

Quantitative microscopic comparisons were made between microstructures in
undeformed and deformed regions using SEM and TEM techniques in the W-Mo-Si
alloys. The lamellar spacing in lamellar grains systematically decreased by 15-35% with
increasing stress, during the first 0.2-0.5% strain at the early stage of primary creep. More
refinement of lamellar spacing occurs at lower temperature and higher stress. The
refinement process is a consistent microstructural feature during primary creep
deformation in all investment cast TiAl alloys. The refinement occurred by mechanical
twinning (easy mode deformation) or/and on shear transformation parallel to lamellar
boundary. Primary creep in TiAl alloys is divided into two stages, an initial rapid process
that refines lamellar spacing through strain levels between 0.2 and 0.5%, and a second
primary creep process, where no lamellar refinement occurs.

The stress exponent and activation energy for creep in the lamellar microstructure
depend on the direction of the stress tensor. In comparing activation energies of the early
process in primary creep in the W-Mo-Si alloy with creep in the soft orientation in a PST
crystal, the activation energies are small, near 150 kJ/mol. This indicated that the early
stage deformation in TiAl alloys correlates closely with PST crystal creep in easy mode
deformation. From simulations of lamellar refinement, computed local shear strains are
similar or higher than effective shear strain. There are some correlations between the
computed local shear strain and microstructural changes during primary creep

deformation.

To My Parents

Ik-Soo Seo, Dong-Pil Kim, and Sangshin Suh

iv

ACKNOWLEDGEMENTS

I would like to express my sincere appreciation to my academic advisor, Professor
Thomas R. Bieler and to Dr. Sung-Uk An at Korean Institute Science and Technology

(KIST) for their effective guidance and encouragement to finish this research.

I also wish to thank the members of my committee, Professor David Grummon,
Professor Matrin A. Crimp, Department of Materials Science and Mechanics, and
Professor F. William Cambray, Department of Geological Science. They have been
provided many suggestions and comments on this work, and the facilities necessary to

pursue this research.

Particularly, I would like to express my sincere appreciation to Howmet
Corporation, Whitehall, Michigan for the financial and material support to make this
research possible. I would like to express my thanks to Mr. D.E. Larsen. and Mr. P.A.

McQuay for their contributions in pursuing this research.

I would like to express my gratitude to my wife, Sooyun Chung, my son,
Younghun (Freddie Y.) Seo, my family, Dr. Freddie L. Poston and Charlotte Poston, for
their never-ending love, patient support and encouragement, during my entire graduate

study.

 

Table of Contents

xi

 

LIST OF TABLES

LIST OF FIGURES

 

CHAPTER 1

/

xiii

 

1.1 INTRODUCTION

1.2 RESEARCH STRATEGY

 

1.2.1 Rationale and goals

CHAPTER 2: CREEP THEORY
2.1 DEFINITION OF CREEP

2.2 STAGES OF CREEP

00

CD

10

 

2.3 STRESS AND TEMPERATURE EFFECTS

12

 

2.4 POWER LAW RELATIONSHIP

14

 

16

 

2.4.1 Power law breakdown

 

2.4.2 Transitions in n value

18
18

 

2.4.3 Transitions in Qc value

 

2.5 DIFFUSIONAL CREEP PROCESS

20

22

 

2.5.1 Nabarro-Herring Creep

23

 

2.5.2 Coble Creep

 

2.5.3 Diffusional Creep and Grain Boundary Sliding

2.6 DISLOCATION CREEP PROCESSES

26

28

 

28

 

2.6.1 Dislocation Glide Rate-controlled Creep

30

 

2.6.2 Dislocation Climb Rate-controlled Creep

2.7 DIFORMATION-INDUCED TWINNING

33

 

2.8 CREEP IN SOLID SOLUTION ALLOYS

35

 

vi

2.9 STRESS AND TEMPERATURE CHANGE CREEP

 

(TRANSIENT CREEP)

2.10 PRIMARY CREEP IN METALS

 

2.11 DEFORMATION MECHANISM MAP

 

CHAPTER 3: OVERVIEW OF TITANIUM ALUMINIDES

3.1 INTRODUCTION

 

3.2 TITANIUM ALUMINIDES

50

 

50

 

3.2.1 Crystal Structure and Microstructure

 

3.2.2 Dislocation Structure and Deformation Mode

56
66

 

3.2.3 Creep Behavior in Gamma Titanium Aluminides

67

 

3.2.3.1 Al-rich single phase gamma Titanium aluminides

 

3.2.3.2 Ti rich fully lamellar gamma Titanium aluminides

67
7O

 

3.2.3.3 Ti rich duplex gamma Titanium aluminides

77

 

3.2.4 Precipitates and Alloying Additions

77

 

3.2.5 Primary creep behavior in TiAl

CHAPTER 4: EXPERIMENTAL PROCEDURE
. 4.1 COMPRESSION AND TENSION CREEP TESTS

 

 

4.1.1 Compression creep test (ATS creep tester)
4.1.2 Tension creep test

4.2 THE CREEP CURVE FITTING

 

 

4.2.1 The 9 projection concept

 

4.2.2 Modification of 9-projection concept

 

4.2.3 Modification for this analysis

4.3 MEASUREMENT OF LAMELLAR SPACING

 

4.4 A STASTISTICAL ANALYSIS OF MEASURED CHANGES
IN LAMELLAR SPACING

 

vii

CHAPTER 5: OVERVIEW OF RESEARCH

104

 

CHAPTER 6

 

THE CORRELATION OF PRIMARY CREEP RESISTANCE
TO THE HEAT TREATMENT AND MICROSTRUCTURES IN
INVESTMENT CAST Ti-Al GAMMA ALLOYS

6.1 INTRODUCTION

 

6.2 MATERIALS AND EXPERIMENTAL PROCEDURE

116

116

117

 

 

6.3 RESULTS

120

120

 

6.3.1 Primary creep to 0.5% strain

 

6.3.2 Microstructures

6.4 DISCUSSION

 

120

138

 

6.5 CONCLUSIONS

CHAPTER 7

 

EFFECT OF INTERSTITIAL CONCENTRATION AND HEAT TREATMENT
ON MICROSTRUCTURE AND PRIMARY CREEPOF INVESTMENT
CAST Ti-47Al-2Nb-2Mn WITH 0.8 vol.% TIB2

7.1 INTRODUCTION

 

141

143

143

 

7.2 EXPERIMENTAL

7.3 RESULTS AND DISCUSSION

 

 

6.3.1 Microstructures

144

145

I46
148

 

6.3.2 Lamellar spacing

 

7.4 CONCLUSIONS

CHAPTER 8

 

CHANGES IN MICROSTRUCTURE DURING PRIMARY CREEP
IN A TI-47Al-2Nb-1Mn-O.5W-0.5Mo-0.2$i ALLOY

8.1 INTRODUCTION

 

8.2 EXPERIMENTAL

155

156

157

158

 

viii

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8.3 RESULTS 163
8.3.1 Primary creep behavior 163
8.3.2 Microstructures 169
8.4 DISCUSSION 190
8.5 CONCLUSIONS 194
CHAPTER 9 196
SHEAR+COMPRESSION CREEP DEF ORMATION IN
POLYSYNTHETICALLY TWINNED (PST) TI-47Al-2Nb—2Cr
WITH <110> AND <112> ORIENTATIONS
9.1 INTRODUCTION 197
9.2 EXPERIMENTAL DETAILS 198
9.2.1 Specimen preparation and Shear+Compression creep test 198
9.2.2 Data acquisition and analysis 201
9.3 RESULTS AND DISCUSSION 202
9.3.1 Creep behavior 202
9.3.2 Activation energy 205
9.3.3 Stress exponent 207
9.3.4 Microstructure and lamellar spacing 208
9.4 CONCLUSION 212
CHAPTER 10 214
THE EFFECT OF HEAT TREATMENTS ON MICROSTRUCTURES
AND PRIMARY CREEP DEFORMATION OF FOUR INVESTMENT
CAST TITANIUM ALUMINIDE ALLOYS
10.1 INTRODUCTION 215
10.2 MATERIALS AND EXPERIMENTAL PROCEDURE 217
10.3 RESULTS AND ANALYSIS 219
10.3.1 Primary creep behavior 219
10.3.2 Microstructures 223
10.3.3 Lamellar spacing 232

ix

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10.3.4 Effect of Anisotropic Crystal Structure on Deformation 239
10.4 DISCUSSION 240
10.4.1 Lamellar refinement 240
10.4.2 Effects of Heat Treatment on Grain size and Lamellar
Volume Fraction 242
10.4.3 Effects of Heat Treatment on Precipitation, and Dynamic
precipitation 244
10.5 CONCLUSION 244
CHAPTER 1 l 246
THE EFFECT OF CARBON CONCENTRATION ON PRIMARY CREEP
OF AN INVESTMENT CAST Ti-47Al-2Mn-2Nb+0.8vol.%TiB2 ALLOY
11.1 INTRODUCTION 247
11.2 EXPERIMENTAL PROCEDURE 247
11.3 RESULTS AND DISCUSSION 249
11.3.1 Microstructures 249
11.3.2 Lamellar spacing 257
11.3.3 Primary Creep behavior 259
11.4 CONCLUSIONS 267
CHAPTER 12 269
MECHANISMS OF LAMELLAR REFINEMENT DURING PRIMARY CREEP
OF NEAR-GAMMA DUPLEX TiAl
CHAPTER 13 277
SUMMARY
APPENDIX 281
PROGRAM CODES FOR SIMULATION OF LAMELLAR REFINEMENT
REFERENCES 288

 

List of Tables

CHAPTER 2

 

Table 2.1 Characteristic features in Class I and Class II creep behavior 36

Table 2.2 Different and Q: values associated with dislocation and diffusion creep

 

 

 

 

 

 

 

occuring with pure metals 47
CHAPTER 3
Table 3.1 Comparison of Gamma TiAl with Titanium alloy and Superalloy ------------- 49
Table 3.2 Creep data on TiAl alloys 72
Table 3.2 (cont’d) 73~76
Table 3.3 Alloying effects in gamma TiAl alloys 78
CHAPTER 6
Table 6.1 Observation of the effects of heat treatment on three TiAl alloys from

optical microscopy 122
Table 6.2 Average and range of dimensions describing lamellar microstructures ------- 137
CHAPTER 7
Table 7.1 Chemical compositions of investment cast Ti-47Al-2Nb

-2Mn-0.8TiB2XDTM 145
CHAPTER 8
Table 8.1 Curve fit parameters in several experimental conditions 165
Table 8.2 Summary of Precipitate Composition and Shapes (at% of element) ---------- 182
Table 8.3 Diffusion distances of Ti in TiAl 192

 

xi

CHAPTER 9

 

 

 

 

 

Table 9.1 The conditions for each Shear+Compression creep test 200
CHAPTER 10

Table 10.1 Nominal chemical composition of investment cast TiAl alloys (at%) ------- 218
Table 10.2 Grain sizes of four investment TiAl alloys (unit: pm) 228
Table 10.3 Average lamellar spacing and reduction % of TiAl alloys 237
CHAPTER 11

Table 11.1 Al and C content in as-HIPed investment cast alloys 248
Table 11.2 The lamellar voltune fraction and grain size in TiAl alloys 252
Table 11.3 Curve Fit Parameters in Creep tests shown in Figure 11.10 261

 

xii

List of Figures

CHAPTER 2

Figure 2.1 Creep curves at high and low temperatures. At high temperatures,
creep deformation can cause considerable changes in dimensions
and failure generally occurs after some time, tp. In contrast,
dimensional changes due to creep are usually extremely small
and failure rarely occurs

 

Figure 2.2.a Schematic representation of creep stages at a high temperature
creep curve

 

Figure 2.2.b Schematic creep strain rate versus creep strain curve at a high

 

Temperature

Figure 2.3 (a) Temperature effect on creep curves (b) Log é vs. 1/T for Ni-

67% Co-Ale3 showing the variation in activation energy above
and below 0.5 Tm

11

ll

 

Figure 2.4 (3) Stress effect on creep curves (b) Log 53 vs. Log 6 for aluminium
and aluminium-magnesium alloys at 523 K showing the variation
in stress exponent, n, values

 

Figure 2.5 The temperature-compensated secondary creep data (or Z) for
Polycrystalline copper. The activation energy of 130 KJ/mole
is used for this plot

15

17

 

Figure 2.6 A plot of the grain-size compensated secondary creep rate as a
function of stress for polycrystalline copper

 

Figure 2.7 Schematic illustration of the transitions in n value occuring
when the stress dependences of steady-state creep rate for
pure metals at temperatures above 0.4 Tm are described
using a power-law relationship

19

19

 

Figure 2.8 The variation with temperature of the activation energy for
the creep of polycrystalline copper

 

Figure 2.9 The variation of the diffusion coefficient with temperature for the
single crystals and the polycrystalline silver

 

Figure 2.10 Schematic diagram of (a) Nabarro-Herring creep and
(b) Coble creep on an ideal grain

21

21

24

 

xiii

Figure 2.11 Accommodation of diffusional creep by GBS: (a) An original
configuration of equiaxed grains. (b) The result of diffusional
creep without participation of GBS. (c) Simultaneously with

diffusional creep, GBS occurs as shown by the offset of

 

a marker line

 

 

version of the model

 

 

 

and (b) the viscous glide

 

to be dominant

CHAPTER 3

 

Figure 3.1 Phase diagram of binary Ti-Al system

 

Figure 3.2 Lattice structure of (a) Llo-TiAl and (b) Dolg-TI3AI

on the basal plane (0001) in D019

 

27
Figure 2.12 Sequence of events which may lead to boundary sliding and
migration. S, slip spacing; C, subgrain diameter 27
Figure 2.13 Weertman’s models of creep: (a) Dislocations form pile-ups,
the leading dislocations of both pile-ups climb and annihilate.
(b) Arrangement and motion of dislocations assumed in a future
32
Figure 2.14 Schematic diagram of twinning and slip after applied stress 34
Figure 2.15 Relationship between strain and strain rates for pure metals,
solid solutions of Class I (dislocation-drag control), and solid
solutions of Class II (substructure control) 37
Figure 2.16 Schematic drawings of the instantaneous elongation and
contraction upon small changes in cases of (a) the free-flight glide,
40
Figure 2.17 Schematic version of a deformation map for a polycrystalline metal
illustrating the stress/temperature regimes within which
dislocation and diffusional creep processes may be expected
46
51
52
Figure 3.3 (110)y directions and the atomic arrangement on a (11 1) plane
in L10 and <1120>a2 directions and the atomic arrangement
54
Figure 3.4 The stereographic projection of (111)y and (0000012 show the
six possible orientation variants of (110)y with respect to
54

 

<1 120>a2 matrix

xiv

Figure 3.5 (a) duplex and (b) fully lamellar (c) near-gamma microstructures
in TiAl alloys 57

 

 

Figure 3.6 The thin twin morphology near the twin tip 59

Figure 3.7 The formation sequence of TiAlsp, TI3AI, Ti3Alsp and TiAlT due
to high temperature deformation 60

 

Figure 3.8 (a) The maximum Schmid factor for deformation in a given crystal
direction is plotted for ordinary and super dislocations, and for
true twinning, which depends on deformation mode. (b) There are
no operative ordinary dislocation or true twinning systems
near <001] for tension, and none near <110] for compression 64

 

Figure 3.9 Yield stress of PST crystals as a function of angle between
compression axis and the lamellar boundaries 65

 

Figure 3.10 Comparison of steady state creep data of TiAl with different

 

 

microstructures 69
Figure 3.11 Primary creep curve of TiAl alloys 80
CHAPTER 4
Figure 4.1 The diagram of ATS creep tester with data acquisition system 82

 

Figure 4.2 Creep testing involves holding test specimens at constant load
and temperature until some specific experimental parameter is met ------------ 85

Figure 4.3 The B-projection concept envisages normal creep curves in terms
of a decaying primary component (8p) and an accelerating
tertiary component (81“) 88

 

Figure 4.4 Measurement of lamellar spacing in (a) backseattered electron
image in scanning electron microscopy (SEM) mode and
(b) bright field image in transmission electron microscopy
(TEM) mode 94

 

Figure 4.5 Frequency of a; spacing with different number of bins (a) 10 bins,
(b) 15 bins, (c) 20 bins in W-Mo-Si alloys 98

 

Figure 4.6 Frequency of a; spacing with different number of bins (a) 5 bins,
(b) 10 bins for the natural logarithm scale in W-Mo-Si alloys 99

 

XV

Figure 4.7 Normal and log-normal probability plots of W-Mo-Si alloy for
undeformed and deformed cases 100

 

Figure 4.8 Frequency of 7 thickness with different number of bins (a) 10 bins,
(b) 15 bins, (c) 20 bins in Ti-47Al-2Nb—2Cr 101

 

Figure 4.9 Normal and log-normal probability plots of Ti-47Al-2Nb-2Cr alloy
for undeformed case 102

 

Figure 4.10 Frequency of 7 thickness with different number of bins (a) 5 bins,
(b) 10 bins for the natural logarithm scale in Ti-47Al-2Nb-2Cr alloy ------ 103

CHAPTER 5

Figure 5.1The or; spacing decreases with respect to increasing stress at similar
strains at 650 °C 110

 

Figure 5.2 The amount of lamellar refinement (% reduction of or; spacing)
increases with the volume fraction of lamellar grains 1 12

 

Figure 5.3 Higher lamellar fraction retards the strain rate at the transition between
the early and later creep processes 112

 

Figure 5.4 Comparison of lamellar refinement history with primary creep rate
shows that the early process operates by lamellar refinement, whereas
the later process does not alter the lamellar spacing appreciably

 

with known transition strain in the XD+C alloy 1 14
CHAPTER 6
Figure 6.1 Phase diagram illustrating heat treatment and creep condition ---------------- 119

Figure 6.2 Time to 0.5% creep strain with respect to heat treatment.
The lines go through the average of 3 values, the high
and low values are indicated by additional symbols 121

 

Figure 6.3 Optical microstructures of deformed Ti-45Al-2Nb-2Mn-O.8TiBz
a) as-HIPed, as polished, b) as-HIPed and etched,
c) 900 °C/20hrs heat treatment, and d) two-step heat treatment ---------------- 124

Figure 6.4 Backseattered electron images of Ti-45Al-2Nb-2Mn-0.8TiB2

a) as-HIPed undeformed, b) as-HIPed deformed,
c) 900 oC/20hrs heat treatment, and d) two-step heat treatment ---------------- 125

xvi

 

Figure 6.5 Bright field images of as HIPed Ti-45Al-2Nb—2Mn-0.8TiB2
afier creep test a) Fine mechanical twins observed in
(12 grains, b) Fine mechanical twins observed in lamellar regions,
0) dislocations interactions with a TiBz particle 127

 

Figure 6.6 Optical microstructures of deformed Ti-47Al-2Nb-2Mn-0.8TiB2
a) as-HIPed, as polished, showing TiBz particles, b) as-HIPed
and etched, c) 900 °C/20hrs heat treatment, and
d) two-step heat treatment 128

 

Figure 6.7 Optical microstructures of deformed Ti-47Al-2Nb-2Cr
a) as-HIPed, as polished, b) as-HIPed and etched,
c) 900 °C/20hrs heat treatment, and d) two-step heat treatment ---------------- 130

Figure 6.8 Backseattered electron images of deformed Ti-47Al-2Nb-2Cr
a) as-HIPed, b) 900 °C/20hrs heat treatment, and
c) and d) two-step heat treatment 132

 

Figure 6.8 Low magnification backseattered electron images of
Ti-47Al-2Nb-2Cr illustrate dendritic segregation of
aluminum, e) as-HIPed undeformed, f) as-HIPed deformed,
g) 900 °C/20hrs heat treatment,undeformed, and
h) 900 °C/20hrs heat treatment,deformed 133

 

Figure 6.9 Bright field images of as HIPed Ti-47Al-2Nb-2Cr after
creep test a) commonly observed relationship between
coarser and finer lamellar regions exhibit a 170° about
a <110> axis relationship, b) subgrain boundaries develop
in larger 7 regions 134

 

Figure 6.9 Bright field images of Ti-47Al-2Nb-2Cr after creep test
c) on lathes resist absorption in a boundary where an equiaxed
7 grain is consuming a lamellar grains (900°C/20hrs heat treatment),
(1) particles are located at grainboundary, and e) inside a y grain
afier two-step heat treatment 135

 

Figure 6.10 Refinement of lamellar feature was observed with increasing
heat treatment conditions, and resulting from deformation 136

 

Figure 6.11 Inverse pole figures for 3 heat treatments 3) Schimid factor,
b) as-HIPed, c) 900 °C/20hrs heat treatment,
(1) two step heat treatment after creep test 139

 

xvii

CHAPTER 7

Figure 7.1 Time to 0.5% creep strain with respect to heat treatment.
The lines go through the average of 3 values, high and low

 

values are indicated by symbols

Figure 7.2 Strain rate-strain plot of the creep deformation for the specimen

 

with the middle creep time

Figure 7.3 Optical microstructure of creep deformed (a) medium 0

147

147

149

 

and (b) high 0 alloys

Figure 7.4 Backseattered electron (BSE) images of after creep deformation
of (a) medium 0 and (b) high 0 alloys

150

 

Figure 7.5 Lamellar spacing distribution of high 0 alloys before and after
creep deformation

 

Figure 7.6 Plot of strain-rate ratio vs. % reduction in 0Q spacing and 7 thickness
in low and high 0 alloys

153

154

 

Figure 7.7 Relationship between fraction of specified lamellar spacing
and time to 0.5% creep strain

 

CHAPTER 8

Figure 8.1 A Ti-Al phase diagram near the stoichiometric TiAl composition --------

Figure 8.2 Geometry of shear+compression deformation conditions

154

----159

161

 

Figure 8.3 Primary creep strains are illustrated a) at low strains (0.2%)
in as-HIPed shear+compression specimens,

164

 

b) at larger strains (1%)

Figure 8.4 Primary creep strains to 0.5% in heat treated tensile specimens
with different temperature and stress

 

Figure 8.5 Differentiated curve fits (strain rate vs. strain) of data
in Figure 8.3 and 8.4

 

Figure 8.6 Microstructures of Ti-47Al-2Nb-1Mn-0.5W-0.5Mo-0.2Si:
(a) As-HIPed condition before deformation (shear vector
was subsequently horizontal), (b) Heat Treated tensile gage

166

166

 

section deformed at 760 °C, 138 MPa

xviii

170

 

Figure 8.7 X-ray data obtained from the heat treated tensile gage section
tested at 649 °C, 276 MP3 171

 

Figure 8.8 BSE image of Ti-47Al-2Nb-1Mn-0.5W-0.5Mo-0.2Si:
(a) As-HIPed condition before deformation (b) Specimen tested
at 760 °C, 138 MPa (1010 °C/50 hours heat treatment) 172

 

Figure 8.9 TEM image of lamellar and equiaxed grains from the specimen
tested at 760 °C, 138 MPa (1010 °C/50 hours heat treatment) ----------------- 174

Figure 8.10 Dislocation interactions with particles in lamellar regions.
The below image is an enlargement of the precipitate (a) indicated
by the arrow (760 °C and 138 MPa) 176

 

Figure 8.11 Precipitates in equiaxed grains are surrounded by dislocations
(649 °C and 276 MPa), (b) an enlargement of the arrowed
precipitate in (a) 177

 

Figure 8.12 BSE images of specimen a) undeformed, b) deformed at 760 °C,
138 MPa (1010 °C/50 hours heat treatment) 178

 

Figure 8.12 BSE images of specimen c) deformed at 760 °C,138 MPa
(1010 °C/50 hours heat treatment) d) undeformed
as-HIPed condition 179

 

Figure 8.13 Size distribution of precipitates in lamellar grains (tested at
760 °C, 138 MPa) 180

 

Figure 8.14 Lamellar refinement process is illustrated as indicated by arrows.
Diffraction patterns in boxes a, b, and c illustrate true twin relationship----183

Figure 8.15 Lamellar spacing was refined after creep deformation
(760 0C and 138 MPa): (a) undeformed grip section and
(b) deformed gage section 184

 

Figure 8.16 a), b) Distribution of 0L2 spacing and 7 thickness in 750 °C and
138 MPa, and c) (12 spacing in 649 °C and 276 MPa specimens in
undeformed grip and deformed gage section of heat treated
tensile creep specimens 186

 

Figure 8.17 Average a2 spacing and y thickness in tensile creep specimens
a) 760 °C and 138 MPa, and b) 649 °C and 276 MPa before
and after creep deformation 187

 

xix

 

Figure 8.18 a) Frequency of or; spacings based on about 500 counts in 25
lamellar regions after compression-shear deformation at 750 °C
and 77 MPa, b) or; spacing resulting from different effective
strains at 650 °C and 750 °C 188

 

CHAPTER 9

Figure 9.1 Shear+Compression fixture used for creep deformation of
5x5x3 mm PST crystal bricks. PST crystal is shown in
soft orientation 199

 

Figure 9.2 Creep deformation of <110> PST crystal in Ar (with curve fitting)
(a) strain vs. time curve (b) strain vs. strain-rate curve 203

 

Figure 9.3 Creep deformation of <112> PST crystal in air (with curve fitting)
(a) strain vs. time curve (b) strain vs. strain-rate curve 204

 

Figure 9.4 3) Temperature and b) Stress dependence of minimum creep
rate for PST crystals. Numbers next to datum points refer to
Figures 9.2 and 9.3 206

 

Figure 9.5 Backseattered electron images of deformed <112> PST crystal
(1.38% shear strain at 760°C and 13 MPa) 210

 

Figure 9.6 (a) or; spacing distribution, and (b) average a; spacing and
thickness with different primary creep strain levels in the
<112> PST crystal 211

 

CHAPTER 10

Figure 10.1 Differentiated curve fits of creep data exhibit a decreasing strain rate to a
saturation strain between 0.1 and 0.5% strain, whereafter the decrease in
strain rate is less rapid 221

 

Figure 10.2 Microstructures of As HIPed 45A] XD (a) optical micrographs before

 

 

creep test. 224
Figure 10.3 Volume fraction of lamellar grains in TiAl alloys with respect to

heat treatment 226
Figure 10.4 (a) optical micrographs of As HIPed 47A1XD before creep test ----------- 226

Figure 10.5 Microstructures of As HIPed 47-2-2 (a) optical micrographs before
creep test (b) BSE images after creep test 229

 

XX

Figure 10.6 Distribution of or; spacing and 7 thickness in TiAl alloys before

 

and after creep deformation (a) high 0 (b) 47-2-2

Figure 10.6 Distribution of a or; spacings and 7 thickness in TiAl alloys before and
after creep deformation (c) WMoSi alloys under tension mode ----------

Figure 10.7 The plot of or; spacing respect to different effective strains and different

234

---23 8

 

loads at 650 °C and 760 °C

CHAPTER 1 1

Figure 11.] Optical macrographs of Ti-47Al-2Nb-2Mn+0.8 vol%TiB2
alloys having (a) Low carbon, (b) 0.065 wt% carbon,
(c) 0.11 wt% carbon

 

Figure 11.2 Optical micrographs of Ti-47Al-2Nb-2Mn+0.8 vol%TiB2
alloys having (a,b) Low carbon, (c) 0.065 wt% carbon,
((1) 0.11 wt% carbon

238

250

251

 

Figure 11.3 Backscattered electron images of Ti-47Al—2Nb-2Mn
+0.8 vol%TiB2 alloys having (a) Low carbon,
(b) 0.065 wt% carbon, (c) 0.11 wt% carbon. The arrows

 

indicate the thin layers of or; phase

Figure 11.4 Distribution of thin or; layers surrounded by y grains in TiAl alloys -----

Figure 11.5 a and b: Backseattered electron images of Ti-47Al-2Nb-2Mn
+0.8 vol%TiBz alloys having 0.065 wt% carbon

 

Figure 11.5 c: Precipitates along the grain boundaries of y grains in
creep tested Ti-47Al-2Nb-2Mn+0.8 vol%TiB2 alloys

254

---255

255

256

 

having 0.065 wt% carbon

Figure 11.6 Backseattered electron images of Ti-47AI-2Nb-2Mn+0.8 vol%TiB2
alloys having 0.11 wt% carbon

256

 

Figure 11.7 Dislocation activity near TiB particles in creep deformed
Ti-47Al-2Nb-2Mn+0.8vol%TiB2 alloys having 0.065 wt%
carbon. White arrows show bowing, and black arrows

 

show pinning points

Figure 11.8 Distribution plot of a; spacing in undeformed grip and deformed

258

 

gage section of tensile creep specimens

258

Figure 11.9 (a) Creep times to 0.2% and 0.5% strain as a function of carbon
concentration for specimens deformed at three deformation
conditions. (b) Creep strain to 0, 20, 30, 50, and 100 hours
in Ti-47Al-2Nb-2Mn+0.8 vol%TiB2 alloys. Lines go through

 

the average of 3 samples

Figure 11.10 Creep strain rate as a function of creep strain in Ti-47Al-2Nb-2Mn

260

 

+0.8vol%TiB2 alloys

Figure 11.11 The relationship between lamellar vol. fraction and strain rates
with carbon concentration. On this and subsequent plots, the size
of the symbol represents carbon concentration

262

262

 

Figure 11.12 The relationship between lamellar vol. fraction and initial strain
rates and strain rate at minimum and transition points. Symbol size
represents carbon concentration (small, medium, and large size

264

 

symbol: Low carbon, 0.065 wt%, and 0.110 wt% carbon)

Figure 11.13 m spacing as function of creep strain

264

 

Figure 11.14 Lamellar refinement and strain rates at transition point as
a function of lamellar vol. fraction

266

 

CHAPTER 12

Figure 12.1 The distribution of 0L2 spacing is shifted to a greater fraction of
finer spacing as a result of early stage primary creep deformation.
A simulation based on the undeformed room temperature grip
that generates 0.2 um thick or; layers within thicker y lamellae
matches the deformed distribution

273

 

Figure 12.2 Lamellar laths configuration in undeformed and deformed case
from simulation

274

 

Figure 12.3 The relationship between computed local shear strains and

 

reduction of a; spacing

Figure 12.4 The relationship between computed local shear strains and

275

 

strain rates at transition points

Figure 12.5 The relationship between computed local shear strains and

 

volume fraction of lamellar grains

Figure 12.6 The relationship between strain rates and strains at transition point

275

276

276

 

with volume fraction of lamellar grains

xxii

CHAPTER 1

1.1 INTRODUCTION

Gamma titanium aluminide (TiAl) based materials have been developed as
candidates for high temperature structural applications, such as heat engines, turbine
engine blades, aerospace and automobile components, because they have low density,
low coefficients of thermal expansion (CTE), high thermal conductivities, high specific
strengths and stiffness at elevated temperature, and high resistance to oxidation and
creep. TiAl alloys are beneficial in the design of components operating at high
temperature, because they are light weight, exhibit minimum thermal fatigue, and can
operate at high temperatures with external cooling. However, these alloys have lower
yield strengths, ultimate tensile strengths and room temperature ductilities than the
current titanium alloys and superalloys used in these high temperature applications.

Several investment cast TiAl alloys (Ti-45Al-2Nb-2Mn+0.8TiB2 XDTM, Ti-47Al-
2Nb-2Mn+0.8TiB2 XDTM, Ti-47Al-2Nb-2Mn+0.8TiBz )0)“ with different amounts of
interstitial elements (oxygen and carbon), Ti-47Al-2Cr-2Nb, and Ti-47Al-2Nb-1Mn-
0.5W-0.5Mo-0.ZSi alloy) have been developed by Howmet Corporation. Selected heat
treatments and the addition of interstitial elements to the alloys have been used to reduce
the costs of processing and to improve primary creep resistance, especially primary creep
resistance as measured by the time to 0.5% creep strain. Most of these TiAl alloys have a
duplex microstructures which have better ductility than lamellar structures. However,
duplex h structures display poorer creep resistance than lamellar structures. Most studies

of creep of TiAl alloys have focused on minimum creep rate conditions or stress-rupture

. The primary creep properties, however, are important in practical applications.
)n of ordinary and super-dislocations are the major deformation mechanisms in
chanical twinning, however, is also an important deformation mechanism at
terature. Limited research has been conducted on deformation behavior of TiAI
rly stage of creep, and the creep mechanisms of these alloys are not well
(:1.

chapter two, the main creep mechanisms for pure metals and alloys at high
re are reviewed. In chapter three, titanium aluminides are reviewed in terms of
ructure, microstructure, deformation modes, deformation twinning, creep
and alloying effects.

:search methodology is presented in chapter four. An overview of research
presented in chapter five, with the detail of these results being presented in
ix through twelve.

chapter six, the primary creep resistance (time to 0.5% creep strain) is analyzed
wo-phase TiAl alloys. Two of the alloys containing TiB; particulates produced
)TM process and a third alloy without TiB; are compared. Because the nature of
reep is affected by changes in the microstructure that arise from different heat
3 and compositions, the details of primary creep are explored.
re effect of interstitial concentration on the primary creep deformation and
ctures of investment cast TiAl alloys is presented in chapter seven. The
ctures and the primary creep deformation modes are analyzed in terms of grain

ibution of or; and 7 thickness, and a2 thickness.

In chapter eight, microstructural changes during primary creep of Ti-47Al-2Nb-
1Mn—O.5W-0.5Mo-0.ZSi alloys as determined by analysis of the volume fraction of
lamellar and equiaxed microstructures, the distribution of y thickness and (12 spacings,
mechanical twinning, and precipitates are presented.

In chapter nine, stress and temperature change creep tests were carried out under a
condition of combined compression and shear stress on carefully oriented PST crystals of
Ti-47Al-2Cr-2Nb alloy with <110> and <112> directions parallel to the principle shear
stress axis. Based on the measured stress exponents and activation energies, possible
creep mechanisms are suggested. By understanding the fundamental mechanisms of
creep deformation of polysynthetically twinned (PST) crystals, an understanding of the
fundamental basis of deformation of the lamellar colonies present in polycrystalline TiAl
alloys can be established.

The effects of different strengthening treatments on primary creep resistance of
four alloys are presented in chapter ten. The changes in the different lamellar and
equiaxed microstructures; 7 thickness, a; spacing, mechanical twinning; interstitial
levels, precipitate size, composition, and shape; and dislocation interactions with the
precipitates are quantified for each alloy.

Ti-47A1-2Nb-2Mn+0.8TiB2 x1)TM alloys with higher interstitial element
concentrations were developed for the enhancement of primary creep resistance and
process cost reduction by Howmet Corp. The effect of carbon concentration on primary
creep deformation and microstructures of investment cast and HIPed TiAl alloys with no

further heat treatment are compared to the same alloy with low interstitial content in

chapter eleven. Grain size, distribution of a2 spacing, and volume fractions of lamellar
grains with respect to carbon concentration are also analyzed.

From previous studies have shown that the lamellar refinement is a common
microstructural feature in the primary creep behavior of TiAl alloys. In chapter twelve,
lamellar refinement is simulated using a set of lamellar spacings that represente an
undeformed lamellar spacing frequency distribution. The lamellar spacing distribution
obtained from the simulation is compared with the measurements of lamellar spacings in
TiAl alloys. Local shear strains in the lamellar microstructure are computed to examine
how different deformation histories affect the resulting deformed distribution.

In summary, the primary creep deformation behaviors of several TiAl alloys are
characterized in terms of lamellar grain size, lamellar spacing, volume fraction of
lamellar grains, and other factors. The primary creep mechanisms are suggested from all

studies.

1.2 RESEARCH STRATEGY

This research focuses on primary creep in the duplex microstructure TiAl alloys,
concentrations the relationship between creep deformation and microstructural
characteristics: precipitates, twinning, lamellar spacing, dislocation activity, and volume
fraction of lamellar and equiaxed grains. were applied to examine The effects of several
heat treatments on primary creep resistance in five investment cast TiAl alloys were
investigated. The effects of heat treatment on characteristics of precipitates, distribution
of lamellar grains and equiaxed grains, lamellar spacing, volume fraction of lamellar and

equiaxed grains, and orientation distribution of grains were examined. These variables in

turn effect the deformation process. By using stress and temperature change techniques,
this investigation eventually lead to a phenomenological understanding of primary and
secondary creep mechanisms at a particular stress condition and assisted in the

development of alloy design criteria.

1.2.1 Rationale and goals

The substantial progress in understanding creep behavior of single or two-phase
titanium aluminides has been made as summarized in the chapter three. Some important
and interesting questions have emerged from previous work and these questions have to
be answered in order to develop a better understanding of creep deformation. A brief
summary of these is given.
1) If heat treatment gives better primary creep resistance in TiAl alloys, what kind of
microstructural features can be influenced through the heat treatment? Some authors
showed that the reduction of lamellar spacing leads to lower creep strain rate by blocking
moving dislocations. Could the lamellar spacing be altered by heat treatment, and how
sensitive is the lamellar spacing to alloy composition?
2) Addition of carbon in TiAl alloys resulted in improvement of creep properties.
Therefore, understanding the relationship between interstitial concentration and heat
treatment in the duplex TiAl alloys with respect to primary creep is important. The
precipitates in the lamellar microstructure increase creep resistance. The mechanism by
which the precipitates enhance primary creep resistance is not well understood. If
precipitates grow during creep then models for strain assisted growth need to be

examined.

3) Deformation twins were frequently observed in secondary or tertiary creep, and in
some observations of primary creep, the amount of mechanical twinning strongly
depends on stress and temperature. The role of deformation twins in the early stage of
creep is not well understood. The deformation twinning parallel to lamellar planes is
called easy mode deformation, but the twins that intersect lamellar boundaries are hard
mode deformation. If easy mode twinning occurs in primary creep, it is necessary to
determine the contribution of easy mode twinning to the early stage of creep strain with
different stresses and temperatures.
4) The creep mechanism in PST crystals has not been examined. Thus it is important to
determine the creep mechanism in PST crystals with different orientations, because the
deformation mode of the lamellae is highly orientation dependent. By comparing creep
in PST crystals and polycrystals in the same test conditions it is also helpful to
understand how lamellar microstructures contribute to the creep mechanism. For
example, if stress exponents or activation energies that govern PST deformation are also
observed in polycrystals, then the role of deformation in lamellar microstructures on
creep of duplex materials can be identified.
5) What kinds of deformation structures occur at the different levels of primary creep
strain as function of strain, for example, 0.2%, 0.5%, 1.0% and 1.5%? Analyzing changes
in the deformed microstructures is essential to understand primary creep mechanism.
Although primary creep is practically more important for many candidate
applications for TiAl, most of the research has been focused on the secondary or tertiary
creep in TiAl alloys. Thus, the goals of this research are to investigate the relationship

between microstructure and primary creep mechanisms in TiAl alloys having different

compositions. First, the effect of composition, microstructure, lamellar spacing, and heat
treatment on primary creep properties will be investigated. Next, the role of deformation
twinning on primary creep deformation with different stress, temperature, and creep
strain will be described by observation of deformed microstructures. Finally, creep
behavior in PST crystals with different orientations will be compared to polycrystals in
similar test conditions in order to understand the contribution of lamellar microstructures
on duplex materials to the creep mechanism. Therefore, the primary creep mechanism in

duplex TiAl alloys will be rationalized by analysis of experimental observations.

CHAPTER 2

C REEP THEORY

2.1 DEFINITION OF CREEP

There are many practical examples where materials are required to survive for
long periods under load at high temperatures, such as turbine blades of jet engines and
power generators. Thus when assessing the resistance of materials to deformation and
failure over long times under load at high temperatures, particular attention must be given
to a phenomenon known as “creep”. Creep may be defined as the process by which
plastic flow occurs when a constant stress is applied to a material for a prolonged period
of time.

As illustrated in Figure 2.1, the shape of a plot of creep extension versus time
usually differs at low and high temperatures [1]. At low temperatures, dimensional
changes due to creep are usually extremely small and failure rarely occurs. In contrast, at
high temperatures, creep deformation can cause considerable changes in dimensions and
failure generally occurs after some time. The time to fracture, tp, decreases with
increasing temperature and with increasing applied stress.

Under low-temperature creep conditions (T<0.3Tm), for many different crystalline
materials, it has been found that the creep strain increases with log(time) as shown in

Figure 2.1 [2]. On this basis, the increase in creep strain with time can be expressed as

8:8

(r)!

— £0 = a, log((azt +1) (1)

 

4

Time to fracture, tF

 

Change in length —-)

 

Figure 2.1.

Low temperature /
creep curve(T<0.3Tm)

 

 

Failure occurs after a time, tF, termedl
/ creep fracture or creep rupture

Typical creep curve
at high temperature (T>0.4”l}n)

 

Time or log (Time) -—)

Creep curves at high and low temperatures. At high temperatures, creep
deformation can cause considerable changes in dimensions and failure
generally occurs after some time, tF. In contrast, dimensional changes due to
creep are usually extremely small and failure rarely occurs [1,2].

where Sun is total specimen strain, 80 is the virtually instantaneous strain, at and (12 are
constants for the material which depend on stress and temperature, and t is time. [2]

Differentiating equation 1 gives the creep rate at any instant as

é=£=fl (2)
dt a2t+1

which is consistent with the fact that the creep rate decreases continuously with

increasing time during low-temperature tests [2].

2.2 STAGES OF CREEP

While the creep curves recorded at low temperatures are well described by
equation (1), in the high-temperature regime (DOATm), the creep curves are commonly
described by reference to different regions of the curves in Figure 2.2.a [3]. Afier the
initial strain on loading, the creep rate usually declines during the “primary stage” until
an ostensibly constant creep rate is reached during the “secondary stage”, after which the
creep rate increases again during the “tertiary stage” which leads to fracture. From Figure
2.2.b, the primary or transient stage, the creep rate continuously decreases with time; that
is, the slope of the curve diminishes with time [4]. This indicates that the material
experiences an increase in creep resistance or strain hardening. For the secondary stage,
sometimes termed steady-state creep, the rate is constant; that is, the strain-time plot
becomes linear. This is ofien the stage that is of the longest duration. The constant creep
rate is explained on the basis of a balance between the competing processes of strain

hardening and recovery (work softening). Recovery is the process whereby a material

10

 

 
   
 

Secondary creep rate
és =A8/At

 
  
   

Creep strain, a

 

 

 

 

Primary t At tem— .__, — —>
e444 _ --. *v-r ITertlary *
M ...3. ,.3,
Secondary
- , if - .
Instantaneous deformation, 80
J.
Time, t tr
Figure 2.2.a Schematic representation of creep stages at a high temperature creep
curve [3].
Rupture

   

 

-w

.93”

C3 I

H

.5

t3

3::

‘0 Secondary creep rate

3 ' éu=Ae/At=constant ‘

93. ;

U ‘zr:.,_ l. >< , .-,,L ’ *m— ~ ~ View ,f,fi >
Prlmary Secondary Tertiary

 

 

 

Creep strain, 8

Figure 2.2.b Schematic creep strain rate versus creep strain curve at a high
temperature [4].

ll

becomes softer by various mechanisms, i.e. dislocation glide, climb, cross slip, etc. and
acceleration of the strain rate retains its ability to experience deformation. Finally, in the
tertiary stage, there is an acceleration of the strain rate and ultimate failure that is
frequently termed rupture. Microstructural changes in the tertiary stage occurs such as
gain boundary separation, the formation of internal cracks, cavities, voids, and neck
formation in the tertiary stage. These all lead to a decrease in the effective cross-sectional

area and an increase in strain rate.

2.3 STRESS AND TEMPERATURE EFFECTS

Figure 2.3.a shows the effect of temperature on the creep behavior at the same
stress [5]. As the test temperature increases, the creep strain increases at the same creep
time. When this kind of experiment is undertaken, it is usually found that a linear
relationship is obtained by plotting In a", against (lfI) as shown in Figure 2.3.b [6]. Thus
the secondary creep rate increases exponentially with temperature according to the

equation

In ésoc —(—;-j (3)

where é, is steady-state creep rate and T is temperature [7]. This form of temperature

dependence is found for many thermally activated processes, including diffusion,

oxidation and etc.. The creep rate can be defined to obeys Arrhenius’s Law, so that

 

,- =Kt exp-[1%] <4)

12

 

 

 

 

 

 

 

 

 

 

o = constant
on
.5“
re
1::
V)
a.
G)
G) ‘
H
U
\
T 1ncreasmg
Time, t
(a)
10": j
i 96.52 MN/m 1
C 82.73MN/m Z
_A F J
'33, 68. 94MN/m
-7._ a
3 1° t 137.90MN/m 1
m : 55.15 MN/m .
9 ~ 1
we r _
5'- t
" K
ow 8
:.0616 r 24 l3MN/m 3
a 1
'4 _ 22°58 MN/m gradient=-Q /2.303R j
. 13.79 MN/m ° ,
-R d(ln e )
1 Q = 5
c 2.303d(1/T)
100.9 1 1.1 1.2 1.3 1.4 1.5 1.6
(1000/1) K
(b)

Figure 2.3 (a) Temperature effect on creep curves (b) Log 6' vs. 1/T for Ni-67% C0-
Ale3 showing the variation in activation energy above and below 0.5 Tm [5,6]

13

where K; is a constant, O0 is called the activation energy for creep and R is the universal
gas constant (8.31 J/mole K) [8]. From Figure 2.3.b, the magnitude of activation energy
for creep can be evaluated from the slope of In 6‘, versus (l/T) plot. From O.D. Sherby
and RM. Burke’s study about the creep behavior of pure metals at high temperatures, the
values of Qc have frequently been found to be close to the activation energy for lattice
self diffusion, QSD [9]. The fact that Qc often equals QSD indicates how the important
diffusion is for high-temperature creep conditions (T>0.4Tm). The presence of diffusion
accounts for the transition from low-temperature to high-temperature creep behavior.

A similar effect of different stress on creep behavior at the same temperature can

be found in Figure 2.4.a [5]. Its dependence on stress can be written

a =K,o—" (5)

where K2 is a constant and n is material constant denoted as the stress exponent [10]. This
expression, sometimes called Norton’s Law, provides the basis for the “power law”
relationships which have been widely used to describe high-temperature creep behavior.

A plot of the log (6.“) versus the log (0') yields a straight line with slope of n as

illustrated in Figure 2.4.b for aluminium and aluminium-magnesium alloys 523 K [11].

2.4 POWER LAW RELATIONSHIP
From the equations (4) and (5), steady strain rate, 6*, , varies with both stress and

temperature. Thus steady state creep behavior can be described through their product by

multiplying equations (4) and (5) together,

14

 

 

 

 

 

 

 

 
     

 

   
   

 

 

 

 

 

 

T = constant
00
.5“
cc
3—
4.0
(I)
Q.
cu
cu
H
U
f \
Y
0' increasmg
Time,t
(a)
1(y'.1:s.,,...,...,...,...,..,,,CL1
- d(|né) ‘
104? gradient=n=————§— j
A i d(lno) 3
.. i n=3.2 ,
lm » 4
v
, M105 - 1
0° 3
0 .
H
g 1
Q1 10.6 g 11
0 r 1
2 1
U . U Al .-
107g 0 Al—0.76at%Mg 4
: A Al-l.66at%Mg j
o Al-2.93 at% Mg ~
10'sLlilririirnl.i.1..rirLrlr..
0.6 0.8 1 1.2 1.4 [.6 1.8 2
Log stress 0’ (MN/m)

(b)

Figure 2.4 (a) Stress effect on creep curves (b) Log 5' vs. Log 6 for aluminiturr and
aluminium-magnesium alloys at 523 K showing the variation in stress
exponent, n, values [5,] 1].

15

. ,1 Q n Q‘
8.=KK0'ex———‘=A0'ex—-—‘ 6
.t 12 P (RT p RT ()

where A is a constant. In some cases, there is no steady-state rate region in the creep

curve, and the minimum strain rate 8'," is usually used instead of a", [12]. This power law

relationship appears to provide a good description of steady-state creep behavior.
However, the values of n and Qc can vary in different stress and temperature regions

when the power law is used to describe steady state creep behavior.

2.4.1 Power law breakdown

By multiplying the first and third sides of equation (6) by exp(Qc/RT),

 

é_.ex chzAa" =Z (7)

The Z value is known as the Zener-Hollomon parameter [13] or the temperature

compensated secondary creep rate [14,15]. In the plot of log[ésexp(Q/RT)] against

log(o/ E) as shown in Figure 2.5 [16], the curvature of the line (indicated by an arrow)

at the highest stresses is observed. This nonlinear increase in n value with increasing
stress in tests of very short duration is often referred to as “power-law breakdown”.
Under power-law breakdown conditions, it shows an exponential dependence of the

secondary creep rate on stress and temperature as

16

 

 

 

 

 

 

 

 

 

B 1

‘ .

4 _ x 4

L -1

, v .1

5' ° ‘

Q 3 . -
0’ 1

Z . -

N

0 L 1

own 1

i—I: 2 ’— fl

u l

3 1

i

1 ” l

0 +4 %l A i r 1+1 m L I A i 4 L L l 4 1+1 1

0.8 1 1.2 1.4 1.6 1.8 2

Logm[(o/E)xl05]

Figure 2.5 The temperature-compensated secondary creep data (or Z) for polycrystalline
copper. The activation energy of 130 KJ/mole is used for this plot [16].

17

8 = Bexp- Qg—Tw (8)

Where V' = VRT and B is constant [17]. It offers a good description of steady creep rate

behavior in the power-law breakdown region.

2.4.2 Transitions in n value

Creep behavior with stress is well described by a power-law equation except for
the power-law breakdown region at very high stresses. From B. Burton and G.W.
Greenwood’s research, Figure 2.6 shows a transition from n=5 to n=1 with decreasing
stress in a plot of the grain-size compensated secondary creep rate as a function of stress
for polycrystalline copper [18]. Figure 2.7 also indicates n~1 low stress regime, at n~4,5
at high stress regime, and power law breakdown at very high stress regime for pure metal
at temperatures above 0.4Tm [19]. These different values of n are correlated with different

deformation mechanisms (see Figure 2.2).

2.4.3 Transitions in Qc value

A similar to transition in n value, transition in the activation energy, QC, occurs at
different temperature regimes instead of different stress regimes. At a creep temperature
around 0.4Tm, the activation energy for creep of polycrystalline copper is close to that of
lattice self diffusion, Q50. In contrast, at creep temperatures around 0.4Tm and just below,
Qc values are about half those for lattice self diffusion (see in Figure 2.8) [20]. Moreover,

when the creep tests are conducted at high stresses when ns4-6 or low stresses when ngl ,

this transition from QCEQSD to Qc—EO.5QSD has been found as the creep temperature is

decreased toward 0.4Tm [21].

18

 

-7
1O _YY_TYt Y T 1' rY fiVfi Mfi VVVVV r

Y YTTTYr ‘7 j rTW—TY

 

1 0’8
10f?
.8 :
E 1010 E

 

1O 3 114111 1 u 1 111111 4 1_1 111111 4 1 1 111111 1 +414 1141]

01 1 10 100 1000
-2
Stress MNm

Figure 2.6 A plot of the grain-size compensated secondary creep rate as a function of
stress for polycrystalline copper [18].

 

Power law breakdown
at very high stresses

Log e

High stresses, n~4,5

 

 

Low stresses, n~l

Logo ——->

 

Figure 2.7 Schematic illustration of the transitions in n value occuring when the stress
dependences of steady-state creep rate for pure metals at temperatures above
0.4 Tm are described using a power-law relationship [l9].

l9

These transitions in n and Qc are commonly explained on the basis that different
mechanisms determine creep behavior in different stress-temperature regimes. This

suggests that a change in n and/or Qc signifies a change in creep mechanism.

2.5 DIFFUSIONAL CREEP PROCESS

High-temperature creep behavior can be described by using a power law equation.
The Qc term in the power law can be correlated with diffusion process. Diffusion is
clearly important in determining creep behavior, since diffusion normally occurs due to
the presence of vacancies within the crystal lattice at high temperatures. Thus vacancy
movement is the dominant diffusion mechanism in most metals and alloys. For example,
the movement of vacancies in a pure metal by which copper atoms move in the copper
lattice is called self-diffusion, and the activation energy for this process is the activation
energy for self-diffusion, QSD. As mentioned before, the activation energy for creep is
frequently found to be equal to or less than that of self-diffusion through low and high
temperature ranges. The “preferential diffusion path” can explain why the magnitude of
the activation energy for creep, Q, can be smaller in different temperature regimes. The
magnitude of the activation energy for diffusion depends on the bond energies between
the moving atom and its neighbors. Surfaces provide “easy paths” for diffusion, since
surface atoms have fewer neighbors and therefore fewer bonds than the atoms in crystals.
Similarly grain boundaries and dislocation cores should also offer rapid diffusion paths,
characterized by activation energies lower than that for lattice diffusion. Figure 2.9 shows
evidence for a lower activation energy in polycrystalline silver at different temperature

ranges as compared to single crystals, due to the existence of grain boundaries [22].

20

220 I I 7 I T I I I I

 

210— _
200- ://////I-:
190e ,

180e I -
170— -
160i 1/

150~ ’ —
140~

130p . . _.

Activation energy ( KJ molle)

120~ e

 

 

110 1 1 1 l_ 1 1 L g 1

600 700 800 900 1000 1100
Temperature, K

 

Figure 2.8 The variation with temperature of the activation energy for the creep of
polycrystalline copper [20].

 

'26 I I I I I I

 

0 Single crystals

 

+ Polycrystals

 

Log D

 

 

 

l I l l l

008 0.0009 0.001 0.0011 0.0012 0.0013 0.0014
1/T

Figure 2.9 The variation of the diffusion coefficient with temperature for the single
crystals and the polycrystalline silver [22].

 

-40
0.0007 0.0

21

2.5.1 Nabarro-Herring Creep

As shown in Figure 2.10.a [23], a stress is applied to a cube-shaped grain in a
polycrystal deforming at high temperatures. If vacancies are moving away from
boundaries experiencing tensile stress, these vacancies diffuse into the boundaries under
compression. This results in a flow of vacancies toward compressive regions and a
counter flow of atoms toward tensile regions. Continued vacancy transfer and counter
flow of atoms causes time-dependent extension of the crystal in the tensile direction,
simultaneously reducing the cross-sectional area of the grain at right angles to the tensile
axis. This diffusional creep process is called Nabarro-Herring Creep. For Nabarro-

Herring creep, the creep rate is given by [23,24]

, .1 D.“ or;

where .1V is the flux of vacancies diffusing down the concentration gradient per second per
unit area, (1 is the average grain diameter, 01W is a geometrical constant (~10), DSD is the
lattice diffusion coefficient, Q is the atomic volume, k is Boltzman’s constant, and T is
temperature.

The power law relationship, equation (6), is fully consistent with the form of

equation (9) as indicated by

 

- Q Ds‘n [00']
_ = A n —- C = , —‘-— —— 10
5.. 0' eXP (RT an” d2 kT ( )

22

By inspecting equation (10), 6‘3 varies directly with the applied stress, i.e. in terms of the

power law relationship, n=1 when creep occurs by stress-directed flow of vacancies

through the crystal lattice. Furthermore, .9, depends on DSD, the lattice self-diffusion

coefficient, since DSD is
é'OCDsI) =DoexP" £82 (11)
.S 1 RT

with Do being a constant and QSD the activation energy for self-diffusion in the crystal
lattice. The Nabarro-Herring creep mechanism is dominant at higher temperatures and

low stresses.

2.5.2 Coble Creep

Coble creep involves vacancy transfer from tensile boundaries to those under
compression along the grain boundaries [25]. Sometimes Coble creep is named grain
boundary diffusion creep. Consider the transfer of vacancies only along a narrow zone
having a width 63 to the grain boundary. The area of grain boundary zone intersecting a
unit area of a polycrystal of average grain dimension, d, is given by (5/d). However,
diffusion along this narrow grain boundary zone is determined not by lattice self-
diffusion but by grain boundary diffusion. Under these conditions, the creep rate

associated with Coble creep can be modified from equation (9) as given in equation (12)

23

 

+6

4? 11.

 

(a) 0'

 

 

 

 

4 e , Vacancy flow

* ‘ *- Atom flow
18

 

,-, 4» ,
I} \ ,
,1 A If [
t ~75

 

 

7 :I 117 ,
(b) 91

Figure 2.10 Schematic diagram of (a) Nabarro-Herring creep and (b) Coble creep on an
ideal grain [23,25].

 

24

 

é =K D08 51(93 :K DO, 6,,er : PL 0035300 (12)
2 d2 d kT 2 d3 kT 7r d3kT

where K2 is a constant (~40), Don is the grain boundary diffusion coefficient, and Bc is a
constant (=148) [25]. From equation (12) it follows that the creep rate again increases
linearly with stress (i.e. n~1 in equation (6)), and is inversely proportional to the cube of
the mean grain diameter. The temperature dependence of the creep rate is practically
identical to that of the grain boundary diffusion coefficient (i.e. Qc < QSD). As illustrated
in Figure 2.9, diffusion along grain boundaries (preferential paths) becomes important as
lattice diffusion becomes progressively more difficult as the temperature decreases
toward about 0.4 Tm. When power law creep behavior leads to an n value of unity and an
activation energy for creep which is less than the value of lattice self-diffusion in tests
carried out at low stresses at temperatures around 0.4 Tm, Coble creep can be considered
to be the dominant creep process.

Both Nabarro-Herring and Coble mechanisms, operating in a parallel way, can

contribute to the diffusional creep. The creep rate is then described by the equation

 

e 3090 D,~=D,I[I+SD”6”] (13)

which expresses a sum of creep rates described by equation (1 1) and (12), and where Day
is the effective diffusion coefficent, and the constant B is 14 [26]. The grain boundary
diffusion will contribute to the creep rate to a greater extent, if the average grain diameter

dis lower, and the ratio DB/DL is greater.

25

2.5.3 Diffusional Creep and Grain Boundary Sliding

As described in the diffusional creep of polycrystals, atoms are transported from
boundaries subjected to compressive stress, to those subjected to tensile stresses. This
results in changes in the shape of individual grains. Burton suggested a simple
geometrical consideration as shown in Figure 2.11 [27]. Figure 2.11.a shows the original
configuration of equiaxed grains. Diffusional creep without any participation of grain
boundary sliding (GBS) results in the formation of voids due to incompatibility along
longitudinal and horizontal the boundaries (Figure 2.1 lb). For the compatibility to be
preserved, GBS has to occur simultaneously with the diffusional elongation of individual
grains (Figure 2.11.c), showing a shift of a marker line across a grain boundary. An
experiment using a bi-crystal of tin explained the grain boundary movement during creep
[28]. Gifkins also suggested that it is possible that grains slide past each other along their
boundaries in addition to the climb of dislocations during secondary creep [29,30]. The
sequence of events which may lead to grain boundary sliding and migration is shown in
Figure 2.12 [30]. The dislocation pile-up and relaxtion by climbing toward the boundary
controls the rate of slip along the grain boundary, which is not constant with time. As a
consequence of this process, the activation energy for grain boundary slip may be
identified as that for steady-state creep. After climb, dislocations are able to give rise to
grainboundary migration, which is proportional to the overall deformation when sliding
has temporarily ceased.

The role of GBS in diffusional creep differs from that in dislocation creep. In the
former, GBS is an indispensable pre—requisite of diffusion, while in the letter, dislocation

creep, GBS need not occur at all if a sufficient number of glide and climb systems

26

A

 

‘ | L ’ Marker I i
. 1'1 . \
‘x/Nt / me TenSIIe‘r \ I,
1 I i I Istress. 1 j I
/ \ _ / \\
r . I a... [» y y

(C)

Figure 2.11 Accommodation of diffusional creep by GBS: (a) An original configuration
of equiaxed grains. (b) The result of diffusional creep without participation of
GBS. (c) Simultaneously with diffusional creep, GBS occurs as shown by the
offset of a marker line [27].

 

{gran},
Boundary Vacancies , fi
1- 1- '1' [SI 1 1
t |- 7' 1: . .
II: I- v 'v‘ ”we
1 H317 1' ' .
i [- 6711) I- 'C11111 1
he s n 1

(a) 3 (b) I (c) “ (d) I (e) 1
‘ I
I

1 f [
pa, 1
g J’gon .1 C Sub-
M 13am,” ‘ 13%

Figure 2.12 Sequence of events which may lead to boundary sliding and migration. S,
slip spacing; C, subgrain diameter [30].

27

operate for the Von Mises criterion to be met. From several studies by Raj and Ashby
[31] , Beere [32], Speight [33]and Aigeltinger and Giflcins [29], diffusional creep can be
considered as diffusion which is accomodated by GBS, or GBS which is accomodated
by diffusion. Any consideration of mutually independent contributions of the GBS and

the diffusion is meaningless [31].

2.6 DISLOCATION CREEP PROCESSES

When the temperature is between 0.4Tm and 0.7Tm and at intermediate/high stress
levels (104<o/G<10'2), the dominant mechamism of creep is the rate of dislocation
motion by glide or climb. This regime is known as dislocation creep. Dislocation creep
processes start from two ideas of the processes that recovery is the creep rate controlling
process, and that dislocation glide is the creep rate controlling process, respectively.
Recovery aspects are related to nonconservative dislocation motion, e.g., dislocation
climb, in which obstacles are circumvented and/or by which dislocations are annihilated.
Work hardening aspects of creep are related to dislocation glide. Hardening mechanisms
include dislocation interactions, dislocation pile-ups against barriers, and Kear—Wilsdorf
(KW) locking by thermally activated cross slip of segments onto other slip planes

observed in some ordered materials.

2.6.1 Dislocation Glide Rate-controlled Creep

In pure metals, a possible model of dislocation creep controlled by the glide of
screw dislocations with jogs dependent on diffusion was first considered by Mott [34]
and later by Raymond and Dorn [3 5], Barrett and Nix [36]. Barrett and Nix assumed that

the non-conservative motion of jogs on screw dislocations can be connected with both the

28

emission and absorption of vacancies and that any screw segement contains either
vacancy emitting or vacancy absorbing jogs [36]. Under this condition, the creep rate is
determined from the following equation
b2 Gb 0' 3
6'" = D 47: ,————— — 14
I. _/ a3 kT ( G] ( )

where DL is the coefficient of lattice self-diffusion, [3 is the number of atoms in the unit
cell, a is the lattice parameter and Ij is the distance between jogs. This equation also
predicts the dependence of the creep rate on the third power of the stress.

Weertman presented the first theory of creep controlled by viscous dislocation

glide in solid solution alloys, and assumed not a specific type of interaction of solute

atoms with dislocations [3 7]. Creep rate is described by the following equation,

éz

0'3b2 : 27r(1—V)G[_o_']3 (15)
GAB A G

where A is a constant dependent on temperature and on the mechanism of interaction of
solute atoms with dislocations, B=Gb2/21r(1- v), b is the length of the Burgers vector, and
0' is the applied stress. Thus, the creep strain rate depends on the third power of the

stress.

Cottrell and Jaswon analyzed the dislocation creep involved in the drag of an
atmosphere formed around a dislocation due to the elastic interaction between the
dislocation and solute atoms [38]. From their analysis, the modified equation for the

creep rate is expressed:

29

, 27r1—v TD. 3
‘9: 26513); [g] (16)

where Ea =(r—r0)/r0 is the misfit parameter; r and r0 are radi of solute and solvent

atoms, respectively, D3 is the diffusion coefficient of the solute and C = G, exp“ WM] / kT]

is the solute concentration in the atmosphere; WM is the solute atom—dislocation

interaction energy.

2.6.2 Dislocation Climb Rate-controlled Creep

Weertman proposed a creep theory based on the assumption that dislocations
form piles-up, and the leading dislocations of pile-ups climb and annihilate in the pure
metals as shown in Figure 2.13.a [39]. Under this assumption, the creep rate is expressed

as

D Gb3 0' ‘5
e=A—8————- — 17
b3'5M0‘5 kT [a] ( )

where M is Frank-Read source density. The activation energy of creep is close to the
activation enthalpy of lattice diffusion, and the creep rate depends on the 4.5-th power of
applied stress. The theory predicts not only the usual temperature dependence but also the
frequently observed stress dependence of steady state creep rate. With a different
assumption in this theory, Weertman and Weertman modeled dislocations that do not

form pile-ups, but climb in groups, and found the creep rate again proportional to the 4.5-

30

th power of applied stress [40]. Again, Weertman derived another version of model from
the idea of dislocation motion and rearrangement in Figure 2.13.b [41,42]. The creep rate

is presented as the following equation, which depends on the third power of stress [42].

. Gb 3
a = Dixie—15%] (18)

The previous Weertman creep models were considered based on the assumption
that the creep rate is controlled by a recovery process dependent on dislocation climb
occuring by means of lattice diffiision. In contrast, Nabarro considered individual links of
the network as dislocation sources: while absorbing or emmiting vacancies they bow out;
the dislocation density increases due to this bowing [43]. The increase in dislocation
density is compensated by the annihilation of links of opposite signs which meet during

climb. The Nabarro analysis led to the equation,

00 0' 3 4G “
é=DA — — In —— 19
" ”bZkTIG] (no) ( )

The value of the logarithmic term changes with stress very slightly, so that the creep rate

practically varies with the third power of stress.

31

 

L

S

4 1---“,
TTTT'T‘L .Lv_l_.L.l.
F-R h,
A TTTTT'L'L ‘4'"-
‘ F-R
(a)
.l.
.L S J_ ,
T T .L' .L l
T-‘- T TS -'- I
T S ‘1'“ .L T T ‘LT h
J.
S .L J- J-. i
T T J.” T
SJ- ..-L , .T.8:TS J- J-
T J. .L T
.T
(b)

Figure 2.13 Weertman’s models of creep: (a) Dislocations form pile-ups, the leading
dislocations of both pile-ups climb and annihilate. (b) Arrangement and
motion of dislocations assumed in a future version of the model [39,41].

32

2.7 DEFORMATION-INDUCED TWINNING

Deformation or mechanical twinning is a mode of plastic deformation that occurs
by shear as the result of applied stresses. The shear associated with mechanical twinning
is uniformly distributed over an entire deformed volume rather than localized on discrete
slip planes. When a crystal deforms plastically by twinning, there occur atomic
displacements, which give rise to rotated crystal bands within a grain. Hexgonal metals,
such as Zn and Mg, behave in this way when deformed at ambient temperatures, while
BCC metals, such as iron and Nb, show this behavior when deformed at subambient
temperatures [44]. Deformation twins have been observed in a few FCC metals such as
silver and copper-aluminium at low temperatures [45].

The results of plastic deformation by twinning are very different from that of slip.
First, the twinned region of a grain has a mirror image of the original lattice with respect
to the twin plane, while the slipped region has the same orientation as that of the orginal
unslipped grain as shown in Figure 2.14. Second, twinning consists of uniform shear
strain, while slip consists of a multiple shear displacement on a single slip plane. Third,
the twinning direction is polar, while the slip direction can be positve or negative. In
general, the stress necessary to form twins is greater than but less sensitive to temperature
than that necessary for slip [44]. This stress for initiation of twinning is much larger than
the stress necessary for its propagation [44].

In high temperature creep conditions, deformation twinning is rare, but it is
observed in some ordered materials. Yamaguchi, Jin and Feng suggested that mechanical
twinning is an important and common deformation mechanism at high temperatures in

TiAl intermetallics [46,47,48,49,50], and sensitive to temperature, stress, strain rate, and

33

v-fl-Wl (SQ DO CD

1 , eve) or.) 01.96 Twin
1 - NGQOGQ
* ooooooee——fl~,
1 QgggogMatrlx
I I ......
‘ ......
1 ¢-ww—
i

1 vfimi~~i+
1 ........
,1 0.000000
,, poeoooooOgmmm

: ‘ . . . . . . . .

1 «~ ’~—-

Figure 2.14 Schematic diagram of twinning and slip after applied stress.

34

crystal orientation [46,47,50].

2.8 CREEP IN SOLID SOLUTION ALLOYS

Sherby, Burke, and Langdon classified the solid solutions into two classes
according to the value of stress exponent n of the steady state creep rate [51, 52]. For
Class I alloys (class A), the movement of dislocations is controlled by the drag of solutes,
and the parameter n is approximately 3. For Class 11 alloys (class M), the creep is
controlled by climb of dislocations, as in the case of pure metals, and n is approximately
5. Other characteristic differences in Class I (A) and Class II (M) creep behavior are
noted in Table 2.1 [53]. Figure 2.15 shows the different deformation behavior between
strain and strain rate for pure metals, and Class I (A) and Class II (M) type in solid
solution alloys [54]. In a pure metal, the initial creep rate is very high since the existing
dislocations can freely move. The decrease in strain rate after some deformation is due to
an increase density of dislocations. For Class I (A) alloys, the initial creep rate is low
because there are a lot of solute atoms pinning each dislocation. As the deformation
proceeds, the dislocation density increases and the number of solute atoms per unit length
of dislocation decreases. Thus the creep rate increases with deformation. For Class II (M)
alloys, a control of the dislocation drag by solutes is the initial process. However,
dislocation climb becomes the main rate-controlling process after some deformation.

Thus creep rates decrease after it reaches the peak.

35

Table 2.1 Characteristic features in Class I and Class II creep behavior [53]

 

Class I
creep
behavior

no instantaneous plastic strain after stress application

‘inverse’ primary creep behavior

 

the higher creep rate after the stress reduction than the steady state

creep rate

 

the steady state creep rate not depend on or only slightly depends on
the stacking fault energy

 

no significant extent formation of dislocation substructure

(cell or subgrain structure) during creep

 

generally slightly curved, and homogeneously distributed dislocations,

slight tendency of network formation

 

dislocations with predominant edge orientation

 

Class II
creep
behavior

 

instantaneous plastic strain after stress application

‘normal’ primary creep behavior

 

the lower creep rate after the stress reduction than the steady state

creep rate

 

the steady state creep rate strongly depend on the stacking fault energy

 

well defined dislocation substructure (cell or subgrain structure)
during creep, and not necessary well developed cell structure in the

case of low stacking fault energy

 

generally slightly curved, and homogeneously distributed dislocations,

slight tendency of network formation

 

 

random orientation of dislocations and not prevailed edge orientation

 

36

 

 

 
   

 

10'2 r ‘1‘ ( 0', T: constant ) 2
q, 10'3 L," /Pure Metal _
in! 1
'5 .'
fl 4.'
:10- * f
0-
N
h
VII-1 -5 L
m 10 Class II ~

  
 

Solid Solution

 
 
 
  

 

 

 

 

-6 fl
10 Class I
Solid Solution
10-7 1 1 1 1 1 1
0 0.1 0.2 0.3
Strain

Figure 2.15 Relationship between strain and strain rates for pure metals, solid solutions of
Class I (dislocation-drag control), and solid solutions of Class II (substructure

control) [54].

37

2.9 STRESS AND TEMPERATURE CHANGE CREEP (TRANSIENT CREEP)

The transient creep behavior after stress changes generally depends on the
material and on the amount and the sign of the stress change. The behavior during sudden
stress changes during steady-state creep at high temperature is thought to depend on the
type of glide motion of dislocations [55-57]. Oikawa and Sugawara suggested that there
are two types of transient creep behavior in pure Al and Al alloys: the free-flight glide in
which dislocations can glide freely and the viscous glide in which dislocations glide in a
viscous manner [58]. Figure 2.16 shows the transient creep behavior of the pure metal
and the alloy after increments and reduction of stress [59].

For the free-flight glide deformation in Figure 2.16.a, as stress increases,
instantaneous strain is followed by transient deformation in which ds/dt is much greater
than the previous steady-state rate, and it decelerates to a new (higher) steady state value.
This transient creep behavior consists of elastic and plastic components. But with a stress
decrease, an instantaneous elastic contraction is followed by transient deformation in
which de/dt is much less than the previous steady-state creep rate and accelerates to a
new (lower) steady state stage. This type is known as ‘normal’ transient creep behavior.
As indicated above, the transient creep behavior in a pure metal is governed by the
dislocation climb processes, i.e. dislocation glide contributes strain but the creep rate is
controlled by the rate of dislocation climb processes in which n is generally close to 5.
These glide-climb creep processes which are associated with the formation of
substructure allow rearrangement and annihilation of dislocations to occur. The effective
stress (the difference between the applied stress and the internal stress; stress field around

the dislocations due to the distortion of the crystal) causes dislocation movement after a

38

stress increment, and the substructure rearrangement and annihilation of dislocations
associated with the decrease of effective stress causes deceleration of the creep rate. On
the other hand, after a stress reduction, the coarsening of the dislocation substructure
results in an increase of the effective stress for thermally activated glide that causes
acceleration of the creep rate.

For the viscous glide deformation in Figure 2.16.b, instantaneous plastic strain
does not occur when the applied stress is increased suddenly, and the instantaneous strain
is elastic behavior. In contrast to free-flight dislocation glide, dislocations remain
homogeneously distributed with no formation of subgrain structure during the primary
creep in this type and the dislocation glide becomes the slowest process due to the
viscous drag of mobile dislocations by the solute atmosphere, where n is equal to 3. Thus
the solute-drag model can explain the inverse of normal primary creep behavior. The
creep rate is limited by the mobile dislocation density which is low prior to creep. As
subsequent stress increases, the mobile dislocation density increases gradually and
therefore it leads to a higher creep rate. After a stress decrease, the moblie dislocation
density is higher than the equilibrium level, but the excess dislocations have not yet
glided to annihilation sites so the strain rate decelerates to the steady state value.
Weertman presented a glide-controlled model which assumes that the creep rate depends
on the mobile dislocation density and the average dislocation velocity that is determined

by the rate at which the solute atoms can move along with the dislocations [60, 61].

39

 

 

 

 

 

(a) (b)
(+A0')
+A
5:: ( 0)
.2
‘66
OD ,
g: y -1 __ x
2 A1P I 1+
IL] 41“" ; Al 1"?“
Al =A1 1 1 ‘Al+
Y,__c, L- _ Y- ,- Ex— _, #‘W_
_W (_AG _ T ( Ac)
Time

Figure 2.16 Schematic drawings of the instantaneous elongation and contraction upon

small changes in cases of (a) the free-flight glide, and (b) the viscous
glide [59].

40

2.10 PRIMARY CREEP IN METALS [62]

Primary creep has not been studied as extensively as secondary creep. Since the
investigation of primary creep is complicated by variations in the initial loading process,
which is very dependent on operator and machine variables. Primary creep is commonly

modeled with equations of the form

8 = so +al log,(a2t+l) or
8:80 +At" or

s = so + e(l — exp(— rt» (20)

at lower and higher temperatures, respectively, where so represents the initial strain upon
loading. Quantity of so can not be measured or defined easily, since it represents a
combination of elastic and rapid plastic deformation. Primary creep is closely connected
to microyielding [63], anelasticity [64-66], and viscoplasticity [67-69]. Andrade and
Jolliffe [70] observed that the exponent a is near 1/3 in the middle equation above for
many materials. Alternatively, stochastic approaches to modeling heterogeneous
deformation in materials have been proposed by Daehn [71] and by Bagley [72], that
reproduces the empirical nature of various primary creep models.

In a study of room temperature primary creep in Ti alloys, correlations between
microstructural features and changes in the rate constant a have been identified for one
and two phase alloys [73, 74]. When loaded at about 80-90 % of the yield stress, a one
phase alloys exhibit three stages with increasing strain and time. The a value is initially

close to 0.2, which correlates with rapid dislocation motion across grains optimally

41

oriented for easy slip as slip bands on the primary slip system. And then a value for the
second stage increases to 0.4 near a strain of 0.1%, where slip transfer across grain
boundaries is observed. Exhaustion of slip bands resulting from slip transfer accounts for
the return of the a value to 0.2 at a strain near 0.5%. In two phase Ti alloys [74], the
initial a value is 0.8, and it decreases to a value of 0.2 after a strain near 0.5%. The
initially high value is attributed to the much smaller size of the microstructure, where slip
transfer must occur at a lower overall strain. These observations indicate that when slip
transfer is precluded, primary creep strains are dependent on the slip distance to a barrier,
and that this strain process will be in the rage between 0.1- 0.5%.

At elevated temperatures, the role of diffusion further complicates the details of
dislocation motion, creation, and annihilation. With higher temperatures, diffusion affects
primary creep behavior by processes of anelasticity and dislocation creep processes
identified in steady state creep investigations. Anelastic flow is complex, and rheological
modeling is commonly used [64]. Primary creep strains are commonly in the range of
0.01-0.05, with lower values being observed at lower temperatures [67]. In low stress
diffusion creep conditions, one study suggests that primary creep strain may be
completely anelastic with a value of about 5 times the elastic strain [65]. Primary creep
strains were shown to be anelastic, as unloading resulted in recovery of the primary creep
strain (similar to pure FCC metals where the activation energy implies diffiision
controlled deformation [75]). With larger strains in the steady state regime, anelastic
recovery was greater than the primary creep strain.

In a study of an austenitic stainless steel, dislocation densities measured as a

function of strain indicated that the dislocation density increases and then decreases as

42

secondary creep is approached [76]. The geometrical perfection of dislocation networks
increases with greater strains. Internal friction measurements on single crystal A1 at 0.5
Tm indicate a similar result [77]. Argon and Bhattacharya [78] identified two stages of
primary creep in pure Ni deformed between 0.54 and 0.62 Tm, the first related to dynamic
recovery with a time constant of about 100 s, followed by a static recovery process with a
time constant of about 1043. The internal stress was found to increase with strain to a
plateau value at strains larger than the value where minimum creep was observed. They
developed scaling laws that provided results consistent with Andrade's observed t“3 law.
The composite model has been developed for primary creep [79, 80], in which the back
stress is not constant, but evolves to the steady state value.

Primary creep strains and creep strength are very sensitive to orientation [81]. The
deceleration of the creep rate and the evolution of dislocation structure is the result of
Orowan bowing and dislocation pileups imposed by the geometry of the reinforcements
[82]. When solid solution strengthening is present (e.g. lnconel 617 [83]), inverse creep
followed by a normal creep transient is observed, reflecting interactions between solute
atom drag effects on initial dislocation multiplication, and development of dislocation
substructure. In contrast to the relatively large primary creep strains in single crystal
superalloys, primary creep strains in aluminum strengthened with incoherent
reinforcements are less than 20% of the values measured in pure aluminum [84].

Heat treatment can affect primary creep significantly. If precipitation occurs, it
can remove solute atoms from solution that alters primary creep kinetics [85]. A smaller

particle spacing is beneficial, particularly when the distance between them becomes small

43

enough to prevent Orowan looping, which significantly reduces primary creep strain, and
causes incubation times before primary creep strains are observed [86].

As mentioned above, it is apparent that a wide variety of phenomena have been
observed in primary creep. Though much of it can be rationalized in terms of known
deformation mechanisms and microstructural observations, it is apparent that
composition and processing history have a large impact on the processes of primary
creep. Also, the details are commonly specific to the specific class of materials in which

they are observed.

2.11 DEFORMATION MECHANISM MAP

As described before, the dominant creep mechanism depends on stress and
temperature conditions. Thus the deformation mode will change and cause different creep
behaviors under different stress and temperature conditions. Information about which
creep mechanism controls deformation under various conditions can be obtained from
construction of ‘Deformation Mechanism Map’ with two main creep mechanisms;
dislocation creep and diffusion creep (Coble creep and N-H creep). The deformation
mechanism maps, first introduced by Weertman [87] and Ashby [88], are a graphical
description of creep, representing the ranges in which the various deformation modes are
rate-controlling steps in the stress versus temperature space. Figure 2.17 shows a simple
deformation map for a pure polycrystalline metal [89]. The boundaries are formed by
seting two equations for different creep mechanisms equal to each other.

In general, the dislocation and diffusion creep are considered as an independent

process, and both of them contribute to the overall creep rate. Under different stress and

44

temperature conditions, one of these processes will dominant in creep deformation. This
creep process is defined as dominant mechanism, and a different dominant creep
mechanism results in different values of stress exponent n and activation energy QC as

indicated in Table 2.2 [90].

45

10

o/E

10

10

-8

10

0

 

 

fl , Ideal Strength

Dislocation Glide

Dislocation Creep

 

 

Coble Creep #
Nabarro
-Herring
Creep
I L l L I
0.2 0.4 0.6 0.8 1.0
T/T

Figure 2.17 Schematic version of a deformation map for a polycrystalline metal
illustrating the stress/temperature regimes within which dislocation and
diffusional creep processes may be expected to be dominant [89].

46

Table 2.2 Different and QC values associated with dislocation and diffusion creep

occruingwith pure metals [90].

 

 

 

 

 

 

 

 

 

 

Creep process Temperature Stress n values QC value
high temperature above Intermediate/
. . . >3 ~er)
dislocation creep ~0.7Tm high
low temperature ~04 to intermediate/
. . , >3 QCORE
dislocatron creep 0.7Tm high
high temperature
diffusion creep above
low ~1 051)
(N abarro- ~0.7Tm
Herring)
low temperature
~04 to
diffusion creep low ~1 QGB
0.7Tm
(Coble creep)

 

Q50: the activation energy for self diffusion

QCORE: the activation energy for self diffusion along dislocation core,
ofien termed pipe diffusion
Q03: the activation energy for grain boundary diffusion

47

 

CHAPTER 3

OVERVIEW OF TITANIUM ALUMINIDES

3.1 INTRODUCTION

Titanium Aluminide intermetallic compounds have been studied and developed
for use as structural materials at high temperatures, such as heat engines, turbine engine
blades, aerospace and automobile components, since it has a high specific strength at high
temperatures [91-93]. From the comparison of properties provided in Table 3.1 [94,95] it
is evident that near gamma TiAl alloys are better than current Titanium alloys and
superalloys in terms of density, oxidation, and creep properties. The melting temperatures
of the two materials are essentially the same and y-TiAl has a distinct advantage when
comparing density, coefficient of thermal expansion (CTE), and thermal conductivity, but
it has a lower yield strength, ultimate tensile strength and room temperature ductility.
Having lower density, CTE, and thermal conductivity are all advantages in the design of
jet engines since they give light weight components, minimum thermal fatigue and allows
higher operating temperatures with external cooling.

In the first part of chapter three, titanitun aluminides will be reviewed in terms of
crystal structure, microstructure, deformation mode, deformation twinning, creep
behavior, and alloying effects. Then a proposed research plan will be presented with the
research goal of determining the effect of heat treatments on microstructures and primary

creep deformation of four investment cast titanium aluminide alloys, and to

48

 

i 785

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

: mm £36268 Rene:
. . 03b : eeaeeaxm
w 3 w 2 BEEF—1.3 3205000
92: NTm 2 A15 H :30 - 5:325
21m m; 2 exit Eoom - cat—:25
as: m a 0% Ge Susie - s 82
co? coo 9% 66V @020- .522
SN e: o. _ EB sanctum
82 825m 33: £325 29:2:
o3 28sz 32v eueesm 26;
69 Rd 2:: season
e8 ow 22 EB 33662 when;
3 $22 $4 $35
. . . 8. 78 as none
33:38:” N: + xEmE o o m EEHEVA 13.9.: 3:8meon 882:6 Embo
8: 9.: Ge ass mega:
3:833 omen: Z 2:.-.» their? Hanoi

 

 

 

 

 

$6.43 6:236 nee 6% 5:58 5? 2E eseao we eseaseo 2 2.3

49

compare primary creep deformation in these alloys with creep in polysynthetically

twinned (PST) Crystals.

3.2 TITANIUM ALUMINIDES
3.2.1 Crystal Structure and Microstructure

In general, there are two basic lattice structures (phases) found in Ti-Al alloy
system near the stoichiometric composition in Figure 3.1[96,97]. One is the 7 phase of
TiAl and the other is 012 phase of Ti3Al. As indicated in Figure 3.2.a [98], the y-TiAl has
L10 lattice crystal structure where the titanium and aluminum atoms alternately stack in
the sequence of ABCABC on (002) planes and it is a face-centered tetragonal structure in
which c/a ratio is 1.02. The anistropic crystal structure causes more complicated slip
system in the L10 lattice [99]. From Figure 3.2.b [100], the 012 (Ti3Al) has D019 crystal
lattice structure where the stacking sequence is ABABAB and it is the hexagonal closed-
packed structure (HCP) with a larger value of basal plane axis (2a). The system is often
off-stoichiometric with Al atoms in Ti sites and therefore, this results in a slightly
disordered lattice structure.

TiAl has a wide composition range, but mainly on the Al-rich side of the
stoichiometric composition as shown in Figure 3.1. The microstructure of Ti-rich
contains a large volume fraction of Ti3Al (012 phase) and a small volume fraction of TiAl
(7 phase). Therefore, as Al concentration increase, the volume fraction of TiAl increases
and the volume fraction of Ti3Al decreases in the equilibrium condition. With higher
concentration of Al (> 50 at% ), an equiaxed type of microstructure with single phase (y

phase) can be obtained. In Ti-rich two-phase microstructures, the volume fraction of each

50

Temperature ( 3C )

Weight%ofAl
20 25 30 35 40 45 5

 

 

1600 T L
r
1400
L
- NL
1200 Duplex
NG
1000

   
    

800

 

 

 

I I 1

30 40 4547 50
Atomic % of Al

 

Figure 3.1. Phase diagram of binary Ti-Al system [96,97].

51

 

 

 

 

 

 

 

Figure 3.2 Lattice structure of (a) Llo-TiAl and (b) DOtg-Ti3Al [98,100].

52

phase depends on composition and processing history. Upon solidification of
stoichiometric or Ti-rich composition, a two phase microstructure having a lamellar
microstructure develops from the or phase. Three types of lamellar structures, called type

I, II and IH (see below), have been identified through several microstructural development

studies [48,49,101].

TYPE I: a—>a+yl,—)L(a/y)—>L(a2/y)

TYPEII: a—>a2—->a2+y,,, -—>a,+7,,—>L(a2/y)

ht'ulmg con/mg

TYPE III: 7m +a2"——9ym +011) “’7’”. +a,, —>L(y/a)————) L(a2/y)

ym—W—Waym +a,,, —>ym +a,, —->L(y/a)—""+””'15'—>L(a2 /y)

The type I and II lamellar structure is typical formed by the growth of y (TiAl) plates into
the or or 012 (Ti3Al) phase in two phase regions such as (1+7 and 012+y (see Figure 3.1). The
transformation proceeds by TiAl plates that form on the basal plane of the Ti3Al and then
thicken at the expense of the Ti3Al phase. This phase transformation causes the lamellar
structure, composed of alternating laths of TiAl and Ti3Al phases. The orientation
relationships between the y (TiAl) phase and the 012 (Ti3Al) phase are {11111141 //
(000011311, <110>T1Al // <1120>mm as shown in Figure 3.3 [102,103]. The type III
lamellar structure is formed by Ti3Al plate growing into TiAl phase. This lamellar
structure has the same orientation relationship between the TiAl and Ti3Al as type I and

II. However, since all <1120> directions in Ti3Al are equivalent to one

53

[110] w [A [1710] _°‘

/
4

 

    

C . B 1’)
______ .7 f" 7‘ _
[011] [101] [1120] [2110]
TiAl Ti3A1
(111) (0001)

Figure 3.3. (110)}! directions and the atomic arrangement on a (11 1) plane in L10 and

<1120>a2 directions and the atomic arrangement on the basal plane (0001) in

D019 [103].
[130lr/<1170>ae , , *- V [0'1'1]y/<112'0>a,

Variant A1 f . Variant F
[10T]y/<Il20>or, i -. 2W 2 1 2 _ i g g [HEW/(0001201._%[I01]y/<1120>or,
Variant D , Variant C
[01I]y/<11f0>ot,~ , - , ‘1T1011/<1120>a.
Variant E Variant B

Figure 3.4. The stereographic projection of (l 1 l)y and (0001)ar2 show the six possible
orientation variants of (110)}! with respect to <1120>a2 matrix [18].

54

another (see Figure 3.3), y plates have the same crystal orientations within a grain in the
type HI, and y plates crystal orientations in the type I and II vary from plate to plate. The
TiAl lamellae are often twinned. Yamaguchi showed that there are six possible
orientation variants corresponding to the six possible orientations of [110] on (111) in the
gamma phase with respect to <1120> on (0001) in the 012 phase as shown in Figure 3.4
[104-106]. TiAl/TiAl or TiAl/Ti3Al/TiAl lamellar interfaces with the {111}<112] true
twinning relation are found with greater frequency than those with orientation
relationships other than the true twinning [107]. True twinning-type lamellar interfaces
are more energetically favorable than interfaces of other types during the or —> y
transformation. Domains within a gamma plate can have any of the 6 orientations. The
factors controlling the lamellar structure are (1) thickness of TiAl lamellae, (2) size of
domains in TiAl lamellae, (3) the volume fraction, distribution and thickness of Ti3Al
lamellae, (4) size of equiaxed gamma grains and lamellar grains, (5) the volume fraction
and distribution of lamellar grains [108,109]. These factors are closely related to strength,
ductility, and creep properties in TiAl with Ti-rich and near stoichiometric compositions.
By processing and heat treatment conditions various microstructures in gamma
alloys can be achieved. Four types of microstructures have been classified: i) near-
gamma, ii) duplex, iii) nearly lamellar, and iv) fully lamellar microstructures [110]. The
near-y structure is obtained by an annealing heat treatment at temperatures just above the
eutectoid temperature (see Figure 3.1) after a hot-working process and is normally
characterized by coarse gamma grains and banded regions consisting of fine 17 and 012
grains [110]. The fine duplex microstructure, which is composed of fine 7 grains, fine

lamellar grains and some fine 012 particles, can be obtained through a combination of hot-

55

working and subsequent heat treatment at temperatures where the volume fractions of
equiaxed and lamellar microstructures are roughly equal. The nearly lamellar and fully
lamellar microstructures are composed of coarse lamellar grains and a small volume
fraction of fine 7 grains that can be obtained by heat treatment of as-cast ingots near the
alpha transus [110]. A typical near gamma, duplex and fully lamellar microstructure of
TiAl alloys are shown in Figure 3.5 [111,112].

In zone melting growth of crystals of TiAl with a nearly stoichiometric
composition, if the crystal growth rate is appropriate, crystals consisting of a single set of
lamellae can be grown [113]. These crystals are called as polysynthetically twinned
crystals or PST crystals, since as-grown crystals contain numbers of thin parallel twin
lamellae having only six orientations. PST crystals can be used to investigate the nature
of deformation behavior in lamellar TiAl microstructures with a single lamellar

orientation.

3.2.2 Dislocation Structure and Deformation Mode

As mentioned before, the ordered intermetallic compound TiAl has a face
centered tetragonal structure (f.c.t.: an Llo-type superlattice). It consists of alternating
titanium and aluminum (002) planes stacked normal to the c-axis as shown in Figure
3.2.a. At the stoichiometric composition, the c/a ratio is 1.02, and tetragonality increases
up to 1.03 as aluminum concentration increase and decreases to 1.01 with decreasing
aluminum concentration [114-116]. As indicated in the Figure 3.2.a the slip planes in
gamma TiAl are the {11]} close packed planes and the possible slip directions are <110],

<112] and <101]. The nomenclature <101] indicateds that the first two indices can be

56

 

Figure 3.5. (a) duplex, (b) fully lamellar, (c) near-garnma microstructures in TiAl alloys
[111,112].

57

reversed to get a similar crystallographic direction, but the third indice must remain in the
third position. The 1/2<110] dislocations in this slip system are called ordinary
dislocations. <101] dislocations are superdislocations because they can be dissociated
into two l/2<101] super-partial dislocations. 1/2<112] dislocations are superdislocations
whose Burgers vector is the second shortest of the Burgers vectors lying on {111} planes.

Twinning is an important deformation mechanism in TiAl. Only true twinning on
{111}<112] can occur since the L10 structure is no longer preserved by twinning on
{111}<12I] [117-119]. Some studies indicated that the mechanical twins nucleated at
grain boundaries of triple points and propagated by emission of 1/6<112] partial
dislocations [120-122]. Unlike normal or super dislocations, a twin leaves a stacking fault
behind that is as long as the twinning dislocation has traveled as shown in Figure 3.6
[123]. Thus, the microstructure is refined by increasing the area of twin interfaces as
twinning deformation proceeds. Shechtman et al. reported that TiAISF (SF: stacking
faults) form at an early stage of deformation and develop into twins [124]. Since TiAlgp
and Ti3Alsp have the equivalent stacking sequence to the Ti3Al structure (D019) and TiAl
(L10) structure, respectively, this suggests that TiAlsp could act as the nucleation sites for
the Ti3Al and Ti3Alsp be source of nucleation of TiAlme/TiAlTw,n [124-126]. This leads
to the formation of the Ti3Al layers between TiAlMam and TiAlthn during deformation
[125,126]. The sequence of formation of Ti3Al layer between TiAlmam and TiAlmn are
shown schematically in Figure 3.7 [125]. These observations suggest that Ti3Al can be
formed by mechanical twinning during deformation. Jin and Seo also showed that 012

(Ti3Al) layers form between deformation twins after creep deformation [47,127].

58

 

    
 

1

 

 

 

Twinning
direction

Twinning
dislocation lines

 
  
 
  

Twinning
planes

 

Twinning
direction - Incoherent twin interface
Coherent twin interface

(a)

[111] Twinning direction [112] Stacking sequence
+ ' ’ ' ’ ' 1 After twinning Before twinning
I . O . C 4 -- B
- . - . 1 . ‘. . B -l A
110 ' .W2
1 1‘ +1112] 0 o x o M t . A >- c
, - +0 0 .‘ o , laJ'xc' 1 B
(111) . , 116L112]. I-LJ. ,, . . . . As his A
111 1/6 112 . . .

1m) ------ ll”; >0 0 o Twm B >77 c
1 ,7) 2 ,l/6[1712] ,2 , . . C. L, B
(111) 7 71/61112L..L . . 7 L, A ,2, A
. . . . . M atrix C A C
o o o ' B L, B
o o o A e— A

(b)

Figure 3.6. The thin twin morphology near the twin tip [123].

59

[M l

i

l TiAl lTiAlsp

i

T i 3A1 lTiAlSF

Irirl i
.3 I .3 . SF

Ti3Al lTiaAlgrl T iAlr lTigAngThAl lTiAlsF

Figure 3.7. The formation sequence of TiAlgF, Ti3Al, Ti3AlSF and TiAlT due to high
temperature deformation [ l 25].

 

TiAl l

 

 

l TiAl l’l’iAlsp TiAl I

 

 

l TiAl I'l‘iAlsp TiAl l

 

 

I TiAl I'l'iAlsp TiAl I

60

From several TEM studies [124,128-132], ordinary 1/2<110] dislocations are
much more active than <101] super dislocations up to 700 °C. The motion of many of
<101] super dislocations has been reported to be impeded by the pinning of the trailing
1/6<112] partial by an intrinsic pinning mechanism. The pinning of <101]
superdislocation makes them sessile dissociations due to forming superlattice intrinsic
stacking fault (SISF) and superlattice extrinsic stacking fault (SESF) which leads to
formation of faulted dipoles and loops [129,130]. These dipoles and loops are commonly
observed not only in Al-rich binary compounds, but also in Al-rich ternary TiAl alloys
[132].

The most significant differences in deformation mode between TiAl coexisting
with Ti3Al and Al-rich TiAl, are the amount of twinning activity and the difference in
mobility of 1/2<110] dislocations. Twinning on {111}<11§] systems in Al-rich TiAl

becomes the most active mode of deformation only at temperatures higher than 800 °C,

while the {111}<11§] twinning is active even at room temperature in the Ti-rich two-
phase compounds. Although dislocations with 1/2<110] are observed in the single phase
(Al-rich) compounds and in the two-phase compounds (Ti-rich), Vasudevan et al.[l33]
indicated that the increase in mobility of 1/2<110] normal dislocations that are observed
in Ti-rich TiAl arise from fewer impurities or interstitials in the 7 matrix. The non-
stoichiometric Ti3Al apparently absorbs interstitials and impurities that interfere with
normal dislocation motion [133], since the solubility of these elements in Ti3Al is much
larger than in TiAl. This is one reason why two phase compositions have better ductility.
Thus the presence of the a; phase in TiAl has an important effect on deformation

mechanisms.

61

The anomalous hardening behavior in single crystal and polycrystalline forms of
single phase Al rich gamma Ti-Al alloys have been investigated by several researchers
[134-140]. Three regimes are characterized by soft and hard mode deformation, and the
data between (273 K~1073 K) is identified with the anomalous hardening. Both <101]
superdislocations and l/2<110] ordinary dislocations exhibit anomalous hardening as
well as orientation dependence. This anomalous hardening can be explained by the
dissociation process and cross-slip reaction of <101] type superdislocations from primary
slip plane onto other octahedral slip planes below 650 K (an octahedral lock mechanism),
whereas between 650 K and 1073 K, cross slip occurs from the primary octahedral plane
onto a cube plane [135-140]. The nature of hardening by ordinary dislocation is not
understood yet. However, the pinning point and cusp formation mechanism and a local
pile-up is probably due to the presence of immobile dislocations which lie in multiple slip
planes as suggested by Viguier [137,138] and Whang [135].

The effect of crystal orientation on deformation mode was examined by Bieler and
Seo using an analysis of the Schmid factor for <110] ordinary dislocations, <101] super
dislocations, and true <112] twinning deformation systems as shown in Figure 3.8 [141].
In tension, orientations close to <441] (near tail of arrow) are likely to twin. They
indicated that in the twinned orientation, the Schmid factor for ordinary dislocations is
small, and only superdislocations have a high Schmid factor, so twinning increases the
volume fraction of crystals where only super dislocations can operate. Luster and Morris
observed that superdislocations were active only in crystals having a Schmid factor
greater than 0.35, and they only nucleated from interfaces where efficient slip transfer

from neighboring crystals was possible [142]. Ordinary dislocations and true twinning

62

were observed in grains where the Schmid factor was larger than 0.25. From these criteria
indicated above, deformation in tension has only 19 vol% of a set of randomly oriented
crystals with 4 Schmid factors operative, while in compression, 28 vol % have 4 systems
active, as described in Figure 3.8.b.

From deformation behavior of PST crystals studied by Yamaguchi, Inui, and
colleagues [143-147], yield strength and tensile elongation at the room and high
temperature strongly depend on the angle (1), where 4) is the angle between lamellar planes
and the loading axis. When the lamellar boundaries are parallel or perpendicular to the
loading axis, the yield stress is high, but it is very low for intermediate orientations as
described in Figure 3.9 [147]. For parallel or perpendicular orientations, shear
deformation proceeds on {111} planes that intersect the lamellar boundaries (the hard—
type of deformation). For intermediate orientations, shear deformation occurs parallel to
the lamellar boundaries (easy-type of deformation) and hence large strains can be
obtained. Thus the strength of TiAl with lamellar structure is thought to be related to the
hard mode of deformation, i.e. shear deformation across the lamellar boundaries, and its
ductility must be associated with the easy mode of deformation, i.e. shear deformation
parallel to the lamellar planes. At room temperature, deformation of PST crystals occurs
by {111}<112] ordered twinning and slip on {111}<110]. These twinning and slip
systems are still operative at high temperatures. However, the tendency for deformation
twinning was found to decrease at high temperatures [144]. Few <101] superdislocations
are observed in PST crystals of TiAl deformed at room temperature, while at 800 °C
<101] super dislocations are frequently observed. This is indicative of an increased

activity of <101] slip at high temperatures. Dislocations with 1/2<112] are found in PST

63

    
    

“0 <ioiland (211]
Superdislocations

Ordinary <1 10]
Dislocations ‘

 
 
 
 
 

Schm i d
Factors

True Twinning

True Twinning a 6?] Compression /.r”\\

e.......__-._..-..._..._.... . .N

Number of <1 10] and (1)) Number of <1 10] and

(112] systems (112] systems 1 10
with m>0.25 i
Tension

199424 ‘

    

    
   

:100

Figure 3.8. (a) The maximum Schmid factor for deformation in a given crystal direction
is plotted for ordinary and super dislocations, and for true twinning, which
depends on deformation mode. (b) There are no operative ordinary dislocation
or true twinning systems near <001] for tension, and none near <110] for
compression [141].

64

 

 

 

 

 

E600— ; a
a L _j-
_ A
(I) ,
(I) if: N
a 400 — . 2
E—‘ {g g .. ,
Q T X l , ,
5 200 - A — - ~ x
_ x—v—e a:
.9 A D
O 1 Bl C l l
0 30 60 90

4) (deg)

Figure 3.9. Yield stress of PST crystals as a function of angle between compression axis
and the lamellar boundaries [147].

65

crystals deformed at room and high temperatures, however their contribution to

deformation is not high even at 800 °C. The activity of all these dislocations depend

highly on the orientation of given domain with respect to the stress axis [145,146,148].
From this review of deformation behavior in Al-rich, Ti-rich 2-phase, and PST

crystals (lamellar structure), possible deformation systems are l/2<110] ordinary

dislocations, deformation twinning with l/6<112] dislocations, <101] super dislocations,
and 1/2<112] super dislocations. The dominant deformation system strongly depends on
the concentration of Al, the presence of Ti3Al phase, service temperatures, domain
orientations, stress level, and lamellar plane orientations during plastic deformation. The
deformation behavior at high temperature with constant loads for a long term, that is
creep deformation, is not as comprehensively explored in the literature, as the next

section shows.

3.2.3 Creep Behavior in Gamma Titanium Aluminides

Creep is commonly analyzed using an empirical equation which is relates the
stress and creep strain rate by é = A0" exp— (Q/ RT) where A is constant depending on

microstructure, Q is the activation energy of the rate limiting process, and n is the stress
exponent which has particular values that are commonly associated with well described
mechanisms. Creep in gamma titanium aluminides strongly depends upon microstructure
and alloying compositions. From these two points of view, creep behavior in gamma
titanium aluminides is reviewed based on three categories: (1) Al-rich single phase or Ti-
rich near equiaxed gamma Titanium aluminides, (2) Ti-rich lamellar gamma Titanium

aluminides, (3) Ti rich duplex gamma Titanium aluminides.

66

3.2.3.1 Al-rich single phase gamma Titanium aluminides

The creep behavior in these alloys shows different behavior under high,
intermediate, and low stresses as shown in Figure 3.10 [149]. The stress exponent,
n=dlogé mm/dlogcr, is the slope in Figure 3.10. For Ti-50-53Al, it changes from 3.5 to 7.5
to 4.7 with increasing 0'. Creep in TiAl commonly exhibits a minimum strain rate at a
strain below 5%, which is used in the empirical equation. Oikawa [150] reported that in
the high stress regime (251 MPa), the activation energy for creep varies between 350-600
kJ/mole for different stress ranges, which is much larger than the 150 kJ/mole value for
inter-diffusion reported by Ouchi. Lu and Hemker also reported that activation energy for
creep in single phase Ti-51Al-2Mn is about 440i15 kJ/mol which can be ascribed to
either diffusion-related climb or the thermally activated motion of ordinary dislocations
[151]. He mentioned that the fraction of grains containing twins increase with creep strain

at 550 °C under an applied stress of 550 MPa.

3.2.3.2 Ti rich fully lamellar gamma Titanium aluminides

The creep behavior of forged Ti-49Al-1V-0.120-0.07C (at%) with three different
microstructures was investigated by Worth et al. [152]. Creep strain at 760—815 °C in air
with a constant load of 87.5 MPa is highest in the duplex microstructure, intermediate in
the near equiaxed gamma microstructure, and lowest in the fully lamellar microstructure
for identical conditions. The stress exponents were found to be 3 to 5 for all
microstructures, indicating that possible creep mechanism under this condition is
dislocation controlled. Most creep studies on Ti-rich TiAl have been focused on the fully

lamellar microstructure. Since they provide the best creep resistance as shown in Figure

67

3.10 [149,153,158,159,160]. Creep of fully lamellar Ti-48Al alloys was studied by J.
Triantafrllou et al. [153] at 760 °C in air under 240 MPa using stress increment tests. In
Figure 3.10 the stress increment tests for fully lamellar structures exhibit an increasing
stress exponent with increasing stress from n=1 at low stresses to n=10 at high stresses,
suggesting a transition from diffusional controlled creep at low stress to dislocation climb
controlled creep or power law breakdown (dislocation glide creep) at high stress. He also
suggested that the reduction of lamellar spacing may be the microstructural parameter
controlling dislocation creep as evidenced by the bowing of dislocations between adjacent
lamellae and leading to a lower creep strain rate as indicated in Figure 3.10. In Ti-48Al-
2W alloys with ~330 um lamellar grain size, a similar stress dependence on the stress
exponent was found (n=0.6-8.2) [154]. In a ternary Ti-48Al-2Cr alloy crept at 800 °C in
air under stresses of 145 to 330 MPa for 100 to 500 hours, the lamellae were deformed by
twinning and dislocations which consist of tangled screw dislocations and curved
ordinary dislocations, but the contribution of superdislocations appears to be marginal
[155,156]. The creep deformed microstructure consisted of subgrain boundaries found in
the equiaxed grains which is thought to trigger dynamic recrystallization observed at the
tertiary creep stage.

The effect of grain size on creep behavior of fully lamellar Ti-48Al-2Cr-2Nb was
studied using uniaxial compression tests at 850 °C with a true stress of 250 MPa [157].
The creep rate of the fully lamellar gamma TiAl alloy is independent of grain size in the
region where dislocation creep is the dominant deformation mechanism. The major

microstrucutral variable influencing the creep resistance of a fully transformed gamma

68

 

10 r ' ‘ I T T] Y T l i r Tfi
a
.,

 

   
  
   
  

 

 

 

 

 

 

 

 

: l
: A Ti-SOAl] Ref. 149 Eqwaxedv
" I Ti-SlAl
4 0 Ti-53Al 827°C
10 g' +Ti-48Al(0.12nm) _— """""""""""""""""""""""""""""""""""""""""""""" g
: +Ti-48Al(0.45um) Ref 133 .
3;; ‘ +Ti-48Al-2Cr-2Nb
\ -5 F ' -— Hofmann & Blum
C, 10 g — — Wheeler et al.
Q) C ""‘"" Soboy ejo & Lederich —
a t =
z...
c: 10.5 r
.8 F:
L": t
m t
E 10-7 5‘
E
.a 10'8 E‘
S 5 Fully lamellar
10'9 F ......................................................
F Stree exponent i
1040 r . i r . r .
100 1000
Stress (MPa)

Figure 3.10. Comparison of steady state creep data of TiAl with different microstructures
[149,153,158,159,160].

69

TiAl alloy is the (1.2/Y lath spacing. Hoffmann and Blum have reported that the lamellar
microstructure transforms dynamically into a globular one in many parts of the sample
with softening in Ti-48Al-2Cr-2Nb at high temperatures [158].

Nieh and Wang studied creep behavior of Ti—47Al-2Cr-2Nb alloy with a fine
grained fully lamellar structure at 760 °C, 140 MPa single or multiple stress increases
[159,160]. For a given temperature, the higher the applied stress, the greater is the
primary creep strain. Creep data from Nieh suggested that the creep of the alloy at 760 °C
may be glide-controlled in the high stress region (>250 MPa) but recovery-controlled in
the low stress region (<250 MPa). When deformed subsequently in the low stress region,
creep would not commence for a certain period of time. This prestraining effect is
believed to be related to a slow recovery process which controls the deformation in the
low stress region. They suggested that that prestraining technique could be used
effectively to reduce primary creep of TiAl alloys for engineering applications.

Huang and Kim [161] investigated creep in Ti-47Al-1Cr-1V-2.5Nb alloy at
nominal stresses of 138, 207 and 276 MPa in air. The stress exponents range from 4.5 to
5.7. The results indicated that deformation took place in almost all cases in the gamma

phase only and the (12 plates were almost inact. A dislocation cell structure is also

developed in gamma grains during deformation which led to dynamic recrystalization.

3.2.3.3 Ti rich duplex gamma Titanium aluminides
Creep in two phase gamma Titanium aluminides with duplex microstructure has
not been investigated as much as in TiAl with lamellar microstructure [47,152,157,162-

164], since the creep resistance of the duplex microstructure is always inferior to that of

70

lamellar microstructure. From the study of microstructural effects on creep in Ti-48Al-
2Cr-2Nb, Ti-49Al-1V and Ti-48Al-1V-0.3C, the creep resistance with a fully lamellar
structure was higher than that with a duplex microstructure [152,157,164]. This suggests
that the decrease in creep resistance of the duplex structure might be due to an increase in
dislocation mobility in the gamma phase.

Bieler et al. [165] showed that stress exponents in Ti-47.4Al-2.2Cr-l .9Nb are 3 in
the low stress region (<200 MPa), and 8 in the high stress region (> 200 MPa). In
compression+shear mode, creep in Ti-47Al-2Cr-2Nb, the stress exponent 3 at 760~774
°C and the activation energy 257 kJ/mole (in Ar) and 574 kJ/mole (in air) were found by
stress and temperature change experiments. The low stress exponent and the high
activation energy suggest that dynamic recrystalization occurs, and dislocation networks
form, but glide is rate limiting [166].

The differences in creep behaviors between Al-rich single phase TiAl and Ti-rich
two phase TiAl alloys can be easily seen in Figure 3.10. Al-rich alloys creep much faster,
and well defined transitions are evident. There are two transition points in the stress
exponent in Al-rich TiAl. On the contrary, in Ti-rich TiAl alloys, the stress exponent
increases smoothly as stress increases until power law breakdown region is reached.
Although some creep mechanisms were suggested, no systematic view in the creep
mechanisms has been developed, and the role of deformation twinning has not been
examined, other than in the work of Jin [47,120,121]. The creep properties of several

TiAl alloys having different compositions and microstructures are summarized in Table

3.2 [138,152,154,155,l58,159,16l,162-l64,167-189].

71

 

 

 

 

 

 

 

 

05m £me 9:: 9.le wamz 81 on Tmmuxfi 380 558E800
NS $6385 5;» 3.0.3.526 E: V nma Ac 0.8 V;
as gusto sesame A .2 e g » eases. 3, E 8 a.
.5 8m A53 Rm: o a 83:3 was compass
at 8o 3; Re 978 3.32: v are
81 8789:28 $602: a
«.2 8m 4 cow. New wagons Emacs Ba
x. + 3:083 2§<m.om-fi.
3% “88559838 ace R” an 8:858 is 683
”2 2: Ea Buaowoeo: am 3 z z . a
a 320 atom Re 2: €28 a a + 3:052 -c.
:5. 84 E §_o§_
$2 Defiance 52: 25 NR 2 ET 8? age so swam 23083249
829285 a was Sn 2: -fi
320 EEEEE nonwE m 3 mg. 893 > Bwamm
“WWW MMMHW coo w haw EN @098 commmanoo
_ a c z z a w - . 2
3 sense Ea 823836 8.7 m2? Re ME 32 2 am 5: _< Eases? mm o3
. . . o2 -fl
mo 28:0 boson:
@3855 Gov 3&5
mom “5888 O s 95H mmobm 5530905555822 AXBKEE

 

 

 

 

 

 

 

 

.5_-§.§-§.6&2.

 

w2.mfl.§.~2.mm: £21.. 2:. S 5% 320 mm 03¢

72

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

w: Ammzwgmv Goons 3% an? m. :82 535 6.8:,8538
:1? 6088 h v 2: 32 -255
398.3: Auoomt 3»
2: 6033 m
om m 6&3 3 m _ a: E? Amssoooomuflmamboomm :
E 8” won 2: 830255 321 H
E Sam; we: 32 :Az 32 magooooa .
3 wen we 8%0255
Km? 81 27.6 88:32:88
0: 5 ea oomzfi 8 88:3 -nzm.m-_<m.$fi
votomno Mauvvmfi
0 Q
s . a 3:83 £8 . .
a mg: mmfi g 5335:? ”3%
. UK: 78: a 898 .
80:82me a? M
£83 o3
. 825 8 26 % 8 owcmno o: 380
m: h. m coo oomzom @8825 UooowVH 868888
c v own 282% means: 8 magma 83¢
on a 83 Roz 8? a: £78ch 320
.5 8w 9 8: 5 cm 8 3353 530588
. 88063-:
@080
m2 own Ab 5w v Sol 81 ovzmmnxu 80588800
5 8 ,9; 878 g Bfisfi . . -
E :8 3 :.L

 

$83 2 28

 

73

 

.8 8 2a
aofiEeEm

 

 

 

 

 

 

 

 

ofifiogfifiv @8825 030 -m E? .59, 3+
.3 BEBEBEV Ab a o? Ann—ZoomAv 5 fl 32 v a < 2%-?
. . Ammwéoomzowvv oooz
Q: 3805 £53 a G 6 RM 9% cm
. Aameéowvv_
a? 02a 8:828
.3 d 8:268 8898-3 Nae 80> m+
coast? V35 388-3 13:13
988
a: 08 3 8» cm 33% 8832888
528?:
8038036
8888 833 m 08 83%
cam 88:32qu 8 M8 8 cmmz > 32903:? 3 8 - -—
.oM: 380m @0388 28 8% ““8an m . cow 9: 398085 08m 88m UN 2: .H
m: “was?“ 3 883% m a _ o s 8828335
£05883 £8838
328$ 3 E :13 8a: 8?. 8.58% .
.— u u—
o: mom 8 d m.m2m.~ 2: E 42 um _ 2% .9
Goon” 3 c3: 2:2 BEE
z .- .;
$8 o3 own 3 own 3: 8+§u§xo+23 E m .5 M3 p
86305 .
m
V: as 533 £88888 W ”W Wm o3 mm”, d 26. _-_<v.$-fi
Q88 mo 528.8
820 c3 38
SECQ o 880 was . z 2 xo a: .. 8
N2 . M H . v 3m 2K «.8: _ a 928 28x9

 

3:888 @820

 

 

 

 

 

 

 

 

6.23 an 23

74

 

c288 5332me

33.8 2:

 

 

 

 

 

 

 

_ . .
n: 3% 2: 292 38.8 m 8w 8N d Mm” fimw
”25 mafi :5 33 Q to .
onwz ommz 83%: mmobméocsbxoiv - - -
E: omm 3m .2: oNEoonofcomEtowE nZN HUN 1&3 C.
commumamboe
3:383 3
23026 52:an «m2 ommzoofi hmaz 032 E: ommzoMnc L - ~
mm“ 05 85 23035 8m me haw com um nZN Um _<wv..,~
Egg :2 2:
mo :ofiwESMmfiah 05
832% nwE E @080
8:803 £883 EV a; Ev a? 8? $333982 296 -2539
m2 32 an Q88 scumbag 3v com 3v M mom .2: \ Em onmemQEE n m .
3563-325 _ Q
@083 S
2 z 81 n
a: omozem 6°86 S o%%u 2% xommau pz~¢om-_<$¢
Scoot fl:
:EENESmboe
03888 98 .
a: gauge 23 am WWW“ ET 2? 35 pz~-:2m-_<$¢
.2052: 5an o . och vdo 3:25— 3:me mam .
a=o§_ 2: €23 6085 w v
$.53 30% @3232
8&8
35w gouge—m6 Ami 8m Wm m $2
§ fiasnéoésgflm 8&8 o3 o: d £296 s 33:3 om.o->_-_<$¢

 

oogmmme gouge

 

 

>90— HOM N6

 

 

 

 

 

 

 

75

 

Hmon—

 

 

 

 

 

 

 

 

commfibu SEE Am a v o3 7 0528095 .81 mnw own—Eu 88 :85
o3 3 3:838 mam oomzowm Am d m coo" oovzmm voxflsco :ommmoaaoo
\vafi
a: 5% 9.95 ESE ca 3.2 9&2 _ xzae Emboeoo: mmޤo>v$>~
:25 be, a £25 608%; CE Eguooofiéougfi -2213
E .
. . azm m
E w m 25 mm. d ->_.§-_<$fi
32 88
. £7
a: 35:8 8% a do fl can we §_o§_ £3 €82 nz~¢uN-_<$¢
3:2 cmmvv a E n .2; .
60:05:00 DEED
fizzmfitma
a: o3 6&3 OE wwwm 8255 mm 33>?
22¢
:1 ”Tana N
I u o o>
:3: 32m ooo%o mm 7: 83:3 mwxmwfi 0+
2: SS .
u . So: a: EU -
Rafi. m _ 25 @223 @9338 2a 2: E
a 81 . 38o
2: A nuance 3 cow: 8? ushwwmewm 538558
N u a: u m m \2 :1 H

 

 

 

 

 

 

 

 

 

ease mm 2an

76

3.2.4 Precipitates and Alloying Additions

The alloying effects observed in gamma alloys are described in Table 3.3 [190].
The most important alloying elements for creep deformation are Al, Si, W, Ta, B, C, and
0. With less Al, more (12 phase is produced, which results in poorer creep resistance [40].
The formation of silicides by the addition of Si in Ti-47Al-1Cr-an-l .5Nb-O.2 alloys led
to high creep strength [191]. In Ti-48Al-1V containing C, the creep resistance is higher in
the high carbon alloy than those of the low carbon alloys, and the influence of carbon on
creep is pronounced in the equiaxed and duplex microstructures [164]. The addition of W
and B refined the microstructure and reduced the defects such as low angle grain
boundaries, misoriented and kinked lamellae which is related to thermal instability prior
to aging [192]. In general, the precipitates act as barriers to dislocation gliding as well as
dislocation rearrangement. The interface between the matrix and precipitates act as sinks

or sources for vacancies.

3.2.5 Primary creep behavior in TiAl

The primary creep deformation may be defined as a transient deformation and an
alloy hardening period before reaching the minimum creep rate. Compared with
secondary (or minimum) or tertiary creep studies in TiAl, few studies concerning primary
creep behavior in TiAl alloy have been conducted. Here are some results of primary creep
behavior from the past investigations [47,152,] 55,164,193,194,195]: See and Bieler et al.
reported that transition point of creep curve (indicated by an arrow) was observed at the
very early stage of primary creep (~0.25% strain) with different stress and temperature

conditions as shown in Figure 3.11 [195]. Hsiung and Nieh [193] investigated the

77

Table 3.3 Alloying effects in gamma TiAl alloys [190]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Elements Reported effects
Al It strongly affects ductility by changing the microstructure. Best ductility
occurs in the range of 46—50 at%
B Add >0.5 at% refines grain size, and improve strength and workability.
C Carbon-doping increases creep resistance and reduces ductility
Add 1-3 at% increases the ductility of duplex alloys
Cr Add >2 at% enhances the workability and superplasticity
Add >8 at% greatly improves the oxidation resistance
Er Changes the deformation substructures and increase the ductility of
single-phase gamma
Fe Increases fluidity, but also the susceptibility to hot cracking
Mn Add 1-3 at% increases the ductility of duplex alloys
Mo Improves the ductility and strength of a fine-grained material
Improves the oxidation resistance
Ni Increases fluidity
Nb Improves greatly the oxidation resistance and creep resistance slightly
P Improves oxidation resistance
Si Add 0.5-1 at% improves the creep and oxidation resistance
Improves fluidity, but reduces the susceptibility to hot cracking
Improves the creep and oxidation resistance
Ta Increases the susceptibility to hot cracking
V Add 1-3 at% increases the ductility of duplex alloys and reduces the
oxidation resistance
Greatly improves the oxidation resistance
W Improves the creep resistance
0 stabilize on phase

 

78

 

deformation structure of crept Ti—47Al-2Cr-2Nb alloy at 760 °C using stresses of 138 and
518 MPa. The deformed structure from a stress of 138 MPa shows dislocations in soft
lamellar grains at interfaces and in {111} slip planes. On the other hand, the deformed
structure from a stress of 518 MPa includes mechanical twins and dense boundary
dislocations in both soft and hard grains. It appears that both twins and boundary
dislocations are barrier to the moving dislocations. Es Souni et al. [155] reported subgrain
boundaries in equiaxed gamma grains in Ti-48Al-2Cr deformed at 800 °C, 145 MPa to a
strain of 0.05, which supports the climb-controlled model. Worth et al. [65,77] also
reported that the dislocations in the lamellar layers were pinned by the lamellar
boundaries in the Ti-49Al-1V alloy deformed at 815 °C, 87.5 MPa to a strain of 0.04. Jin
[47,194] showed that very fine mechanical twins in equiaxed grains triple points and
subboundaries in equiaxed and lamellar grains were observed in Ti-47.4Al-2.2Cr-1.9Nb
alloy deformed to a strain close to the end of primary creep, and that a possible
deformation mechanism is low temperature dislocation climb controlled creep.

The primary creep process in gamma titanium aluminides is not well understood.
The possible creep mechanisms suggested by various authors include recovery controlled,
climb controlled, stress-assisted diffusional creep mechanism at low stresses, and glide-
controlled plus twinning and dislocation creep mechanisms at high stresses. The role of

deformation twinning in primary or secondary creep behavior is still not well described.

79

0006ffi"lf"1I'rifirrrfifiTTfifirr'li
l W-Mo-Si alloy Heat treated (1010°C/50hrs)

 

 

_ 760°C, 138 MPa, 0.5%
""" 649°C, 276 MPa, 0.5%

0.001

1

0.005 Tensile creep ", j

: l

00 0.004 P q
.5 0.003 ’ " 1
§ : a
0.002 Temp, 0, a i

* i

l

J

 

 

 

_l

 

 

0 Alia..i.l .1....1..li
010° 5105 1106 1.5106 2106 2.5106 3106

Time (sec)

L Li LLLPI

 

Figure 3.11. Primary creep curve of TiAl alloys [195].

80

CHAPTER 4

EXPERIMENTAL PROCEDURE

4.1 COMPRESSION AND TENSION CREEP TESTS

Creep tests of polycrystalline TiAl (Ti-47Al-2Nb-1Mn-0.5W-O.5Mo-0.ZSi) alloys
and polysynthetically twinnned (PST) crystals were conducted on an ATS series 2710
creep test machine with a PC based data acquisition system. All tensile creep tests were

conducted in the creep laboratory at Howmet Research Corporation, Whitehall, MI.

4.1.1 Compression creep test (ATS creep tester)

The ATS system is designed to perform either creep testing (constant load) or
stress relaxation testing (constant strain). The machine has a lever arm ratio of 20:1 to
provide a high load capacity and loading rate. As shown in Figure 4.1, the creep machine
consists of three main control systems, which are the creep frame control system, the
fiimace control system, and the data acquisition system. A load signal and displacements
from two transducers, A and B, are conditioned in the ATS control system. Temperatures
can be adjusted by controlling three heating zones through the fumace control system.
Data are collected from the automated data acquisition system consisting of a computer
and data acquisition board (DAB).

The creep frame controller unit, which was made by ATS and Electronics
Instrmnent Research Corp. (EIR), allows automatic or manual control mode for constant

load or stress creep test. The three furnace heating zones are controlled by a fumace

81

 

 

 

 

  
 
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

 

LC» 20;1 .. _| . H
‘1“ A I I ll ‘ I! ‘
'3' ,1 ‘ H ' l ‘1 1
T ‘ ,1 L , 7 , l ‘
U
L] ‘ 1
fl Furnace control system
5 1‘ {7,6139} 0 r
i ‘ ‘ , ,, 7 c. ‘
, 760.50]
C I} 1:. ‘7: 0 ‘ 0-
‘7 W ,760.40\
r, , , J
ta ' w]
L} I
[W o I I
f: ‘3. Data acquismor
[J “ system
L.

 

 

 

 

 

 

 

Frame control syste ’ ,3

load , ,
450.501 A452”)

[’1 811504751

 

Figure 4.1 The diagram of ATS creep tester with data acquisition system.

82

control unit (LFE model 2010). The furnace is designed to reach a maximum temperature
of 1200 °C and the temperature gradient between top and bottom of the chamber can be
controlled by adjusting the three heat zones. K-type thermocouples were used for
temperature measurement, and the temperatures are displayed on the panel of the furnace
control unit. The one analog output channel from the unit is monitored by the data
acquisition system.

The strain is measured by linear variable capacitance transducers (LVCT) which
perform precise and stable specimen gaging. A small change in displacements in the
tranducer causes a change in capacitance. Two LVCTs were connected to the specially
designed extensometer for compression creep tests, and they directly measure the change
in length during creep test. The measurement from the LVCT and load cell are input into
the control unit and show on a display panel. The analog outputs of strain and load
transfer to the data acquisition system through the control unit, and are limited on
-10~+10 V and the initial position of LVCT is adjusted to be within this limit.

The data acquisition system consists of a PC computer and a National Instruments
AT-MIO-l6L-25 board. This is a high-performance multifunction analog, digital and
timing I/O board for a PC. Five analog input channels are scanned to monitor the values
of displacements (A&B), temperatures of the top and bottom of the specimen, and loads
are recorded through the data acquisition program developed by Wu [196]. A sampling
rate for the testing is 5000 pts/sec for 5 channels, i.e. 1000 pts/sec-channel. The noise
effect caused by fluctuations of loading or other noise is smoothed by two levels of
averaging, first by averaging 1000 points acquired in a second, and then about 20 such

points are averaged, and the datum point is kept if it differs from the datum that was most

83

recently kept. The details of hardware and software in this system are explained in

reference [197].

4.1.2 Tension creep test [198]

Every tension creep test was conducted at the Howmet Corporation, Whitehall,
MI. In 1995, Howmet installed a creep test system with a closed-loop control system for
one of its highest volume activities, creep-and-stress-rupture testing. An example of data
from Howmet is shown in Figure 4.2. The closed-loop system includes an IBM-Pentium-
compatible host computer and an OPTO G4LC32SX Mistic Model 200 local computer to
control and monitor all facets of the testing process. Both units are configured to allow
interaction with the operator. The local computer controls multiple test stands, and is
linked to individual stands by G4 brick I/O data-acquisition interfaces.

To eliminate the risk of losing data because of a local computer failure, the host
computer backs up all test data at present intervals. The main windows from the host
computer displays real time data for all 96 test stands supported by the system. The local
computers are connected to all of the extensometer inputs from the creep test stands, and
temperatures inputs. The local computers are also linked to end-of-test inputs, weight
raising relays, and temperature controllers. The host and local systems exchange data
through RS 422 communication lines.

The creep testers are designed to permit experiments at temperatures ranging from
540 to 1100°C (1000 to 2000°F), and stresses ranging from 100 to 1030 MPa (15 to 150
ksi). Resistance-heated tube or split furnaces are used to ease loading and unloading of

the specimens. Two thermocouples are connected to the top and bottom of the specimens

84

HOWUET CORPORATION

Tm Numb-r- ] Wat-1.809.011 my” “we,” no: .3,
M W‘ M Tau 1.0.01.9 0.09 (at): 0.02
m” 3"“ 001407 cmp At 20 11am (X): 0.11
9101111101110» H 01.09 N 20 noun (1‘): 0 12
000011100 111mm. 15 Sutup rm; 00.142710112111107 01009 A! :0 boon (x): 0.10
rut ”rampant-t (1:): 1200 Tee: SIM 11m: 10:20:32 1011211007 61ch 00 00010 at): 0.15
snot Lem 0181; 40 T001311» Tim: 1000-001111011007 7110.110100 00010 (x: 0.17
mean 1.10.11 (10:; 70.3 r... 1. 0.111 01...: (1010): 12.0
0.111. 041.9. 1.90.0.1: 1.011 T-OUMXM (1000): 100.0
“humid“: CJH 11010 In 05% 0009 (In) All!
10100111110011.1- 00000 11.1.10 matcmpom); 01.
10111.1 menu... (10): 0.0000 mun 2.0: 0000 (In): 1110
1.11101 nu 01mm 11¢ 0.2010 Final 8a 01m: (to: “den 01 Nu (x): 0.0
011111.111 Long. 11.): 1.704 Final Llagih (1.): new (at): 0.0
find 40:10.1 01000 (x): 0.42
301101 10.1.1110 T.“ E] T'"" 1:] El 1:] 111110 to Fuller. (In): 090.5
Upload 1111-0. 1 z 0 4 s 0 r 0 0

811000 (1:30 40.0

L000 (LBS) 70.3

M More 0.0

End Hon-0 0005

LhAlBtrm 0100!!) MS
Ala-appointing” In corrocbd ”Mamaw bytho'lkfiushd Guy. LaotipuASTME-IJS coda-11.1.3

HOWMET CORPORATION
Team-1110:: 311000; 40 x01 T0010: 1m r
Creep Curve

0.5 '1’

0.4-e

Creep (96)

0.1—

 

 

00 l l l l 11
0 200 400 000 000 1000
Time (Hrs)

Figure 4.2 Creep testing involves holding test specimens at constant load and temperature
until some specific experimental parameter is met.

85

to monitor the test temperature. Extensometers are used to measure the strain (or creep),
and measurements are collected and recorded every 15 minutes after the applied load is

rerached

4.2 THE CREEP CURVE FITTING

In order to predict a long term creep behavior from short term creep data, some
constitutive equations are used, such as the power law equation which relates és (steady
state strain rate) to stress and temperature by the power law relationship and polynomial
expressions which have proposed by Conway [199]. However, in the power law equation
different values of n and Q are observed as the test conditions are varied, and it is
difficult to interpret the polynomial coefficients in terms of physical meaning in the
polynomial equations. Evans and Wilshire have proposed the 0-projection concept, which
can not only accurately describe the shape of the creep curve but it also has some relation

to the physical meaning of creep deformation [200].

4.2.1 The 0 projection concept

Based on the micromodelling of the dislocation processes occurring at high
temperatures in primary and secondary stages, the following constitutive equation is

described as

e = a, + 6, {1 —exp(— 6,t)}+ at (1)

where a, is the initial strain upon loading, and the second and third terms describe

86

primary and secondary creep stages, respectively [200]. 93 is the secondary (or

minimum) creep rate, and the parameters 19, and 02 are determined by curve fitting.

Differentiation then shows that the creep rate decays as

.9 = 9,92 exp(— 9,z)+ .9 (2)

And the primary creep rate can be rearranged as

9,, = 9,92 exp(— 9,1) = 92 (9I — 9,) where 9,, = (9' — .9). (3)

Equation (3) indicates that the primary creep rate (9],) decays linearly with increasing
primary creep strain (9],) under the rate constant given by the parameter. 8,. 0, is a

parameter of the strain-like quantity, and 612 is a rate parameter as shown in Figure 4.3.a

[201]. From Equation (1), a creep curve is described as a continuously decaying creep
curve, i.e. a curve characterized by a creep rate which decreases gradually until a steady

state creep rate is eventually attained (9,.) at very long times as the primary strain
approaches 6,.

In a similar way it is possible to define a tertiary strain 9,. as 9 — 9,, where

8,. =9—9p =63{exp(194t)—1} (4)

and by differentiating equation (4), the tertiary creep strain rate is given by

87

. (a) primary stage
1 8p = 01 {l- exp(-02 t)}

 

   
 

 

f
00
.s“
33 t
(D l
8.. 0l quantifies
2 l 02 determines the total
0 the curvature primary strain
of primary creep y! .
Time, t +

(b) tertiary stage”
1 8T = 03 {exp(04 t)-1 }

 
   
  

04 determines

the curvature
of tertiary creep

 

the tertiary
creep strain

Creep strain, a --->

 

Figure 4.3 The 0-projection concept envisages normal creep curves in terms of a
decaying primary component (8,) and an accelerating tertiary component

(81) [201]-

88

5.1 = 0394 exp(6l4t) = 64 (93 + 87') (5)

The tertiary creep rate (97.) increases nearly linearly with increasing creep strain (9,.)
indicating that the tertiary stage also obeys first-order reaction-rate kinetics, with 194
being the rate constant. The meanings of the tertiary parameters 193 and 64 are also

indicated in Figure 4.3.b [201].
All of the creep curve including primary and tertiary stages can be described by

the following equation [199],

9 = 9, + 61,{1— exp(— 621)}+ 193 {exp(94t)— l} (6)

where 611 and 63 define the strain magnitude with respect to time, and 192 and 194 describe

the curvatures of primary and tertiary stages of creep, respectively. The second term in
this equation represents the strain hardening term that dominates in primary creep, and
the third term describes the strain weakening term that dominates in tertiary creep. The
secondary stage is explained on the basis of a balance between the competing processes
of strain hardening and strain weakening. It agrees qualitatively with theoretical
interpretation of creep deformation by dislocation motion. However, since the secondary
stage is only a transition between primary and tertiary creep processes, a true steady state

creep rate can not be estimated by this equation.

4.2.2 Modification of 0-projection concept
The accurate measurement of initial strain, a, , is not easy even when high-

precision test methods are used. It is because 9, is determined by calculating only the

89

difference in strain immediately before loading and immediately after loading. Therefore,
Maruyama et al. [202,203] presented the same equation as the 0-projection concept in

order to get a better interpretation but with the variables having a different meaning.

8 =90 +A{l—exp(—at)}+B{exp(,Bt)—l} (7)

90 is an adjustable parameter instead of initial strain in the original 0-projection model.

A, 01, B and B are the parameters for curve fitting. 01 represents the rate constant of the
strain hardening process and [3 represents the rate constant of strain weakening process.
Maruyama and Oikawa [203] have also proposed a modified and simplified equation of

the form,

.9 = 9, + A{1— exp(— at)}+ B{exp(at)— 1} (8)

and 90 , A, B and 01 are parameters to be determined by the regression analysis. It has

only one rate constant 01 instead of the two rate constants in equation (7). Actually,
equation (7) having two rate constants, on and B, may be more suitable for all materials
due to the complicated mechanisms during creep deformation. Microstructural changes

often lead to changes in the rate-controlling mechanism during the creep process.
4.2.3 Modification for this analysis ’

For the experiments described here, only the primary and secondary stage of TiAl

creep deformation have been considered, but especially the primary stage creep. The

90

tertiary stage creep and rupture life are beyond the scope of this research. The primary
creep tests under shear+compression and tensile modes were conducted to examine
primary creep behavior in detail. The initial elastic and plastic strain components were
not recorded, since zero time represents the time when the creep load is reached.
Therefore, only the second term of equation (1) as the constitutive equation was used for
analyzing creep curves of the primary stage by using a modified mathematical model in

the form of

apnmm W,” = e(1— exp(— rt)) (13)

which describes a process with a rate constant r that slows exponentially with time to
reach a saturation strain of e. Most of the experiments showed a transition in the
downward curvature that suggested a change in the dominant deformation mechanism. If
the transition in deformation phenomena results from two process that slow down
exponentially at different rates, where one saturates earlier than the other, then these two

process may be independent and therefore additive, and thus represented by

Sprimary creep 2 e1 (1 _ exp(_ ’11)) + 82 (1 _ exp(— th)) (14)

This equation was used as a fitting function using KaleidaGraphW software to represent
the experimental data in a form that can be easily differentiated. The parameters for el
and r1 were determined for the experiments deformed at a given stress and temperature to
the lowest strains, and these parameters were used as initial values in the least square

fitting process for the larger strain experiments.

91

In stress and temperature change experiments for polycrystalline and PST TiAl
alloys, another modified equation from equation (1) was used to account for various
starting times for a particular deformation condition following a stress or temperature
change [204]. The modified equation for transient creep after stress and temperature

increase is

9:90+A{1—exp(—a(t—to))}+és(t—t0) (15)

where 90 is an adjustable reference strain parameter, to is the starting time for a particular

loading condition. 95 is the steady state (or minimum) strain rate, A and a are curve
fitting parameters. In the case of transient creep after stress and temperature decrease, the

constitutive equation is

9:9,+Bexp(—9(t—z,))+9,(t—zo) (16)

Differentiating equations (15) and (16),

9 = A 01 exp(— a(t - t0 )) + 3.1 for positive temperature or stress changes (17)

and

9 = —B,6 exp(-— a(t — to )) + 9 for negative temperature or stress changes (18)

Equation (17) implies that the creep rate during the primary stage decreases gradually
towards a “steady state” value, 9 S , with times greater than to, and equation (18) indicates

a gradual increase in creep rate towards a “steady state value”.

92

4.3 MEASUREMENT OF LAMELLAR SPACING

Scanning electron microscopy (SEM) and Transmission electron microscopy
(TEM) were used for measuring lamellar spacing in TiAl alloys before and after
deformation. In the SEM (25 kv accelerating voltage, 9mm working distance), the
backseattered electron technique was used for imaging in order to measure the 012
lamellar spacing as shown in Figure 4.4.a. The ythickness and 012 thickness were
measured by using the dark field method in the TEM (by tilting the foil to so that a
<110> direction in a lamellar plane was parallel to the electron beam). For example, after
012 laths were indentified by using the diffraction spots diffracted from 012 under the Dark
field mode, these 012 were sketched on the Lab notebook. Then bright field images were
taken at the beam direction very close to the <110> as indicated in Figure 4.4.b and the
012 can be marked on the micrograph for the measurement. A ruler was put across the
micrograph perpendicular to the lath boundaries as shown in Figure 4.4. The distance
from one 012 lath to the next 012 lath is denoted as 012 spacing, and the distance from one 7
interface to the next 7 interface or 012 lath is called as the y interface spacing. The average
values of 012 spacing measured in SEM before and after creep deformation were
computed from 250-500 012 spacing counts in more than 20 lamellar regions. The average
values of 7 interface and 012 thickness before and after creep deformation were computed
from at least 100 y interfaces, and about 50 012 laths in about 6 different regions near grain
boundaries. The 012 measurements are upper bounds of the actual spacing, since the
lamellar plane normal direction was not determined to make the parallax correction
needed to determine the actual 012 spacing. Despite this systematic error, however, this

method provided a way to compare 012 spacing in specimens with similar microstructures.

93

 

ruler

Figure 4.4 Measurement of lamellar spacing in (a) backseattered electron image in
scanning electron microscopy (SEM) mode and (b) bright field image in
transmission electron microscopy (TEM) mode.

94

After these measurements, distribution curves of lamellar spacing were plotted with the

same bin sizes using KaleidaGraph’“ software.

4.4 A Statistical Analysis of Measured Changes in Lamellar Spacing

In this section, a statistical analysis of the raw data from the measurement of 012
spacing and 7 thickness is examined using data shown in Figures 8.16.c and 10.9.b. These
two figures show the smallest changes in lamellar spacing, so if these can be shown to be
statistically reliable, then all other plots are also statistically reliable. These data are used
to illustrate how variations in statistical presentation affect the perception of the lamellar
spacing distribution. Also, an analysis of the average spacing is presented. The statistical
analysis was conducted using MIN ITAB TM sofiware.

As shown in Figure 4.5, a different number of bins results in a different
distribution curves for lamellar spacing. Figure 4.5.b shows that using a line curve is a
better way to present the lamellar spacing distribution than a histogram, since relative
differences in area under the curve for different bin sizes make changes easy to see. Thus
all other plots use the line type presentation. Using a larger number of bins causes a
bimodal type of lamellar spacing to be presented as shown in Figure 4.5.c. Although
different bin sizes show different distribution curves, the higher frequency in the smaller
bin size and the lower frequency in larger bin size after deformation is evident regardless
of the number of bins. This indicates that more finer lamellar spacings are formed after
deformation. The difference of average lamellar spacing before and after deformation is
about 0.15 pm (see also Figure 8.16.c). There is a 95% chance that the change in the

average value was between 0.1 and 0.2 mm from a confidence interval computation.

95

In Figure 4.6, the natural logarithm of every lamellar spacing was taken and
plotted in a similar way. Depending on the number of bins, different distribution curves
before and after deformation can be seen in Figure 4.6.a. Similar to Figure 4.5.c, a
bimodal type of distribution curve with the log scale can be obtained with a higher
number of bins.

From the comparison of two probability plots in Figure 4.7, both normal and log
normal descriptions of this distribution provide a linear relationship, but the normal
distribution plot has more outlying data at small spacings. In the log-normal plot, the
reduction of lamellar spacing after deformation can be seen by a translation of the
deformed distribution (marked by gray lines in Figure 4.7.b) to the left, indicating a
systematic shift to smaller thicknesses. Except at very low spacings, the 95% confidence
levels do not overlap, indicating that the shift in lamellar spacing is statistically certain.

In one case of changse in y thickness, there is 4% reduction of lamellar spacing
after deformation (see Figure 10.9). As with the 012 distribution, a different number of
bins affects the appearance of the distribution, but a shift toward finer spacings is
apparent in all three variations in Figure 4.8. In Figure 4.8.c, a bimodal type distribution
can be seen in a 7 thickness with a larger number of bins, which is similar to the effect in
the 012 spacing distribution. In the normal probability plot in Figure 4..9.a for 7 thickness
distribution, there is no linear relationship. The same data obey a linear relationship in the
lognorrnal probability plot (see Figure 4.9.b). The 7 thickness distributions using the
natural logarithm of the y thickness is plotted in Figure 4.10, and there is no obvious

difference in the distribution before and after deformation

96

The 95% confidence interval for the average values of the undeformed and
deformed 47-2-2 y thickness included zero, indicating that a change in the lamellar
spacing for this specimen is statistically insignificant. However, no other specimen
showed such a small change in lamellar spacing. The distribution data presented in every
chapter is reliable, since the reduction of lamellar spacing after deformation was similar
to or greater than the example analyzed for the 18% in 012 spacing. This statistical
analysis shows that the data are better described using a log-normal distribution.

The plots shown in chapters (6-11) use a linear scale and a rather small number of
bins, but the above analysis shows that the trends that are apparent in these plots are

statistically reliable.

97

 

60 fivrvrvwW‘vafivVTYfivaYYYr‘VYYYV

  
    
 
      
     

 

 

 

[ '1 .‘ Heat Treated Tensile ]

50 i— ' ‘ 049 'c.270 MP0 1

1 I

A 9 + Undeformed
°\° 4 0 [— ' '6' ' Deformed '1
v _ 1
31 : 10 bins 1
E 30 ~ 7
0.1 ~ 1
= ‘1
=1- * 1
E 20 5 1

L AL

10

 

 

 

 

 

 

 

Ot ,
0 0.5 1 1.5 2 2.5 3
(a) 012 spacing ( um )
40*...r...r..r1fi..,...... 4
35 ’ Heat Treated Te“ 3
; 049’C,270Mr. j
30 l ‘ __ Undeformed j
? t ' Deformed :
\ » " ' 1
3., 25 r . 1
a. ’ 15b1ns
:
0.1
:1
8
1...
[it

 

 

 

 

 

0.67 1.2 1.7 23 28

 

   
  

 

 

 

 

 

(b) .
012 spacing( um )
35.,0
' 1
: Heat Treated Te-ile j
30 5 «woven». 1
i 1
A 25 f +Undeformed J
\° r 1
a“ C --A--Derormed ]
B. 20 f j
E i 20 bins 1
00 . 4
=1 15 9 1
5" ; 1
E 10 I.
5 _
oi .
0 0.5 1 _1.5 2 2.5 3
(C) 012 spacrng( um )

Figure 4.5 Frequency of 012 spacing with different number of bins (a) 10 bins,
(b) 15 bins, (c) 20 bins in W-Mo-Si alloys.

98

 

6 o T 1— l I Y Y Y T— v r l 1 V Y r T T r T T ‘F Y J
1 Heat Treated Tessie <

50 l 049 c.2715 MPa —e— undeformed L

 

 

 

Frequency (%)

 

 

 

 

 

 
    
  
 

0r 1 l L 1 . 1 . 1 r #4 1 r n i 1 rAQ-c-iu._ u
.06 -o.4 -o.2 o 0.2 0.4 0.6
(a) In (012 spacing, um)
30‘VVfififvv‘Tr—v rj-vv-Irvv‘vvvtvfir].
; 9‘ Heat Treated Tede
25 F f 049'c, 210 MPa L

 

—9— Undefonned
‘ 'A' ' Deformed

10 bins

20

 

erWj—Yfr
-
a

  

10'—

Frequency (°/o)

 

 

 

.;

OLE... ,
-O.8 -0.6 0.4 02 0 0.2 0.4 0.6 0.8
(b) In (012 spacing, um)

Figure 4.6 Frequency of 012 spacing with different number of bins (a) 5 bins,
(b) 10 bins for the natural logarithm scale in W-Mo-Si alloys.

99

normal probability plot for undeformed & deformed

 

 

 

 

 

 

in WMoSi (649°C, 276 MPa)
' , ,7 'Uadet‘orm
- Defined
fl -
fl .
w .
0 .
N .
E £2
a 3:
to .
5 a
I .
ob as m 1.3
(a) a: simian u m)
lognormal probability plot for undeformed & deformed
in WMoSi (649°C, 2‘76 MPa)
' ; : , ‘ - um
- Muted

t ltlatlt )-

Percent
- .5 8858888 88 8

I

 

 

 

 

'1 i i I10
(1))“ a: spacing( in")

Figure 4.7 Normal and log-normal probability plots of W-Mo-Si alloy for undeformed
and deformed cases.

100

f

 

 

I 1' fi'

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

35 : {T r :
[ .' \ nun-zmno :
30 :— 1" \ Tension Creep j
A 25 ,_ ’I A +Undeformed _‘
°\° » .' ‘\ ' 'A' ‘ Deformed
v L .’ x ,
g~ 2° f ; IO bins «
a ,. ' s :
» 4* ‘ -
g 15 ~ ; j
5" r ,' r J
0 t I ‘t j
E 10 f': A\ T
t: i
5 11' j
o A A A A J 4 J—L 4 l 4 A --- j
0 0.5 1 1.5 2
(a) 7 thickness (um )
30 _ w T , 7 v 1 I Y . ' ff r 4
L 4
: Ti-47Al-2Nb-2Cr j
25 * TenioaCreep _.
r A ..
,2 : A” \ +Undeformed :
o\ 20 L f “‘ "A“Deformed ‘1
v r a. 4 4
B1 i ; a 15 bins i
g 15 r ; “ -«
0 P l t q
5 1 .' ‘.
3 10 l " "
I- 1” .' Is. ,
kl » ,‘ A .
5 f j
1
0 L A 4 4 A I J
0 0.5 1 1.5 2
(b) 7 thickness (pm )
20 ' r ' T I I r l v r v
t 1
» Ti-47Al-2Nb2Cr l
L 53‘ TensionCreep 4
t ,' . 1
15 _ ,' 2‘ a
’; r 4 1‘ +Undeformed «
°\ 1 .' ~ "A"Deformed t
v r : - '4} .
3“ ' : '1 20 bins *
= 1° 1 : '. r
o F' : \| f
5 r ' - 1
3 1 -' ‘- :~. 1
‘- r 5 i 4
h 5 F: :t o ‘i
t : ‘. I. A ‘
r ' u. A, <
r A ‘
I A. A‘ +
o ‘1 L 1 i 1 1 4 J_ 1 J 1 DA
0 0.5 1.5 2
(c) 7 thickness (um )

Figure 4.8 Frequency of 7 thickness with different number of bins (a) 10 bins,
(b) 15 bins, (c) 20 bins in Ti-47Al-2Nb-2Cr.

101

normal prob ability plot for 47—2—2

 

 

 

 

 

 

 

 

 

 

(undeformed)
. (z'
99
Q?
90
so
H 70
5 %
CL ‘0-
l I
c' i
(a) 1 thislmawl:fim)
lognormal probability plot for 47—2—2
(undeformed)
W .1
95 -1 ~
03 ..
an .
70 4
E $:
s i 3
D- 10 .1
6 .1
I ..
O1 10
(b) 7 fllicklcn(pm)

Figure 4.9 Normal and log-normal probability plots of Ti-47Al-2Nb-2Cr alloy for
undeformed case.

102

7 O Y T l T I r T Y I I Y T l' T

 

1
l
l

   
     
 
 
 
  
    

 

 

 

1
60 } Ti-47Al-2Nb-2Cr J
E Tension Creep i
50 7 j
: +Undebnned 3
4O } --A~-Debrmqj —]

5 bin size

Frequency (%)

 

 

 

 

20

1o ;
o i 1
~ 1
-10 %1 4 1 1 1 L a 1 1 1 1 1 1 I 4 g 4 1 ‘
-1 -O.5 O 0.5 1

(a) log (y thickness)
40 1 f7 ' i r Y T r ifi fY—r l ' T Y 7 T ' ' ' r1" ffi _‘
1
35 ” Ti-47Al-2Nb—2Cr :
Tension Creep

 

—0— Undebnned
- 'A' ' Debnned

 

 

lO bin size

Frequency (%)

  

1 LJLll4iAlJLJLLLiLLlll l

 

 

 

4.5;“ ~11 1:0.5 0‘ 10.5 1 1.5
(b) log (y thickness)

Figure 4.10 Frequency of 7 thickness with different number of bins (a) 5 bins,
(b) 10 bins for the natural logarithm scale in Ti-47Al-2Nb-2Cr alloy.

103

CHAPTER 5

OVERVIEW OF RESEARCH

In this chapter, an overview of the research results from chapter 6 t012 is
presented. Gamma TiAl based alloy has been developed for high temperature structural
applications, since it has low density, low coefficient of thermal expansion (CTE), high
thermal conductivity, high specific strength and stiffness at elevated temperature, and
high resistance to oxidation and creep [94,95].

Yamaguchi, Inui, and colleagues [143-147] show orientation dependence (the
angle (1) between lamellar planes and the loading axis) of yield strength and tensile
elongation at room and high temperature in PST crystals. For deformation parallel or
perpendicular to lamellar planes, shear deformation proceeds on {111} planes that
intersect the lamellar boundaries (the hard-type of deformation that causes a high yield
stress). At intermediate orientations, shear deformation occurs parallel to the lamellar
boundaries (easy-type of deformation that causes a low yield stress) and hence large
strains can be obtained for intermediate orientations (see Figure 3.9). Twinning is an
important deformation mechanism in TiAl. Some studies reported that TiAl stacking fauns
form at an early stage of deformation and develop into twins during deformation [124].
TiAl stacking fan", could act as the nucleation sites for formation of the Ti3Al and Ti3A1 stacking
faults are sources of nucleation of TiAl.,m,,x/TiAlT,,,in {124-126}. These observations suggest
that Ti3Al can be formed by a process similar to mechanical twinning during

deformation.

104

Luster and Moris suggest that deformation modes are restricted to twinning or
dislocation motion parallel to the lamellar interface in lamellae thinner than about 0.5
um, and superdislocation is nucleated from interfaces when efficient slip transfer from
neighboring crystals was possible [142].

The differences in creep behaviors between Al-rich single phase TiAl and Ti-rich
two phase TiAl alloys can be easily seen in Figure 3.10. Al-rich alloys creep much faster,
and well defined transitions are evident. There are two transition points in the stress
exponent in Al-rich TiAl (n changes from 3.5 to 7.5 to 4.7 with increasing 0') [149]. On
the contrary, the stress exponent increases smoothly as stress increases until power law
breakdown region is reached in Ti-rich TiAl alloys [153]. Most of the studies in creep
have focused on minimum creep rate conditions or stress-rupture properties. However,
primary creep property is important for practical applications. Only limited research has
been accomplished concerning deformation behavior in the early stage of creep on TiAl
and there are still many mysteries regarding the creep mechanisms of TiAl. Thus, the
goals of this research are to investigate the relationship between microstructure and
primary creep mechanisms in TiAl alloys having different compositions. By comparing
creep in PST crystals with different orientations and polycrystals in similar test
conditions, the contribution of deformation in lamellar microstructures to the creep
mechanism on duplex materials can be examined. The primary creep mechanisms in
duplex TiAl alloys can be rationalized by analysis of experimental observations.

For the research, several investment cast TiAl alloys (Ti-45Al-2Nb-2Mn+0.8TiB2
XDTM, Ti-47Al-2Nb-2Mn+0.8TiB2 XDTM, Ti-47Al-2Nb—2Mn+0.8TiB2 XDTM with

different amounts of interstitial elements (oxygen and carbon), Ti-47Al-2Cr-2Nb and Ti-

105

47Al—2Nb—1Mn—0.5W-0.5Mo-0.ZSi alloy) have been developed and tested at Howmet
Coporation. All cast investment TiAl alloys were HIPed at1260 °C/ 172 MPa for 4 hours,
and then several kinds of heat treatments were applied. Conventional tensile and
compression+shear creep tests were conducted at various temperatures, stress, and strain
conditions. For study of deformation in lamellar microstructure, PST crystals with
different orientations were creep deformed with stress and temperature changes creep.
The creep behavior of these alloys and the influence of heat treatment are summarized,
and details of these studies are found in chapter 6-12.

Ti-45Al-2Nb-2Mn-0.8TiBz did not show much difference in primary creep
resistance after various heat treatments because no difference in the microstructures was
caused by heat treatment. The creep resistance was poor, due to very fine microstructures
and large angular (12 grains (see Figure 6.4). Some limited improvement in the Ti-47Al-
2Nb-2Mn—0.8TiBz alloy with low 0 was obtained with heat treatment. The lamellar
volume fraction was about 25%, and this correlated with a smaller decrease in creep rate
upon loading. Having 40% lamellar grains (see Figures 7.3 and 7.4), and less angular a2
grains in The Ti-47Al-2Nb-2Mn-0.8TiB2 alloy with high 0 resulted in better creep
resistance than low and medium 0 alloys. The heat treatments (especially the two step
HT) significantly improved the creep resistance of the Ti-47Al-2Nb-2Cr alloy. This heat
treatment led to the smallest volume fraction (18%) of lamellar grains, a homogeneous
microstructure, the least amount of pockets of small equiaxed grains, large equiaxed
grains, and precipitation of some small particles (see Figures 6.7 and 6.9), all of which
correlated with better primary creep resistance, in this alloy, and similar effects were

observed in other alloys, too. Texture analysis indicates that the desirable texture to

106

minimize mechanical twining is to have crystals oriented with the tensile axis near 001 '
and away from 441 (see Figure 6.11). This texture is strengthened with heat treatment
and it may contribute to the excellent creep resistance obtained from the two-step heat
treatment. Ti-47AI-2Nb-1Mn—0.5W-0.5Mo-0.ZSi alloys show the best creep resistance
(~300 hrs to 0.5% creep strain) due to presence of precipitates (see Figures 8.12. b and c).
The precipitates (including TiSSi3, but also precipitates rich in M0 and W atoms) were
nucleated by heat treatment, and they grow with the assistance of strain or transformation
of (12 to y. The precipitates vary in size from 0.05 to 2 pm, and larger precipitates are
found in the equiaxed regions. The nucleation and growth of these precipitates interferes
with dislocation motion, and they harden the microstructure. This precipitation hardening
mechanism accounts for much of the excellent creep resistance of this alloy when it was
heat-treated. Similar to W-Mo-Si alloys, the primary creep resistance of investment cast
Ti-47Al-2Nb-2Mn-O.8TiBz was improved by the addition of some limited amount of
carbon. Fine precipitates ranging from 0.08 pm to 0.2 pm are observed preferentially in
or; layers of lamellar laths, in equiaxed a2 grains, and near grain boundaries of equiaxed
grains and in lamellae (see Figure 11.5). The precipitates can be a barrier to moving
dislocations during high temperature deformation, and their presence in a2 may impede
slip transfer through the 0L2 phase. The addition of carbon in this alloy results in a grain
size refinement, and the Iarnellar volume fraction is reduced from 70% to 30% by adding
more than 0.065 wt% C (see Figures 11.1 and 11.2). In 0.110 wt% C alloys, thick layers
of a; are observed, which are similar to angular 0L2 grains that correlate with degraded

creep resistance, compared to the 0.065 wt% C alloy (see Figure 11.3).

107

Two kinds of creep tests (tensile and compression+shear) were conducted, and
they show similar creep curves in different loading condition at the early stage of creep.
An empirical equation was used in a curve fitting process to characterize the data in a

convenient functional form, as follows [200,204].

8 = el (1 — exp(— rlt» + 62 (1 — exp(—— r2!» (tensile creep)
a = so + A{1— exp(— a(t — tO ))} + if (I — to) (for stress and temperature increase)

a = so + B exp(— ,6(t — to )) + $50—10) ( for stress and temperature decrease)

Creep strain rate vs. strain was obtained by differentiating the curve fit equations (see
Figures 8.3, 8.4, 8.5, 9.2, 9.3, 10.2, and 11.10). All creep tests show a 2-step primary
creep process. The transition points or saturation points (where the first and faster
deformation process is exhausted) can be seen in any strain rate vs. strain plot of creep
data. The strains at the transition point depend on the temperature and stress. A higher
stress at the same temperature or a higher temperature with same stress cause the
saturation strain to shift toward smaller strain. The transition strain and strain rate at the
transition strain also depend on composition and microstructure features such as lamellar
volume fraction and deformation twining (see Figures 11.10, 11.12, and 11.14).

The on; spacing and y thickness (>200 and 100 measurements for (12 and 7,
respectively) were measured from more than 20 and 6 different regions near grain
boundaries, respectively, before and after deformation. As shown in distribution plots
(see Figures 7.5, 8.16, 8.18, 9.6, 10.9, and 11.8), there is a shift of the frequency

distribution toward higher frequencies at lower spacing values due to small creep strains

108

in every case. For example, the average a; spacing in 0.065 wt% C changed from 1.17 to
0.84 pm due to a small creep strain of 0.42% (see Figure 11.8). This reduction of 0L2
spacing can be explained indirectly by easy mode deformation twinning parallel to
lamellar boundaries or/and strain assisted transformation to on; that occurs by a process
similar to mechanical twinning. The average (12 spacing is plotted with respect to
effective strain and stress in Figure 8.18.b. There is about 23% reduction in the on;
lamellar spacing with strain that saturates at effective strains between 0.2 and 0.5%,
whereafter the spacing remains relatively constant, or increases slightly. More refinement
of lamellar spacing occurs at lower temperature and higher stress (the amount of
refinement at 750 °C is about half of the amount that occurred at 650 °C). For example,
The amount of refinement at 650 °C increases with increasing stresses (Figure 5.1). The
refinement process at the early stage of creep results in rapid drop of creep rate, because
refinement of lamellar spacing will reduce the probability of dislocation motion on {111}
planes not parallel to the lamellar interface [142].

As shown in Figure 3.8 [141], in tension, orientations close to <441] (near tail of
arrow) are likely to twin. This figure indicates that in the twinned orientation, the Schmid
factor for ordinary dislocations is small, and only superdislocations have a high Schmid
factor, so twinning increases the volume fraction of hard orientations where only super
dislocations can operate. Therefore, twinning can orientational hardening in addition to
hardening that results from refinement.

The relationship between the initial strain rate, the strain rate at the transition
point, and minimum strain rate and lamellar volume fraction is shown in Figure 11.12.

All of these creep strain-rates decrease linearly as the volume fraction of lamellar grains

109

 

1.6 TYrTYIVrer] TTTTT I IIIII IlTIffrlffiTT

WMoSi alloy (As HIPed) ]

Compression+Shear creep at 650 °C l
14 2

(pm)

 

mwmg

d

 

 

 

 

 

 

 

”N . Undeformed,
as V 650 °C/77 MPa(y=0.26%)
0'8 b ’ 650 °C/115 MPa(y=0.24%) “
A 650 °C/154 MPa(y=0.28%)
06 11111 l 11111 liiiinlllgmillnglml 11111
0 30 60 90 120 150 180
(b) Stress (MPa)

Figure 5.1 The a; spacing decreases with respect to increasing stress at similar strains
at 650 °C [205].

110

increases. As shown in Figure 5.2, more lamellar spacing refinement occurs with a higher
volume fraction of lamellar grains and with lower temperature and higher stress. Figure
5.3 shows how the strain rate at the transition between the early and later processes (when
the early process is exhausted) correlates closely with the volume fraction of lamellar
grains [62]. The best primary creep resistance occurs when the transition occurs at a
lower strain rate. This set of studies shows that microstructural variance has a large
impact on primary creep resistance, but that the lamellar refinement process occurs
similarly in all microstructural variations. Thus a higher volume fraction of lamellar
grains is beneficial for creep resistance because a larger volume fraction of lamellar
interfaces will block overall dislocation motion during deformation.

From the investigation of PST crystals with different orientations (<110> and
<112>), the orientation dependence of the activation energy for creep and the stress
exponent was examined. The activation energy was near 135 kJ/mol, when shear
deformation occurred in the <110> direction and it was higher, near 185 kJ/mol for the
<112> direction. These values are about half of the 290 kJ/mol activation energy
measured for Ti diffusion, so mechanisms dependent on pipe diffusion or grain boundary
diffusion are plausible. Furthermore, the amount of primary creep strain was several
times larger in the <112> direction than in the <110> direction (see Figure 9.4.a), and this
greater amount of strain occurred at a much lower stress. The stress exponent of n=5.3 at
higher shear stress regions (see Figure 9.4.b) in both directions suggest that viscous
dislocation glide (n=3) controlled by pipe diffusion (n+2) in this microstructure and
lamellar orientation. Similar to the lamellar refinement in TiAl alloys, refinement of

lamellar spacing of 13 and 23 % reduction during a strain of 0-1 .4 % creep was also

111

35 I T 1 T l 1 Y T j f fl V l

 

30

   
 

higher stress and
lower Immature

 

Reduction of or2 spacing (%)

 

 

 

 

 

 

20L— I XDhighO a
A XDhigh HIPed 76° C’ ‘
t 138MPa ‘
lower stress and A XD 10W HIPed
15: highertcnperaturc o XDlowC :
,—-'L'! o XDMC 650°C/ .
-‘A"fl 276MPa ,
. XDhighC
10b 4 1 ‘ ‘ ‘ 1 - . . l - . . t . . . q
20 40 60 80 100

Vol. fraction of lamellar grains (%)

Figure 5.2 The amount of lamellar refinement (% reduction of a; spacing) increases with
the volume fraction of lamellar grains.

 

 

 

 

    

 

 

 

. As HIP 1185°C 5 As HIP [260°C . 0.02 at% C
v +8 hrlOlO°C . +8 hrlOlO°C o 0.21 at%C
7 +50 hr lOlO°C A +50 hr 1010°C O 0.36 at%C
: 'Ii-47Al-2Nb—2Mn-XD 3
_m _ \v Q ,
g x \ 138 MPa 760°C
5 r ‘ (v: \ o ‘
.5 v ‘ \ \ ' .\‘ Q NWC
E 1048 t ‘ \ f 1
m L. 0' s
g ; 276 MPa 650°C ,
E r .
E l w
10-9 1 1 4 l r r A l 1 1 1 l L A r l A J l
0 20 4o 60 80 100

Vol. fraction oflamellar grains (%)

Figure 5.3 Higher lamellar fraction retards the strain rate at the transition between the
early and later creep processes [62].

112

observed. The reduction of lamellar spacing contributes to strain in the early primary
stage, and it also hardens the crystal. Near 1.38% shear strain, hard mode deformation
(cross twinning) was activated in some lamellar laths (see Figure 9.5).

Some specimens of as-HIPed W-Mo-Si were deformed in the same manner as the
PST crystals, but with monotonic loading histories (see Figure 8.2). The lamellar spacing
was measured in the polycrystals as a function of strain. A strong correlation between
exhaustion of the early primary creep process and the exhaustion of lamellar refinement
is evident in Figure 5.4 (note that the strain scales for the creep rate and lamellar spacing
are the same). In comparing activation energies of the early process in primary creep
experiments in the W-Mo-Si alloy with creep in the soft orientation in a PST crystal
(Qc=140 and 137~185 kJ/mole in the W-Mo-Si, and PST crystals, respectively), the
activation energies are similar, near 150 kJ/mol (see Figure 8.5). The second process has
an activation energy close to 300 kJ/mol that is similar to values observed in minimum
creep conditions [206]. This suggests that early mechanism at the primary creep in
polycrystals is correlated with creep kinetics of easy deformation in PST crystals that
causes lamellar refinement.

The lamellar refinement process has been simulated to determine how much
twinning or shear transformation contributed to the reduction of lamellar spacing after
deformation. The thickest lamellae are subdivided in a manner that changes the frequency
distribution to match the deformed distribution by assuming that strain induced on; shear
transformations with a thickness of 0.2 pm occur. When several variation of the detailed
deformation histories are considered, the resulting deformed distribution, and the

resulting local strains in the lamellar microstructure are computed. As shown in Figure

113

 

-5
10 t r 1 T f] T r T l 1' l T T T I T I I T I Y fi‘r T T 7 T l l

 

 

 

L Tl-47Al-2M:-1Mn-0.6W-0.6Mo-0.28| l
)-
_\ \ T / 750°C ,
- 5 n 140 ¥ L-
1 0 3 tra—x.
E \y._ lem ol 77 MP” l
‘ s P

As -l-IPed
\ “ 3t: Shari-Compression

E\
i \ 96' ts ‘ . ‘ \ 650°C

P In ..~ b
\ \‘ / .--.-.A ‘ 1“ MP8 l-

, sso'c
650°C 276NPa ' 77mm ———1.4

 

Strain Rate, s'1

_L
O
G
K
8.
0
g
I
l
a"
a"

 

 

   
  
  

Tensile Tests w ith HT 50 hr 1010°C

 

 

 

 

 

 

ofi-osm-ozsu alloy 1‘2 N9
m
————— 221—“q E
012 spacing In As-l-lPed Shear+Com pression :1 g-
)- CG
_ Q--\: ------------------ -o 0.21 at% c L f;
‘ L'\ Low Carbon 0'8 a
I \>--——--——o-——--——--——-o —--——--—0
r. O ......................
E O 0.36 at% e V
1mm “2 mm _0 6
l Tl-47Al-2hb-2Mn-XD+C ~

 

 

L41_LllJL1llllllllllllllllllll
0 0.002 0.004 0.006 0.008 0.01 0.012
Effective Strain
Figure 5.4 Comparison of lamellar refinement history with primary creep rate shows that

the early process operates by lamellar refinement, whereas the later process

does not alter the lamellar spacing appreciably with known transition strain in
the XD+C alloy [62].

114

12.1, the simulated lamellar spacing distribution was quite close to the actual spacing
distribution from the measurements. The computed local strains (0.7%) in lamellar
microstructure were higher than the specimen (0.5%). There is a correlation between the
computed local shear strain and the lamellar volume fraction, creep strains, and strain at
transition points in most TiAl alloys. Higher lamellar volume fraction results in a lower
calculated shear strain, which leads to more lamellar refinement and then lower creep
strain.

From this research, primary creep properties depend on the composition, heat
treatment, and microstructural features such as lamellar volume fraction, lamellar
spacing, grain size, and the presence of precipitates. There are two stages between the
transition point in primary creep, such as early and later stage, where creep rate is
dropped rapidly, and slow down, respectively. There is a strong correlation between
exhaustion of the early primary creep process and the exhaustion of lamellar refinement
due to mechanical twinning parallel or/and a; shear transformation to lamellar planes
(easy mode deformation). The early stage deformation in TiAl alloys correlates closely
with PST crystals creep in easy mode deformation. There are some correlations between
the local shear strain in lamellar microstructures with primary creep behavior, such as
resulting from lamellar refinement are about 30% larger than the measured global strains,
indicating that lamellar microstructures are softer than the equiaxed grains during the

early stage of primary creep.

115

CHAPTER 6

THE CORRELATION OF PRIMARY CREEP RESISTANCE TO
THE HEAT TREATMENTS AND MICROSTRUCTURES IN
INVESTMENT CAST Ti-Al GAMMA ALLOYS

ABSTRACT
Cast gamma titanium aluminides are gaining acceptance as replacements for superalloy
and steel components in aerospace, automotive and industrial applications. Components
cast from these alloys can operate at temperatures up to 815 °C at half the weight of the
components they replace. Most applications of cast gamma components require good
creep resisitance to meet component life requirements. Four heat treatments were
developed and applied to investment cast Ti-45Al-2Nb-2Mn+0.8v%TiB2 XDTM, Ti-47Al-
2Nb-2Mn+0.8v%TiB2 XDTM, and Ti-47Al-2Nb-2Cr alloys in an effort to enhance creep
properties with a decrease in heat treatment time compared to current methods. Results
show that the creep resistance of Ti-47Al-2Nb-2Cr alloys can be significantly altered, up
to 3 times, through heat treatment. Some improvement was obtained in Ti-47Al-2Nb-
2Mn+0.8v%TiB2 XDTM and heat treatment had no effect on Ti-45Al-2Nb-
2Mn+0.8v%TiB2 XDTM alloy. The variation or lack of variation, in creep resistance with
heat treatment can be explained by differences observed in the microstructures and

textures produced by the various heat treatments.

§.l INTRODUCTION

Gamma titanium aluminide can be heat treated to obtain four types of

116

microstructures [110]. The best creep resistance has been observed in fully lamellar
microstructures, but the duplex microstructure provides desirable room temperature
ductility [92,161]. The effect of alloying elements such as Cr, Nb, V, Mn, W, Mo, and Si
on the properties of two phase Ti-Al alloys has been investigated [143,188,207-210].
Most of this research has focused on minimum creep rate conditions or stress-rupture
properties. However, the primary creep resistance is important for practical applications.

Gamma based TiAl exhibits primary creep strains that are typical for metals, where the
minimum creep rate is reached after about 1% strain. In an effort to decrease the creep
rate during primary creep deformation, additions of W, Mo, and Si have increased the
time to reach 0.5% strain due to a dynamic precipitation process [211,212]. In related
studies, observations of lamellar refinement were made that suggest that primary creep
consists of a significant amount of mechanical twinning that occurs as an easy mode of
deformation that hardens the microstructure at low strains [50,165,212,213]. In this
chapter, the primary creep resistance (time to 0.5% creep strain) was analyzed in three
two phase TiAl alloys that do not rely on the refractory elements for primary creep
resistance. The nature of primary creep strain in these materials is explored in terms of
microstructural changes that arise from different heat treatments. Two alloys containing
TiBz particulate produced by the XDTM process and a third alloy without TiB; are

compared.

6.2 MATERIALS AND EXPERIMENTAL PROCEDURE
Three investment cast gamma TiAl alloys, such as Ti-45~47Al-2Nb—2Mn—O.8TiB2

(0.8 TiBz refers to volume fraction) and Ti-4_7Al-2Nb—2Cr, were cast. All cast materials

117

were then hot isostatically pressed (HIPed) at 1260 0C for 4 hours at 172 MPa (2300 °F,
25 ksi) in order to eliminate casting porosity. The heat treatments used were in the a2+y
phase field at 900 or 1010 °C (1650 or 1850 °F) for 10 or 20 hours, and these alloys were
cooled under the static argon atmosphere (SAC). A two step heat treatment was also used
_ consisting of 1177 °C (in the or+y phase field) for 5 hours and then in the airy field for 10
hours at 1010 °C (2150 and 1850 oF), respectively, and cooled in static argon. These heat
treatments are illustrated on the phase diagram in Figure 6.1 [96,97]. The phase
boundaries are not accurate for the quaternary alloys of this study. The specimens were
subjected to room temperature tensile tests and creep tests at 760 °C, 138 MPa (1400 °F,
20 ksi). The creep specimens were cooled under load once 0.5 % strain was reached.

For optical and SEM analysis, deformed samples were obtained from the gage
section of the creep deformed specimens. Some undeformed samples were also cut from
the grip section of tensile test specimens. The samples were mounted, mechanically
polished, and etched in Kroll's reagent (2ml HF, 6 ml HNO3, and 92 ml H20). For TEM
investigation, 0.7 mm thick slices were cut such that the sample normal was parallel to
the tensile axis. 3mm diameter discs were cut from the slices using an electrodischarge
machining (EDM) and ground to about 0.1 mm thick. The disks were thinned using a
twin jet electropolishing system with a 10% sulfuric acid and methanol solution at -25°C
and 20 volts. These foils were examined in a Hitachi H-800 transmission electron
microscope operated at 200kV. For SEM analysis, a Cam Scan 44FE SEM was used at
25kV to obtain backscattered electron images that show phase contrast. Texture

measurements were made on selected specimens and analysis was made using the popLA

software [214].

118

Temperature ( °C )

1600

1400

1200

1000

800

Weight % of Al

 

 

 

 

 

 

 

 

 

20 25 30 35 40 45 50
L 2900
2700
B
. , 2500
on . ‘S e 2300
" 2100
66 1900
or, ; 3 Y
02+“)! ..‘ 1700
reep Test in 1500
' ~ ' .- ' '+ 1300
30 40 4547 50
Atomic%ofAl

( do) ammrodutol

Figure 6.1 Phase diagram illustrating heat treatment and creep condition [96,97].

119

6.3 RESULTS
6.3.1 Primary creep to 0.5 % strain

The time to 0.5 % creep of the three titanium aluminide alloys varies with heat
treatment as shown in Figure 6.2. The average value of three samples is plotted with the
line, and high and low values are also plotted. From this plot, it is apparent that the 45Al
composition is insensitive to heat treatment. The 47Al samples exhibit different responses
to heat treatment. With TiBz, the heat treatment for 10 hours at 900 °C is damaging to
primary creep resistance, but otherwise there is no significant improvement with
increasing time at temperature in heat treatment, other than a smaller range of values with
the two step heat treatment. For the alloy without TiBz, the creep resistance is
substantially improved with increasing amounts of time and temperature in heat
treatment, but the variability in the samples is not significantly improved with heat

treatment.

6.3.2 Microstructures

The samples chosen for microstructural observation were the middle performance
specimens. The microstructural variations in these specimens are summarized in Table
6.1. Details of the microstructure for each combination of composition and heat treatment
are described in more detail below.
Ti-45Al-2Nb-2Mn-O.8TiB;

As shown in Figure 6.3.a, optical micrographs of the crept specimen show that
many TiBz particles (indicated by arrow) are distributed randomly. There are several

types of particles such as equiaxed, needle, and irregular shaped precipitates with spacing

120

 

 

 

 

 

 

 

 

200 4 7 I I I E i
180 :_-.. —0—Ti-45Al-2Nb-2Mn-O.8T1123 ......... .......... . -..6 ............... 3:
E ; .. --a- Ti-47Al-2Nb-2Mn-0.8TiB ........ ______________ j
5160 : --A--T1-47A12Nb-2Cr i ' 1
:—.-....---...--.......-................-..........fi:.. ..... ’ .......-i ................ J
140 p 3 g g a p47 2 2 -
8‘ E E E I A
2 120 5W. -......... ,....: ....... .’.’ ..-.., ............... 3
U L ........................................................... ‘3... of ....................... s
\c 100 [:1 1% ' or 1
6
"3
e
3 z
E C
F" E
0

As HIPed 900°C/ 900°C/ lOlO°Cl 1 177°C/
thrs/SAC 20hrs/SAC thrs/SAC 5hrs/to/
1010°C/

Heat Treatment thrs/SAC

Figure 6.2 Time to 0.5% creep strain with respect to heat treatment. The lines go
through the average of 3 values, the high and low values are indicated by
additional symbols.

121

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

msoocom08om 36on $3 msoocflobuom msoocoabfim 81888885
a; 9.2 amoflzm :3 oflzm 81 876 6% saw
$mw $3 $3 $9. a. $ 3883 HUAZN1<$1H
81 87% :3 879. a: 87% a: 87% as 58
$3 $9 $3 $~m 6% $ 3:083
88% .285 owhfi 3853803 wobuom mmoq gonzowEBom 838888;”.
wfiommm 81 $22 8:86 @8883 wioamm 81 Omzom NM: .1.
a: 91: :1 91: :3 c2: 81 $28 6% saw £5568
$3 $16» $3 $9. a. $ 3883 573-2811.
E: 822 81 87% 51 879. :1 082% 6% as»
$3 $8 $mm $3 6% $ 8:083
8:6qu 83 3:83 83 maceaowoaom 83880886
88630 381 18w wcmoanmm 81 omzom mm;
81 omzom 81 Omzom 81 omzom 81 2 BE 8me NmEtwdéEm
$8 $3 :3 s s 3353 62022-8
81 87% a; 22% 81 873 an em 9% 568
$2 $3 $3 a. $ 3:083
U<m>85<
06 oz: 91981356 coo U<m\mEo<uo ooa BE: m<
88506 R:

 

 

 

 

 

 

$188828 30310 88m 98% 1.3 H 025 no 80838 ”no: mo acute 65 mo 8083.530 .6 2an

122

of 30-50 pm. As the annealing time increased no distinct change in the particles was
observed, but in the case of two-step heat treatment there appeared to be fewer particles.
From the micrographs of etched samples, a very fine duplex microstructure was observed
in all specimens, regardless of heat treatment. There is a homogeneous distribution of 20-
30 um equiaxed and 30-100 um lamellar grains. The volume fraction of lamellar grains
increases slightly increased after heat treatment from as-HIPed to 900 °C/20hrs, but no
difference between the 900 °C/20hrs and the two-step heat treatment was observed.

From analysis of backseattered images of undeformed and deformed specimens
obtained in SEM, there are no distinguishable changes in the microstructure resulting
from creep deformation (Figure 6.4.a and b). TiBz particles are distributed randomly, and
they are located inside gamma grains, in grain boundaries between gamma (dark phase)
and (12 (bright phase) grains and in lamellar grains, but not inside on; grains, as shown in
Figure 6.4. Single phase y and a2 grains represent about 40% of the volume fraction. The
a2 grains have large and angular shapes, but the y grains tend to be more round and
equiaxed. This indicates that equiaxed y grains nucleate in several places in a given parent
on grain, and that some of the parent on grains transform into lamellar grains, but others do
not. Very fine mechanical twins are observed in some on; grains in both deformed and
undeformed specimens (see arrows in Figure 6.4). These finely winned regions also
exhibit contrast indicating formation of extremely fine lamellar microstructures
consisting of both on; and y lathes. Since they are much finer than the obvious lamellar
grains, they are not likely to have resulted from thermally induced transformation, so they
are more likely to be a result of local stress concentrations [47,50,213]. These mechanical

twins were observed in all heat treatments.

123

. l
3

V - a t. [‘3‘
V . \f‘:‘-;»“§.
1:? VA;
\

f ,
...,

 

Figure 6.3 Optical microstructures of deformed Ti-45Al-2Nb-2Mn—0.8TiB2 a) as—HIPed,
as polished, b) as-HIPed and etched, c) 900 °C/20hrs heat treatment, and
d) two-step heat treatment.

124

.8088: “no: 36-95 G 98 45888 8o: mEoQoo coo C
45.88% Bani“. 3 £08808: 315-3 Q NmfiwdEENbZ~1<mYF 10 8mg: 5886 wouoxmomxomm v.0 oSmE

 

125

Since the 0L2 grains have an angular shape, stress concentrations are more likely to be
focused at triple points in 0:2 grains, where the fine mechanical twinning is most often
observed. The fine twinned regions are also observed in TEM images. They occur in the
angular shaped (12 grains, as illustrated in Figure 6.5.a. Regions of extremely fine
mechanical twinning are also observed in some lathes of the lamellar grains (Figure
6.5.b). The width of the mechanically twinned region appears to be larger in regions with
a higher stress concentration. Dislocations are also observed in the vicinity of TiB;

particles, in Figure 6.5.c.

Ti-47Al-2Nb-2Mn-0.8TiB2

The TiBz particles in the as-HIPed alloy Figure 6.6.a appear larger (10-70 pm)
than in the 45 Al specimen in Figure 6.3.a, but they have a similar spacing.
Microstructures of as-HIPed specimens after creep deformation exhibit a heterogeneous
microstructure illustrated in Figure 6.6.b. There are clusters of small equiaxed grains in
some regions where there are fewer lamellar grains, and conversely, few fine 7 grains are
observed in regions containing lamellar grains. After heat treatment in Figure 6.6.a and d
the microstructures exhibit fewer small grain clusters. The volume fraction of equiaxed
grains changes, and after the two step heat treatment, the lamellar volume fraction
becomes largest. The maximum grain size of lamellar grains is about 200 pm. As time
and temperature of heat treatment increase, microstructures became more homogeneous.
The equiaxed grain size distribution becomes more homogeneous, and fewer clusters of

finer y grains were observed.

126

a

.2083 NEH e. 513 888885 8280036 3 .82on 3:08: 8 83.530 833 188888 081 3

88w no 8 83.530 835 188888 081 838186 8898 NmfiwdéENAZm

-238 88¢ a co 3925 22c 535 3 2%;

 

127

 

Figure 6.6 Optical microstructures of deformed Ti-47Al-2Nb-2Mn-0.8TiB2 a) as-HIPed,

as polished, showing TiB; particles, b) as-HIPed and etched, c) 900 °C/20hrs

heat treatment, and d) two-step heat treatment.

128

Ti-47Al-2Nb-2Cr

Although no TiB; is present in this alloy, there are second phases present (Figure
6.7.a). However, in contrast to the microstructures containing TiBz, the as-HIPed
microstructure exhibits very large 300-400 um lamellar grains and about 200 um pockets
of equiaxed 10-20 pm grains (Figure 6.7.b). The microstructure is extremely
heterogeneous, and it exhibits some of the same types of heterogeneities (clusters of small
equiaxed grains) observed in Figure 6.6.b. With heat treatment, the volume fraction of the
lamellar grains decreases with increasing time and temperature, and the microstructure
becomes less heterogeneous. The pockets of equiaxed grains become fewer and there are

fewer fine grains in the pockets.

In the backseattered electron SEM images of this alloy there is a smaller volume
fraction of angular equiaxed a2 grains between gamma grains and a2 laths are thining in
lamellar grains (Figure 6.8). The very fine mechanical twins seen in the 45 A] sample
(Figure 6.4) were not observed in the smaller on; equiaxed grains in the 47 Al specimens.
Very small bright phases can be seen either in or; equiaxed grains or boundaries of
gamma grains (indicated by arrows) and these phases are [52 phase which is produced by
addition of Cr [215]. They have an orientation relationship of (0001)or2//{111}y//{110}
[32, and <112O>a2//<1T0>y//<111>132 [215]. Since Cr is known to strengthen the on; and
7 phases, this may account for the lack of mechancal twinning observed in the 0L2 phase in
the Ti-47Al-2Nb-2Cr alloy [215]. There appears to be a decrease in the amount of B;
phase with the increasing amounts of heat treatment (Figure 6.8.c and d). At lower

magnifications, the distribution of large equiaxed grains correlates with lower aluminum

129

 

Figure 6.7 Optical microstructures of deformed Ti-47Al-2Nb-2Cr a) as-HIPed,
as polished, b) as-HIPed and etched, c) 900 °C/20hrs heat treatment, and
d) two-step heat treatment.

130

content regions (which are darker in Figure 6.8.e). With deformation, some dynamic
grth of a2 regions drain the surrounding region of titanium, in the lamellar region
illustrated in Figure 6.8.f. Also there appears to be less segregation since there are fewer
dark(alumium rich) regions after creep deformation (Figure 6.8.e~h). TEM investigations
indicate that a 170° angle relationship between lamellae is frequently observed in the as-
HIPed and heat treated conditions. Typically, one of the two lamellar lathes has a finer
spacing as shown in Figure 6.9.a. Subgrains developed in larger regions of lamellar grains
during high temperature deformation in Figure 6.9.b. No subgrains had developed fully in
equiaxed grains even though there were regions exhibiting a dislocation density greater
than 2 x IOU/m2 in Figure 6.9.e. From Figure 6.9.e, it appears that an equiaxed grain is
consuming a lamellar region, but the (12 lamellae resist consumption. After the two-step
heat treatment, dynamic precipitation was observed in grain interiors and interfaces in
Figure 6.9.e, and this may account for some the exceptional creep resistance.

By using the dark field method and tilting techniques, the spacing of 7 thickness,
a2 spacing and the thickness of a2 laths in several lamellar grains were measured. The
data was obtained by measuring about 100-3 50 laths and taking the average as shown in
Figure 6.10 and Table 6.2. The spacing generally decreases with more heat treatment. In
the case of 7 thickness, there was not much reduction of spacing from heat treatment, only
about 8%. However, a 4% reduction of y thickness after creep test was observed in the
900 °C/20hr heat treated alloy. The refinement of the lamellar spacing could result from
mechanical twinning parallel lamellar boundaries since this is an easy mode of
deformation against loading. Measured (X2 spacing shows that on the contrary to the

previous case, the spacing decreases about 5% and 18% from the As HIPed condition

131

80888 88: 33-95 3 28 G
Ea 80888821 85800 com 3 4515-3 Q. Cmb 78-251? 888% 10 8928 8886 wobzmomxowm We Eswi

c1om Avooo

\\\\\ \ko

...:Sm Nmooo .34 o.

81”.
\
7M‘\

.....s.\

 

132

.woqtomogaogmob “no: 3:85,. com E was
.woccowucnsscugmob “so: mEoQUo coo Aw 628803 335-5 C 62883:: domain 8 .825833
mo :osmwoaom 2:653 0653: HUN-n—Zm-fi<\.v-§ mo momma: 8:020 @2838on cosmoficwmfi 304 We 25me

 

133

 

“63-1700

"r4 I”

- "zuhIH-Ks‘ng ‘_ L,

 

Figure 6.9 Bright field images of as HIPed Ti-47Al-2Nb-2Cr after creep test a) commonly
observed relationship between coarser and finer lamellar regions exhibit a 1700
about a <110> axis relationship, b) subgrain boundaries develop in larger
7 regions.

134

 

Figure 6.9 Bright field images of Ti-47Al-2Nb—2Cr after creep test c) (12 lathes resist
absorption in a boundary where an equiaxed 7 grain is consuming a lamellar
grains (900 °C/20hrs heat treatment), d) particles are located at grainboundary,
and e) inside a y grain afier two-step heat treatment.

135

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.6 _
I Ti-47Al-2Nb-2Cr
A 0.5 7 i
E § 3
3- 0.4 — _
v : 2
m : 1
3 0.3 5 1
5 i 3
3 0.29 3
fl - .
H -4
>_ 0.1 : j
0' - 4
Al HIPed 900'Cl 900‘CJ
20hrllSAC Zohfl/SAC
(Land-tanned) (dammed)
Heat Treatment
1.5
Ti-47AI-2Nb-2Cr
A 1.2 — _
E :
1 :
v 0.9 ~ f
E. 0.6 E s
n t
N t ‘
5 0.3 L ;
t
L
A! HIPOd 900'Cl WO'CI
ZOhI'I/SAC 20hI'I/SAC
(undefomd) (untamed) 1010-a
He at Treatment ‘°'""s‘°
0.15,
Ti-47AI-2Nb-2Cr
A 0.12'
E
:x.
V ,
m 0.09
3 .
:1
'3 006
IE .
a
N
5 0.03
0.

 

 

 

A: HIPed OOO'CJ BOO'CJ 1| '
20hnI$AC ZOMIISAC fihuflol
(undeformed) (deformed) 1010‘CJ
mun/SAC
Heat Treatment

Figure 6.10 Refinement of lamellar feature was observed with increasing heat treatment
conditions, and resulting from deformation.

136

Table 6.2 Average and range of dimensions describing lamellar microstructures.

 

 

 

 

 

 

 

 

 

900 °C/ 900 °C/ 1 121713524) /
Ti-47Al-2Nb- 20hrs/SAC 20hrs/SAC o
2Cr AS HIPed (undeformed) (deformed) 1010 C/
10hrs/SAC
0.518 um 0.493 um 0.475 um 0.473 pm
7 thickness
0.05~2.43 um 0.07~2.0 um 0.083~1.65 pm 0.125~1.98p.m
1.064 um 1.289 mm 1.007 um 0.874 mm
a2 spacing
0.55~1.68 pm 0.1~5.25 um 0.15~2.41J1m 0.25~2.42 pm
0.127 pm 0.113 mm 0.102 um 0.090 pm
a; thickness
' 0.04~0.415 pm 0.04~0.25 pm 0.06~0.166 pm 0.01~0.15 um

 

137

 

to the one and two-step heat treatments, respectively. After deformation, there was a 22%
reduction of the 0.2 spacing, and a 10% reduction in the a; thickness. This may result
from nucleation of new a2 lamellae initiated by mechanical twinning, since a mechanical
twin has the crystal structure of the 0L2 phase for three layers in each interface [213]. The
heat treatment and creep deformation cause the lamellar scale to become more refined.
Figure 6.11 shows inverse pole figures for 3 heat treatments. Figure 6.11.a shows that
mechanical twining has a low schmid factor near 001, and increasing heat treatment tends

to increase crystal orientations that are hard to twin.

6.4 DISCUSSION
Ti-45Al-2Nb-2Mn-0.8TiBg

As shown in Figure 6.2, primary creep resistance does not improve much and only
a slight increase occurs after heat treatment from the as-HIPed condition. There is also no
appreciable change in microstructure. The poor primary creep resistance can be most
clearly correlated with the large amount of fine mechanical twinning observed in the
microstructure, particularly in the a; grains. Heat treatment did not affect the rate of
generation of these fine mechanical twins. The fine microstructures can also account for
poor creep resistance, since diffusional processes are accelerated with more interfacial

area.

Ti-47Al-2Nb-2Mn-O.8TiB;
The as-HIPed condition gives more creep resistance than the 45Al composition,

but after heat treatment at 900 °C for 10 hours the creep resistance decrease substantially.

138

.88 @020 3% 80:33 Re: @on o>5 9
JESSE“ “no: ESQ? cow 6 69:39.. 3 £88.“ 2838 Am fiaoEfiob “so: m 8m 35%: 20m 352: :6 oBmE

 

So as ., «8

 

8.- Inn-a
In. an ab 89 ‘0.

       

I83... 5
£3.35 2:.- 3-
.1.8.3: 19.:
86¢.— 3Iauw lulu-2.5

 

139

Heat treatment for 20 hours at the same temperature and the two step heat treatment
provides slightly improved creep resistance. The microstructure of this alloy has a larger
scale and the (TiBz) particles are bigger and are more evenly distributed than in the 45Al
alloy as shown in Figures 6.3 and 6.6. The particles were often observed with dislocation
tangles nearby. Thus the primary creep resistance in all heat treatments are slightly
improved except for the 900 °C/10hrs case, where the only explanation is a less
homogeneous microstructure and a higher lamellar volume fraction. With further heat
treatment, the microstructure becomes more homogeneous and the number of small
equiaxed grains decreases. However, it appears that the benefit gained by reducing the
number of small grains was offset by the increasing the lamellar volume fraction after the

two—step heat treatment.

Ti-47Al-2Nb-2Cr

This alloy shows the best primary creep resistance due to the fact that heat
treatment provides substantial benefits (Figure 6.2). As time and temperature of heat
treatment increase the creep resistance increases. The two-step heat treatment provides
the best 0.5% creep resistance of about 140 hours. This microstructure has the smallest
volume fraction (18%) of lamellar grains, a homogeneous microstructure, the least
amount of pockets of small equiaxed grains and bigger equiaxed grains, which all
correlate with better primary creep resistance. In general, the lamellar microstructure
gives a lower minimum creep rate [161]. However, this is not true for primary creep
resistance, where initial deformation occurs easily in the lamellar microstructure. By

comparing Figure 6.8 with 6.4, fine mechanical twins are found in the angular equiaxed

140

a2 grains, but not in the or; regions in the 47Al material. One difference is that the (12
grains in the 45 Al material are larger and more complicated in shape, and this may lead
to larger stress concentrations and a geometrical condition that permits mechanical
twinning to accomodate the stress concentration. Figure 6.11 shows that resistance to
mechanical twinning occurs when crystals are oriented near 001 and away from 441, and
this desirable texture is strengthened with heat treatment. The increased volume fraction
in a hard orientation for twinning may contribute to the excellent creep resistance
resulting from the two-step heat treatment. Figure 6.10 shows that lamellar grains were
refined after deformation. The correlation of refined lamellar spacing with better creep
resistance indicates that the contribution of the lamellar microstructure to the strength can
arise from a Hall-Petch relationship that reduces slip distances, and/or from a sub-grain
strengthening effect, if a twin interface can be considered effectively similar to a low-
angle subgrain boundary. Since the Hall-Petch effect is associated with a low temperature
strengthening mechanism, and the large equiaxed grain size is consistent with better
diffusion creep resistance, then the primary creep resistance at 760 °C arises from a
particular mixture of low and high temperature strengthening mechanisms. Consequently,
deformation at a different temperature may result in a different balance of strain arising

from low and high temperature mechanisms.

6.5 CONCLUSIONS
The effect of different heat treatments on primary creep resistance in three
investment cast TiAl alloys were studied. Ti-45Al-2Nb-2Mn-0.8TiB2 did not show much

difference in primary creep resistance because of no difference in the microstructures

141

resulted after various heat treatment. The creep resistance was poor, due to very fine
microstructures and very fine mechanical twining observed in a; grains. Some limited
improvement in the Ti-47Al-2Nb-2Mn-0.8TiB2 alloy was obtained with heat treatment.
However, the heat treatments and especially the two step case significantly improved the
creep resistance of the Ti-47Al-2Nb-2Cr alloy. This heat treatment led to the smallest
volume fraction (18%) of lamellar grains, a homogeneous microstructure, the least
amount of pockets of small equiaxed grains, and large equiaxed grains, which all
correlate with better primary creep resistance. Texture analysis indicates that the desirable
texture to minimize mechanical twining is to have crystals oriented near 001 and away
from 441. This texture is strengthened with heat treatment and it may contribute to the

excellent creep resistance obtained from the two-step heat treatment.

142

CHAPTER 7

EFFECT OF INTERSTITIAL CONCENTRATION AND HEAT TREATMENT
ON MICROSTRUCTURE AND PRIMARY CREEP
OF INVESTMENT CAST Ti-47AI-2Nb-2Mn WITH 0.8vol.% TiBz

ABSTRACT
Most research about creep has focused on minimum creep rate or stress rupture
properties, but primary creep is important for practical applications. In this chapter, we
compare how interstitial content and heat treatments of shorter duration than used in past
studies affect the microstructure and primary creep resistance of a particular alloy
containing TiB; particles. More time and temperature in heat treatment homogenizes the
microstructure and reduces the scatter of creep times to 0.5% strain. Increasing interstitial
content increases the time to 0.5% creep, and it stabilizes the a; phase. Preliminary
evidence for large mechanical twinning strains parallel to lamellar interfaces shortly

following loading is provided.

7.1 INTRODUCTION

Most research on TiAl alloys has focused on room temperature mechanical
properties, oxidation resistance, and yield strength [133,216-220]. Not many studies have
been made on the effect of interstitials (hydrogen, nitrogen, carbon, and oxygen) on creep
properties of TiAl. These interstitials strongly affect mechanical properties and

microstructure development, and consequently, deformation mechanisms [221].

143

Investment cast Ti-47Al-2Nb-2Mn+0.8TiB2 XDTM alloys are under development
at Howmet Corp. Several less expensive heat treatments have been investigated, yielding
some improvement in primary creep resistance [127]. Alloys with higher interstitial
element concentrations were also investigated to examine effects on primary creep and
cost reduction. In this chapter, the effect of interstitial concentration on primary creep
deformation and microstructures of investment cast TiAl alloys is examined and
compared to heats of the alloy with lower interstitial content. Microstructures and primary
creep deformation mode were analyzed in terms of grain size, distribution of a2 spacing

and "y thickness and a; thickness.

7.2 EXPERIMENTAL

Three heats of investment cast near-y TiAl with the nominal composition of Ti-
47Al-2Nb-2Mn-0.8Ti82XDTM were used with low interstitial contents typical of normal
production capability, and higher interstitial compositions (denoted hereafter as low,
medium and high 0, as shown in Table 7.1). All materials were hot isostatically pressed
(HIPed) at 1260 °C for 4 hours at 127 MPa (2300 °F, 25 ksi) to eliminate casting porosity
and to establish the duplex microstructure. The alloys with higher interstitial content were
subsequently heat treated using a standard schedule of 1010 °C for 50 hours and cooled in
a static argon atmosphere (SAC). The low 0 alloys were heat treated using several shorter
times and/or different temperatures indicated in Figure 7.1 (2-step is 5 hr. 1177 °C, then
10 hr. 1010 °C, details in. [127]). These specimens were then subjected to creep tests at

760 °C, 138 MPa (1400 °F, 20 ksi). Data were collected in 15 minute intervals following

144

application of load. The creep specimens were cooled under load once 0.5% creep strain
was reached.

Samples for microscopy were obtained from the grip section of a deformed room
temperature tensile specimen (for the undeformed microstructure) and the gage section of
creep deformed specimen for the deformed microstructure. A modified Kroll's reagent
were used for etching [222]. Microscopy was conducted in a Neophot-21 optical
microscope equipped with a LECO 2001 image analysis system, a J SM 6400V seaming
electron microscope (SEM) operated at 25 kV, and a Hitachi H-800 transmission electron
microscope (TEM) operated at 200 kV. TEM specimens were sliced from the specimens
and the final thinning process was performed in a double jet electropolishing system
using a 10% sulfuric acid + methanol solution at -20°C. All microscopy samples were

prepared so that the sample normal was parallel to the tensile axis.

Table 7.1 Chemical compositions of investment cast Ti-47Al-2Nb-2Mn-0.8TiBZXDTM

 

(at%) Ti Al Nb Mn B Cr Si Fe 0 N H

LowO 48.54 46.19 1.87 1.86 1.15 0.02 0.04 0.07 0.203 0.026 0.012
MediumO 48.55 45.43 2.02 1.95 1.14 0.02 0.04 0.10 0.227 0.032 0.012
HighO 47.66 47.16 1.81 1.88 1.07 0.02 0.04 0.03 0.289 0.037 0.012

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7.3 RESULTS and DISCUSSION

The time to 0.5% strain for the three alloys varies with heat treatment as shown in
Figure 7.1. The average value of three samples is plotted with the line, and high and low
values are plotted with symbols. The heat treatment for 10 hours at 900 °C apparently
damaged primary creep resistance but there was an improvement with longer heat

treatment time and/or higher temperature. A smaller range of creep times is obtained with

145

the two step heat treatment. For the high O alloys, creep resistance is significantly
improved with increasing amounts of O.

The alloys with medium and high O exhibit creep times to 0.5% that are more
than twice as long. The scatter is similar in magnitude for all alloys except the low 0
alloy afier the two-step heat treatment. Figure 7.2 shows a power law curve fit of the
creep data. All alloys exhibit a rapid decrease in strain-rate up to about 0.1. The creep
rates of the higher 0 alloys were lower and decreasing more rapidly at 0.5% strain. At
low strains, the creep curves for both higher 0 alloys cross at a strain of 0.1%, indicating

that the high 0 alloy had a faster initial creep rate.

7.3.1 Microstructures

Figure 7.3 shows optical micrographs of alloys with medium and high 0. The
microstructures of low 0 alloys are similar to that of the medium 0 alloy [127]. The
medium 0 alloy has a lower volume fraction of lamellar grains than the high 0 alloy
(22% and 40%) which is contrary to the common observation that lower aluminum (note
Table 7.1) causes more lamellar grains. The average lamellar volume fraction of low 0
alloys is about 25%. This result is consistent with the observation that oxygen is an alpha
phase stabilizer in Ti alloys [143,223]. The alloy with high 0 has 100-200 urn lamellar
grain sizes and 25-100 mm equiaxed grain sizes. The low 0 alloys have 20-200 um
lamellar grains and 10-50 pm equiaxed grains [127]. The higher 0 levels have fewer
small grains, which correlates with improved creep resistance [127].

Backscattered electron (BSE) images of the higher 0 alloys after creep

deformation in Figure 7.4 illustrate differences in or; morphology (white phase, see

146

 

250 T T I I 1 r

 

 

 

 

- —a—Ti-47AI-2Nb-2Mn-o.8riB 2 A
200 :"'~+Ti-46Al(medium0)with T1132 ------ n, --------------- ------------- i
» +Ti-47Al (high 0 )with r1132 i
‘ A
[50, ........................................................... ............................ j
g 1
r i «

100

Time to 0.5% Creep (hrs)

 

 

 

 

ASHIPed 900°C/ 900°C/ IOIO°C/ 1010°Cl 1177°CI
thrs/SAC 20hrs/SAC IOhrs/SAC SOhrs/SAC 5hrs/to/

1010°C/

Heat Treatment thrs/SAC

Figure 7.1 Time to 0.5% creep strain with respect to heat treatment. The lines go through
the average of 3 values, high and low values are indicated by symbols.

 

6 T 7 T 7 V f *Y Y I fir Y Y I
. I j M

\ Ti-4‘7Al-2Nb-2Mn+0.8TiB2

L L L414

2’ (Us)
3.

 

 

1
1

0 4

a -— LowO .

m

a --

0 10‘. .._ ------------ -

2 ; -.\'---.-_~_ 900’C/20hrs/SAC— .

U "x._\ "rortrCIsohrs/SAC (medium 0):

""1010C/50hrs/SAC (high 0)

 

 

 

1 -9 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 O 0.002 0.004 0.006 0.008 0.01

Strain s

Figure 7.2 Strain rate-strain plot of the creep deformation for the specimen with the
middle creep time.

147

arrows). The largest Otz regions in the high 0 alloy are similar to the scale of the lamellar
microstructure, but angular shaped 3-7 pm (12 grains (arrow) were commonly observed in
the medium 0 alloy (it also has lower aluminum). Angular or; grains correlate with faster
primary creep [127].

In the three alloys, two types of boride particles were found, needle and round.
They have dark halos in Figures 7.3 and 7.4; the round type is 1.5-3.5 mm, and the needle
type is 3-60 pm long. The needle particles have been identified as TiB and equiaxed
particles as TiB; [224]. The heat treatment or interstitial content did not affect the size
and distribution of these particles. From analysis of SEM BSE images and TEM images,
these particles appear to provide nucleation sites for lamellar lathes (fine lamellar
colonies were frequently found near particles), and they also block and entangle

dislocations.

7.3.2 Lamellar spacing

The a; spacing was measured by using BSE imaging and about 400 or; laths were
used to characterize the spacing. These measurements are upper bounds of the actual
spacing, since the lamellar plane normal direction was not determined to make the
parallax correction needed to determine the actual a; spacing. However, this is a
systematic way to compare a; spacings in specimens with similar microstructures. The
TEM was used for measuring the true 7 thickness and 012 thickness (by tilting the foil to
so that a <110> direction in a lamellar plane was parallel to the beam) and more than 100

laths for y thicknesss and 50 a2 laths were counted in several lamellar grains or regions

148

 

Figure 7.3 Optical microstructure of creep deformed (a) medium 0 and (b) high 0 alloys.

149

 

Figure 7.4 Backscattered electron (BSE) images of after creep deformation of
(a) medium 0 and (b) high 0 alloys.

150

for this measurement.

The 012 spacing was reduced in these alloys after creep deformation and this
reduction varies with heat treatment and interstitial content [127]. The average 012 spacing
before deformation in the alloy with high 0 is 1.37 pm, which is nearly as large as the
1.57 um spacing of the low 0 alloy which showed the poorest creep resistance. The
average 012 thickness is larger (0.2 pm) in the high 0 alloy than the low 0 alloy (0.15
pm). This also suggests that oxygen stabilizes the 012 phase.

The distribution of 012 spacing and 7 thickness before and after creep deformation
are shown in Figure 7.5. From the two plots, the reduction of the 0L2 spacing and
7 thickness after deformation at the early stage of creep can be easily visualized since the
dashed line shows the highest frequency of lamellae with fine spacing. The reduction in 7
thickness is greater than the 012 spacing. An easy mode of deformation twinning parallel
to lamellar boundaries accounts for the lamellar refinement. Mechanical twinning will
rotate the crystal to a hard orientation where only superdislocations are operative [141].
The reduction of 012 spacing could occur by nucleation of a2 laths in deformation twin
interfaces [120], and/or due to a strain induced phase transformation [125].

During the first 0.5% strain, more reduction in the strain-rate correlates with
more reduction in lamellar spacing in the low 0 alloy (Figure 7.6). The high 0 alloy does
not follow the same relationship between the reduction in spacing and the reduction in
strain-rate: The high 0 alloy exhibits a larger reduction in strain-rate, and a greater
amount of reduction in the 7 thickness. The high O alloy exhibits more reduction of

7 thickness than the alloy heat treated at 900 °C for 10 hours, which gave poor creep

151

resistance. If the reduction in lamellar spacing occurs by mechanical twinning, then more
twinning appears to cause a greater amount of strain at early times. The effect of the high
0 is that more refinement in the y thickness occurs in primary creep, but it occurs more
slowly.

From the TEM investigation, several regions of tangled dislocations were
observed in equiaxed grains and wider y lamellar laths. Luster and Morris showed that
several slip systems were activated in thicker lamellar laths, but that only twinning or slip
parallel to the lamellae occurred in thinner lathes [142]. From our TEM observations,
dislocation density was very low in lamellar lathes smaller than 0.5 um wide. Thus
lamellar laths less than 0.5 mm were defined as “thin”, and laths and between 0.5 and 1.0
pm as “thicker” in the plot in Figure 7.7, which indicates that before deformation, more
laths with finer spacing gave better primary creep resistance, and conversely, more of the
thicker laths result in poorer creep resistance. Dislocation motion is apparently hindered
by more lath boundaries in the finer lamellar laths during the initial part of primary creep
deformation. However the relationship in Figure 7 .7 is not observed in the high 0 alloys
or in the low 0 alloy with poor creep resistance. It appears that interstitials also act
strongly as obstacles to generation of dislocations or to dislocation motion. If so, then this
may force a greater amount of strain to be accomplished by mechanical twinning, as

indicated by the greater amount of refining of the y thickness in the high 0 material.

152

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50 fig, , -.-,W,
r I
. ,' ‘. Ti-47Al-2Nb-2Mn+0.8°/.Til3 with high 0 ,
40 L Q
’ .‘ ‘. +Undeformed ‘
a? , .' '. ~
°\ . .' ‘1 ‘ '9‘ ' Deformed <
v 30 _ ,' ', 1 a
>} i : .1 “
u . . .
g : - 1
A 4
3- 20 - ; a
2‘3 L :' '. 2
Ln , ; 1 .
.; A, .
10 1—.' “ _,
.0 “-'A‘-‘ ‘
O_'mm111.11n1n11111111‘.:1“I’1'A111v111vL11n14
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
y thickness ( pm)
50 .--.(....,-. .,....,....,-m,....,....,....,....,..-.,....
" 1
I Ti-47Al-2Nb-2Mfl+0.8%TiBz with high 0 .
40 - —
C A" , —°—Undefomred
F; . . “
Q ~ x ' “A" ‘ Deformed 4
- 1 A .4
5* 30 . ’ .
g » .' x 1
0 * .' 1
5 > :
5‘ 20 — . . 4
2 . ‘ .
p“ t , .‘ .
10 ~ 4
. , 1‘ J
041L11111111H111111111.1114111‘11~1A:1-1-1- A 1 ml :11...1
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

0 0.5 .
a, spacmg ( 11m )
Figure 7.5 Lamellar spacing distribution of high 0 alloys before and

after creep deformation.

153

 

30

 

 

 

 

 

 

 

 

 

 

. . . . , . - . . , . . . . , . . . . , - .
1 Open symbols: 9, Ti-47A1-2Nb-2Mn+0.8Til;
. Closed symbols: 7 ‘
251. 7 _
L V V
\° * 0 AsHIPed
°.,, 20~ —
:5 : D 900°C/10hrs/SAC
'09 * 0 900°C/20hrs/SAC
\ r
75 15f V 1010’C/50hrs/SAC - ‘
.‘E » A TwoStcp [3,,
5 10- 9 900°CIZOhrs/SAC ,, '
ow P _
- ' 1010’C/50hrs/SAC
' Q.
5 r ’1 O . _
r A,
or 1 1 1 1 1 J 1 1 1 1 l 1 L 1 I 1 1
0 10 20 30 40 50

% Reductionin spacing

Figure 7.6 Plot of strain-rate ratio vs. % reduction in 012 spacing and 7 thickness in
low and high 0 alloys.

 

 

 

 

 

 

 

50 ‘ ' T’ V j ‘ ' ' l ' ' ' ' l

L Large symbol: 0.5 ~ 1 um Ti-47A1-2Nb-2Mn+0.8TiI;
A >
°\° L Small symbol: < 0.5 pm
V 40 ~ ‘ —
an “ lOld’C/SOh ‘
._ [S
E. [ As 11le 1:1 9000020 (m high 0 A]
m . 1
a 30 ~ g 1
= t [:1 Two Step ‘ . 1
0 _ .(
E 1 900°C/10hrs 1
s * 4
5 2° ' i
i b ,
m 900°C/20 hrs
"5 ~ Pa
= 10 1‘ 1' Two Step
,3 1 .’ lOlO°ClSOhrs A .
g r a )1 high 0
E t900°Clthrs .’ As HIPed

1

o L 4 L g P 1 1* 1 1 1 1 1 4 1 I 1 11
0 50 100 150 200

Time to 0.5% creep strain

Figure 7.7 Relationship between fraction of specified lamellar spacing and time
to 0.5% creep strain.

154

7.4 CONCLUSIONS

A 10-40% reduction in lamellar spacing was observed within the first 0.5% creep
strain, depending on alloy composition and heat treatment. In low 0 alloys, some limited
improvement of creep resistance was obtained with heat treatment. The lamellar volume
fraction was about 25%, and this correlated with a smaller decrease in creep rate upon
loading. The high 0 alloy exhibited 40% lamellar grains (in spite of having higher Al
content) indicating that O stabilized the or; phase. A large decrease in creep rate was
observed upon loading, and this caused a 40% decrease in 7 thickness. Large reductions

in y thickness can be accomplished by mechanical twinning parallel to the lamellar plane.

155

CHAPTER 8

CHANGES IN MICROSTRUCTURE DURING PRIMARY CREEP
IN A Ti-47AI-2Nb-an-0.5W-0.5Mo-0.ZSi ALLOY
ABSTRACT
Cast gamma titanium aluminides are gaining acceptance as potential replacements for
superalloy and steel components in many applications. One particular alloy with W, Mo
and Si additions has shown exceptional primary creep resistance. Quantitative
microscopic comparisons were made between microstructures in undeformed and
deformed regions in creep specimens deformed to strains between 0.1 and 1.5 % strain,
using optical microscope, scanning electron microscope (SEM), and transmission
electron microscope (TEM) techniques. As-hot isostatically pressed (“hipped”) and heat-
treated (1010 °C for 50 hours) conditions were compared. The as-hipped specimen had a
higher lamellar volume fraction, and it crept more than 100 times faster. The lamellar
spacing in the lamellar grains systematically decreased by 15 to 35 %, with increasing
stress, during the first 0.1 to 2 % strain. Precipitates containing W, Mo, and/or Si were
observed in the deformed gage and undeformed grip sections of the heat-treated
specimens. Precipitates are nucleated by heat treatment, but during creep deformation, a
more homogeneous and faster growth process occurs in the gage section than in the aged
but undeformed grip section. The gage section had a 35 % higher precipitate volume
fraction, but their average size was smaller. A lower volume fraction of lamellar grains
and the presence of precipitates account for the excellent creep resistance in the heat

treated alloy.

156

8.1 INTRODUCTION

The Ti-Al intermetallic alloys are potential replacements for superalloys in some
high-temperature applications in heat engines, turbine engine blades, aerospace and
automobile components since it has high specific strength at high temperatures
[92,93,128]. The microstructure is affected by composition and heat treatment, and creep
properties are sensitive to microstructure. Gamma titanium aluminide can be heat treated
to obtain four types of microstructures [110]. The best creep resistance has been observed
in fully lamellar microstructures [161]. However, the duplex microstructure is preferred
for many applications because it provides a desirable combination of room temperature
ductility and toughness [92].

The additions of Cr, Nb, V, Mn, W, Mo and Si, in two phase Ti-Al alloys affect
many material properties. 1-3 at% of Cr, V, or Mn can improve the ductility of duplex
alloys [143,225-229], but V generally reduces the oxidation resistance[230]. Nb greatly
enhances the oxidation resistance [231-233] and slightly improves the creep resistance
[232]. The addition of 0.5-1 at% Si enhances creep strength, the oxidation resistance, and
room-temperature fracture toughness of 48at% Al two phase alloys [188,208,234—236]. Si
also increases fluidity, and reduces the susceptibility to hot cracking [237]. Mo provides a
good balance of strength and ductility in TiAl having very fine equiaxed gamma grains
with the 012 and 0 phase [210]. Addition of W greatly improves oxidation resistance, and
enhances creep resistance [188]. Most of these studies have focused on either room
temperature properties, minimum creep rate conditions or stress-rupture properties.
However, primary creep resistance is important for practical applications. Gamma based

TiAl exhibits primary creep strains that are typical for metals, where the minimum is

157

reached after about 1% strain. In an effort to decrease the creep rate during primary creep
deformation, additions of W, Mo, and Si have significantly increased the time to reach
0.5% strain. A dynamic precipitation process has been indicated in references 231 and
212. Observations of lamellar refinement during primary creep suggest that a significant
amount of mechanical twinning occurs parallel to lamellar interfaces as an easy mode of
deformation, and this hardens the microstructure [127,205,212,238,239]. In this chapter
we analyze microstructural changes during primary creep in Ti-47Al-2Nb-1Mn-0.5W-
0.5Mo-0.2$i alloy by characterizing the volume fraction of lamellar and equiaxed
microstructures, the distribution of 7 thickness and 02 lath spacing, examination of
mechanical twinning, and precipitates with various size, composition, and shapes that

interact with dislocations.

8.2 EXPERIMENTAL

The Ti-47Al-2Nb-1Mn-0.5W-0.5Mo-0.2Si alloy was investment cast, hot
isostactically pressed (HIPed) at 1260 0C (2300 °F) for 4 hours at 127 MPa in order to
eliminate casting porosity at Howmet Corp. Whitehall, MI. The cast 16 mm diameter test
bars were heat treated in the two-phase, 012 + y, field at 1010 °C(1850 °F) for 50 hours and
air cooled to obtain a duplex microstructure, by the Howmet Corp. The heat treatment is
illustrated on the binary phase diagram (see Figure 8.1 [96,97]). Tensile creep specimens
with a gage length of 25 mm and 5 mm diameter were machined from the test bars, and
creep tests were conducted at the Howmet Corp. One creep specimen (000248) was
deformed at 760 °C (1400 °C) at 138 MPa (20ksi) and the other (012784) was deformed

at 649 °C (1200 °F) at 276 MPa (40ksi). The specimens were loaded in a few seconds,

158

Temperature (°C )

1600

1400

1200

1000

800

20

Weight % of Al

25 30 35 40 45

50

 

 

1

08+7

Creep est “—N'

1

$43

I

L

\

-HIP’ed
-1 Heat

Y

 

treatment

   

- 2900

2700
2500
2300
2100
1900
1700
1500

 

1300

 

30

40 4'547 50
Atomic % of Al

( do) 011112.10de J.

Figure 8.1 A Ti-Al phase diagram near the stoichiometric TiAl composition [96,97]

159

and data acquisition of one datum each 15 minutes started just after loading. The
experiment was stopped when about 0.5% strain was reached, and cooled under load.
Some additional 5x5x3 mm brick shaped specimens were cut using EDM
machining from an as-HIPed slice of a 28-mm-square casting gate from a different
casting for shear+compression creep tests. The shear+compression experiments were
deformed on a 45° inclined stage in an lnconel 718 compression cage fixture developed
for evaluating creep in a PST crystal [165,240], as illustrated in Figure 8.2. The top and
bottom platens were free to translate in the horizontal direction. Shear+compression creep
tests were conducted at 650 and 750 °C in air with equal shear and compression stress
components ( 1: and CC ) of 61, 91.5 and 122 MPa. They were loaded slowly, to the
desired load, in about 3 minutes. Data were recorded using a PC based data acquisition
system developed for creep deformation that recorded temperature, load, and
displacement. Displacements were measured with an extensometer connected to the
compression cage platens with two water cooled external superlinear variable capacitor
(SLVC) tranducers that were accurate to about 1pm. Displacement was converted to

shear strain under the assumption of simple shear in the middle of the specimen, where
displacement d is related to shear strain 7 = J2d /dt, where t is the thickness of the

specimen. The principle stresses were determined with Mohr’s circle, and they are
illustrated in Figure 8.2. The maximum shear stress is 45° from the principle stress state,
and it has a value of 1.12 t. The effective stress for for this loading condition was

computed to be 0", = 1.261. Since the specimen has compressive stress component with a

magnitude equal to the shear stress, there may be some reduction in thickness due to

160

 

 

 

 

Figure 8.2 Geometry of shear+compression deformation conditions.

161

deformation. Supposing that all the strain occurred due to compression, then a shape
change indicated Figure 8.2.b would occur, and the magnitude of the compressive strain
is more than 6 times larger than the shear strain computed by assuming only simple shear
(for the measured displacement). If a mixture of shear and compression strain occurred,
the resulting shear strain is very nearly the same as obtained using the assumption of only
simple shear, as illustrated in Figure 8.2.b. Consequently, the computed simple shear
strain is a lower bound on the strain that occurred in the middle of the sample. In order to

compare results with tensile experiments, the effective strain was estimated as
.9, = 7/ J2 . Tests were stopped at shear strains of about 7 = 0.2, 0.5 and 1.5% to measure

microstructural changes in the middle of the specimen. With such small strains,
mechanical measurement of the specimen shape change was not resolvable.

For microstructural analysis of the tensile creep specimens, slices were cut from
the middle of the gage and the grip sections of the deformed specimen. For analysis of
different phases, polished (unetched) specimens were examined in a J SM-6400V
scanning electron microscope (SEM) at 25kV, using the backscattered electron (BSE)
technique. LECO 2001 image analyzer software was used to measure the size distribution
and volume fraction of precipitates. For optical microscopy, a modified Kroll’s reagent
( 10ml Hydrofluoric acid, 5m] Nitric acid, 35ml Hydrogen peroxide, and 100m] Water, 5
sec., immersion [222]) was used to reveal the microstructure. Lamellar volume fractions
were obtained by outlining the shapes of lamellar regions on a piece of transparent plastic
covering an optical micrograph, coloring them in, and digitizing the image for image
analysis. For micro-analysis by transmission electron microscopy (TEM) the Specimens

were prepared as follows. First, 0.7 mm thick slices were taken from the specimen using a

162

diamond saw, and 3 mm discs were produced using an EDM cutting machine such that
the foil normal was parallel to the tensile axis. Then, after mechanical polishing, thin foils
for electron microscopy were prepared by the twin jet polishing method using a solution
of methyl alcohol and 10% sulfuric acid at 20 volts and -25°C. These foils were examined
in a Hitachi H—800 TEM operated at 200kV using a double tilt goniometer. Energy
dispersive X-ray analysis (EDAX) was conducted on a Philips CM 30 scanning
transmission electron microscope (STEM) at Korea Institute of Science and Technology
(KIST). X-ray diffraction measurements were made using a Scintag XDS-2000

diffractometer with pole figure goniometers.

8.3 RESULTS
8.3.1 Primary creep behavior

The shear+compression experiments were conducted to examine primary creep
behavior in detail. Figures 8.3 and 8.4 illustrate how the creep strain varies with time for
the as-HIPed shear+compression and heat-treated tensile deformation conditions
investigated. Since zero time represents the time when the creep load is reached, the
initial elastic and plastic strain components are not recorded in Figures 8.3 and 8.4. All
samples indicate a rapid initial strain rate that decreased rapidly at a strain between 0.1
and 0.3% effective strain. It is common [241] to characterize primary creep deformation

using a mathematical model in the form of

8primary creep : e( 1 -CXp(-I't))

163

 

 

 

 

 

 

 

 

 

 

0.0025.vx.(v.4.(.rrfi.....(.4..r.r..
- o-Si alloy (as HIPed) 4
_ 8 Shear+compression
0.002 ~ A 3..“ 1
00 V "I 1
j WW WW 4
. . 1
.5 ~ ' - $7 8 f, 4
g 0.0015 4 E _’,_.-v--—;. j
. , ./' . ’
3 ' §--."-o no viii-"e“ . 1
> ' 0 I”
'5 ~ J1, ..-’ o J
o 0.001 ‘ y. W T C’ Ger/11> ’ ’ 8. 4
g ‘ v Y: 1
Ffl - I" w" ‘0— 750 77 0.22% ‘
I, /f ,v’ '74-'650 154 (1296 i
0.0005 — s
f f..,* f "-650 us 016% .
0 1'
” .3 $7 "*"6fl)77(M8% ‘
z f’Wv
o 31 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 141 1 444
( 500 1000 1500 2000 2500 3000
(a) Time (sec)
0.01244 . . ~ 11111 (
1
, W-Mo-Si alloy (as HIPed) . 1
‘ R Shear+compressi0n 2' 1
Ah , . ;: 1
V M w" ‘7 J
$3 39 'v "77"“ "V713 --"-=r' 1
:: g :39 :7 v Hg VL—Ww'vww 7v 1
.2. . § .‘a-t; “.818“: 1
‘ “9%; 19.41" V V V 1
.-.- v . J
$.35?sz 1
M“ ° Temp, 0, a 1
:3 —e—750°C, 77 MPa, 0.98% l

"A"650°C,154 MPa, 0.97%
-..... 650°C, 77 MPa, 0.85%

 

 

 

 

1 a = a(l-exp(-bt))+c(l-exp(-dt))
01:4,.4 1,141,114,, 444441141 1 111 1111
0 1105 2105 3105 4105 5105 6105
) Time (sec)
Figure 8.3 Primary creep strains are illustrated a) at low strains (0.2%) in as-HIPed
shear+compression specimens, b) at larger strains (1%).

164

Table 8.1 Curve fit parameters in several experimental conditions

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Matl Mode, *
°C, CMPa 6] r1 62 r2 R R
A22?) 377%; 0.0028287 1.9156e-3 0.097655 1.1681e-5 0.996 0.985
AsggPISS-ZC, 0.0016365 1 .7067e-3 0.0086616 2.5426-5 0.990 0.974
As-HIP S+C,
650, 115 0.0015921 7.0764e-3 0.930 0.980
A8821? $7“; 0.0050585 4.7166e-6 0.0038589 3.1558e—4 0.904 0.797
HT Tensile
6 49, 276 0.0020913 4.6051e-5 0.032154 3.1788e-8 0.998 0.851
HT Tensfle 0.0013451 2.126e-5 0.066612 5.1683e—7 0.998 0.971
760, 138
* R for c=d=0

165

 

 

 

 

 

 

 

 

 

0.006 ............................
L -
[ W—Mo-Si alloy Heat treated (1010 oC/50lmz) j
0.005 _ Tensile Creep j
1
* 1
w 0.0045 1
5 1 1
g 0.003 ~ 1
m . __ 1
0.002 e -
1 \ 760°C, 138 MPa, 0.5% 1
..A-. o 0 1
0.00“ 649 C, 276 MPa, 0.5/o 7 j
s = a(1-exp(-bt))+c(l-exp(-dt))
GD 1 L 1 4 1 1 4 1 1 1 1 1 L 44 4 1 4 1 1 1 4 1 1 1 .
0 10° 5105 1106 1.5106 210‘3 2.5106 310°
Time (see)

Figure 8.4 Primary creep strains to 0.5% in heat treated tensile specimens with
different temperature and stress.

 

  
 

 

 

 

 

 

 

 

 

 

 

105E....fi.,,,.............-, :
W-Mo-Si alloy 1
T°COMP
6,. \-§. 140 a
10' 17'“*'-‘-:.~~\‘ ———750 77 -.
g ‘1 kJ/mole T\ :
'70 1 \. As-HIPed """ 65° '54
0 . - \ 310 .
m _ l SheaH-compresswn --—— 650 115
a _ l "'7--_._ ..... - ......... 650 77
M E -.‘ 1 : ‘- ‘ ---------650 77 3
.E 1 1" . ‘ Tension
g ----- 649 276
U) 2
—750 138 3
i
1
Difl‘erentiated Curve Fit 1
é=ab exp(-bt)+cd exp(-dt) :1
”.1141111.114411111111111.1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
Strain a

Figure 8.5 Differentiated curve fits (strain rate vs. strain) of data in Figure 8.3 and 8.4.

166

which describes a process with a rate constant r that slows exponentially with time to
reach a saturation strain of e. If the transition in deformation phenomena results from two
process that slow down exponentially at different rates, where one saturates earlier than
the other, then these two process may be independent and therefore additive, and thus

represented by

8primary creep = 61(1-CXP(-I'1t)) + C2(1 -exp(-r2t))

This equation was used as a fitting function using Kaleidagraph software to represent the
experimental data in a form that can be easily differentiated. The parameters for el and r1
were determined for the experiments deformed to the lowest strain, and these parameters
were used as initial values in the least square fitting process for the larger strain
experiments. These fitting parameters are provided in Table 8.1. From the R values, these
fitted curves are accurate representations of the primary creep data, and they are more
accurate than a curve fit using only one saturation process. The lowest R value obtained
was for data taken over many days where room temperature oscillations caused
differential thermal expansion strains in the extensometer transducer assembly. This
function was then differentiated to represent the creep rate in a smooth manner as shown
in Figure 8.5, where the bolder lines represent the creep rates of the heat treated tensile
specimens. The effect of these two exponential components of deformation is that a
saturation strain is reached for the early el-rl process (i.e., the value of el) after which
creep deformation proceeds at a lower strain rate.

The shape of the strain-rate vs. strain curves in Figure 8.5 is similar for all

167

specimens, but the details of the shape vary with stress, temperature, heat treatment, and
microstructure. At 0'e = 77 MPa, in the as-HIPed shear+compression specimens, the
initial strain rate is about 6 times faster at 760 °C than at 650 °C, and the transition from
the early to the later process occurs at a similar strain, near 0.4%. Since these early
processes are nearly parallel at the same stress, an apparent activation energy of 140
kJ/mol is obtained. There is a greater decrease in strain-rate in the early process at the
lower temperature, so the later process is on the order of 100 times slower at the lower
temperature. The apparent activation energy at the beginning of the later process is near
310 kJ/mol. With a stress 2 times higher (154 MPa) at 650 °C, the early process is faster,
but it decelerates much more rapidly, and the transition to the later process occurs at a
lower strain, near 0.2%. The later process appears to be nearly parallel for the two 650 °C
specimens, so a stress exponent of n z 3 is estimated. At a stress of 115 MPa, between the
two other values, the initial creep rate is as high as the specimen at 154 MPa, but it
decelerates more rapidly (also evident in Figure 8.3.a). This behavior indicates that the
early deformation process at 115 and 154 MPa is different from the early deformation
process at 77 MPa.

Similar phenomena are observed for the heat-treated alloy deformed in tension.
The specimen deformed at a lower temperature and higher stress (649 0C, 276 MPa)
deformed quickly up to 0.2% strain, where the strain rate slowed considerably. At the
higher temperature and lower stress (750 °C, 138 MPa), the trend was similar, but the
change was less dramatic, leading to a crossover indicated by arrows in Figures 8.4 and
8.5. The stress was higher by a factor of 2, but unlike the comparable data for the as-

HIPed material, the initial creep rate of the higher temperature lower stress specimen was

168

comparatively slower for the heat-treated material. Also, the early process exhausted at a
strain half as large as in the as-HIPed condition, but this is probably due to the higher
stress used in the heat-treated specimens, and the differences in microstructure. This
indicates that the heat treatment was very effective in retarding the creep rate, since, in

spite of a 60% higher stress, the creep rate was more than 100 times slower.

8.3.2 Microstructures

Optical microstructures for the as—HIPed and heat treated materials are
qualitatively very similar, as shown in Figure 8.6. The volume fraction of lamellar grains
in the heat treated and creep deformed tensile specimens was about 21% and the as-HIPed
specimens had about 59% lamellar grains. In both, equiaxed grains between 5-200 11m
were observed, and lamellar grains are on the order of 100-500 um. Many smaller
equiaxed grains are interspersed among the lamellar grain boundaries. Groups of small
eqiuaxed grains are produced during HIP treatment due to large local strains associated
with pore closure. The only differences between the two tensile specimens after
deformation was that there were more twins in the 649 °C, 276 MPa specimen. There was
no other obvious change in the optical microstructure after creep deformation.

Small amounts of [3 phase were observed in backscattered electron (BSE) image
and in X-ray diffraction measurements. The strongest [3 phase peak (110) was evident in
X-ray data shown in Figure 8.7, but higher index [5 peaks were not detected in any of
several diffraction scans on various specimens. The more diffuse peaks are those of the
smaller volume fraction phases. In BSE images in Figure 8.8, the brightest phase and the

gray phase are [3 and a; phase (Ti3Al), respectively, and they are located in lamellar

169

 

Figure 8.6 Microstructures of Ti-47Al-2Nb-1Mn-0.5W-0.5Mo-0.ZSi: (a) As-HIPed
condition before deformation (shear vector was subsequently horizontal),
(b) Heat Treated tensile gage section deformed at 760 °C, 138 MPa.

170

 

 

Intensity, counts per second

 

 

 

 

fi' I I I V ‘r r V I fifi l T I U

 

Heat Treated Tensile 1
649°C, 276 MPa 1
gage section

 

 

Figure 8.7 X-ray data obtained from the heat treated tensile gage section tested

at 649 °C, 276 MPa.

 

171

 

 

Figure 8.8 BSE image of Ti-47Al-2Nb-1Mn-0.5W-0.5Mo-0.2Si: (a) As-HIPed condition
before deformation (b) Specimen tested at 760 °C, 138 MPa (1010 °C/50 hours
heat treatment).

172

boundaries and grain boundaries of equiaxed grains. The dark phase is 7 phase (TiAl).
The [3 phase (indicated by arrows) were found in grain boundaries between 17 grains and in
lamellar grains or in lamellar colony interfaces in both the as-HIPed and heat treated
specimens. The B phase has dimensions between I and 10 um, and it is typically angular
in shape, or elongated when found in lamellar interfaces. The images in Figures 8.8.a and
8.8.b are typical for comparable regions microstructure in both tensile specimens.

From TEM analysis, Figure 8.9 shows a composite image of a typical region
including an equiaxed and a lamellar grain. A number of twins on two twinning planes
(white arrows) are evident in the 7 grain that appear to have initiated from the lamellar
grain boundary. A smaller black arrow shows some more rounded 1-2 pm phases near the
lamellar-equiaxed grain boundary. Rounded phases with sizes below 2 pm are denoted
hereafter as precipitates, and these are considered as different from the [3 phase, since they
are not only found in interfaces associated with the larger microstructural features. In
summary, the important differences between the as-HIPed and heat-treated
microstructures are higher lamellar volume fraction in the as-HIPed material and the

presence of fine precipitates in the heat treated materials.

a. Precipitates

In higher magnification images from TEM and BSE images, there are many small
precipitate phases 2 um or smaller in the heat-treated microstructure, that are not
observed in the as-HIPed microstructure. Precipitates have various shapes with
dimensions ranging from 0.05-2 pm. They are evident in lamellar, equiaxed, and interface

regions of the deformed microstructure. In lamellar regions they tend to be

173

 

Figure 8.9 TEM image of lamellar and equiaxed grains from the specimen tested at
760°C, 138 MPa (1010 °C/50 hours heat treatment).

174

smaller, and reside in lamellar interfaces, and many of them have a needle shape (Figure
8.10). In equiaxed grains, precipitates are larger, and tend to have an aspect ratio of 2-3
(Figure 8.11). Precipitates always have a high density of dislocations in their vicinity.

In the BSE images in Figure 8.12, they have a brightness (high Z atom content)
similar to the [3 phase, but they are much smaller. They are mostly found in the on; phase
of lamellar structures. No such precipitates are evident in the as—HIPed lamellar regions as
indicated in Figure 8.12.d. The precipitates in lamellar interfaces vary from 0.05-0.45 pm.
These precipitates have been identified as Ti5Si3 or [3 in several studies [96,154,185,235].
In comparing the fine precipitates in Figures 8.12.a and 8.12.b, there are more of them in
the deformed sample than in the undeformed grip section of the tensile specimen. Figure
8.13 shows the distribution of sizes obtained from precipitates in at least 5 lamellar
grains. The area fraction of precipitates in the deformed sample is 35% higher. These
observations taken together indicate that precipitation is nucleated by heat treatment, and
that a more homogeneous and faster growth process occurs during deformation than in
long term static aging at creep temperatures.

On a larger scale, the precipitates are inhomogeneously distributed; some areas
seem devoid of precipitates, while in other regions they are readily found. STEM analysis
with EDAX of precipitates in various parts of the microstructure indicated some variance
in the composition. These measurements are probably sensitive to the size of the
precipitate, but in a qualitative sense, there does appear to be more than one composition
of precipitate, as some precipitates had significantly different amounts of the refractory
elements, and different primitive cell sizes measured in convergent beam electron

diffraction [212]. In the larger precipitates, EDAX analysis indicated a higher

175

 

Figure 8.10 Dislocation interactions with particles in lamellar regions. The below image
is an enlargement of the precipitate (a) indicated by the arrow (760 °C and
138 MPa ).

I76

 

Figure 8.11 Precipitates in equiaxed grains are surrounded by dislocations (649 °C
and 276 MPa), (b) an enlargement of the arrowed precipitate in (a).

177

 

Figure 8.12 BSE images of specimen a) undeformed, b) deformed at 760 °C, 138 MPa
(1010 °C/50 hours heat treatment).

178

 

Figure 8.12 BSE images of specimen c) deformed at 760 °C,138 MPa (1010 oC/50
hours heat treatment) d) undeformed as-HIPed condition.

179

 

 

 

 

 

 

 

50
Heat Treated Tensile .
760°C, 138 MPa 1
40 ~ _
. ' 1
”\3 '1 —— undeformed grip
a — ' ‘I .4
: 30 1 f ‘. """ deformed gage
o ; .
E ~ 1 ‘.
a 1 a
5' 20 f -' X‘ “
h 1" “\
10 h " “‘x —
0 O 0.05 0.1 0.]5 0.2 0.25 0.3 0.35 0.4 0.45 0.5

Diameter of precipitates (um)

Figure 8.13 Size distribution of precipitates in lamellar grains (tested at 760 °C,
138 MPa).

180

atomic fraction of W, Mo, and Si. Since the silicon peak is adjacent to the aluminum
peak, it is difficult to obtain a meaningful estimate of the silicon composition. A summary
of the shape and composition of several kinds of precipitates is given in Table 8.2. Figure
8.10.a shows a cross twin in the hard direction within a lamellae. Such twins were rarely

seen in these primary creep deformed specimens, though they are more commonly seen at

higher strains [47,50,165].

b. Lamellar Spacing

Twinning propagation parallel to the lamellar boundaries is an easy mode of
deformation that does not alter the appearance of lamellar microstructures [38]. Figure
8.14 shows a lamellar region where the twin relationship between lamellae is shown at
location a. In this figure, two very narrow lamellae are evident in regions marked b and c,
where the thin layer has a different contrast from the surrounding lamellae on either side
that have the same contrast variations. The similarity on either side of the narrow twin
suggests that the thin twin formed in the middle of a pre-existing y lath. The thin layer is
tapered, and thinner toward the lefi side of the image, suggesting that it was growing from
right to left by a mechanical twinning mechanism such as that described by J in [120,121].
The effect of propagating mechanical twins parallel to the lamellar interface is that two
new interfaces are created, and the thickness of the y lamellae are reduced.

Figure 8.15 shows that the lamellar spacing was refined during creep deformation
in the tensile heat treated specimens. The lamellar colony in the undeformed grip section
has a lamellar spacing of approximately 111m, whereas a lamellar colony in the deformed

gage section has a typical spacing less than 0.5um (Figure 8.15.b). Although there are

181

Table 8.2 Summary of Precipitate Composition and Shapes (at% of element)

 

 

 

 

 

 

 

Type of precipitate Ti Al Mn W Nb Mo Si
7
A// 45.5 46.0 0.5 0.5 2.9 0.6
ymatrix
Y
Ag/ 54.2 35.0 0.8 3.4 2.3 4.3 *
plate type
r
fl/Ul/ 51.3 39.6 0.8 2.4 2.8 3.2 *
needle type
55.2 27.6 1.2 6.1‘ 2.9 7.1 *
rectangular type
C
54.2 30.0 1.4 6.0 2.6 6.0 *
spherical type
@ 54.1 29.9 1.4 6.0 2.6 6.0 0.2
needle type

 

 

 

 

 

 

 

 

 

 

* Si composition is less than 0.1 at% from Energy Dispersive X-ray analysis.

182

 

Figure 8.14 Lamellar refinement process is illustrated as indicated by arrows. Diffraction
patterns in boxes a, b, and c illustrate true twin relationship.

183

 

Figure 8.15 Lamellar spacing was refined after creep deformation (760 °C and 138 MPa):
(a) undeformed grip section and (b) deformed gage section.

184

variations in the lamellar spacing in duplex microstructures, these observations of
lamellar refining in the creep deformed lamellae are typical. This indicates that
mechanical twinning is an important deformation mode in lamellae of TiAl alloys, as
described in Reference 50,141 ,240,244.

The change in lamellar spacing in the heat-treated tensile specimens was
characterized statistically. The a2 spacing was determined by measuring the distance
between two org/y interfaces and the 7 thickness was determined by measuring the
thickness of y laths. In regions near grain boundaries, 250-400 0L2 spacing counts in more
than 20 lamellar regions and at least 100 y lamellar interface spacings in about 6 different
regions yielded the distributions shown in Figure 8.16. There is a shifi of the frequency
distribution toward smaller spacings due to small creep strains. This refinement of
spacing is consistent with previous studies that include other alloys [141,205,212,239].
The average values are presented in Figure 8.17, where a 34% reduction in or; spacing
and a 24% reduction in 7 thickness occurred for the specimen deformed at 760 °C, 138
MPa.

The lamellar spacing in the as-HIPed shear+compression samples was
characterized from about 500 measurements of a; spacing in BSE images of 25 lamellar
regions that are near grain boundaries, before and after creep deformation. The average
lamellar spacing in three undeformed specimens provided average values between 1.32
and 1.43 pm. This 8% variation is representative of this quantification method, which
also neglects effects from lamellar plane normals that are not on the surface of the

specimen. The distribution of or; spacings is represented in Figure 8.18.a for the average

185

 

7o YVVVrVVYYrIYYI'YYfIlTVY‘T‘V'VYrVY'VrYYTY'VVYYIVIY1
b

 

 

: Heat Treated Tensile 3
60 F ° 7 60°C, 138 MPa 7
50 _— :1. '1. +Undefi>rmed 3
t --A--debrmed

 

Frequecny (%)

  

 

 

w

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

 

 

 

 

 

 

 

 

 

 

 

 

(a) a2 spacing (um)
60 .-..“...,...-T....rfi..,....r....,.r..,r...,... 4
* 9‘ Heat Treated Tensile 1
5°." ;' 760°C,138 MPa i

A 3 :

.\° 4°. 5 . 1

v ’ \

>. .1 1

:1 - 1 <

o 30~ : g a

0 ' 1.

i- 1 ' x

0 > ',

I: 20: g 7
10_ . ~
ou11111.111111.1111.11.11.1111.1111.1.'l‘. -L‘4L‘1J 41

o 0.5 1 15 2 2.5 3 35 4 45 5
(b) 7 thickness ( um)

7o3........r....,....,--..,afi4,......--.,.Wfi,...4
: avcmcspadng .
60L 11.111.111.111 11.111.511.111 Heat Treated Tensile j
: 33‘1"" “7““ 649°c,276 MPa 1
1' Is 1
50’— l“ J
g : 1' ‘. +Undefirned 1
g 1 l \ J
I 40; : 5. "A"D°""“°d i
I: : 1' ‘1 I
9 .: '. .
E- 30L-' ‘1 ‘
0 1' 1
E 20}; 1
15' 9
10 '0 j
2.-.. .
ou1111111111114111111‘1-”1111.311...:AAAAJ311111A11

o 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

(c) 01, spacing (#111)

Figure 8.16 a), b) Distribution of a; spacing and 7 thickness in 750 °C and 138 MPa,
and c) 012 spacing in 649 °C and 276 MPa specimens in undeformed grip and
deformed gage section of heat treated tensile creep specimens.

186

 

 

 

 

 

 

 

undeformed Heat Treated Tensile
760°C, 138 MPa

A I deformed
E

‘5 _ H g
1' . 3
v l 57
'g I )1) ...‘i -
a 1 * 4 _
m , and
DD 1 1 .
E ’ 1 _ 4

0.5 r . .

>
< :.-’

0 ———.———_—————

(a) y thickness

2 .

- “dammed Heat Treated Tensile

A I deformed 649°C, 276 MPa
E
:1. 1.5
v
.§ }
m i
0 1
50
L“ .
3
<1: 05

 

 

 

 

0 (b) y thickness 012 spacing

Figure 8.17 Average 012 spacing and 7 thickness in tensile creep specimens a) 760 °C
and 138 MPa, and b) 649 °C and 276 MPa before and after
creep deformation.

 

50 r‘l'leTTI'YTVT‘rfiI‘Y’Y‘r‘TIYTYTTrY—VYY—VYYfirTT—TT'IY‘Y"TYYY

As HIPed WNbSi alloy (750 “077 MPa) 1

 

V I‘Vfijfi'

40 [Q +Undeformed :

, 0— e= 0.22% j

30 j 58""? x -13- s= 0.46% 1
‘ -~A- - e= 0.93%

 

 

 

Frequency (%)

 

 

 

 

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

 

 

 

 

 

 

 

 

(a) 012 spacing (pm)
1W6 fiercrmfirmrm,WWW”Wm
' 1
1 w-Mo-Si alloy (As HIPed) A
A
E
:1.
v
.g
m
as" 1 +750°cn7 MPa j
’ "'“fi 650 °C/154 MPa 4
0.8—
0 ”°‘ 650 °C/115 MPa 1
+650°Cl77MPa
0.6 ii.ii1..ili.i.. .ii..rirmm.ii.rnir...i

 

 

0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01
(b) Effective strain, 8

Figure 8.18 a) Frequency of 012 spacings based on about 500 counts in 25 lamellar
regions after compression-shear deformation at 750 °C and 77 MPa, b)
012 spacing resulting from different effective strains at 650 °C and 750 °C.

188

of three undeformed specimens, and after three levels of strain at 750 °C and 77 MPa. In
every case, there is a nominally similar shift of the frequency distribution toward smaller
spacings due to small creep strains. The average on; spacing is plotted with respect to
effective strain and stress in Figure 8.18.b. There is about 23% reduction in the or;
lamellar spacing with strain that saturates at effective strains between 0.2 and 0.5%,
whereafter the spacing remains relatively constant, or increases slightly. At a strain of
0.2%, the amount of refinement at 750°C is about half of the amount that occurred at
650°C. The amount of refinement at 650°C increases with increasing stresses. At 0.2%
strain, where the reduced spacing saturated at 23% at 154 MPa, only 16% reduction was
observed at lower stresses (Figure 8.18.b).

Contrary to the observations in the as-HIPed shear+compression specimens, the
reduction in a2 spacing was smaller (18%) in the 650 °C 276 MPa tensile heat treated
specimen, but this specimen had a lower undeformed lamellar spacing value. The
deformed lamellar spacing (0.61 pm) in this higher stress specimen is much smaller than
the spacing for the deformed 760 °C 138 MPa specimen (1.03 um). Unlike the
shear+compression specimens, the undeformed lamellar spacing was measured in the grip
section of the creep specimens, which experienced the same thermal history with
presumably no stress. The shear+compression specimens all came from the same physical
region of the casting, but nothing is known about the relative positions of the two tensile
specimens in the casting, where the initial lamellar spacing may have differed due to
differences in cooling rates, or other unidentified process parameters. In summary, the
lamellar spacing is reduced more quickly and saturates at a smaller value with increasing

StI'CSS.

189

8.4 DISCUSSION

The lamellar structure is modified in three ways due to small creep strains, by a
reduction in the 7 thickness [127,239], which can only occur by mechanical twinning
[121], as well as a reduction in the a; spacing and thickness due to a strain induced phase
transformation [125,127]. The refinement of lamellar spacing presented above is mostly
based on or; spacing changes during primary creep deformation. These changes correlate
with reductions in 7 thickness, as indicated in 8. 15 through 8.17 and in Reference 127.
Measurements of a2 spacings is the most convenient way to characterize lamellar
refinement, and it serves as an indirect representation of lamellar refinement that also
occurs by mechanical twinning in the y laths.

Mechanical twinning parallel to lamellar interfaces accounts for the lamellar
refinement during initial creep deformation. From Figures 8.17 and 8.18, the larger
reduction in lamellar spacing with lower temperature and higher stress is consistent with
the observation that more twinning was observed with lower temperature deformation in
equiaxed grains [50], due to greater self locking behavior of ordinary dislocations at
lower temperatures [245]. Since mechanical twinning is limited more by nucleation
processes that require large stress concentrations, a lesser stress is also consistent with
less twin nucleation [121]. Since mechanical twinning parallel to lamellar interfaces
correlates closely with the early process of primary creep deformation, then it is apparent
that for a given stress, much of the primary creep strain is taken up by saturation of the
easy mode of deformation twinning in lamellar regions. Accommodation of this easy
mode of deformation in surrounding y grains by mechanical twinning is evident in Figure

8.9.

190

The rate parameters in Figure 8.5 for the early process of n = 3 and an activation
energy of about 140 kJ/mol is similar to that observed in creep of PST crystals oriented
for shear deformation on planes parallel to the lamellar interfaces [165,240]. This low
activation energy is consistent with pipe diffusion processes as well as thermally activated
glide of ordinary dislocations [245]. Once this initial easy mode has saturated, the relative
contribution of mechanical twinning to the total strain is substantially reduced. The
apparent activation energy is increased to a value near lattice diffusion, where the rate
limiting process for dislocation motion changes. This transition in deformation mode can
be easily correlated by comparing Figures 8.5 and 8.18.

Mechanical twinning and dislocation motion are co—deformation mechanisms in
the early stages of creep. Twin interfaces and or; lamellae are well known to provide
barriers to dislocations, so both mechanisms provide a hardening contribution to the
alloy. Luster and Morris [142] have shown that deformation modes are restricted to
twinning or dislocation motion parallel to the lamellar interface in lamellae thinner than
about 0.5 pm. Furthermore, mechanical twinning rotates the crystal into an orientation
where only superdislocations have a high Schrrlid factor for deformation [141]. Thus
more twinning refinement results in a harder microstructure, and consequently a greater
reduction in strain rate at the saturation of the early process (Figures 8.5 and 8.18).

J in has shown that the interface of a y/y twin boundary actually represents 4
atomic layers with the same stacking arrangement as the (12 phase [120]. If the twin
interface represents a nucleating process for precipitation of 0L2, then these initially 4 layer

on; films may have grown in thickness, and contributed to the measured reduction in

191

average or; spacing. Table 8.3 shows how typical diffusion distances vary with time and

temperature [246].

Table 8.3 Diffusion distances of Ti in TiAl

 

 

 

 

 

 

 

 

 

 

 

T, °C, condition t, hr J5, ,um
1260 HIP 4 16.36
1010 Heat Treatment 50 6.26
760 Creep 0.5% 289 0.554
760 Creep 0.2% 50 0.231
649 Creep 0.5% 800 0.120
649 Creg 0.2% 50 0.030

 

For the 760 °C specimen, the average 7 thickness is nearly equal to the a; spacing, and
the typical diffusion distance at the end of the experiment is 54% of the measured or;
spacing. Since there is a thermodynamic driving force to move Ti out of existing 7 layers
(see Figure 8.1), the twin interface provides a sink for the rejected Ti atoms. The typical
diffusion distance is 20% of the average 0L2 spacing for the 650 °C specimen. This is
consistent with the final 7 thickness being smaller than the 0:2 spacing, since fewer Ti
atoms could reach a newly created twin interface.

The heat treatment and strain-induced precipitation process during creep
deformation in this Si-containing alloy result in the formation of silicide precipitates
[185,212]. Two mechanisms for silicide precipitate formation were reported by Noda et
al. [185]. These precipitates could be produced during solidification through the eutectic

reaction:

L —> ,B+ Ti,Si3

192

or nucleated at the interface of y and (1.2, through the eutectoid reaction:

a2 —> y +Tz'5Si3

where the 0L2 layer was thinned by consumption via the eutectoid reaction. Since there
were no precipitates in the as-HIPed condition, the formation of precipitates during heat
treatment in this alloy is more likely to come from the or; eutectoid reaction. However,
this mechanism cannot be the only one to account for reduction of (12 thickness since non-
Si bearing alloys also exhibit a reduction in or; thickness [127,239].

The solute atoms and precipitates interfere with dislocation motion. The existence
of precipitates in lamellar interfaces inhibit interfacial dislocation motion [154,247]. The
fact that more precipitates and twins were observed in the deformed gage sections
suggests that precipitates nucleated during heat treatment grow faster and more
homogeneously with the assistance of strain. It is known that W in solution will cause
dislocation drag [188]. With deformation, a dislocation impeded by a precipitate will
provide a pipe through which the heavier elements can travel more rapidly to the
precipitate. The dislocations sweeping through the matrix can thus accelerate and
homogenize precipitation by collecting solute elements and then piping them to growing
precipitates. This phenomena can account for the more homogeneous and rapid growth of
precipitates in the strained material as well as the larger size in the equiaxed grains, where
the dislocation pipes sweep a larger volume than they could in the lamellar regions.

The precipitates interfere with deformation processes throughout the

microstructure. They have various shapes, and although some precipitates have been

193

identified as silicides, it is also possible that some are B phase, since they have similar Z
content and many morphologies are observed. W and Mo are known as a [3 phase
stabilizers [169,247,248]. The presence of precipitates clearly reduces the creep rate
compared to the as-HIPed condition, as illustrated in Figure 8.5, which shows that the
creep rates are more than two orders of magnitude smaller due to the heat treatment that
causes precipitation. However, the lower creep rate may also result from a smaller
volume fraction of lamellar grains because a smaller volume fraction of material would
deform by the early mechanical twinning process. The shapes of the creep deformation
curves indicate that the same physical mechanical twinning processes occur in both

microstructures, regardless of lamellar volume fraction and the precipitate content.

8.5 CONCLUSIONS

TEM analysis of the microstructures of creep deformed Ti-47Al-2Nb—an-O.5W-
O.5Mo-O.28i alloy shows a number of microstructural changes resulting from heat
treatment and creep deformation. Refinement of lamellar regions occurs by mechanical
twinning and a strain assisted transformation process. Mechanical twins are introduced
into both lamellar and equiaxed grains starting from grain boundary interfaces. A rapid
initial mechanical twinning process parallel to lamellar interfaces initially hardens the
microstructure, and a higher stress increases the amount of mechanical twinning and
hardening that occurs. Less mechanical twinning occurs at 760 °C than at 650 °C at the
same stress.

Precipitates nucleated by heat treatment grow with the assistance of strain or

consumption of a2. Many precipitates containing W, Mo and Si are found, but there is

194

considerable variety in the composition, crystal structure, shape, and size, which all
depend on their location in the microstructure. After heat treatment and a 0.5% creep
strain, the precipitates vary in size from 0.05 to 2 pm, and larger precipitates are found in
the equiaxed regions. The nucleation and growth of these precipitates interferes with
dislocation motion, and also hardens the microstructure. Precipitate growth is faster and
more homogeneously observed in deforming regions than nonstrained regions, because
dislocation pipes provide transport paths for solute elements that homogenize precipitate
growth. This precipitation hardening mechanism accounts for much of the excellent creep

resistance of this alloy when it was heat-treated.

195

CHAPTER 9

SHEAR +COMPRESSION CREEP DEF ORMATION IN
POLYSYNTHETICALLY TWINNED (PST)
TI-47Al-2Nb—2Cr WITH <110> AND <112> ORIENTATIONS

ABSTRACT
Stress and temperature change creep experiments under a compression shear stress state
were conducted on Polysynthetically Twinned (PST) crystals prepared from a cast and
HIPed Ti-47Al-2Cr-2Nb alloy. Specimens with <110> and <112> directions parallel to
the principle resolved shear stress in a shear+compression deformation mode, and the
deformation characteristics are compared. Transient creep of the PST crystal afier stress
and temperature changes (at 9.5~34 MPa and 760~806 °C) exhibited ‘normal’ behavior
and indicates Class M type creep deformation in TiAl. The stress exponent n changes
from 3.1 to 5.3 with increasing strain. The activation energy Q = 137 kJ/mole at 28 MPa
for the <110> PST indicates that dislocation glide aided by pipe diffusion is a likely
mechanism of creep deformation in this deformation regime. The activation energy (185
kJ/mole at 13 MPa) of the <112> PST is higher than the <110> PST crystal, but primary
creep in the <112> direction occurs much faster than for the <110> direction at a higher
stress. Primary creep at 9.5 MPa in the <112> direction caused a 25% refinement of
lamellar spacing that hardened the microstructure, indicating that mechanical twinning
parallel to the lamellar planes provided an easy initial deformation path during primary

creep. At strains near 1%, the creep strength of the two orientations becomes similar.

196

9.1 INTRODUCTION

Gamma Titanium Aluminide (TiAl) based materials are candidates for high
temperature structural applications because of their low density, high specific strength
and stiffness at elevated temperature, and high resistance to oxidation and creep. For
service at high temperature, it is very important to understand the creep behavior of TiAl
based alloys. However, only limited research has been accomplished and there are still
many mysteries regarding the creep mechanisms of TiAl. The wide range of values for
the apparent stress exponent (n=2~8) and activation energy (80~6OO kJ/mole)
[161,163,169,178,183,249] suggest that complicated mechanisms of deformation are
involved in TiAl creep behavior. Polysynthetically twinned (PST) crystals provide a
fundamental unit of microstructure of the lamellar colonies present in polycrystalline
materials. A study of creep in PST crystals should assist in analyzing and developing
theories for creep prOperties of near-gamma alloys. Experiments which remove the effects
of microstructure are very helpful to examine the fundamental mechanisms of creep in
TiAl.

Stress and temperature change creep tests under compression and shear stress on
carefully oriented PST crystals with <110> and <112> directions parallel to the principle
shear stress were conducted on the alloy Ti-47Al-2Cr-2Nb. The results of stress and
temperature change creep tests were analyzed using the power law equation and a
modified O-projection concept equation to examine creep deformation [250]. These
experiments covered a limited range of stresses and temperatures so that stress exponents
and activation energies can be measured for nominally similar microstructures to examine

the fundamental mechanisms for creep deformation. Since lamellar refinement has been

197

proposed as an important primary creep mechanism [165], microstructures of PST
crystals oriented in the <112> direction were characterized at an early stage of primary

creep.

9.2 EXPERIMENTAL DETAILS
9.2.1 Specimen preparation and shear+compression creep test

The PST crystal was grown from a rod of Ti-47Al-2Cr-2Nb that was sectioned
into a quarter-circle using EDM machining. The crystals were grown [165] using the
ASGAL optical floating zone furnace at the University of Pennsylvania in the LSRM
facility. The crystal were grown in a flowing argon environment using the same material
as seed and feed crystals. The seed and feed was translated vertically through the hot zone
to obtain a growth rate of 5 mm/hr. This PST crystal was oriented using a Laue camera,
and a 5 x 5 x 3 mm brick was cut with a diamond saw so that the {111} lamellar plane
was parallel to the 5 x 5 mm surface as shown in Figure 9.1, and the <110> and <112>
directions were normal to one of the other faces. By using an X-ray diffractometer with a
texture goniometer, the deviation from the desired {111}<110> and {111}<112>

orientation was determined to be less than 2°.

A compression cage fixture that imposes a shear traction 45° to the compression
axis was developed for evaluating creep in PST crystals under maximum resolved shear
stress state [165], as shown in Figure 9.1. The {111} lamellar planes were parallel to the
45° surface from the loading axis, and normal dislocation <110> directions or twin
dislocation <112> directions were consequently parallel to the principle resolved shear

stress. Shear and compression stresses of the same magnitude were simultaneously

198

 
   

Specimen
111

 

Figure 9.1 SheaH—Compression fixture used for creep deformation of 5x5x3 mm
PST crystal bricks. PST crystal is shown in soft orientation.

199

imposed on two PST crystals crept at various temperatures and stresses as shown in Table

9. l in air and argon atmosphere using an ATS series 2710 creep testing machine.

Table 9. 1 The conditions for each Shear+Compression creep test.

 

 

 

 

 

 

 

 

Specimen <110> PST <112> PST
Microstructure Fully lamellar Fully lamellar
Test temperature 760 ~ 806°C 760 ~ 806°C
Test stress 1,0- = 22 ~ 34 MPa 1,6 = 9.5 ~ 27 MPa
Final strain y = 0.068 7 = ~ 0.09
Test atmosphere Ar Air

 

 

 

The specimen shape change was measured using external extensometry connected to the
compression cage and SLVC transducers located outside of the furnace, providing a
displacement resolution of 1 pm. They were loaded to the desired stress in about 3
minutes. The creep specimens were cooled under 80% of the creep load at selected small
strains of 0.6%, 1.38% for the <112> PST crystal, and at the final strain. The specimen
load, temperature, and displacement-time data were recorded by a PC based data
acquisition system developed for creep tests. Assuming simple shear deformation, i.e., no
compression strain in the hard direction, 7 = (2)”2 Al/t, where Al (mm)= 2.54x10'4 AV
(mV) and t=3 mm thick sample, the shear stress and strain data were obtained with the
aid of spreadsheet and plotting programs.

For microstructure analysis, a J SM-6400 scanning electron microscope was used
at 25 kV in the backscattered electron mode. X-ray analysis was conducted by using a

Scintag XDS 2000 X-ray diffractometer with a texture goniometer.

200

9.2.2 Data acquisition and analysis

The automated data acquisition system consists of a PC with a National
Instruments AT-MIO-16 board. A resolution of 5 mV provided a shear strain resolution
of 0.04%. Multiple-channel scanning was used to obtain 1000 points per second for each
channel of data, and these values were averaged. An average value was retained when the
value was significantly different from the previously stored datum.

Based on the physical understanding of creep deformation in primary and
secondary stages, the 0-projection concept constitutive equation is described in [250]. A
variation of the 0-projection constitutive equation was used to analyze creep data. A
modification of this equation was used to account for various starting times for a
particular deformation condition following a stress or temperature increase [251]. The

modified equation for transient creep after stress and temperature increase is
8=so+A{1-6XP(-Ot(t-to))}+és(t-to) (1)

where so is an adjustable reference strain parameter, to is starting time for a particular
loading condition. a, is the steady state (or minimum) strain rate, A and on are curve
fitting parameters. In the case of transient creep after stress and temperature decrease, the

constitutive equation is
8=So+B€XP(-B(t-to))+és(t-to) (2)

Differentiating equations (1) and (2),

201

Aaexp(-a(t-to)) +35 (3)

m.
H

and

-BB eXp(-a(t-to)) + 8's (4)

03.
ll

Equation (3) implies that the creep rate during the primary stage decreases gradually
towards a “steady state” value, é S , with times greater than to, and equation (4) indicates a

gradual increase in creep rate towards a “steady state value”.

9.3 RESULTS AND DISCUSSION
9.3.1 Creep behavior

In general, the stress change experiment is used to evaluate substructure evolution
during creep deformation. The temperature change experiment is seldom used because it
is not easy to quickly control the temperature fluctuation after temperature changes. From
our observations after temperature changes, the temperature generally fluctuated in the
range of Tc (creep temperature) i5°C for a half hour before Tc became stable. This effect
may have caused uncertainties in the interpretations of transient behaviors, but the
subsequent steady state values are unaffected.

The creep curves of <110> and <112> PST crystals with stress and temperature
changes are shown in Figures 9.2 and 9.3. The lines through the data represent the curve
fit to equations (1) and (2). As the stress or temperature increases, an instantaneous strain
is followed by creep which decelerates toward the new (higher) steady-state rate. With

stress or temperature decreases, a reduction in strain is followed by transient creep which

202

 

 

 

 

 

  

0.09 ' r
PST 110
0.08 ’
8068C
0.07 O 790 C : 9, M
0.06 760°C ' w * 760°C
a. ,
.... : 8
I- .
u 7
M 6'
s. 0.04 3W”
3 4t , ., 5' 1
.fl , ., ,- ___ ___-_ ,_ a , ._, ,_
0.03 _
CD 25 MP3 3' 1.0—28 MPa 28 MPa
0.02 .- :‘ . .
‘ 2' stress changing hlstory
0.01 l' ' '
m: 22 MP3 temperature changing history .
0‘ i I 1 1 l A 4 l . . 1 1 . 1 1
0 2105 4105 6105 8105 1106 1.2106
(a) Time (sec)
10'5 . ....... 1 - T f
PST 110
10'6

A
O.
‘1

w
“17' M

 

Shear strain rate, ( s'l )

 

 

 

10‘8
11'
10'9
1
10-10 . 1 1 1 . . 1 1 1 _ 1
0 0.02 0.04 0.06 0.08
(b) Shear strain, 7

Figure 9.2 Creep deformation of <110> PST crystal in Ar (with curve fitting)
(a) strain vs. time curve (b) strain vs. strain-rate curve.

203

 

   

 

 

 

 

 

 

 

 

PST 112
0.00
no ....1.... .2 .. ”...... ..
100T m“c 790°C NVC 760"C 15:.
co. . , ‘1 “rm
2! . I
005 5 '1’ 7 . ’: 13.113?“ ”" N
‘ W 7001: 12.1.1.0- i .2 '3
0.04 ' 113:1“, 1310‘. H)”. ‘ IO U :7.
003 ~11 4‘ 2 l 3
0.02 m . 95 Hr. sinus changing history
0 0! *- |~33°b 31mm temperature clmiging history
4— 0.6“: slrm‘n
610° 2105 410‘ 6105 3105 1 10‘ 1.510‘ 210‘ 2.510‘ 310‘ 3.510“
(a) Time (sec)
10'5 . . . .
PST 1 12
A 10‘ , .
'm ‘ 12
v
,. 1 ‘ 4 6 L8— 14
8 10‘7 , . ’
2 2 10 ,
fl ._
'a‘ 3 15
:- s
45 10
:- / 7 13
5 9 1“
CI) 10'9 1 1 4
L D
10'10 g ‘ I ._._._L J_. 1
0 0.02 0.04 0.06 0.08

(b)

Shear strain, y

Figure 9.3 Creep deformation of <1 12> PST crystal in air (with curve fitting)
(a) strain vs. time curve (b) strain vs. strain-rate curve.

204

accelerates to a new (lower) steady state. This typical form of “normal” transient behavior
is consistent with the TEM observation of dislocation network observed by many
researchers, such as Jin et al. [194] in polycrystalline TiAl. They found that the creep
deformation mode in TiAl is pure metal type (Class M) and subgrain boundaries were
observed both in equiaxed gamma grains and within gamma laths in lamellar grains in the
creep deformed specimen of duplex microstructure. This indicates that stress and
temperature have a similar influence on the substructure changes.

By comparing creep curves in the <110> and <112> PST crystals, primary creep
deformation in the <112> oriented PST crystal is much faster than the <110> oriented
PST crystal, even though the stress is much lower. Primary creep processes lasted until
about 3.5 % shear strain was reached in the <112> specimen as compared to 1 % shear
strain in the <110> specimen. Since mechanical twinning has the highest resolved shear
stress in the <112> direction it apparently has a lower flow stress in creep than ordinary

dislocations, which have the highest resolved shear stress in the <110> direction.

9.3.2 Activation energy

From the slope of a plot of log é, vs. (1/T), the apparent activation energy for
creep can be calculated. Figures 9.2.b and 9.3.b show that the transient creep was usually
exhausted so that a well defined steady state value could be obtained from the curve fit
parameter in equations 1 and 2. The activation energy QC: 137 kJ/mole for <110> PST
crystal and QC: 185 kJ/mole was obtained in the temperature range of 760~806 °C, as
shown in Figure 9.4.a. Polycrystal samples from which the PST crystals were grown

exhibited much higher activation energies between 229 ~ 574 kJ/mole in samples

205

 

1 0'6

—L
q
N

Minimum strain rate, (s'l )

 

f v r l f

Activation energy for PST TiAl

 

 

 

0 PST <112> *
g PST <110> Qc = 137 KJ/mole J
...... for r,o=28 MPa
37;; ........ 6' <110> PST

   

Qc =185 KJ/mole

 

 

 

 

 

 

 

 

 

 

for t,o=13 MPa 7
1043L <112> PST 5 .~
9 A 9.2 A l 9.4 A 9.6 l 9.8 10
4 1
(a) 10 /T (K )
10‘6
Stress exponent for PST TiAl
- PST <112>:n= 1.6~5.1
A 1
“TV, _; PST <110>:n= 3~5.3
v l . , l0'
o“ 10'7 2' 3 '
a .1. ‘
h
.E 3 1'
a + o
E “a n = 1.6 ' l4 ‘
5 10 ' ' 0 12 0 ' “
E n= 3.2 10 13
. OE .
E 11
10'9 . ,. . .
8 9 10 20 30
(b) Shear stress, 1 (MPa)

Figure 9.4 a) Temperature and b) Stress dependence of minimum creep rate for PST
crystals. Numbers next to datum points refer to Figures 9.2 and 9.3.

206

deformed in the same way [251]. The lower activation energy may be due to the presence
of an easy mode of deformation that is not present in polycrystals which exhibit activation
energies typically between 250-450 kJ/mol. The activation energy of the <110> PST
crystal is similar to the activation energy of 150 kJ/mole for interdiffusion in y—TiAl
which was reported by Ouchi et al. [252]. Pipe or grain boundary diffusion is commonly
understood to occur with an activation energy of about half of the lattice diffusion energy,
which is consistent with the measured values in the PST crystal. Also, an activation
energy value of 130 kJ/mole for thermally activated glide of screw dislocations in y—TiAl
at room temperature was reported by Appel et al. [245]. However, the activation energy
of <112> PST crystal is higher than that of <110> PST crystal, which indicates that the
creep deformation mechanisms are either dependent on stress, or significantly different in

these orientations.

9.3.3 Stress exponent

According to the power law equation, at each temperature, the stress exponent is
given by the gradient of the log é s vs. log a plot. This gradient is shown for each steady-
state strain-rate that resulted from the stress changes as the experiment progressed, as
shown by the lines in Figure 9.4.b. Thus, by comparing the minimum creep rates
following stress changes in the strain range of 1~3% and 6~7%, the stress exponents
n=3.1 and n=5.3 were observed in the <110> PST crystal, and n=1.6, 3.2, and 5.1 were
observed at strains of 3~4%, 6~7%, and 7~9% in the <112> PST crystal, respectively.

Furthermore, the <112> PST crystal is softer than the <110> PST crystal, since these data

207

lie above the trend of the <110> PST crystal data. Both specimens exhibited a hardening
trend with increasing strain.

The stress exponent of n=3 is characteristic of dislocation glide creep and n=5 for
dislocation climb creep. In the <110> PST crystal orientation, there are two possible ways
to explain a stress exponent of n=3. One is to assume that steady state was not reached in
the strain range of 1 ~3%, so as to discredit this measured value. However, minimum
strain rates are commonly observed in the range of 1-3% strain in many studies. In the
case of the PST crystal deformed in the <112> direction, n=1.6 was measured in the range
of 3 ~4% strain. Due to the large primary creep strain in the <112> specimen, the lack of
an early steady state may account for the low n value. Stress exponent values increased
with stress, such that both crystals appear to approach n=5.1~5.3 mechanism at larger
strains.

From the interpretation of the steady-state creep rate and the fitting parameters of
equations (1) and (2), the stress exponent of n=5.3 for <110> PST crystal can be
rationalized as being due to the dislocation glide (n~3) aided by pipe diffusion which
increases the stress exponent by 2. The occurrence of pipe diffusion is supported by the
low activation energy Qc=l35 kJ/mole. In <112> PST crystal, the stress exponent of
n=3.2 may result from dislocation glide creep that is not dependent on pipe diffusion, due
to the lower stress or strain where n=3 was measured. Also, the activation energy may be
stress dependent, since a higher activation energy may be needed for pipe diffusion if the

stress is lower.

9.3.4 Microstructure and lamellar spacing

208

Microstructures were examined in the <112> PST sample. Figure 9.5 shows
backscattered electron images of a deformed (about 1.38 % shear strain) <112> PST
crystal. Very fine lamellar microstructures consisted of y (darker phase) and 012 laths
(brighter phase). Especially fine lamellar features are found in 012 lathes > 1 um, as
indicated in the upper left side of Figure 9.5.a. The 012 spacing (distance between 012-
y interfaces) varies between 0.025~4.25 um. Cross twinning was only observed in the
1.38% shear strain specimen in Figure 9.5.b. This indicated that the easy mode of
deformation twinning parallel to lamellar interfaces preceded cross twinning.

The lamellar spacing was characterized from about 500 measurements in more
than 20 lamellar regions from the side <110> surface in the undeformed and deformed
samples. There is an increase in smaller 012 spacings with strain, as shown in the
distribution in Figure 9.6.a. The average on; spacing decreased from 1.1 to 0.7 pm in
Figure 9.60 There is a 13% reduction in lamellar spacing at 0.6% and 28% reduction at
1.3 8% shear strain. However, the average 012 thickness does not change much during
primary strain. This refinement of spacing is consistent with previous studies, and the
amount of reduction depends on the composition, interstitial content, heat treatment, and
deformation temperature and stress [111,217,141,205,212,239,253]. Figure 9.3 shows
that the minimum strain rate at loading condition 2 decreases by one order of magnitude
even after increasing the stress from 9.5 to 11 MPa. A similar phenomena occurs at the
increase from 18 to 20 MPa. This anomalous reduction in creep rate is probably the result
of further reduction of lamellar spacing that further hardens the microstructure. This
indicates that initially, twinning strain in the <112> direction is very easy, but with the

consequent lamellar refinement, the <112> crystal hardens dramatically, as compared to

209

shear direction

'\

cross twins

 

Figure 9.5 Backscattered electron images of deformed <112> PST crystal
(1 .38% shear strain at 760°C and 13 MPa).

210

so c we ,.-c,.s..e_-

 

 

 
  
 
  
  
   

 

 

 

 

 

 

 

 

 

 

 

 

 

PST 112 l
* ““- Undeformed ‘
40 - . 1
average (12 spacmg: 1.05 um .
>> "I"Deformed (0.6% strain) l
g 30 ., average (12 spacing: 0.91 pm j
g +Deformed (1.38%\strain)
5 average a2 spacing: 0.74 um 1
r...
m 20
10 ~
0 1 1 1 1 1 . . 1 , L A , .
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
(a) 012 spacing ( um )
1.2 a
I PST 112 _ __ . l
1-1 .7 Average 0L2 spacmg
’ 1
E 1 , ‘I'Average 012 thickness 9
:5.
V 0.9
on
.E‘.
g 0.8 1
Q.
m
o 0.7 .-
ea ------------------ I -----------------------
E '1 i i
3 0.6 [
0.5
0.4 a . . - . . - m - . . . . . .
0 0.5 1 1.5 2
(b) Shear strain, 7 (%)

Figure 9.6 (a) or; spacing distribution, and (b) average 012 spacing and thickness with
different primary creep strain levels in the <112> PST crystal.

211

the <110> crystal.

Luster and Morris have shown that only deformation modes parallel to the
lamellar planes are operative in y lamellae thinner than about 0.5 pm, which makes thin
lamellae more resistant to deformation [142]. Improvements in creep resistance have been
correlated with initial lamellar spacing that was reduced by heat treatment [127]. These
results indicate that refining the lamellar spacing is desirable for improving primary creep
resistance. Recent results indicating that primary creep deformation is retarded by pre-
straining [159] can be easily understood in terms of the lamellar refinement observations

discussed above.

9.4 CONCLUSION

‘Normal’ transient creep after stress and temperature changes were observed in
the <110> and <112> PST crystals. This indicates that Class M type creep deformation
and subgrain structures were formed during the transient creep processes. The stress
exponent of n=5.3 at 22~34 MPa and activation energy QC: 137 kJ/mole (in Ar) at
760~806°C for the <110> PST crystal suggest that dislocation glide is aided by pipe
diffusion in this microstructure and crystallographic orientation. For the <112> PST
crystal, the stress exponent of n= 3.1 at 13~l8 MPa and Qc=185 kJ/mole (in Air) at
760~806 °C suggest that the creep deformation mechanism is limited by dislocation glide.
However, at a higher shear stress glide limited by pipe diffirsion can account for an
increase in the stress exponent to n=5.3. The activation energy of the <112> PST crystal
at 13 MPa is higher than that of <110> PST crystal at 28 MPa, suggesting that the

apparent activation energy may depend on stress. The refinement of lamellar spacing

212

(13~23 % reduction) during primary stage was observed. Reduction of lamellar spacing
contributes to strain in the early primary stage, and also hardens the crystal. After 1.38%

shear strain, hard mode deformation (cross twins) was activated in some lamellar laths.

213

CHAPTER 10

THE EFFECT OF HEAT TREATMENTS ON MICROSTRUCTURES AND
PRIMARY CREEP DEFORMATION OF FOUR INVESTMENT CAST
TITANIUM ALUMINIDE ALLOYS

ABSTRACT
Several heat treatments were developed and applied to investment cast Ti-45Al-2Nb-
2Mn(at%)+0.8v%TiB2 XDTM, Ti-47Al-2Nb-2Mn(at%) + 0.8v%TiB2 XDTM, Ti-47Al-
2Nb-2Cr(at%), and Ti-47Al-2Nb-lMn-0.5W-0.5Mo-0.28i alloys in an effort to enhance
creep properties and decrease heat treatment time compared to current practices. Results
show that the creep resistance of WMoSi can be improved by 10 times with heat
treatment, and 47-2-2 can be improved by 3 times, and the XD alloys can be improved
slightly, or not at all with lower Al level. The variation, or lack of variation, in creep
resistance with heat treatment can be explained by differences observed in the
microstructures and textures produced by the various heat treatrnents. Quantitative
microscopic comparisons were made between microstructures in undeformed and
deformed regions in creep specimens deformed to strains between 0.1 and 1.5% strain,
using optical, SEM and TEM techniques. The lamellar spacing in lamellar grains
systematically decreased by 15-35%, with increasing stress, during the first 0.1-2% strain.
Precipitates containing W, Mo, and/or Si nucleated and grew at a faster rate in the
deformed part of the specimen, as compared to the grip section, with approximately twice

the volume fraction in the deformed region, and the precipitates were generally smaller

214

and more homogeneously dispersed. These observations indicate that strain assisted

nucleation of precipitates accounts for much of the excellent creep resistance of the

WMoSi alloy.

10.1 INTRODUCTION

Cast near gamma titanium aluminides are gaining acceptance as potential
replacements for superalloy and steel components in aerospace, automotive and industrial
applications since it has high specific strength at high temperatures [92,93,128].
Components cast from these alloys can operate at temperatures up to 815 °C (1500 0F) at
half the weight of the components they replace. Most applications of cast gamma
components require good primary creep resistance to meet component life requirements.

The microstructure is an important factor that is sensitive to composition, heat
treatment, and strongly affects creep properties. Gamma titanium aluminide can be heat
treated to obtain four types of microstructures [110]. The best creep resistance has been
observed in fully lamellar microstructures [161]. However, the duplex microstructure is
preferred for many applications since it provides desirable room temperature ductility
[92].

The addition of Cr, Nb, V, Mn, W, Mo and Si, in two phase Ti-Al alloys is also
an important factor in order to control the material properties. Cr can improve the
ductility of duplex alloys with additions of 1~3% at%. [225,226]. The additions of 1~3
at% V and Mn are the most common elements for improving plasticity in two phase TiAl
alloys [143,227-229] but V generally reduces the oxidation resistance[230]. Nb greatly

enhances the oxidation resistance [231-233] and slightly improves the creep resistance

215

[232]. The addition of 0.5~1 at.% Si enhances creep strength, the oxidation resistance,
and room temperature fracture toughness of 48 at% Al two phase alloys [208,234-236]. Si
also increases fluidity, and reduces the susceptibility to hot cracking [237]. Mo provides a
good balance of strength and ductility in TiAl having very fine equiaxed gamma grains
with the 012 and [3 phase [210.254]. Addition of W greatly improves oxidation resistance,
and enhances creep resistance [169].

Most of these studies have focused on either room temperature properties,
minimum creep rate conditions or stress-rupture properties. However, primary creep
resistance is important for practical applications. Gamma based TiAl exhibits primary
creep strains that are typical for metals, where the minimum is reached after about 1%
strain. In an effort to decrease the creep rate during primary creep deformation, additions
of W, Mo, and Si have increased the time to reach 0.5% strain due to a dynamic
precipitation process [231,23 8]. In several related studies, observations of lamellar
refinement indicate that primary creep deformation includes a significant amount of
mechanical twinning parallel to lamellar interfaces that occurs as an easy mode of
deformation that hardens the microstructure at low strains [50,165,212,213]. In this
chapter, the primary creep resistance (time to 0.5% creep strain) is compared between
four alloys having different kinds of strengthening strategies, including ceramic particles,
solid solution elements and precipitates, which affect the ways the microstructure can be
manipulated by heat treatment. After primary creep to 0.5% we quantify changes in the
different lamellar and equiaxed microstructures, 7 thickness and a; spacing, mechanical
twinning, interstitial composition, precipitate size, composition, shapes, and dislocation

interactions with precipitates.

216

10.2 MATERIALS AND EXPERIMENTAL PROCEDURE

The investment cast gamma TiAl alloys used are described in Table 10.1. Some
specimens with varying interstitial content were investigated and they are hereafter
identified as med-O or high-O alloys. All materials were hot isostatically pressed (HIPed)
at 1260 °C for 4 hours at 172 MPa (2300 °F, 25 ksi) in order to eliminate casting
porosity. The heat treatments used were in the 012+y phase field at 900 or 1010 °C(1650 or
1850 °F) for 10, 20 or 50 hours, and cooled in a static argon atmosphere (SAC). A two
step heat treatment was also used consisting of 1177 °C (in the or+y phase field) for 5
hours and then in the 012+y field for 10 hours at 1010 °C (2150 and 1850 °F), respectively,
and cooled in static argon. These heat treatments all provided duplex microstructures.
These specimens were then subjected to creep tests at 760 °C, 138 MPa (1400 °F, 20 ksi)
and 649 °C, 276 MPa (1200 °F, 40ksi). The creep specimens were cooled under load once
0.5 % strain was reached. Some additional 5x5x3 mm brick shaped specimens of Ti-
47Al-2Nb-1Mn-0.5W-0.5Mo-0.28i alloy were cut using EDM machining from a piece of
an as-HIPed casting gate for compression-shear creep tests. They were deformed on a 45°
inclined stage in a compression cage fixture developed for evaluating creep in a PST
crystal [165].

Compression-shear creep tests were conducted at 650 and 750 °C with shear and
compression stress components of 61, 91.5 and 122 MPa in air using an ATS series 2710
creep testing machine. Tests were stopped once 0.2-1.5% creep strains were reached. The
creep data were recorded by a PC based data acquisition system developed for creep tests.

The specimen shape change was measured using external extensometry connected

217

 

an no 885 ._o> ..

 

 

 

 

 

 

 

 

 

 

 

 

no no _ N No 2. 835 5023
a; m N a. 833m ex :5.
3o 23 33 $2 2: 8.0 a: a: 3.0 2.: 8.? 0.3:
o; mad 33 RS ...: So 3; 8d :5 3.3 mm? 0-88
6 so:
:3 23 £3 83 m2 mod 2: 2: :3 2.9. «2:. 935.
N N t. 8:28 3-:
an : z o m 3 o: co :2 z a 2 fl 83558..

 

 

 

 

 

 

 

 

 

 

Ao\o~mv PAC—mm ~<£1 ammo “GOEHmeCm MO GomwmmOQEOU ROMEDSO fimfimeoz —.O~ D—Dmy—r

218

to the compression cage and SLVC transducers located outside of the fumace. Specimen
load, temperature, displacement, and time data were reduced, assuming simple shear
deformation with the aid of spreadsheet and plotting programs.

For optical and SEM analysis, deformed samples were obtained from the gage
section of the creep deformed specimen. Some undeformed samples were also cut from
the grip section of tensile test specimens. A modified Kroll’s reagent [222] was used to
reveal the microstructure in this material. For analysis of different phases, polished and
unetched specimens were used in a J SM-6400V scanning electron microscope (SEM) at
25kV, using the Backscattered electron (BSE) technique. For micro-analysis by
transmission electron microscopy (TEM) the specimens were prepared as follows: 0.7
mm thick slices were taken from the specimen using a diamond saw. 3 mm discs were
produced using an EDM cutting machine such that the foil normal was parallel to the
tensile axis. After mechanical polishing thin foils for electron microscopy were prepared
by the twin jet polishing method using a solution consisting of methyl alcohol and 10%
sulfuric acid at 20 volts at -25 0C. These foils were examined in a Hitachi H-800
transmission electron microscope operated at 200kV. Texture measurements were made

on selected specimens and analysis was made using popLA software [214].

10.3 RESULTS AND ANALYSIS

10.3.1 Primary creep behavior
Tensile creep

The time to 0.5% creep of the four alloys varies with composition and heat

219

treatment as shown in Figure 6.2 and 7.1. The average value of three samples is plotted
with the line, and high and low values are also plotted. From this plot, it is apparent that
the 45 Al composition is insensitive to heat treatment. The 47 Al (low 0) samples exhibit
different responses to heat treatment. With TiBz, the heat treatment for 10 hours at 900 °C
is damaging to primary creep resistance, but otherwise there is no significant
improvement with increasing time at temperature in heat treatment, other than a
beneficially smaller range of values with the two step heat treatment. But for the 47A]
with high 0 alloys, creep resistance is significantly improved with increasing amounts of
O. The alloys with medium and high 0 exhibit creep times to 0.5% that are more than
twice as long. The scatter is similar in magnitude for all alloys except the low 0 alloy
after the two-step heat treatment. For the 47-2-2 alloy, the creep resistance is substantially
improved with increasing amounts of time and temperature in heat treatment, but the
variability in the samples is not significantly improved with heat treatment. For the
WMoSi alloy, the primary creep resistance is the best (295 hours). In WMoSi alloy,
tensile creep deformation at low temperature and high stress (649 °C, 276 MPa), the
material deformed quickly up to 0.2% strain, where the strain rate slowed considerably.
At the higher temperature and lower stress, (750 °C, 138 MPa) the trend was similar, but
the change was less dramatic. This effect is clearly shown in Figure 8.5, and 10.1. The
strain rate vs strain curve was obtained by differentiating a least squares curve fit of the
creep data using the equation (a =a(1-exp(bt))+c(l-exp(dt))) [205]. All alloys exhibit a
rapid decrease in strain-rate below about 0.2%. The creep rates of the higher 0 alloys
were lower and decreasing more rapidly at 0.5% strain. At low strains, the creep curves

for both higher 0 alloys cross at a strain of 0.1%, indicating that the high 0 alloy had a

220

 

10 PfifYTrYTrYrYTYfifo IIIIIIIIIIIIIIIII Y IIIIIIIII [WYYV

 

—45'I'1B2 As HIPed

- - 45118:I Two Step

 

—I
0.
N

— 47TiB1 As HIPed

- - 47Ti132 Two Step

 

 

' _47TiBz+MO 1010050115

-._‘ ~ _. _. _ _ _ M47TiBz+HOIOIGOSOhrs ]

Strain rate, 8° (s'l)
3..
‘1” .
’-

....-._____M_. — 4722 As HIPed

’ Transition point “'“ 4722 9001020111: l
* (Saturation strain) ‘

 

 

 

 

 

°90 ‘ ‘ 0.101011 00513303 3031‘ 01010510000 aro‘o‘y‘ retrain; 10.01
Strain 8
Figure 10.1 Differentiated curve fits of creep data exhibit a decreasing strain rate to a
saturation strain between 0.1 and 0.5% strain, whereafter the decrease in
strain rate is less rapid.

 

221

faster initial creep rate. The WMoSi alloy also shows a cross over in the curve creep
around at 0.2% strain due to different stress and temperature conditions shown in Figure
8.5. The most dramatic reductions of strain rate occurred in the alloys having better

primary creep resistance.

Compression-shear creep

For the larger strain compression-shear creep conditions in Figure 8.3.b, the initial
creep rates are similar, but the 750 °C specimen reached 0.8% about 20 times faster than
the 650 °C specimen at the same stress. From Figure 8.5. the initial creep rate and shape
of the creep deformation curve for stresses larger than 77 MPa is similar, indicating a
rapid decrease in the creep rate. At 77 MPa, the strain rate decreases more slowly to an
initial saturation strain that is reached at a larger strain. At the lower temperature, the
saturation of the initial creep rate is reached at a smaller strain, and the decrease in strain
rate is larger.

The initial saturation strains depend on the temperature and stress. A higher stress
at the same temperature or a higher temperature with same stress make the saturation
strain shift toward smaller strain. The curvature of the creep curves is similar at the same
temperature or stress level. A higher stress results in a greater drop in the strain rate at the
saturation strain. The effect of heat treatment on the creep rate for the WMoSi alloy

reduces the creep rate by more than a factor of 10.

222

10.3.2 Microstructures

Ti-45Al-2Nb-2Mn-0.8TiBg

From optical micrographs, a very fine duplex microstructure was observed in all
specimens as shown in Figure 10.2.a, regardless of heat treatment. There is a
homogeneous distribution of 10-50 pm equiaxed and 20-100 pm lamellar grains. The

volume fraction of lamellar grains generally decreases with increasing amounts of heat

treatment (Figure 10.3).

From analysis of backscattered images of undeformed and deformed specimens
obtained in the SEM, there are no distinguishable changes in the microstructure resulting
from creep deformation. Optical micrographs of the crept specimen show that many TiBz
particles are distributed randomly. There are several types of particles such as equiaxed,
needle, and irregular shaped precipitates with a spacing of 30-50 pm, and they are located
inside gamma grains, in grain boundaries between gamma (dark phase) and or; (bright
phase) grains and in lamellar grains, but not inside 012 grains, as shown in Figure 6.4.d. As
annealing time increased no distinct change in the particles was observed, but in the case
of the two-step heat treatment there appeared to be fewer particles. Single phase 7 and 012
grains represent about 40% of the volume fraction. The 012 grains have angular shapes,
but the y grains tend to be more round and equiaxed. This indicates that equiaxed y grains
nucleate in several places in a given parent 012 grain, and that some of the parent 012 region
transform into lamellar grains, but others do not. Very fine lamellar features are observed
in some 012 grains in both deformed and undeformed specimens (see arrows in Figure

6.4.d). These regions also exhibit contrast indicating formation of extremely fine

223

 

Figure 10.2 Microstructures of As HIPed 45Al XD (a) optical micrographs before
creep test.

224

lamellar microstructures consisting of both 012 and y lathes. Since they are much finer than
the obvious lamellar grains, they are not likely to have resulted from thermally induced
transformation, so they are more likely to have resulted from local stress concentrations
[47,50,213]. These features were observed in all heat treatments. Since the 012 grains have
an angular shape, stress concentrations are more likely to be focused at triple points in 012
grains, where fine mechanical twinning is most often observed [47,50,213]. The fine
twinned regions are also observed in TEM images. They occur in the angular shaped 012
grains, as illustrated in Figure 6.5.a. Regions of extremely fine mechanical twinning are

also observed in some lathes of the lamellar grains (Figure 6.5.b).

Ti-47Al-2Nb-2Mn-0.8TiBg

Microstructures of as-HIPed specimens exhibit a heterogeneous nricrostructure
illustrated in Figure 10.4 a. There are clusters of small equiaxed grains in some regions
where there are fewer lamellar grains, and conversely, a few fine 7 grains are observed in
regions dominated by lamellar grains. After heat treatment the microstructures exhibit
fewer small grain clusters. The volume fraction of equiaxed grains changes, and after the
two step heat treatment, the lamellar volume fraction increased slightly as described in
Figure 10.3. As time and temperature of heat treatment increase, microstructures became
more homogeneous. The equiaxed grain size distribution becomes more homogeneous,
and fewer clusters of finer y grains were observed. The microstructures of medium 0
alloys are similar to that of the low 0 alloy shown in Figure 10.4. a. The low and medium
0 alloy has a lower volume fraction of lamellar grains than the high 0 alloy (25%, 22%

and causes more lamellar grains. However, this result is consistent with the observation

225

._
<2
<2

 

1 1 1 1 1

 

—. — Ti-4SAlthb-2Mn—0 8T113 1
l- Ti-47Al-2Nb-2Mn-0 8TiB ‘

A T1-47Al-2Nb-2Cr
Ti-46AI ( medium 0 2 ) with TiB:

 

 

 

CI
I , 0 Ti-47AI( high 0 . ) with TiB I
v

 

1‘1—47Al-2Nb-an- SW- SMO- ZS

 

 

 

 

Volume Fraction oflamellnr structure (%)

 

 

 

 

 

. , ’5’
20 p I. / .............. z
i i i 1
As HIPed 900°C/ 90000 lOlG‘C/ lOl()°C/ “7700
thrs/SAC 20hrs/SAC thrs/SAC 50hrs/SAC 5hrs/lo/
10l0°0
Heat Treatment ”hm/SAC
Figure 10.3 Volume fraction of lamellar grains in TiAl alloys with respect to
heat treatment.

 

Figure 10.4 (a) optical micrographs of As HIPed 47A] XD before creep test.

226

that oxygen is an alpha phase stabilizer in Ti alloys [143,223]. From Table 10.2 the higher
0 levels have fewer small grains, which correlates with improved creep resistance [127].

Backscattered electron (BSE) images of the higher 0 alloys afier creep
deformation in Figure 7.4.a and b illustrate differences in 012 morphology (white phase,
see arrows). The largest 012 regions in the high 0 alloy are smaller than those in medium
0 alloy, and angular shaped 3-7 pm 012 grains (arrow) were commonly observed in the
medium 0 alloy (it also has lower aluminum). Angular 012 grains correlate with faster
primary creep [127]. In the alloys with boron, two types of boride particles were found,
needle and round; the round type is 1.5-3.5 um, and the needle type is 3-60 pm long. The
needle particles have been identified as TiB and equiaxed particles as TiB; [224]. The
heat treatment or interstitial content did not affect the size and distribution of these
particles.

From analysis of SEM BSE images and TEM images, these particles appear to
provide nucleation sites for lamellar lathes (fine lamellar colonies were frequently found

near particles), and they also block and entangle dislocations.

Ti-47Al-2Nb-2Cr

In contrast to the microstructures containing TiBz, the as-HIPed microstructure
exhibits very large (up to 500 um) lamellar grains and about 200 um pockets of equiaxed
10-20 pm equiaxed 11 grains (Figure 10.5 a). The microstructure is extremely
heterogeneous, and it exhibits some of the same types of heterogeneities (clusters of small
equiaxed grains) observed in Figure 10.5 a. As shown in Figure 10.3, with heat treatment,

the volume fraction of the lamellar grains increases and decreases with increasing time

227

Table 10.2 Grain sizes of four investment TiAl alloys (unit: um).

 

 

 

 

 

 

 

 

0.5W-0.5Mo-0.2Si

 

 

Alloys lamellar equiaxed
Ti-45Al-2Nb-2Mn-0. 8TiB2 20~50 10~50
Ti-47Al-2Nb-2Mn-0.8TiB2 20~200 10~50

(low 0)

Ti-46Al-2Nb-2Mn-0.8TiB2 20~200 1 0~50
(medium 0)
Ti-47Al-2Nb-2Mn-0.8TiB2
(high 0) 100~200 25~100
Ti-47Al-2Nb-2Cr 50~500 20~200
Ti-47A1-2Nb-1Mn- 100~500 5~200

 

228

 

 

Figure 10.5 Microstructures of As HIPed 47-2-2 (a) optical micrographs before
creep test (b) BSE images afier creep test.

229

and temperature, and the microstructure becomes less heterogeneous. The pockets of
equiaxed grains become fewer and there are fewer fine grains in the pockets.

In the backscattered electron SEM images of this alloy there is a smaller volume
fraction of angular equiaxed 012 grains between gamma grains (Figure 10.5 b). The very
fine twins seen in the 45 Al sample (Figures 6.4.d, and 6.5.a and b) were not observed in
the smaller 012 equiaxed grains in the 47 Al specimens. Very small bright phases can be
seen either in 012 equiaxed grains or boundaries of gamma grains (indicated by arrows)
and these phases are [3 phase which is produced by addition of Cr. They have an
orientation relationship of (0001) 012 // {111} y// {110} B, and <1120> 012/l <110> y //
<111> [3 [215]. Since Cr is known to strengthen the 012 and 7 phases, this may account for
the lack of mechanical twinning observed in the 012 phase in the Ti-47Al-2Nb-2Cr alloy
[215]. There appears to be a decrease in the amount of B phase with the increasing
amounts of heat treatment.

TEM investigations indicate that a 170° angle relationship between lamellae is
frequently observed in the as-HIPed and heat treated conditions. Typically, one of the two
lamellar lathes has a finer spacing described in the reference [127]. Subboundaries
developed in larger regions of lamellar grains during high temperature deformation. No
subgrains had developed fully in equiaxed grains even though there were regions
exhibiting a dislocation density greater than 2 X10'2/m2. After the two-step heat
treatment, dynamic precipitation was observed in grain interiors and interfaces, and this

may account for some the improved creep resistance [212].

230

Ti-47Al-2Nb-1Mn-0.5W-0.5Mo-0.2Si

The microstructure of the undeformed as-HIPed alloy showed a similar feature as
shown in Figure 8.6.b. Since the strains were small, there was no obvious change in the
microstructure after creep deformation. However, reductions in the lamellar spacing were
systematically observed in higher magnification observations after deformation. Figure
8.8.b shows the microstructure of the tensile crept specimen in the gage section with
equiaxed grains ranging from 5-200 um. The lamellar colonies are on the order of 100-
500 pm, but many have smaller equiaxed grains interspersed among the lamellar grain
boundaries. These groups of small eqiuaxed grains are produced during HIP treatment
due to large local strains associated with pore closure. The volume fraction of lamellar
grains in the two crept tensile specimens are 21:2% and with the balance being equiaxed
grains. Macroscopically the only differences between specimens deformed at 750 °C, 138
MPa and 649 °C, 276 MPa, were the observation of more twins in the larger equiaxed
grains of the deformed sample compared to the microstructure in the undeformed grip
section. Also, there appeared to be more twinning at the lower temperature, higher stress
condition.

Backscattered electron images in Figure 8.8.b indicated that the [3 phase (whitest
phase) were found in grain boundaries between y grains and lamellar grains or in lamellar
colony interfaces in both the as-HIPed and heat treated specimens as observed in a related
study [185,212,236]. As shown in Figure 8.12.b, precipitates in the heat treated alloy after
HIPing are located in various places such as lamellar interfaces, grain boundaries, and
within equiaxed grains, and precipitates were found only in the heat treated alloy. The as-

HIPed deformed specimens exhibited no precipitates. This indicates that subsequent heat

23]

treatment facilitates the formation of the precipitates. The precipitates in lamellar
interfaces vary from 0.05-0.45 pm and tend to be smaller, and many of them have a
needle shape (Figure 8.12.b). In equiaxed grains, precipitates are larger, and tend to have
an aspect ratio of 2-3. Thus the precipitates have various shapes with dimensions ranging
from 0.2-2 um. These precipitates have been identified as Ti5Si3 in several studies
[185,212,236]. After analyzing the distribution of precipitates in at least 5 lamellar grains,
about 35% more precipitates were found in the deformed sample. This observation
suggests that precipitates were formed with the assistance of strain during creep
deformation. These observations taken together indicate that precipitation is facilitated by
heat treatment, and that a more homogeneous nucleation and growth process occurs in the
presence of strain. As with TiB; particles, the precipitates always have a high density of

dislocations in their vicinity.

10.3.3 Lamellar spacing

From TEM analysis, twins appear to have initiated from a grain boundary at the
boundary region between an equiaxed and a lamellar grain and propagated through
equiaxed grains. 250-500 012 spacing counts in more than 20-25 lamellar regions and at
least 100 7 thickness in about 6 different regions near grain boundaries before and after
creep deformation were used to compute the average of 012 spacing and 7 thickness before
and after creep deformation. The 012 spacing was measured by using BSE imaging. These
measurements are upper bounds of the actual spacing, since the lamellar plane normal
direction was not determined to make the parallax correction needed to determine the

actual 012 spacing. However, this is a systematic way to compare 012 spacings in

232

specimens with similar microstructures. The TEM was used for measuring the true 7
thicknesss and 012 thickness (by tilting the foil to so that a <110> direction in a lamellar
plane was parallel to the beam) and about 50 012 laths were counted in several lamellar
regions for this measurement.

In the case of Ti-47Al-2Nb-2Cr alloy, by using the dark field method and tilting
techniques, the 7 thickness, 012 spacing and the thickness of 012 laths in several lamellar

grains were measured. The data was obtained by measuring about 100-350 laths and

taking the average.

Ti-47Al-2Nb-2Mn-0.8TiB;

The 012 spacing was reduced in these alloys after creep deformation and this
reduction varies with heat treatment and interstitial content [127,239]. The average 012
spacing before deformation in the alloy with high 0 is 1.37 pm, which is nearly as large as
the 1.57 pm spacing of the low 0 alloy which showed the poorest creep resistance. The
average 012 thickness is larger (0.2 pm) in the high 0 alloy than the low 0 alloy (0.15 pm).
This also suggests that oxygen stabilizes the 012 phase.

The distribution of 012 spacing and y thickness before and after creep deformation
are shown in Figure 10.6.a. From this plot, the reduction of the 012 spacing and y thickness
after deformation at the early stage of creep can be easily visualized since the dashed line
shows the highest frequency of lamellae with fine spacing. During the first 0.5% strain,
more reduction in the strain-rate correlates with more reduction in lamellar spacing in the

low 0 alloy. The high 0 alloy does not follow the same relationship between the

233

Y! T IYYTrITYTVTY—YTfIV

Vvvr

 

 

 
 

 

 

 

 

 

 

 

 

 

 

50a..fi--..,..fir.
l 9
:"z Ti-4 Al-2 b-2 n+0.8%TiB 2 with high 0
40 '5 Temion Creep ,
:l 3. Q l
’3‘ j ' :' ‘\ +Undeformed (a spacing) l
o\ ' 1 2 «
V 30 [- .' .' Q‘ j
>, :' , g. ‘. "@"Deformed (012 spacing)
g .' 2 : . ‘. ]
a 20 ~o -: + Undeformed (7 thickness) i
E [' : V. ’ ‘0 ‘ ' Deformed (7 thickness) 1
15 ;' i
10 1' I: “0 ‘1
1' " ’ '0. A
0 I. §A_ ..1 r 1 .JAALL.LT:E‘1'rr-;h-_ . .
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
(a) Spacing ( 11m)
50 n 1 1 T I r r
1 Ti-47Al-2Nb-2Cr
1 Tension Creep 0.5% .
40 5 ‘
4— 'Y "9"Deformed
g +Undeformed
v
E
3-
2
k.

 

 

 

1.5 1.75 2

1 1.25

0.75
7 thickness (um)

Figure 10.6 Distribution of 012 spacing and y thickness in TiAl alloys before

and after creep deformation (a) high 0 (b) 47-2-2.

234

reduction in spacing and the reduction in strain-rate: The high 0 alloy exhibits a larger
reduction in strain-rate, and a greater amount of reduction in they thickness. The high 0
alloy exhibits more reduction of 7 thickness than the alloy heat treated at 900 °C for 10
hours, which gave poor creep resistance. If the reduction in lamellar spacing occurs by
mechanical twinning, then more twinning appears to cause a greater amount of strain at
early times. The effect of the high 0 is that more refinement in the 7 thickness occurs in

primary creep, but it occurs more slowly.

Ti-47Al—2Nb-2Cr

The 012 spacing and 7 thickness decrease with heat treatment and after creep
deformation [127]. In the case of 7 thickness, there was only 8% reduction of thickness
from heat treatment and a 4% reduction of 1! thickness after creep deformation in the 900
°C/20hr heat treated alloy as determined from Figure 10.6.b. From the reference 34,
measured averaged 012 spacing shows that the spacing decreases from about 5% to 18%
from the As HIPed condition to the one and two-step heat treatments. From Table 10.3,
after deformation, in the one-step heat treatment there was a 22% reduction of the 012
'spacing, and a 10% reduction in the 012 thickness. This may result from nucleation of new
012 lamellae initiated by mechanical twinning, since a mechanical twin has the crystal
structure of the 012 phase for three layers in each interface [213]. In general, heat treatment

and creep deformation cause the lamellar scale to become more refined.

235

Ti-47Al-2Nb-lMn-0.5W-0.5Mo-0.2Si

a. lamellar spacing in compression-shear creep deformation

The distribution of 012 spacings before and after deformation is represented in Figure
8.18.a. The undeformed representation shows the same trend for three samples
characterized before deformation, and the average is indicated by a bold line. In every
case, there is a nominally similar shift of the frequency distribution toward smaller
spacings due to small creep strains. The average of about 500 observations of 012 spacings
is plotted with respect to effective strain and load in Figure 10.7. There is a 23%
reduction in lamellar spacing with strain that saturates at effective strains between 0.1 and
0.3%, whereafter the spacing remains relatively constant. For the same strain of 0.1%, the
amount of refinement at 750 °C is about half of the amount that occurred at 650°C. The

amount of refinement at 650 °C decreases with decreasing stresses.

b. lamellar spacing in tensile creep deformation

From Table 10.3, there was a reduction in spacing of 34 % in 012 and 24 % in y
thickness for the specimen deformed at 760 °C, 138 MPa. Contrary to the observations in
the compression-shear specimens, the reduction in 012 spacing was smaller (18%) in the
650 °C 276 MPa specimen, in part, due to a lower undeformed lamellar spacing.
However, the deformed lamellar spacing for this specimen (0.9 pm) is smaller than the
spacing for the deformed 760 °C 138 MPa specimen (1.03 pm).

Unlike the compression shear specimens, the undeformed lamellar spacing was
measured in the grip section of the creep specimens, which experienced the same thermal

history with presumably no stress. The compression-shear specimens all came from the

236

..x. 888 @8888 08 82225 go 88 £3525 > was wfiomdm .3 mo 82.9» 338 o
.5862; 85 8 can we? wEomam 8:083 BEE ...... .8086on EEO 8 m :85 808 .3388 88on 8:08a— m 88m ...

 

BEE m<

 

 

 

 

 

 

 

82 mm: o. o8
805-commmoa8oo .x. mm . 81 mm; «mo—2.3
m 0 .
82 mg 0. 2% sense .....M% man ”We a; 2. c $023
m 0 m 0
82 mm: o. 85 use“; mum“ m: ”we” ““2...me ma who” 5023
. m 0 m o
82 mm: u. can Sage :3 No Mam“ mafia” MOW m1 2%“ o 82
Be: a...
.x. em . a: N3 8 so:
82 mm: o. o8 88¢ ox 2:.
E 3 o. 25 a: as 0.. 25
81m; $8.812; $8.81 woo
a: on 0.. com a; on 0. com a: as 0. cos
N O gamma“ a . n . a .
.8525 can . H 5: 83; £2 838 ...: 8125 2:.
803980 53 380 305—85 go 988% 30 82225 >

 

 

 

 

 

 

98:8 13 H mo X. 85868 98 wanm 8:082 ow80>< ma: 2an

237

 

 

70yvrvfil rrrrrrr YTVVYYIYYrv'rvvr'YIVII rIrrfIIYrrr
_ Heat treated WMoSi alloy q
60 i 8 760°C, 138 MPa 1
1 o :o' 1‘ Tension Creep
50 T i '1‘ +Undeformed(012 spacing) T

    
  

"@“Deformed (01’ spacing)

+ Undeformed ( y thickness ) l

' '0‘ ‘ Deformed ( 7 thickness )

 

  

(a)
O

 

 

Frequecny (%)

N
O

10,

 

 

l L I l l A j

0 0.5 1 1.5 2 2.5 3 3.5 44.5 5
(c) Spacing ( 11m)

Figure 10.6 Distribution of a 012 spacings and 7 thickness in TiAl alloys before and after
creep deformation (c) WMoSi alloys under tension mode.

 

 

 

 

 

 

 

 

 

 

1.5 .8..,.-..,.....-...,..+.T....,....,.fl.,.r..,.fl.
. 1
W-Mo-Sr alloy (As HIPed) .
Compression+Shear Creep 1
1.4
A
8
3-
go1.2
"é.
" 1
N ,
5 [ +750°C/77MPa
1 "a“ 650°C/154 MPa
0.8 r .. _ _
1 ’ 650°C/115 MPa
+650°Cn7MPa
06 1LI-Lirir11411Lriia_lriiili AJlALLlJlIAAIlAlJILLLL
' 0 0.001 0.002 0.003 0.004 0.005 0.006 0. 07 0.008 0.009 0.01

Effective Strain, a

Figure 10.7 The plot of 012 spacing respect to different effective strains and different
loads at 650 °C and 760 °C.

238

same physical region of the casting, but nothing is known about the relative positions of
the two tensile specimens in the casting, where the initial lamellar spacing may have
differed due to differences in cooling rates, or other unidentified process parameters.
The 012 spacing and 7 thickness distribution curves before and after deformation are
represented as a bold and a dotted line, respectively, in Figure 10.6.c. As with the
compression-shear specimens, there is a nominally similar shift of the frequency

distribution toward smaller spacing due to small creep strains.

10.3.4 Effect of Anisotropic Crystal Structure on Deformation

Figure 3.8 a shows the maximum Schmid factor for true <112] twinning and
dislocation deformation systems. A high Schmid factor indicates that a deformation
system will operate to carry strain. The operation of true twinning depends strongly on the
orientation of a crystal with respect to the loading axis. In tension, orientations close to
<441] (near tail of arrow) are likely to twin. However, upon twinning, the crystal rotates
to an orientation near <118] (near head of arrow). Once the crystal is twinned, the Schmid
factor for true twinning is zero. The same phenomena occurs in compression, in the
opposite sense. In the twinned orientation, the Schmid factor for ordinary dislocations is
also small, and only superdislocations have a high Schmid factor, so twinning increases
the volume fraction of crystals where only super dislocations can operate. Luster and
Morris observed that superdislocations were active only in crystals having a Schmid
factor greater than 0.35, and they only nucleated from interfaces where efficient slip
transfer from neighboring crystals was possible [142]. Ordinary dislocations and true

twinning were observed in grains where the Schmid factor was larger than 0.25. In

239

lamellae thinner than 0.5 um, only deformation systems with Burgers vectors in the
lamellar plane were operative. These criteria were applied in Figure 3.8. b to determine
the volume fraction of a set of randomly oriented crystals that could deform with 4 or

more deformation systems active, excluding superdislocations. Deformation in tension
has only 19 vol% of crystals with 4 Schmid factors operative, while in compression, 28
vol % have 4 systems active. In contrast, 33 vol % of crystals in a FCC metal have 4
systems with m>0.32 (inverse Taylor factor). These observations imply that compression
of TiAl should be less prone to cracking than tension, since superdislocation activity is

restricted as indicated above.

10.4 DISCUSSION

The above review of several investigations indicate that many microstructure
features affect primary creep. The impact of the following microstructural features will be
assessed for the alloys investigated: Lamellar refinement, reduction of lamellar volume
fraction, removal of small equiaxed gamma grains, effect of heat treatment, and the effect
of dynamic precipitation. With understanding of how these microstructural changes affect
creep properties, optimal microstructural design criteria for creep resistance can be

balanced with other design constraints.

10.4.1 Lamellar refinement
Lamellar refinement occurs in all materials examined. Lamellar refinement can
only occur by a mechanical twinning process, or by a strain induced transformation

similar to twinning [121,125]. These processes cause new interfaces to be created as

240

strain occurs. Most of the refinement occurs in the first 0.1-0.2% of strain, where the
strain rate is high. Comparing Figure 8.5 with 10.7 shows that the initial fast strain rate
drops rapidly as the lamellae are refined, and that the strain rate decreases more slowly
with strain after the initial refinement process has saturated, as indicated by the
discontinuity in the strain rate in Figure 8.5. The amount of refinement depends on the
initial lamellar spacing, stress and temperature. A greater amount of lamellar refinement
occurs in materials with higher initial lamellar spacing, and consequently, the time to
reach 0.5% is reduced, since more of the 0.5% strain occurs in the initial high strain rate
process of lamellar refinement. Therefore, reducing the lamellar spacing will reduce the
initial amount of strain that can be accommodated by lamellar refinement, and it will take
longer to reach 0.5% strain. Heat treatment can be used to reduce initial lamellar spacing;
more heat treatment reduces lamellar spacing [127]. Pre-straining can also refine the
lamellar spacing in exactly the same way as in creep deformation, by taking up the easy
refinement strain before the creep process begins. The most creep resistant TiAl based
alloys developed to date have extremely fine lamellar spacings of 0.1 pm [159].

Lamellar refinement hardens the material in two ways, by orientation hardening
due to twinning, and by restriction of deformation to modes parallel to the lamellar plane.
The volume fraction of hard orientations, where only superdislocations have a high
Schmid factor increases with the amount of strain accommodated by mechanical
twinning. In addition to orientational hardening, creation of more finely spaced interfaces
strengthens the material by a Hall-Petch effect. In addition to strengthening, very fine
lamellaer regions have been shown to deflect cracks propagating across lamallae [255].

The greatest amount of lamellar refinement observed in the studies described above

241

occured in the high oxygen XD alloy. From Figure 10.6.a and Table 10.3, the reduction in
y thickness was 40%. The high interstitial content may restrict ordinary dislocation
mobility more than in the other alloys, and the more difficult operation of ordinary
dislocations would increase the probability for strain accommodation by mechanical
twinning. Even though mechanical twinning has a lower critical resolved shear stress than
ordinary dislocations in binary TiAl single crystals [256], twin nucleation requires a high
stress concentration, which exists in grain boundary regions [120].

The reduction in the spacing between 012 layers and the reduction in 012 thickness
indicated in Table 10.3 indicates that new 012 layers are being formed in the process of
lamellar refinement. This is possible by a strain induced phase transformation that
required subsequent diffusion of Ti into the transformed crystal structure, which has been
proposed by [205]. A true twin interface can serve as a nucleus for this process, since a
true twin interface consists of 4 atomic layers of 012 crystal structure [213]. Strain

accomplished by a moving ledge mechanism has been proposed by Wang and Nieh [159]

10.4.2 Effects of Heat Treatment on Grain size and Lamellar Volume Fraction

Heat treatments shown in Figures 6.2, 7.1 indicate that more heat treatment
improves or does not adversely affect creep resistance (with one notable exception for the
10 hr 900°C in the 47Al XD alloy in Figure 7.1). A major source of this improvement is
due to removal of fine equiaxed grains that provide regions where boundary diffusional
processes are more significant. The presence of the TiB; inhibits grain growth, which

causes a limiting effect on improving creep resistance. However the degrading effect of

242

larger grain size on room temperature ductility is well established, so the improved creep
resistance in the 47-2-2 alloy in Figure 6.2 may occur at the expense of ductility

Since the lamellar grains are the source of the lamellar refinement deformation
mechanism, a higher lamellar volume fraction should cause a greater amount of strain at
the initial high strain rate. In low temperature, low stress creep of duplex microstructures,
the steady state creep rate was observed to increase with lamellar volume fraction [257].
A larger strain at the high strain rate will reduce the time to 0.5%. Evidence for this is
effect is found in Figures 6.2, 7.1, and 10.3. Figure 10.3 shows that the lamellar volume
fraction is affected by heat treatment, though it is difficult to ascertain how intrinsic
microstructural variations may obscure these measurements. The data in Figure 10.3 were
taken from several regions and multiple cross sections of a sample, and they are
nominally similar to the trends observed when fewer observations were used. The most
consistent feature was the differences in volume fraction between the 10 hrs 1010 °C and
the 2 step heat treatment conditions. The 47-2-2 alloy exhibits the best improvement in
creep resistance with the 2-step heat treatment, and correspondingly, the lamellar fraction
was reduced by this heat treatment. The same thing occurred in the 45A] XD alloy, and
the effect of reducing the lamellar volume fraction was a big reduction in the
experimental scatter in the times to 0.5%, but no substantial improvement was measured
with any heat treatment in this alloy. Conversely, the lamellar volume fraction increased
with the 2 step heat treatment in the 47 A1 XD alloy, and this would suggest a degrading
effect on creep resistance. Since the 2 step heat treatment also reduced the number of fine
grains, these two changes apparently balanced each other so that no net improvement in

creep resistance was obtained.

243

10.4.3 Effects of Heat Treatment on Precipitation, and Dynamic Precipitation

The most creep resistant alloy is the heat treated WMoSi alloy. The heat treatment
causes nucleation of precipitates. The growth of the precipitates is affected by the stress
state, where the stressed region of the material stimulates a more homogeneous and faster
growth of the precipitates. Moving dislocations acting as pipes for transport of solute
atoms to precipitates enhance and homogenize their growth. This effect is beneficial for
developing a creep resistant microstructure that does not develop larger harder inclusions
that would degrade ductility. The 47-2-2 alloy also developed fine precipitates with the
two-step heat treatment, but the substantial improvement in creep resistance is also
correlated with a much larger grain size.

The WMoSi alloy in the as-HIPed condition exhibited strain rates more than 10
times faster with a smaller stress, giving it a creep resistance similar to the 45Al XD
alloy. The lack of precipitates in this alloy indicates that most of the silicon is still in
solution, which implies that Si in solution degrades creep resistance. This has also been
demonstrated by Wang and Nieh [258]. They showed that Si in solution would facilitates
ledge mobility in lamellar interfaces, which was proposed as a rate limiting creep

deformation process.

10.5 CONCLUSION
Changes in the microstructure have been analyzed to determine how heat
treatment and alloy composition affect primary creep resistance in several near-y TiAl

alloys.

244

All alloys exhibit lamellar refining that causes an initially rapid creep rate up to
strains between 0.1 and 0.2%. This undesirable rapid creep strain can be reduced by
refining the lamellar spacing by heat treatment or pre-straining.

Lamellar refining occurs by mechanical twinning or similar processes that
transform the crystal orientation by a moving interface. The microstructure is hardened by
a Hall-Petch effect which results in prevention of deformation across thin lamellae when
they are smaller than 0.5 pm, and by orientation hardening, since mechanical twinning
rotates crystals into hard orientations where only superdislocations have a high Schmid
factor.

Heat treatment homogenizes the grain size distribution by removing fine grains.
This results in a more homogeneous creep deformation that can reduce the scatter in some
alloys. There is limited evidence for degraded primary creep resistance with higher
lamellar volume fraction.

Precipitation provides a large benefit for creep resistance. The use of silicon can
be detrimental to creep resistance if it is kept in solution. Precipitation is complex, and
heat treatment can provide appropriate conditions for a desirable homogeneous dynamic

grth of precipitates during creep deformation.

245

CHAPTER 11

THE EFFECT OF CARBON CONCENTRATION ON PRIMARY CREEP OF AN

INVESTMENT CAST Ti-47Al-2Mn-2Nb+0.8vol.%TiBz ALLOY

ABSTRACT
Creep experiments in investment cast XDTM alloys with carbon additions in the as-HIPed
duplex microstructure exhibit up to a 10 fold improvement in primary creep resistance
compared to the same alloy with low carbon. The carbon levels examined are 0.006,
0.037, 0.065 and 0.110 wt% and creep experiments were done at 138 MPa at 760 and 815
°C, and at 276 MPa at 649 °C. A ten-fold increase in time to 0.5% strain was observed
under all conditions tested with a carbon concentration between 0.037 and 0.065 wt%,
with no additional benefit from higher carbon concentrations (in contrast, oxygen levels
of 0.1wt% yielded only a two-fold increase in creep resistance). The microstructures were
investigated in selected tensile creep specimens deformed to about 0.5% strain at 650 °C
and 276 MPa. Primary creep deformation in alloys containing carbon are similar to other
alloys in that primary creep is divided into two stages, an initial rapid process that refines
lamellar spacing through strain levels between 0.2 and 0.5%, and a second primary creep
process, where no lamellar refinement occurs. Carbon refines the grain size, alters
dislocation configurations in a way that suggests that solute pinning occurs, and carbides

are present in the vicinity of interfaces.

246

11.1 INTRODUCTION

Recent studies made on the effect of interstitials (hydrogen, nitrogen, carbon, and
oxygen) on creep properties of TiAl have focused on secondary creep or tertiary creep
properties [164,221,259,260]. These interstitials strongly affect microstructural
transformation, deformation mechanisms and consequently, mechanical properties [221].
Few studies, however, have been made on the effect of interstitials on primary creep
[239,260].

Investment cast Ti-47Al-2Nb-2Mn+0.8TiB2 XDTM alloys with low interstitial
content (about 0.03 at% N, 0.02 at% C, and 0.2 at% 0) have been developed at Howmet
Corp. Recent investigations motivated by reducing the process cost by reducing the
HIPing temperature and/or time and temperature of subsequent heat treatments has
yielded some improvement in primary creep resistance [127,239]. Alloys with higher
interstitial element concentrations were also investigated to examine the effects on
primary creep and process cost reduction. In this chapter, the effect of carbon
concentration on primary creep deformation and microstructures of investment cast and
HIPed TiAl alloys with no further heat treatment are examined and compared to the same
alloy with low interstitial content. Microstructures and primary creep deformation mode
are analyzed in terms of grain size, distribution of 012, volume fraction of lamellar grains,

and carbon concentration.

11.2 EXPERIMENTAL PROCEDURE
Four alloys of near-y TiAl with the nominal composition of Ti-47Al-2Nb-2Mn-

0.8TiB2XDTM were prepared with varying carbon contents. The same master heat was

247

used to make all four alloys, by adding measured amounts of aluminum carbide to the
crucible prior to by VAR (Vacuum Arc Remelting) and investment casting that provided
the compositions given in Table 11.1. The master heat had a low carbon content (0.006
wt%) typical of normal production capability. All alloys were hot isostatically pressed
(HIPed) at 1260 °C for 4 hours at 127 MPa (2300 °F, 25 ksi) to eliminate casting porosity
. and to establish the duplex microstructure. Tensile creep specimens with a gage length of
25 mm and 5 mm diameter were machined from test bars. These specimens were then
subjected to creep tests at 649 °C, 276 MPa (1200 °F, 40 ksi), 760 °C, 138 MPa (1400 °F,
20 ksi), and 815 °C, 138 MPa (1500 °F, 20 ksi). The creep specimens were cooled under

load after creep strains between 0.4 - 4% were reached.

Table 11.1 Al and C content in as-HIPed investment cast alloys.

 

 

 

| th% | 0.006 (low) 0.037 0.065 0.110 |
| C at% | 0.020 0.12 0.21 0.36 l
| A1 at% | 46.1 46.4 46.6 47.0 |

 

 

 

 

For microstructure analysis, only the low carbon, 0.065, and 0.110 wt% alloys
deformed at 649 °C, 276 MPa were examined. Samples were obtained from the grip
section of crept specimens to examine the undeformed microstructure and from the gage
section of crept specimens to examine the deformed microstructure. A modified Kroll's
reagent was used to etch for optical microstructure analysis [222]. For scanning electron
microscopic analysis, samples were cut from the grip and gage section of crept samples
and then electropolished using a 350 ml Methanol, 175 ml Butyl alcohol, and 30 ml
Perchloric acid solution at 20 volt and -50 °C. TEM specimens were sliced from the gage

section and the final thinning process was performed in a twin jet electropolishing system

248

using a 10% sulfuric acid + methanol solution at -20 °C. All microscopy samples were
prepared so that the sample normal was parallel to the tensile axis. Microscopy was
conducted in a J SM 6400V scanning electron microscope (SEM) operated at 25 kV, and a

Hitachi H-800 transmission electron microscope (TEM) operated at 200 kV.

11.3 RESULTS AND DISCUSSION
11.3.1 Microstructures

Figure 11.1 illustrates the grain refinement resulting from addition of carbon. The
lamellar grain size is more refined than the equiaxed grain size (see Table 11.2). Figure
11.2(a), shows the needle and round shaped particles arising from the XD process are
evenly distributed (equiaxed type: TiB; and needle type: TiB) [224]. They are known to
be located inside of y and 012 grains and in grain boundaries from previous studies
[111,127,239]. As shown in Figure 11.2(b-d), all alloys show duplex microstructures
having different amount of lamellar volume fractions. Most equiaxed grains are located
around grain boundaries of lamellar grains. Some equiaxed grains can also be seen in the
middle of lamellar grains. The different grain size and lamellar volume fractions strongly
influence primary creep behavior [127].

The reduction of grain size from the addition of interstitials was similar to that in
a cast gamma TiAl alloy with nitrogen (Ti-48.5Al-1.5Mo-1N) studied by Cho et al. [260].
They suggested that the TizAlN particles suppress the grain growth and refine the
microstructure. Chen et al.(600-700C ppm), Tian and Nemoto(0.5C mol%), and
Whittenberger and Ray (0.025-0.4C at%)[220,261,262] showed that the fine precipitation

of carbides as plate-shaped particles (TizAlC: 0.06~0.2 pm) or perovskite type particles

249

 

Figure 11.1 Optical macrographs of Ti-47Al-2Nb-2Mn+0.8 vol%TiB2 alloys having (a) Low carbon, (b) 0.065 wt% carbon,

(c) 0.11 wt% carbon.

 

Figure 11.2 Optical micrographs of Ti-47A1-2Nb-2Mn+0.8 vol%TiBz alloys having
(a,b) Low carbon, (c) 0.065 wt% carbon, (d) 0.11 wt% carbon.

251

(TI3AICZ 0.008~0.2 1.1m) were formed by cooling after HIPing and/or aging treatment.
Such TizAlC or Ti3AlC precipitates could retard the grain growth. Optical micrographs
shown in Figure 11.2 indicate a smaller range of grain sizes in the higher carbon alloys
(Table 11.2). Also, there is a high volume fraction of lamellar grains in the alloys having
low and 0.065 wt% carbon, but in the 0.110 wt% C alloy the volume fraction of lamellar
grains is lower by a factor of 2 (Table 11.2), and more equiaxed grains are developed at
lamellar grain boundaries. Cho et al. [260,263] also mentioned that the addition of more
than 0.3 at% nitrogen results in the development of some equiaxed grains at the grain
boundaries of lamellar grains. Since the carbides and nitrides have similar stoichiometry,
morphology, and atomic weight, the effect of carbon may be similar to that of nitrogen.
Although interstitials such as carbon and nitrogen can change the 01 transus in the 01 + 1!
region (where the phase transformation from 01 to lamellar grains occurs during cooling),

this cannot account for the large difference in the lamellar volume fraction observed in

the 0.110 wt% alloy.

Table 11.2 The lamellar volume fraction and grain size in TiAl alloys

 

 

 

 

 

 

 

TiAl alloys volume fraction of lamellar lamellar grain size equiaxed grain
grains* size
0.006 wt% C 68i7% 100 ~ 350 pm 5 ~ 30 pm
0.065 wt% C 701-10% 50 ~ 200 pm 3 ~ 20 pm
0.110wt%C 29i10% 50~150 um 3~20 pm

 

 

 

* average values from at least four images having areas of 1.02 x 10'0 11m2 in each image.
The italic number represents standard deviation.

In equiaxed grain regions, thin layers of 012 phase exist in some boundaries of y
grains having less than 10 um grain size in the backscattered electron (BSE) images in

Figure 11.3. These layers are thicker and brighter in the alloys containing more carbon.

252

The greater amount of retained 012 may be due to the presence of the higher carbon
concentration since the solubility of interstitial elements in 012 is greater than that of 7
[264,265]. Fine precipitates are often observed in or near these 012 areas. From Figure
11.4, the thickness of 012 in the alloy with 0.110 wt% is as large as 2 pm, while the
thickness is less than 0.5 pm in the alloys with low carbon and 0.065 wt% C. These
thicker 012 features in the 0.110 wt% C are similar to angular 012 grains which have been
correlated with poorer creep properties from previous studies [11,127,239,253].

Fine precipitates are also observed near grain boundaries of equiaxed grains and
in lamellae. In Figure 11.5 very fine precipitates (see arrow) ranging fi'om 0.08 pm to 0.2
pm are observed preferentially in 012 layers of lamellar laths and in equiaxed 012 grains. A
few precipitates are also observed in y grains. These fine precipitates can be a barrier to
moving dislocations during deformation, and their presence in 012 may impede slip
transfer through the 012 phase. Fine particles also are observed in grain boundaries of
equiaxed grains, as shown in the bright field image in Figure 11.5(c). Chen et al., and
Tian and Nemoto [220,261] observed homogeneous distributions of plate-shaped
(TizAlC) or perovskite type particles (Ti3AlC) in y grains, but such homogeneous
precipitate distributions could not be found in the current as-HIPed alloys. These fine
particles can harden the grain boundaries during high temperature deformation.

Particularly in deformed samples, as shown in Figure 11.6, many deformation
twins are found in gamma grains afier small creep strains. From prior studies
[11,127,239,253], such fine twins are common features in investment cast TiAl alloys

crept only in the primary stage.

253

 

Figure 11.3 Backscattered electron images of Ti-47Al-2Nb-2Mn+0.8 vol%TiBz alloys
having (a) Low carbon, (b) 0.065 wt% carbon, (c) 0.11 wt% carbon.
The arrows indicate the thin layers of 012 phase.

254

 

 

 

 

 

   

100 V I I ' I l ' I I ' I I I ' I I 7 I
Ti-47Al-2Nb-2Mn+0.8 “JP/o TiB 2 1
80 Low Carbon 7
\t; g :5 0.065 wt% Carbon
‘1, 60 L I 0.110 wt%Carbon _
fl ~: ,
0 :3
5 % T
3 4o , i
I- 23:
5
2. - g
i 5

 

 

L
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.75
Thickness of 012 between 7 grains (11m)

Figure 11.4 Distribution of thin 012 layers surrounded by y grains in TiAl alloys.

 

Figure 11.5 a and b: Backscattered electron images of Ti-47Al-2Nb-2MnH).8 vol%TiB2
alloys having 0.065 wt% carbon.

255

 

Figure 1 1.5.c Precipitates along the grain boundaries of y grains in creep tested Ti-47Al-
2Nb-2Mn+0.8 vol%TiB2 alloys having 0.065 wt% carbon.

 

Figure 11.6 Backscattered electron images of Ti-47Al-2Nb-2Mn+0.8 vol%TiBz alloys
having 0.11 wt% carbon.

256

Intensive dislocation activity can be seen in the vicinity of a TiB particle, and
dislocations and bowed dislocations along with their pinning points can be easily seen
((Figure 11.7), indicated by arrows). The general shape of these dislocation lines is
different from the straight screw segments commonly observed in low C alloys [266].
This significant difference in dislocation morphology suggests that carbon dissolved in
the 7 matrix can be a source of pinning points that retard moving dislocations during
deformation by solute/dislocation interaction mechanisms, rather than by a barrier effect

of carbide precipitates.

11.3.2 Lamellar spacing

Lamellar spacing was characterized by tabulating 300 to 370 012 spacing
measurements (i.e., distance between 012-y interfaces using backscattered electron (BSE)
imaging) in more than 20 lamellar regions near grain boundaries. Measurements were
taken from the grip and gage section of crept samples to compute the average 012 spacing
of undeformed and deformed material. These measurements are upper bounds of the
actual spacing, since the lamellar plane normal direction was not determined to make the
parallax correction needed to determine the actual 012 spacing. However, this is a
systematic way to compare 012 spacings in specimens with similar microstructures.

The distribution of 012 spacings in undeformed and deformed material is
represented as solid and dashed lines, respectively in Figure 11.8. In every case, there is a
shift of the frequency distribution toward higher frequencies at lower spacing values after
creep. For example, the average 012 spacing in 0.065 wt% C changed from 1.17 to 0.84

um due to a small creep strain of 0.42%. This reduction of 012 spacing can be explained

257

 

Figure 11.7 Dislocation activity near TiB particles in creep deformed Ti-47Al-2Nb-2Mn+
0.8vol%TiBz alloys having 0.065 wt% carbon. White arrows show bowing,
and black arrows show pinning points.

 

 

   
  
  
  
  
 
       

 

 

 

 

60 r . 47,7 fi‘rfi—r 1 r r 1 1 1 1 1
Ti-47Al-2Nb-2Mn+0.8 vol%TiBz
50 > Creeptested at 649“C/276 MPa
Carbon Average Finalcreep -
(wt%) spacing strain
’3 +11 debm’cd 0006 1046
c\ 40 " ' - “m j
‘;: “V"Debmed 0.006 0.734 urn 1.03%
g +Undeforrred 0.065 1172 um
g 30 j "V“Debmnd 0.006 0.340 pm 0.42% f
g 1 +1111de 0.110 0922.111.
I: “V"Debrrmd 0.110 0713 pm 0.79%

20»

 

 

 

- , - V' v
1.5 2 2.5 3 3.5 4
a, spacing (11m)
Figure 11.8 Distribution plot of 012 spacing in undeformed grip and deformed gage section
of tensile creep specimens.

258

indirectly by the easy mode deformation twinning parallel to lamellar boundaries. This

refinement trend is similar to that found in prior studies [11,127,239,253].

11.3.3 Primary Creep behavior

Primary creep resistance increased with increasing carbon content up to a value of
0.065 wt%, as shown in Figure 11.9(a). This plot shows that an increase in creep
resistance by an order of magnitude can be obtained with small carbon additions at all
temperatures in a systematic manner. Figure 11.9(b) shows data for specimens deformed
at 649°C and 276 MPa (40 ksi and 1200°F) as a function of time. The alloys with low
carbon reach 0.60% creep strain in 100 hours, while the alloys with 0.037, 0.065, and
0.110 wt% C show lower strains of 0.22, 0.15, and 0.25% creep strains, respectively at
100 hours creep time. Thus carbon concentrations up to 0.065 wt% increase the primary
creep resistance increase, but a 0.110 wt% carbon addition results in a decrease in creep
resistance. The 0.037 and 0.065 wt% specimens exhibit the lowest primary creep rates
after 100 hours, but the 0.065 and 0.110 wt% specimens exhibit the lowest early primary
creep rates (at 20 hours). Thus the 0.065 wt% alloy is near an optimum carbon
concentration for primary creep resistance.

The 0-projection concept constitutive equation described in [250,267] is based on
the phenomenological understanding of creep deformation in primary and secondary
stages. A variation of the 0-projection constitutive equation was used to characterize the

present creep data. The equation used for curve fitting the creep data is

8(t)= e.(1—exp(—-r.t))+e2(1-exp(—rzt))+é.t (1)

259

 

 

   
  
    
  

 

 

 

 

 

 

0.8 I T T I I
Creep tested at 649 °C/276 MPa
0.7 r .1
-~><—— Low Carbon
0 6 _ - °- 0.037 wt% Carbon x A
I; ' - -<>- - 0.065 wt% Carbon xLowC
o\ -é>—0.1 10 wt% Carbon x
V
E 0.5 r r
'8
"' 1
g 0.4 -
:1. O
3 o 3 s
1- ' T O 0.110 wt% (3 .
U L o //
.2 _ ...— — —- - ----- 4
° — - 1.75, 0.037 wt% C g
0.1 hag/O/ ----- <§ """"""" 0 .065 wt% c .
45:.- ' . O O O
Q 20 4O 60 80 100 120
(b) Time (hours)
: PrimaryCieep Resistance in I f 1
. Ti-47Al-2Nb-2MrXDAlloys WithCa n .
Discontinued below 0.5%
1000E 4

 

 

 

,_ .E
O a
a
12 5, 100 .5
3 3
0
E l-
: D
O o
\
= .IQ l0; _:
.5 G : '
.8 ° .. -<>—649°C 276 MPa 0.5%
o\ .- _
1- Fl. 6 o v 760°C 133mm 0.5%
° 1? 9 -0-815°C138MPa 0.5% 1
5' —<>—649°c 276 MPa 02% ‘
» v 760°C 138 MPa 02%
1 --~°"-8|5°C l38MPa 02%

 

 

 

 

 

 

0 A I i I g l

' 0 0.02 0.04 0.06 0.08 0.1 0.12
(3) W970 C

Figure 11.9 (a) Creep times to 0.2% and 0.5% strain as a function of carbon
concentration for specimens deformed at three deformation conditions.

(b) Creep strain to 0, 20, 30, 50, and 100 hours in Ti-47Al-2Nb-2Mn+0.8
vol%TiBg alloys. Lines go through the average of 3 samples.

260

where e is creep strain, t is creep time, e1, r1, e2, and r; are curve fitting parameters, and
15‘s is the steady state (or minimum) strain rate. Equation (1) was used as a fitting function
using Kaleidagraph in software to represent the experimental data in a form that can be
easily differentiated. These fitting parameters are provided in Table 11.3. These fitted

curves are accurate representations of the creep data as indicated by the R values. By

differentiating equation (1), the strain rate can be calculated from following equation

$0):errl("r1’)+ezr2(—r2’)+és (2)

Table 11.3 Curve Fit Parameters in Creep tests shown in Figure 11.10

 

 

 

 

TiAl

Alloys 6! n e2 r: R
(:23? 0.0017058 14010053 0.0027975 2471932 09997
(1:23: 0.0008675 111507.? 0.0012788 2310093 09992
(iii? 0.0017167 171107.? 0.0007673 $818013 0,9995

 

 

 

 

 

 

 

 

0 R represents Least Squares Curve fits

The trends illustrated by the data in Figure 11.9(b) are also evident in the strain
rate vs. strain plot in Figure 11.10. The minimum strain rate of 0.065 wt% C is lower
than the low C alloy by more than one order of magnitude. The minimum strain rate of
0.110 wt% C is between the Low C and 0.065 wt% C. The transition points which
indicate a change in the dominant primary creep mechanism [253] are indicated by arrows
in Figure 11.10. A slower strain rate at the transition point implies better primary creep

resistance. This transition point is strongly related to microstructure features such as

261

10*3

 

 

 

 

 

 

p I I I I I j

E Ti-47Al-2Nb-2Mn+0.8 vol%TiB 2 ‘

A Ii". Creep tested at 649 °C/276 MPa .

'3 107 i - I" 1

0 : I. :
H i
a |
'- '1
.8 ':
a '.

‘- 10'8 :' : 1

a E I LowC l

g \ 3.110 wt% C
109 f '3. z:
I ""---0.065 wt% C
I
I
10-10 1 1 1 1 1
0 0.002 0.004 0.006 0.008 0.01
Strain

0.012

Figure 11.10 Creep strain rate as a function of creep strain in Ti-47Al-2Nb-2Mn+

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.8vol%TiB2 alloys.
100 Tjjnwfireffirfirjvrvw'r ‘05
A
c,\" [ Ti-47Al-2Nb-2Mn+0.8 vo|% 1‘18 2 j
v
g . Creeptested at 649 ”0276 MPa ] 10.6
'8 80 — 3
I- i
no - 9‘ .
8 I. 7““ ‘~ +1070:
-'.=. =
a . 1 E-
-- 1:
“5 ‘ I f a
2 1 4 10'9 m'
H :1 r—
0 L 3 v
2 ~ :
I-H 20 r “11010
—°. 1 '9'- vol. fraction .‘ Initial
g 1 Am Transition 3
1 lines go through the average of 3 samples ' I ' Minimum 4
0 _ .-7- gm . 1 1 1 . 1 1 1 1 1 1 1 1 1 1 1 17 10'"
0 0.02 0.04 0.06 0.08 0.1 0.12 014
Carbon concentration (wt%)

Figure 11.1 1 The relationship between lamellar vol. fraction and strain rates with carbon
concentration. On this and subsequent plots, the size of the symbol
represents carbon concentration.

262

lamellar volume fraction and deformation twinning [111,253,205].

As indicated in Table 11.2 and Figure 11.11, the volume fraction of lamellar
grains is affected by carbon concentration, and the impact of carbon on several
characteristic strain rates denoted as initial, transition, and minimum is shown. The
volume fraction of lamellar grains increases slightly with increasing carbon but it drops
dramatically with more than 0.065 wt% C. All characteristic strain rates decrease with
increasing carbon concentration up to 0.065 wt% and then the strain rates increase
slightly with additional carbon. Although the alloys having 0.110 wt% C have the lowest
lamellar volume fraction (~29 %), its minimum creep rate is still much lower than low C
alloys with high lamellar volume fractions.

Figure 11.12 shows that the initial strain rate, strain rate at the transition point,
and minimum strain rate decrease linearly as the volume fraction of lamellar grains
increase. The two higher carbon alloys are on the same line, while the low carbon alloy
exhibits the same slope but the transition strain rate is almost an order of magnitude
faster. This suggests that once a saturation level of carbon is added, the transition strain
rate is not reduced further. A higher volume fraction of lamellar grains is beneficial for
creep resistance, since a larger volume fraction of lamellar interfaces will block
dislocations during deformation. Worth et al. [164,221] indicated that 30% lamellar
volume fraction in a duplex microstructure gave poorer creep resistance than the equiaxed

y microstructure (0% lamellar volume fraction) due to the increase in dislocation mobility

in the duplex alloy 7 phase. The fully lamellar microstructure gave the best creep
resistance because the fine lamellar spacing impeded dislocation motion on {111} planes

not parallel to the lamellar interface. The addition of carbon did not have a dramatic

263

 

 

 

 

 

 

 

 

1
E Ti-47Al-2Nb-2Mn+0.8vol%TiB . Initial :
6 Creeptestedat649 °C/276 MPa A Transition
‘10 I Minimum ‘5
_f'\
$107 3
o .
H
N
in
.5 . l
a 1G 1
1’: 4 1
m L
9
1040 r 1 1 1 a 1 1 1 1 1 1 1 1 1 n 4 1 1 1
0 20 40 60 80 100

Vol. Fraction of lamellar grain (%)

Figure 11.12 The relationship between lamellar vol. fraction and initial strain rates and
strain rate at minimum and transition points. Symbol size represents carbon
concentration (small, medium, and large size symbol: Low carbon, 0.065
wt%, and 0.110 wt% carbon).

14 T 1 l T I T

Ti-47Al-2Nb-2Nh-103 vol% TiB 2

1 .3 4
Creep tested at 649 °C/276 MPa

1'2 ‘7 LowCarbon A

V 0.065 wt% Carbon
V 0.110 wt% Carbon

 

 

 

 

 

   

assumed line

Average 012 spacing (um)

 

 

 

0.9
0.8 ~
0.7 . . . —
Transmon pomt
0.6 l l 1 1 1 L
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
Creep strain

Figure 11.13 a; spacing as function of creep strain.

264

effect on the lamellar spacing, as the 0.065 wt% alloy exhibited the best primary creep
resistance, yet it also had the largest average lamellar spacing (Figure 11.8). As a result of
microstructural analysis in Figures 11.5 and 11.7, the improvement in primary creep
resistance with addition of carbon is probably due to a solute atmosphere drag mechanism
along with precipitates that interfere with slip near or across boundaries. The mild
decrease in creep resistance in the 0.110 wt% C alloy correlates with the small volume
fraction of lamellar grains and the increased presence of angular 012 as described in
Figures 11.3.c and 11.4.

Based on prior work, it is known that lamellar spacing was refined due to easy
mode deformation twining that occurs below the transition strain, whereafter the 012
spacing remains constant [205,253]. Using the strain values at the transition for the
specimens in Figure 11.10, and the measured lamellar spacing, the rate of lamellar
refinement can be estimated as shown in Figure 11.13. The 0.065 wt% C alloy exhibits
the fastest rate of lamellar reduction that correlates closely with the greatest deceleration
of strain rate in Figure 11.10, and hence the best primary creep resistance. Also, these
data indicate that the smallest lamellar spacing does not necessarily lead to superior
primary creep resistance. The greatest deceleration of primary creep occurs with the
highest initial 012 spacing, since an increase in lamellar interfaces due to lamellar
refinement will increase the number of barriers to dislocations during early creep
deformation. This is consistent with the case of oxygen additions in TiAl alloys, where
the high 0 alloy exhibited a greater reduction of the thickness of y lathes (from TEM)
than lower 0 alloys, due to mechanical twinning parallel to the lamellar plane [239].

The relationship between reduction of a; spacing and the strain rate at the

265

 

50 fi.f ""“'T"'fifirv10'7
i Ti-47Al-2Nb-2Mn+0.8 vol%TiB 2 1
4o 1 Creeptestedat649 °C/276 MPa 1
“7,11an
0.110wt%C 310-8
30 " - 1

       
 

 
    

~
~\
‘—
A ‘
-
-

-
-
N ‘
-
“
‘ “

0.065wt%C A 1

V T fl f T I f Y T 7

 

Reduction of or spacing (%)
(r- s) rugod nomsurul in are: "19138

 

 

 

 

 

20
110'9
10? OReduction 1
ATransition
0441111111111111111110-10
0 20 40 60 80 100

Vol. Fraction of lamellar grain (%)

Figure 1 1.14 Lamellar refinement and strain rates at transition point as a fimction of
lamellar vol. fraction.

266

transition point (with respect to lamellar volume fraction) is plotted in Figure 11.14. A
higher lamellar volume fraction results in greater reduction of lamellar spacing and a
lower strain rate at the transition. These two qualities taken together are effective in
reducing the initial fast mode of primary creep.

From the above, a limited amount of carbon (0.065 wt% C) gives the best primary
creep property for several reasons: the carbon dissolved in y grains can impede the
moving dislocations, and precipitates at grain boundaries and interfaces block slip
transfer and harden the grain boundary. The carbon causes a smaller lamellar grain size
that gives the better primary creep resistance similar to extruded fully lamellar alloys
[160]. The addition of carbon does not cause any fundamental difference in the reduction
of lamellar spacing in the lamellar grains due to the easy mode twin deformation, which

is observed in low carbon alloys.

10.4 CONCLUSIONS

The addition of carbon in investment cast Ti-47Al-2Nb-2Mn-O.8TiB2 results in a
grain size refinement, and the lamellar volume fraction is reduced from 70% to 30% by
adding more than 0.065 wt% C. In 0.110 wt% C alloys, thick layers of 012, which are
similar to angular 012 grains that correlate with poorer creep resistance, are observed. In
the 0.065 wt% C alloy, bowed dislocations along with pinning points in y grains and
precipitates at grain boundaries are observed. This suggests that carbon dissolved in y
matrix can be a source of pinning points and can cause drag on the moving dislocations.
Precipitates at grain boundaries and interfaces block slip transfer and harden the grain

boundary. The reduction of lamellar spacing occurs due to the easy mode deformation

267

twinning parallel to lamellar boundaries at the early stage of primary creep in a way that
is similar to other low carbon alloys. Higher lamellar volume fraction results in a greater
reduction of lamellar spacing and a lower strain rate at the transition point that separates
an initial fast mode and a slower mode of primary creep. The increase in lamellar
interfaces impedes dislocation motion on {111} planes that are not parallel to the lamellar

interface. These results indicate a limited carbon addition (0.065 wt% C) to Ti-47Al-2Nb-

2Mn-0.8TiB2 gives the best primary creep resistance.

268

CHAPTER 12

MECHANISMS OF LAMELLAR REFINEMENT DURING PRIMARY
CREEP OF NEAR-GAMMA DUPLEX TiAl

Primary creep has been investigated in a number of near-gamma TiAl alloys with
duplex microstructures from previous studies [111,127,205,239,253]. These studies show
that two stages of primary creep deformation occur, an early and rapid process that
exhausts itself after about 0.2-0.5% strain, and a different process that dominates the
deformation up through the minimum creep rate that is more similar to minimum or
secondary creep deformation. When creep data are fit to functions that are complex
enough to accurately represent the measured strain-time data, these two stages are easily
observed in a strain-rate vs. strain plot. From interrupted creep tests, the early process
causes lamellar refinement, i.e. a reduction of lamellar spacing, and the second process
does not alter the lamellar spacing significantly. The primary creep resistance is
dominated by the rate at which the early process is exhausted. The refinement of lamellar
spacing due to small creep strains could be explained by a reduction in the 7 thickness
[127,239], which can only occur by mechanical twinning [121], as well as a reduction in

the 012 spacing due to a strain induced y-orz shear transformation [125,127].

In this study, a simulation of the refinement process has been made to examine
possible mechanisms that could account for the observed refinement (see appendix). The
lamellar distributions obtained from the simulation are compared with the measurements
in TiAl alloys, and the resulting local strains in the lamellar microstructure are computed

to consider how different deformation histories affect the resulting deformed distribution.

269

lnvestrnent cast TiAl alloys, such as Ti-47Al-2Nb-2Mn+0.8TiB2 XDTM with
oxygen (XD high 0), Ti-47Al-2Nb-2Mn+0.8TiB2 XDTM alloys with different HIP

conditions (XD low HIPed and XD high HIPed), and Ti-47Al-2Nb-an—0.5W-0.5Mo—0.2

Si alloy were tested at 760 °C and 138 MPa creep conditions, and the W-Mo-Si alloy was

also tested at 649 °C and 276 MPa at Howmet Corp. All creep tests were stoped at about
0.5% creep strain. The simulation code is developed using Quick-Basic (see appendix).
The code starts with data from a distribution plot from an undeformed specimen. A set of
100 lamellar spacings that represent the undeformed lamellar spacing frequency
distribution is generated in a random order. The simulation is started by following
assumptions: The thickest lamellae in the undeformed condition are subdivided in a
manner that changes the frequency distribution in the thicker bins to match the deformed
distribution by assuming that strain induced 012 shear transformations parallel to lamellar
planes with a thickness of 0.2 pm occur. No more than one twin can occur in one lamellar
lath. When a lath is divided, one lath from the original bin is removed, and two smaller
laths are added to the appropriate smaller bins. The refinement process starts in the
thickest lamellae and continues until the lamellar distribution matches the deformed
distribution in the larger bins. The simulation is finished, when the last bin that has a
larger fraction than the deformed distribution is reached, as illustrated in Figure 12.1.
From the 0.2 pm thickness of each shear transformation, a cumulative shear displacement
is evaluated, assuming a shear strain of 0.3535 in the 0.2 pm layer. This strain
corresponds to a particular dislocation moving on every other plane in a manner to
generate the a; structure [268]. The output file has a calculated local shear-strain, lamellar

spacings and lamellar spacing frequency distribution at each simulation step, and the total

270

number of transformed layers and lamellae is created. This file is opened with Microsoft
Excel and the distribution plot was created by using KaleidaGraphTM software.

From the simulation, the resulting distribution of thinner lamellae in two different
TiAl alloys match the measured distribution quite closely, as illustrated in Figure 12.1.
From this mechanism described above, the local shear strain in lamellar grains due to on;
shear transformation can be computed from this simulation as illustrated in Figure 12.2.
On the horizontal axis, each original lamellae is indicated by a short vertical line. A place
where the 012 shear transformations took place is indicated by an arrow, and the lamellar
distribution line is displaced up due to this shear transformation in several lamellae. From
the slope where the shear transformation occurs, in the simulated sample overall of strain
in the lamellar region can be calculated as indicated by the dashed line. The computed
local shear strains (>0.7%) in lamellar microstructure were similar or higher than
effective shear strains (Yefizl.0%=200.5% a) from specimens simulated in Figure 12.3.
For computed shear strains less than about 2%, the amount of lamellar refinement is
about the same. One datum suggests that more reduction of lamellar spacing would occur
if more than 2% local shear strain develops in lamellar microstructures.

In general, there are two stages in the primary creep deformation, and the first
stage is rapid drop of creep strain rate, and then the later stage is slower process that
controls the deformation up through the minimum creep rate (see Figures 8.5, 10.1, and
11.10). The transition points which indict a changes in the dominant primary creep
mechanism [253] are indicated by arrows in these Figures. A slower strain rate at the
transition point implies better creep resistance, and this transition point is strongly

correlated with microstructure features such as lamellar volume fraction and deformation

271

twinning [111,205,253,269]. There is also a correlation between the computed local shear
strain and the lamellar volume fraction, creep strains, and strain at transition points in
most TiAl alloys as shown in Figure 12.4, 12.5, and 12.6. A linear relationship between
the transition strain rate and the minimum creep rate in XD alloys vs. computed local
shear strains can be seen in Figure 12.4. This trend is strongly correlated with lamellar
volume fraction as shown in Figure 12.5 and 12.6. As the volume fraction of lamellar
grains increases, the computed local shear strain decreases, and the strain rate at transition
point decreases, too. From experimental measurements, the strain at the transition point
increases with increasing lamellar volume fraction. The higher volume fraction of
lamellar grains implies less lamellar refinement per lamellar grain, but with more lamellar
grains, there is more strain accomplished by refinement, so the transition strain is larger.
However, the observation in Figure 12.3, lamellar refinement not being proportional to
local shear strain is difficult to rationalize. Further analytical efforts in simulation are

required.

272

 

YTTYI’YY1'WYfiTY'Y'l'Y‘YT'V'YTYYY

'l‘i-47Al-2Nb-21\‘In+0.8V"/n'I‘iBz (Higher HIPed) 1

01
O

 

 

 
   
  
 
 

 

’ 1

-_ —— Undefi'ormd 1

4° , +chmed .
L - (>- Sinulation 1
limited twin thickness calm «

01)
O
fiTrfir

last bin sizeto be refined

N
o
If Yfif' I

first bin size to be refined

 

Number ofLaths based on initial set of 100

   

 

1 L l l l

O 0.5 1 1.5 2 2.5 3 3.5

 

(a) 012 spacing( 11m)
50-211

40

I I
l

 
 
 
 
  
  

 

3° ” +Unde1innnd
--O- Debnmd

- (>- Sinulation

 

 

20’

10 lirrited twin thickness oann 1

Number of Laths based on initial set of 100

 

 

 

 

..o.5...1....115..12. 2.5 3 3.5
(b) 012 spacing(um)

Figure 12.1 The distribution of 012 spacing is shifted to a greater fraction of finer
spacing as a result of early stage primary creep deformation. A simulation
based on the undeformed room temperature grip that generates 0.2 pm thick
01; layers within thicker y lamellae matches the deformed distribution.

273

 

Microns of Twin Translation

 

2 If I j T I I I I I j ‘r 1 I I I T

Ti-47Al-2Nb-2Mn+0.8V%TiB 2 (Higher HIPed)

 

| Undeformed l

1 -5 k + Simulation

 

 

0.5 1- , / c’omputed local shear strain=0.0|22

+1 / /
1N /

/
/

 

1 4411‘ / / <—a2 sheartransforrnation

o 20 4o 60 80
Position ( 11m )

Figure 12.2 Lamellar laths configuration in undeformed and deformed case
from simulation.

274

L _1 L

1

 

 

100

 

 

 

 

 

 

 

4o _ , . . . . 1 . . . . r . . 1 . 1 1 1 . 1 ~
: o WMoSi 3
35 _ E XD high 0 o d
A : A XDhigh HIPed 760 C1138 MPa 0 j
°\° . A XDlow HIPed .
v ~ 1
a0 30 f j
a l
a r
1— _
m 25 _ 1
N P'
=5 :
“It _ 2
O 20 1 .
1. -
5 i .
3 r TiAl XD alloys 1
g 15 I M 1
. ‘E‘
"U ’ 1
o ’ i
a: 10 . -
’ 1
1
5 P a 1 A 1 I A r 1 1 l A_ A 1 L 4 4L 1 l 1 I L 1 L i 1
0.005 0.01 0.015 0.02 0.025 0.03

Computed local shear strain for lamellar refinement
from the simulation

Figure 12.3 The relationship between computed local shear strains and reduction of
a; spacing.

 

   

 

 

 

 

 

 

 

10.7 :— r Y Y T f fi f f T I T Y Y Y T Y T I’ Y T Y *Y # T .4
: 1
L 1
i . . .
1 Transmon strain rate
.1“ 10-8 :— 4 . . . .1
i ‘ minimum creep strain rate 1
O“ - 4
o _ ' ‘ ‘ ‘ _ , _
"" . " I TiAl XDalloys O
E o
E 1
V1109 r 2 1
~ 1
' WMoSi (649°C/2 76 MPa) .
O WMoSi
1 I x1) high 0 .1
760 CH 38 MP
A XD high HIPed a
A XD low 11le
10-10 1 1 1 1 L 1 1 1 A 1 1 m 1 1 1 1 1 1 1 1 1 1 L
0.005 0.01 0.015 0.02 0.025 0.03

Computed local shear strain for lamellar refinement
from the simulation

Figure 12.4 The relationship between computed local shear strains and strain rates at
transition points.

275

 

 

 

 

 

 

 

 

 

50 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ,
' WMoSi (649°C/276 MPa) 3
j o WMoSi 4
g: 50 j I )0) high 0 11 j
g 1 A XD low HIPed 1
h 1
on 40 r 1
l- " l
_¢_u » 1
a , 1
51 30~ 1
“a 1 4
1 1
20 “ ‘
g F
11’:
0
E ‘ 1
g 10 r 1
c 1 1
> L
l
0 1 L g P L 1 1 1 11 l 1 1 l 1 1 4 4 l 1 1 1 1
0.005 0.01 0.015 0.02 0.025 0.03

Computed local shear strain for lamellar refinement
from the simulation

Figure 12.5 The relationship between computed local shear strains and volume fraction of
lamellar grains.

 

 

 

 

 

 

 

 

10'7 1 v v v rfi f 1 1 Y T w 1 T v I w v v v T V V ‘ r T ' ‘ ' ' OCDS
L Small sy mbol- Transition strain rate 1
» Large sy mbol- Strain at transition point J
r T 00125 W
m g.
g ~ 0002 a
a 00015 a.
E. a
m G
g B
E ~ 0.001 E
E 1.1
a L
[- I m high 0
~ 0.0005
A XDhigh HIPed 760°C/I38 MPa l
A XD low HIPed j
.1
1 -9 1 1 1 g 1 1 1 1 . 1 1 1 1 1 1 1 1 1 1 1 1 A 1_L 1 1 1 1 1
° 15 20 25 30 35 40 45°

Vol. fraction of lamellar grains (%)

Figure 12.6 The relationship between strain rates and strains at transition point with
volume fraction of lamellar grains.

276

CHAPTER 13

SUMMARY

Changes in the microstructure have been analyzed to determine how heat
treatment and alloy composition affect primary creep resistance in several investment
cast TiAl alloys. Creep behavior in PST crystals with different orientations was
investigated to determine how different orientations affect creep behavior in lamellar
microstructure by changing stresses and temperatures. The creep behavior of these alloys
and the influence of heat treatment are summarized.

Ti-45A1-2Nb-2Mn-O.8TiB2 did not show much difference in primary creep
resistance after various heat treatments because no difference in the microstructures was
caused by heat treatment. The creep resistance was poor, due to very fine microstructures
and large angular 012 grains. Some limited improvement in the Ti-47Al-2Nb-2Mn-
0.8TiB2 alloy was obtained with heat treatment. The XD alloys with high 0 showed
better primary creep property than low 0 alloys. The high 0 alloy exhibited 40% lamellar
grains (in spite of having higher Al content) indicating that O stabilized the on; phase. The
addition of carbon in investment cast Ti-47Al-2Nb-2Mn-O.8TiB2 improved primary creep
resistance, refined grain sizes, and reduced the lamellar volume fraction from 70% to
30% by adding more than 0.065 wt% C. Thick layers of 012, which are similar to angular
012 grains that correlate with degraded creep resistance, are observed in 0.110 wt% C
alloys. The heat treatments (especially the two step heat treatment) significantly
improved the creep resistance of the Ti-47Al-2Nb-2Cr alloy. This heat treatment led to

the smallest volume fraction (18%) of lamellar grains, a homogeneous microstructure, the

277

least amount of pockets of small equiaxed grains, large equiaxed grains, and it caused
precipitation of some small particles in interfaces. The texture is strengthened with heat
treatment (the desirable texture to minimize mechanical twining is to have crystals
oriented near 001 with the tensile axis and away from 441) and it may contribute to the
excellent creep resistance obtained from the two-step heat treatment. Ti-47A1-2Nb-1Mn-
O.5W-O.5Mo-0.28i alloy shows the best primary creep resistance (~3OO hrs to 0.5% creep
strain). Precipitates (size from 0.05 to 2 pm) were nucleated by heat treatment, and they
grew with the assistance of strain or transformation of 012 to 7. Many precipitates in this
alloy have considerable variety in the composition, crystal structure, and shape and size,
depending on their location in the microstructure. The nucleation and growth of these
precipitates interferes with dislocation motion, and they harden the microstructure. This
precipitation hardening mechanism accounts for much of the excellent creep resistance of
this alloy when it was heat-treated. Thus precipitation in TiAl alloys provide a large
benefit for creep resistance.

From quantitative microscopic comparisons between microstructures in
undeformed and deformed regions (with various strains, stresses, and temperatures) using
SEM and TEM techniques, all alloys exhibit lamellar refining that causes an initially
rapid creep rate up to strains between 0.1 and 0.5% during deformation. The reduction of
lamellar spacing is a consistent microstructural feature during primary creep deformation
in all investment cast TiAl alloys, and this refinement process occurs by mechanical
twinning (easy mode deformation) and/or 012 shear transformation parallel to lamellar
boundaries. The refinement process at the early stage of primary creep results in a rapid

drop of creep rate, because of limited dislocation motion on {111} planes not parallel to

278

the lamellar interface, and orientational hardening due to mechanical twinning or up, shear
transformation. Primary creep in TiAl alloys is divided into two stages, an initial rapid
process that refines the lamellar spacing through strain levels between 0.2 and 0.5%, and
a second primary creep process, where no lamellar refinement occurs. More lamellar
refinement occurs with a higher volume fraction of lamellar grains and with lower
temperature and higher stress. The best primary creep resistance occurs when the
transition occurs at a lower strain rate. This research shows that microstructural variance
has a large impact on primary creep resistance, but that the lamellar refinement process
occurs similarly in all microstructural variations. Thus a higher volume fraction of
lamellar grains is beneficial for creep resistance because a larger volume fraction of
lamellar interfaces will block overall dislocation motion during deformation.

From the investigation of PST crystals with different orientations for shear in
<110> or <112> directions, the orientation dependence of the activation energy for creep
and the stress exponent was examined (QC: 135 and 185 kJ/mol for the <110> and the
<112> direction, respectively). These values are about half of the 290 kJ/mol activation
energy measured for Ti diffusion. Furthermore, the amount of primary creep strain was
several times larger in the <112> direction than in the <110> direction, and this greater
amount of strain occurred at a much lower stress. The stress exponent of n=5.3 at higher
shear stress regions in both directions suggests that viscous dislocation glide (n=3)
controlled by pipe diffusion (n+2) can account for secondary creep in this microstructure
and lamellar orientation. Similar to the lamellar refinement in TiAl alloys, refinement of
lamellar spacing was also observed. In comparing activation energies of the early process

in primary creep experiments in the W-Mo-Si alloy with creep in the soft orientation in a

279

PST crystal (Qc=140 and 137~185 kJ/mole in the W-Mo-Si, and PST crystals,
respectively), the activation energies are similar: it is near 150 kJ/mol. The early
mechanism for primary creep in polycrystals is correlated with creep kinetics of easy
deformation in PST crystals that causes lamellar refinement.

The lamellar refinement process has been simulated to determine how much
twinning or shear transformation contributed to the reduction of lamellar spacing after
deformation under some limited assumptions. When several variations of initial lamellar
spacing distribution are considered, the resulting deformed distribution, and the resulting
local strains in the lamellar microstructure can be computed. The simulated lamellar
spacing distribution was quite close to the actual spacing distribution from the
measurements, when refinement was assumed to occur by on; shear transformation into
layers 0.2 pm thick in the thicker lamellae. The computed local shear strains (0.7%) in
lamellar microstructure were similar to or higher than the specimen (1%). The higher
volume fraction of lamellar grains implies less lamellar refinement per lamellar grain, but
with more lamellar grains, there is more strain accomplished by refinement, so the
transition strain is larger. Lamellar refinement not being proportional to local shear strain
is difficult to rationalize. There is a correlation between the computed local shear strain,
the lamellar volume fraction, and the strain at transition points between the early and later

primary creep process in most TiAl alloys.

280

APPENDIX

APPENDIX

PROGRAM CODES FOR SIMULATION OF LAMELLAR REFINEMENT

'program LAMDIST.BAS simulates lamellar refinement process
DIM SPACINGS(20, 500), CUR(250), DST(13, 10)
DEFINT I-N
RANDOMIZE
LUND = 100
TITLES = SPACE$(100)

SPEC$ = SPACE$(8)
KEY OFF

'10 PRINT

' PRINT "choose 0 to exit; "

' PRINT " 1 Normal simulation with true twinning"

' INPUT IOPT

' IF IOPT = 1 THEN GOSUB 100

' IF IOPT = 0 THEN CLOSE : END

' GOTO 10

' input initial and final distributions NUMBIN = number of bins

' B,U,D are bin size and fraction for undeformed, deformed distribution.

100 PRINT "Enter name of normalized distrib file, having 1 line header, binsize, undef,

deform"

' INPUT DIST$
DIST$ = "file name from raw data"

OPEN DIST$ FOR INPUT AS #3
INPUT #3, TITLES: TITLE$ = DIST$ + " " + TITLE$

281

PRINT "Bin size undef deformed starting value"
NUMBIN = O: SUMTHICK = 0!: DIS(O, O) = 0!: SUMUND = 0!: SUMDEF = O!

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

 

' input distribution data

WHILE NOT EOF(3)

INPUT #3, B, U, D

NUMBIN = NUMBIN + 1

DST(O, NUMBIN) = B: IBZ = O

DST(l, NUMBIN) = U * LUND: IUND = 1

DST(2, NUMBIN) = D * LUND: IDEF = 2

DST(3, NUMBIN) = U * LUND: ICUR = 3

SUMUND = SUMUND + DST(IUND, NUMBIN)

SUMDEF = SUMDEF + DST(IDEF, NUMBIN)

PRINT NUMBIN;

FOR I = 0 TO 3

PRINT USING "#####.#"; DST(I, NUMBIN);

NEXT 1: PRINT

WEND

CLOSE #3

DISTOUTS = DISTS + ".OUT"

OPEN DISTOUTS FOR OUTPUT AS #2

PRINT TITLES

PRINT NUMBIN; "Bins,"; LUND; "Lamellae in undeformed problem (LUND)"

PRINT #2, TITLES; NUMBIN; "Bins,"; LUND; "Lamellae in undeformed
problem (LUND)"

PRINT "Sum of undeformed and deformed distributions are"; SUMUND;
SUMDEF; "OK ? ? ";

INPUT K

PRINT #2, "Sum of undeformed and deformed distributions are"; SUMUND;
SUMDEF

282

PRINT #2, " Bin # sumdisp sumshear #twins total #lamellae Distribution..."
' Fill CUR(N) matrix randomly using undeformed distribution
' LUND = # lamellae in undeformed condition
' LBIN = # lamellae in a given bin
'NRAND = random position among available positions in CURO
FOR N = 0 TO LUND
CUR(N) = 0!
NEXT N: I = 0
FOR NB = 1 TO NUMBIN
BINSIZE = DST(IBZ, NB)
LBIN = DST(IUND, NB)
PRINT "NB; LBIN | # rand# -> pos:.."; NB; LBIN;
FOR N = 1 TO LBIN
NRAND = RND * (LUND - I)
PRINT "I"; N; NRAND;
IF I = LUND THEN 145
J = 0
FOR K = 1 TO LUND
IF CUR(K) <> 0 THEN 135
J = J + 1
IF I < NRAND THEN 135
PRINT "->"; K;
CUR(K) = BINSIZE
SUMTHICK = SUMTHICK + BINSIZE
I=I+ l: GOTO 140
135 NEXT K
140 NEXT N: PRINT : INPUT K
145 NEXT NB
SPACINGS(O, 0) = 0!: SPACINGS(1, O) = 0!

283

FORNC =1 TO LUND +2
SPACINGS(O, NC) = SPACINGS(O, NC - 1) + CUR(NC)
SPACINGS(1, NC) = 0!
PRINT CUR(N C);
NEXT NC
PRINT " ---- Sumthick ="; SUMTHICK
' Start looping with strain, refining lamellae in the largest bins first
' LBIN = number of lamellae in given bin, NUB = number of bins
' LCUR = # lamellae in CUR(), LUND = INITIAL # lamellae in CUR()
' NREFIN = number of refinement episodes for given bin
' LREFIN = cumulative total number of refinement episodes
' NSHRMAX = used to determine maximum number of columns for position—shear
output

' Will not subdivide the lamellae in the smallest bin

III"IHHHHIHI""H"I"I""IIIHIHMHHHHHH"IIIHHHIHIIHHI""HHHINIIHHO"""HHHHHHHHH"I""I""HI""9"""0""NIH"!!!"""H'IHHHIH"NH"

LCUR = LUND: LREFIN = O: NSHRMAX = 0
FOR NB = NUMBIN TO 2 STEP -1
NREF IN = 0
IF DST(IUND, NB) < DST(IDEF, NB) - .5 THEN 300
'above prevents refinement of lamellae for bin sizes below the crossover point
FOR NC = 1 TO LCUR
IF DST(ICUR, NB) <= DST(IDEF, NB) + .5 THEN GOTO 200
'above prevents distribution from going below the deformed distribution for current bin
IF CUR(NC) <= (DST(IBZ, NB) + DST(IBZ, NB - 1)) / 2 THEN GOTO 190
'above skips lamellae that are smaller than the current bin size indicated by NB

NREFIN = NREFIN + 1: LREFIN = LREFIN + 1

A : RND * CUR(NC): B = RND * CUR(NC) "1""I"HI"H""IHHONH"H"HHHHHHHHHHHH"UH"
IF A > B THEN ' thick lamellae divided
D(1) = B: D(2) =.2 I"IIN"IHHHHHHHI"IIN"I"""HHHIHHNNN"""H'"

D(3) = CUR(NC) - D(1) — D(2)

284

ELSE
D(1)= A: D(2) =.2
D(3) = CUR(NC) - D(l) - D(2)
END IF
FOR K = 1 TO NUMBIN
PRINT USING "####.#"; DST(ICUR, K);
NEXT K: PRINT " = Current Distrib"
FOR K =1 TO NUMBIN
PRINT USING "####.#"; DST(ICUR, K) - DST(IDEF, K);
NEXT K: PRINT " = Cur-Def Distrib"
PRINT "Bin"; NB; "DST(IUND, NB) & DST(IDEF, NB) ="; DST(IUND, NB);
DST(IDEF, NB);
PRINT "Found lath at NC ="; NC; "at/near"; DST(ISZ, NB)
PRINT "=> D(1)+D(2)+D(3) ="; 0(1); D(2); D(3); "="; CUR(NC); "=
CUR(NC)"

iI"IIHHHHH"HHNNHHHHHHH"HI"""0""H"""Hlllnflflfl'""N"HHHHIHH"I"UNI"!"I"IHIHINIHHIII"""HIHIHIHHHIHH"I"I"HM!"

----- find condition to accept adding one to the minimum sized bin - later
' Adjust distrib; N1 = index for bins NB and smaller
DST(ICUR, NB) = DST(ICUR, NB) - 1
FOR N1 = NB TO 1 STEP -1
FOR I = 1 TO 3
IF N1 = NB THEN
IF ABS(D(I)) > (DST(ISZ, N1) + DST(ISZ, NI - 1)) / 2 THEN 160
ELSE
IF ABS(D(I)) > (DST(ISZ, N1) + DST(ISZ, N1 + 1)) / 2 THEN 165
END IF
IF N1 = 1 THEN GOTO 160
IF ABS(D(I)) < (DST(ISZ, N1) + DST(ISZ, NI - 1)) / 2 THEN 165
160 DST(ICUR, N1) = DST(ICUR, N1) + 1
165 NEXT I

285

170 NEXT NI

IIOIIVIIIIIIII!!!IIIIIVIIIIIIIIIIOIIIIIIOOIIIIIIIIIOII'I'lIIIIIIOIVIIIUOOIIIOIOIIIIIlllII'IIOIIIOIIi'1!!!lVIII!I'1!!!ICOIOIII'OIOIIIUOIlIOIOIIIIUIII'I'Iiilllllllit'lI'!Il"'ll'llll|"!'

' Shift lamellae down in CUR(), put refined lamellae into correct position

compute shear displacements and strain

0'!!!lll'Iiil'IIl'I'VIIVIIOIIIIIIIIOIIUIVIVI'II'Vlllli'lll!O'l'lll'I'lO'Ill'IIO!Il0I!!!III'IOIli'll'l'V'IIIIIIIIIQIOOIIVIIIIIIIIIOI[DOIIIVIIVIIIIIIIIIIIVIIIIIIUIIIIOIIIIIIIIIIIOOIIIIIII

FOR N1 = LCUR + 2 TO NC + 3 STEP -1
CUR(NI) = CUR(NI - 2)
NEXT NI: LCUR = LCUR + 2
CUR(NC) = D(l): CUR(NC + 1) = -D(2): CUR(NC + 2) = D(3)
SUMDISP = SUMDISP + D(2) * .7071: shear = SUMDISP / SUMTHICK
FOR K = 1 TO LCUR
PRINT CUR(K);
NEXT K: PRINT "----"; LCUR; "lamellae"
PRINT "Sum of displacements, Shear strain ="; SUMDISP; shear; NREF IN;
"twins,"; LREFIN; "total"
INPUT K
190 NEXT NC
200 IF NREFIN = 0 THEN 220
PRINT "---> Finished bin"; NB; "writing distribution to file": PRINT
PRINT #2, USING "###.####"; NB; SUMDISP; shear;
PRINT #2, USING "##### "; NREF IN; LREFIN; LCUR;
FOR K = 1 TO NUMBIN
PRINT #2, USING "#####.##"; DST(ICUR, K);
NEXT K: PRINT #2,
' Create cumulative arrays for plotting stages of lamellar refinement
' NPOS is position in lamellare crystal, NTWTR is twin shear translation
SPACINGS(NB + 1, O) = 0!: SPACINGS(NB, 0) = 0!
NPOS = NB * 2: NTWTR = NPOS + 1: IF NTWTR > NSHRMAX THEN
NSHRMAX = NTWTR

286

220

FOR NC = 1 To LCUR

SPACINGS(NPOS, NC) = SPACINGS(NPOS, NC - 1) + ABS(CUR(NC))

IF CUR(NC) > 0 THEN

SPACINGS(NTWTR, NC) = SPACINGS(NTWTR, NC - 1)

ELSE

SPACINGS(NTWTR, NC) = SPACINGS(NTWTR, NC - 1) - CUR(NC) * .7071
END IF

NEXT NC

NEXT NB

it!!!ll'0I!I'l'I'lIOOIIVOOOIVIIIOIIOOOOIIII'O'O'Oii'lll'

' Output results

1lIi!!!I!!!I!IIOVUII...I!!!IO!I!!!I'Oi'VlIIIIIIt'ii'Il'!

300

PRINT "writing shear data...":

PRINT #2, "Bin # Bin size undef deform computed"

FOR NB = 1 TO NUMBIN

PRINT #2, USING "#####.##"; NB; DST(O, NB); DST(l, NB); DST(2, NB);

DST(3, NB)

NEXT NB: PRINT NSHRMAX + 1; "columns printed for shear-position data"
PRINT #2, "Position Displace @Bin 1 @Bin 2 @Bin 3 ";
PRINT #2, " @Bin 4 @Bin 5 @Bin 6 @Bin 7 "
FOR NC = 0 TO LCUR

FOR I = 0 TO NSHRMAX

PRINT #2, USING "#####.##"; SPACINGS(I, NC);

NEXT 1: PRINT #2,

NEXT NC

PRINT " -------- DONE -------- "

END

287

LIST OF REFERENCES

REFERENCES

10.

11.

12.

13.

14.

15.

16.

17.

. R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,

p. 2.

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p. 18.

William D. Callister, Materials Science and Engineering an Introduction, 4th edition,
John Wiley & Sons, Inc., 1997, p. 220.

W. Blum, Materials Science and Technology, Eds. R.W. Cahn, P. Haasen and E. J.
Kramer, vol. 6, VCH, Weinheim, 1993, p. 367.

Harry Kraus, Creep Analysis, Wiley-interscience, 1980, pp. 8-9.

R.E. Smallman, Metallurgy Modern Physical, 4th Edition, Butterworth, 1985, p. 450.
R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p. 20.

R.W. Evans and B. Wilshire, Creep of Metals and Alloys, The Institute of Materials,
1985, p. 79.

OD. Sherby and PM. Burke, Prog.Mater. Sci., 1967, (13), p. 327.

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p. 22.

R.W. Evans and B. Wilshire, Creep of Metals and Alloys, The Institute of Materials,
1985, p. 95.

MarcAndre Meyers, and Krishan Kumar Chawla, Mechanical Metallurgy Principles
and applications, Prentice-Hall, Inc., 1984, p. 662.

C. Zener and J .H. Hollomon, J. Appl. Phys. 15, 1944, p.22.

J .E. Dom, in Creep and Recovery, ASM, Clevland, OH, 1957, p.255.

J .E. Dom, in Creep and Fracture of Metals at High Temperatures; Proc. NPL
Symposium, HMSO, London, 1956, p. 89.

R.W. Evans and B. Wilshire, Creep of Metals and Alloys, The Institute of Materials,
1985,p.81.

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p. 25.

288

l8.
19.

20.
21.

22.

23.

24.
25.
26.

27.

28.
29.
30.
31.
32.
33.
34.

35.
36.
37.
38.
39.
40.

B. Burton and G.W. Greenwood, Acta Metall., 1970, (18), p. 1237.

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p. 26.

P. Feltham and J .D. Meakin, Acta Metall., 1959, (7), p. 614. .

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p.27.

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p. 31.

F .R.N. Nabarro, Report of a Conference on Strengthnof Solids, Physical Society,
London, 1948, p. 75.

C. Herring, J. Appl. Phys., 21, 1950, p. 437

R.L. Coble, J. Appl. Phys., 34, 1963, p. 1679.

Josef Cadek, Creep in Metallic Materials, Elsevier Science Publishing Co., 1988,

p. 210

B. Burton, Diffusional Creep in Polycrystalline Materials, Trans. Tech. Publ., Bay
village, OH, 1977

Puttick and King, J. Inst. Metals, 1952, 81, p. 537. '

A.E. Aigeltinger and RC. Gifldns, J. Mater. Sci., 1975, 10, p. 1889.

RC. Gifl<ins, Fracture, Technological Press, courtesy of John Wiley & Sons.

R. Raj and M. F. Ashby, Metall. Trans, 1971, 2, p. 1113.

W. Beere, Met. Sci., 1976, 10, p. 133.

M.V. Speight, Acta Metall., 1975, 23, p. 779.

NE. Mott, in Creep and Fracture of Metals at High Temperatures, Proc. NPL Symp.,
HMSO London, 1956, p.21.

L. Raymond and J .E. Dom, Trans. AIME 230, 1964, p. 560.

CR. Barrett and W.D. Nix, Acta Metall., 1965, 13, p. 1247.

J. Weertman, J. Appl. Phys., 1957, 28, p. 1185.

A.H. Cottrell and MA. Jaswon, Proc. R. Soc., London, 1949, A199, p. 104.

J. Weertman, J. Appl. Phys., 1957, 28, p. 362.

J. Weertman and JR. Weertman, in Physical Metallurgy, R.W. Cahn (Ed.), North
Holland, Amsterdam, 1965, p. 736.

289

41
42

43.
44.

45.

46.

47.

48.

49.

50.

51

54.
55.
56.
57.

58
59

60.
61.

J. Weertman, Trans. ASM 61, 1968, p. 681.

J. Weertman, in Proc. J.E. Dom Symp. Rate Processes in Plastic Deformation of
Materials, J .C.M. Li and AK. Mukherjee (Ed.), ASM, Metals Park, OH, 1975, p. 315.
F .R.N. Nabarro, Philos. Mag, 1967, 16, p. 231.

MA. Meyers and K.K. Chawla, Mechanical Metallurgy principles and applications,
published by Prentice-Hall Inc., New Jersey, 1984, pp. 284-287.

R.E. Smallman, Modern Physical Metallurgy 4th Edition, Butterworth & Co Ltd,
London, 1985, pp. 284-285.

M. Yamaguchi, and Yukichi Umakoshi, Progress in Materials Science, vol. 34, 1990,
pp. 69-73.

Z. J in, R. Beals, and IR. Bieler, Structural Interrnetallics, edited by R. Darolia, et al.,
TMS, 1993, p. 275.

CR. Feng. DJ. Michel and CR. Crowe, Scripta Metall., vol. 22, 1988, pp. 1481-
1486.

CR. Feng. D.J. Michel and CR. Crowe, Scripta Metall., vol. 23, 1989, pp. 1135-
1140.

Z. Jin, S.W. Chung, and TR. Bieler, Gamma Titanium Aluminides edit by Y-W. Kim
et al., TMS, 1995, pp. 975-983.

. O.D. Sherby and RM. Burke, Prog.Mater. Sci., 1967, (13), p. 325.
52.
53.

P. Yavari, F.A. Mohamed and TC. Longdon, Acta metall. 29, 1981, p. 1945.

Josef Cadek, Creep in Metallic Materials, Elsevier Science Publishing Co., 1988, pp.
161-162.

O.D. Sherby and PM. Burke, Prog.Mater. Sci., 1967, (13), p. 1505.

K. Maruyama and S. Karashima, Trans. Japan Inst. Metals, 16, 1975, pp. 671-678.

K. Abe, H. Yoshinaga and S. Morozumi, J. Japan Inst. Metals, 40, 1976, pp. 393-399.
W.D. Nix, Intern. Symp. On Advances in Metal Deformation, Cornell Univ., Oct.,
1976.

H. Oikawa and K. Sugawara, Scripta Metall., vol. 12, 1978, p. 89.

H. Oikawa and K. Sugawara, Scripta Metall., vol. 12, 1978, p. 85.

J. Weertman, Trans. AIME 218, 1960, p.207.

J. Weertman, Acta Metall., 1977, (25), p. 1393.

290

62.

63.
64.
65.

66.
67.
68.
69.
70.
71.

72.
73.
74.

75.
76.
77.
78.
79.
80.

81.
82.
83.

TR. Bieler, D.Y. Seo, T.R. Everard, and PA. McQuay, Creep behavior of Advanced
Materials for the 213t Century, eds R.S. Mishra, A.K. Mukherjee, and K. Linga
Murty, TMS, San Diego Annual Meeting, March, 1999, in press.

Oehlert and A. Atrens, Acta metall. mater. 42, 1994, p. 1493.

J .C. Gibeling, Acta metall. 37, 1989, p. 3183.

RS. Mishra,, T.K. Nandy, P.K. Sagar, A.K. Gogia, D. Banerjee, Trans IIM (India),
49,1996,p.331.

R.S. Mishra and D. Banerjee, Scripta Metall. et Mater. 31, 1994, p. 1555.

S.V. Raj and T. G. Langdon, Acta metall., 39, 1991, p. 1823.

TH. Alden, Metall. Trans, 18A, 1987, p. 811.

E. Krempl, J. Eng. Mataer. Tech. (Trans. ASME), 101, 1979, p. 380.

EN. da C. Andrade and K.H. Jolliffe, Proc. R. Soc. A254, 1960, p. 211.

GS. Daehn, X.J. Xin and RH. Wagoner, Creep and Fracture of Engineering
Materials and Structures, eds. J .C. Earthman and F. A. Mohamed, (TMS, Warrendale,
PA, 1997), p. 257.

R.L. Bagley , D.E.G. Jones, and AD. Freed, Metall Mater. Trans. 26A, 1995, p. 829.
BC. Odegard and A.W. Thompson, Met al.Trans. 4, 1974, p. 1207.

S. Suri, T. Neeraj, G.S. Daehn, D.H. Hou, J .M. Scott, R.W. Hayes, and M.J. Mills,
Creep and Fracture of Engineering Materials and Structures, eds. J .C. Earthman and
F. A. Mohamed, (TMS, Warrendale, PA, 1997), p. 119.

R.W. Hayes, Scripta Metall. Mater. 29, 1993, p. 1229.

RC. Ecob, Acta Metall. 34, 1986, p. 257.

H. Zhou, Q.P. Kong, Scripta Mater. 34,1996, p. 269.

AS. Argon, A.K. Bhattacharya, Acta metall. 35, 1987, p. 1499.

Orlova, Mater. Sci. and Eng. A , A163, 1993, p. 61.

M. Heilmaier, Modeling the Mechanical Response of Structural Materials, eds. EM
Taleff and RK Mahidhara, (TMS, Warrendale, PA, 1998), p. 137.

V. Sass, U. Glatzel, M. Feller-Kneipmeier, Acta mater. 44, 1996, p. 1967.

TM. Pollock, A.S. Argon, Acta Metal. et Mater., 40, 1992, p. 1.

S-U. An, J. of the Korean Institute of Metals and Materials, 30, 1992, p. 247.

291

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

99.

RS. Mishra and A.K. Mukherjee, Creep and Fracture of Engineering Materials and
Structures, eds. J .C. Earthman and F. A. Mohamed, TMS, Warrendale, PA, 1997), p.
237.

J .D. Parker, B. Wilshire, Scr. Metall., 13, 1979, p. 669.

PR. Bhowal, E.F. Wright, E.L. Raymond, Metal Trans. 21A, 1990, p. 1709.

J. Weertman, Trans. AIME 227, 1963, p. 1475.

M.F. Ashby, Acta Metall., 1972, (20), p. 887.

R.W. Evans and B. Wilshire, Creep of Metals and Alloys, The Institute of Materials,
1985,p.13.

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials, 1993,
p.47.

H.A. Lipsitt, D. Shechtman and RE. Schafrik, Metall. Trans. 6A, 1975, p. 1991.
Y-W. Kim, JOM, vol. 41, No.7, 1989, p. 24.

Y-W. Kim and EH. Froes, High Temperature Aluminides and Interrnetallics, eds.
S.H. Whang, et al., TMS, 1990, p. 465.

J .C. Chesnutt, Superalloys 1992, eds S.D. Antolovich, R.W. Stusrud et al., 1992,

pp. 381-389.

S.C. Huang and J .C. Chestnutt, in Intermetallic Compounds, ed., J. H. Westbrook and
R.L. Fleischer, John Wiley & Sons Ltd. press, 1994, p. 73.

C. McCullough, J .J . Valencia, H. Mateos, C.G. Levi and R. Mehrabian, Scripta
Metall., 22, 1988, p. 1131.

C. McCullough, J .J . Valencia, H. Mateos, C.G. Levi and R. Mehrabian, Acta Metall.
Mater., 37, 1989, p. 1321.

M. Yamaguchi, and Y. Umakoshi, Progress in Materials Science, vol. 34, 1990,
p. 13.

M. Yamaguchi, and Y. Umakoshi, Progress in Materials Science, vol. 34, 1990,
pp. 75-83.

100. M. Yamaguchi, and Y. Umakoshi, Progress in Materials Science, vol. 34, 1990,

p.24.

101. Y.W. Kim, Acta Metall., 1992, vol. 40, No. 6, pp. 1121-1134.
102. SM. L. Sastry and HA. Lipsitt, Metall. Trans. A 8, 1977, p. 299.

292

103

104.
105.

106.
107.

108.
109.

110.
111.

112.
113.

114.
115.
116.

117.
118.
119.
120.
121.
122.
123.

124

M. Yamaguchi, and Y. Umakoshi, Progress in Materials Science, vol. 34, 1990,
p. 65.

M. Yamaguchi, Metals and Technology 60, 1990, p. 34.

Y. Yamabe, K. Sugawara and M. Kikuchi, Proc. 1990 Tokyo Meeting, Japan Inst.
of Metals. Sendai: the Japan Institute of Metals, 1988, p. 409.

Y.S. Yang and SK. Wu, Scripta Metall., 1990, vol. 24, pp. 1801-1806.

M. Yamaguchi, and Y. Umakoshi, Progress in Materials Science, vol. 34, 1990,
p. 66.

Y.W. Kim, Acta Metall., vol. 40, No.6, 1992, pp. 1121-1134.

M. Yamaguchi, and Y. Umakoshi, Progress in Materials Science, vol. 34, 1990,
pp. 73-74.

Y-W. Kim and D.M. Dimiduk, JOM, vol. 43, No. 8, 1991, p. 40.

D.Y. Seo, T.R. Bieler and DE. Larsen, Structural Interrnetallics, edited by M.V.
Nathal et al., TMS, 1997, p. 137.

D.Y. Seo and TR. Bieler, unpublished research

T. Fujiwara, A. Nakamura, M. Hosomi, S.R. Nishitani, Y. Shirai, and M.
Yamaguchi, Phil. Mag. 61, 1990, p. 591.

ES. Bumps, H.D. Kessler and M. Hansen, Trans. AIME, 194, 1952, p. 1609.

P. Duwez and J .L. Taylor, JOM, January 1952, p. 70.

SC. Huang, E.L. Hall and M.F.X. Gigliotti, High Temperature Ordered
Intermetallic Alloys 11, ed. N.S. Stoloff, C.C. Koch, C.T. Liu and O. Izumi, MRS,
Pittsburgh, PA, 1987, p. 481.

J .W. Christian and DE. Laughlin, Acta Metall. 36, 1988, p. 1617.

D.W. Pashley, J .L. Robertson and M.J. Stowell, Phil. Mag. 19, 1969, p. 83.

M.H. Yoo, J. Mater. Res. 4, 1989, p. 50.

Z Jin and TR. Bieler, Phil Mag. A. 71, 1995, pp. 925-47.

Z J in and TR. Bieler, Phil Mag. A. 72, 1995, pp. 1201-19.

S. Farenc, A. Coujou, and A. Coret, Phil Mag A. 67, 1993, p. 127.

Z Jin, Ph.D. thesis, Michigan State University, 1994, pp. 98 and 108.

D. Shechtman, M.J. Blackburn and HA. Lipsitt, Metall. Trans. 5, 1974, p. 1373.

293

125.

126.

127.

128.

129.

130.
131.

132.

133.

134.
135.

136.

137.

138.
139.

140.

CR. Feng, D.J. Michel and CR. Crowe, Scripta Metall., vol. 23, 1989, pp. 241-
246.

M.J. Blackburn, in The Science, Technology and Application of Titanium, eds. R.T.
Jaffee and NE. Promisel, Pergamon Press, London, 1970, p. 633.

D.Y. Seo, T.R. Bieler and DE. Larsen, Advances in the Science and Technology of
Titanium Alloy Processing, eds. I. Weiss, R. Srinivasan, P. Bania, and D. Eylon,
(TMS Warrendale, PA, 1997), pp. 603-622.

H.A. Lipsitt, D. Shechtman and RE. Schafrik, Metall. Trans. A6, 1975, p. 1991.

G. Hug, A. Loiseau and A. Lasalmone, Phil. Mag. A54, 1986, p. 47.

G. Hug, A. Loiseau and A. Lasalmone, Phil. Mag. A57, 1988, p. 499.

EL. Hall and S-C. Huang, High Temperature Ordered Intermetallic alloys III
(edited by CT. Liu, A.I. Taub, N.S. Stoloff and CC. Koch) MRS Symposia
Proceedings, Vol. 133, 1989, p. 693.

S.H. Whang and Y.D. Hahn, High Temperature Aluminides and Intermetallics
(edited by S.H. Whang, C.T. Liu, D.P. Pope and J .O. Stiegler) TMS, Warrendale,
PA, 1990, p. .91.

V.K. Vasudevan, M.A. Stucke, S.A. Court, H.S.Fraser, Phil. Mag. Lett., 59, 1989,
p. 299.

S. Sriram et al., Mat. Res. Soc. Symp. Proc. vol.364, 1995, p. 647.

S.H. Whang, Z.M. Wang, and Z.X. Li, Gamma Titanium Aluminides edit by Y-W.
Kim et al., TMS, 1995, p. 245.

S.H. Whang, Q. F eng, and Z.M. Wang, Proc. of 7th Int. conf. on Creep and Fracture
of Engrg. Mat.& structure edit by J. C. Earthman et a1, TMS, 1997, p. 319.

B. Viguier, et al., Gamma Titanium Aluminides edit by Y-W. Kim et al., TMS,
1995,p.275.

B. Viguier, et al., Mat. Res. Soc. Symp. Proc. vol.364, 1995, p. 653.

Z.M. Wang , and SH. Whang, Sym. on Gamma Titanium Aluminides, TMS
Annual Meeting, Orlando, FL, 1997, To be published in Met. Trans. A.

D.M. Dimiduk, Sym. on Gamma Titanium Aluminides, TMS Annual Meeting,
Orlando, FL, 1997, To be published in Met. Trans. A.

294

141.

142.

143.

144.

145.
146.

147.

148.

149.

150.

151.

152.

153.

154.

155.

156.

TR. Bieler and D.Y. Seo, Deformation and Fracture of Ordered Intermetallic
Materials, eds. W.O. Soboyejo, H.L. Fraser, T.S. Srivatsan, (TMS, Warrendale, PA
1996), pp. 89-100.

J. Luster and MA. Morris, Metall. Trans. A, 26A, 1995, pp. 1745-1756.

M. Yamaguchi and H. Inui, Structural Intermetallics, edited by R. Darolia, et al.,
TMS, 1993, p. 127.

M. Yamaguchi and H. Inui, Ordered Interrnetallics-Physical Metallurgy and
Mechanical Behavior (Proc. NATO Advanced Research Workshop), ed. C.T. Liu,
R.W. Cahn and G. Sauthoff (Kluwer Academic Publishers, Dordrecht, 1992),
p. 217.

H. Inui, K. Kishida et al., submitted to Phil. Mag. A in December 1994.

M. Yamaguchi, H. Inui, and K. Kishida et al., Mat. Res. Soc Symp. Proc. vol.364,
1995,p.3.

T. Fujiwara, A. Nakamura, M. Hosomi, S.R. Nishitani, Y. Shirai, and M.
Yamaguchi, Phil. Mag. 61 , 1990, p. 600.

Z. Jin, G.T. Gray III and M. Yamaguchi, Mat. Res. Sco. Symp. Proc. vol.460, 1997,
p. 189.

Y. Ishikawa, K. Maruyama , and H. Oikawa, Mater. Trans, JIM, 33, 1992,

pp. 1182-1184.

H. Oikawa, Mat. Sci. & Eng, A153, 1992, pp. 427-432.

M. Lu and K.J. Hemker, Acta Metall., 1997, vol. 45, No. 9, pp. 3573-3585.

B.D. Worth, J .W. Jones, and J.E. Allison, Met.and Mat. Trans. A, vol. 26A, 1995,
pp. 2947-2959.

J. Triantafillou, J. Beddoes, and W. Wallace, Canadian Aeronautics & Space
Journal, 42, 1996, p. 108.

J. Beddoes, L. Zhao, J. Triantafillou et al., Gamma Titanium Aluminides edit by Y-
W. Kim et al., TMS, 1995, pp. 959-966.

M. Es-Souni, A. Bartels, and R. Wagner, Mat. Sci. and Eng. A171, 1993, pp. 127-
141.

M. Es-Souni, A. Bartels, and R. Wagner, Structural Intermetallics, edited by R.
Darolia, et al., TMS, 1993, pp. 335-343.

295

157.
158.
159.
160.
161.

162.
163.

164.

165.

166.
167.

168.
169.

170.

171.

172.
173.

174.

175.

D.I. Kimm, and J. Wolfenstine, Scripta Met. et Mat., vol. 30, 1994, pp. 615-619.

U. Hofmann and W. Blum, Scripta Met. et Mat., vol. 32, 1995, pp. 371-376.

T.G. Nieh and J .N. Wang, Scripta Met. et Mat., vol. 33, 1995, pp. 1101-1106.

J .N. Wang, A.J. Schwartz, et al., Mat. Sci. and Eng. A206, 1996, pp. 63-70.

J.S. Huang and Y.W. Kim, Scripta Met. et Mat., vol. 25, No.8, 1991, pp. 1901-
1906.

R.W. Hayes, Scripta Met. et Mat., vol. 29, 1993, pp. 1229-1233.

D.A. Wheeler, B. London, and DE. Jr. Larsen, Scripta Met. et Mat., vol. 26, 1992,
pp. 939-944.

B.D. Worth, J.W. Jones, and J .E. Allison, Met.and Mat. Trans. A, vol. 26A, 1995,
pp. 2961-2972.

T.R. Bieler, D.Y. Seo et al., Gamma Titanium Aluminides edit by Y-W. Kim et al.,
TMS, 1995, pp. 795-802.

CH. Wu, Master Thesis in Michgan State University, 1995, pp. 80-99.

M.F. Bartholomeusz, Q. Yang and J.A. Wert, Scripta Met. et Mat., vol. 29, 1993,
pp. 389-394.

T. Takahashi and H. Oikawa, Mat. Res. Sco. Symp. Proc. vol.213, 1991, p. 721.

P. L. Martin, M. G. Mendiratta, and H. A. Lipsitt, Metall. Trans. A, A14, 1983,
p.2170.

H. Oikawa, Proceedings of High Temperature Aluminides and Intermetallics, ed. by
S. H. Whang, C. T. Liu, D. P. Pope, and J. O. Stiegler, TMS, 1990, p. 353.

H. Nagai, T. Takahashi, and H. Oikawa, J. Mat. Sci., 25, 1990, p. 629.

T. Takahashi, H. Nagai, and H. Oikawa, Mat. Sci. and Eng, A128, 1990, p. 195.

T. Takahashi, and H. Oikawa, Proceedings of Microstructure/Property
Relationships in Titanium Aluminides and Alloys, ed. By W.Y. Kim and R. Boyer,
TMS, 1991, p. 227.

H. Oikawa, and T. Takahashi, Proceedings of Creep and Fracture of Engineering
Materials and Structure, ed. By B. Wilshire and R. Evans, The Institute of Metals,
London, 1990, p. 237.

Loiseau, and A. Lasalmonie, Mat. Sci. Eng, 67, 1984, p. 163.

296

176.

177.

178.

179.

180.

181.

182.

183.

184.

185.

186.

187.

188.

189.

190.

191.

192.

V. Seetharaman, S. Semintin, C. Lombard, and N. Frey, Mat. Res. Soc. Symp. Proc.
voL288,1992,p.513. .

G. Viswanathan and V. Vasudevan, Mat. Res. Soc. Symp. Proc. vol.288, 1992,
p. 787.

R. W. Hayes and B. London, Acta. Metall. et Mater., 40, 1992, pp. 2167.

W. O. Soboyejo and R. J. Lederich, Structural Intermetallics, ed. by R. Darolia, J. J.
Lewandowski, C. T. Liu, P. L. Martin, D. B. Miracle, and M. V. Nathal (TMS,
Warrendale, PA, 1993), p. 353.

M.Es Souni, A. Bartels, and R. Wagner, Acta. Metall. et Mater., 43, 1995, pp. 151-
161.

A Bartels, J. Seeger, and H. Meching, Mat. Res. Soc. Symp. Proc. vol.288, 1992,
p.1179.

C.R. Feng, and K. Sadananda, Mat. Res. Soc. Symp. Proc. vol.288, 1992, p. 1155.
R. W. Hayes and P. A. McQuay, Scripta Metall. et Mater., 30, 1994, p. 259.

D. Shih, et al, Proceedings of Microstructure/Property Relationships in Titanium
Aluminides and Alloys, ed. By W.Y. Kim and R. Boyer, TMS, 1991, p. 135.

TO. Noda, M. Isobe, and M. Sayashi, Mat. Sci. and Eng, A192/193, 1995,
pp. 774-779.

Li, J. Wolfenstine, J .C. Earthman, and E.J. Lavemia, Metall. Trans. A, Vol. 28A,
1997,p.1849.

S. Kampe, J .D. Brayant and L. Christodoulu, Metall. Trans. A, 22A, 1991, p. 447.
P. L. Martin, et al., Proceedings of Creep and Fracture of Engineering Materials
and Structure, ed. By B. Wilshire and R. Evans, The Institute of Metals, London,
1990,p.265

S. Cheng, J. Wolfenstine and O. Sherby, Metall. Trans. A, 23A, 1992, p. 1509.

SC. Huang, Structural Intermetallics, edited by R. Darolia, et al., TMS, 1993,
pp. 299-307.

M. Es Souni, A. Bartels, and R. Wagner, Mat. Sci. and Eng. A192/193, 1995,
pp. 698-706.

R.V. Ramanujan, P.J. Maziasz,and C.T. Liu, Acta. Met., vol.44, No.7, 1996,
p.2611.

297

193.

194.

195.

196.

197.

198.

199.

200.

201.

202.

203.

204.
205.

206.

207.

208.

209.

210.

211.

L.M. Hsiung and T.G. Niech, Scripta Met. et Mat., vol. 36, 1997, pp. 323-330.

Z. Jin, R. Beals, and T.R. Bieler, Mat. Res. Sco. Symp. Proc. vol.364, 1995, p. 505.
D.Y. Seo, T.R. Bieler and DE. Larsen, Data from Howmet Corp. 1995.

CH. Wu, Master Thesis in Michgan State University, 1995, p. 102.

CH. Wu, Master Thesis in Michgan State University, 1995, pp. 48-57.

Colin Thomas and Mark Behrendt, Advanced Materials and Processes, Vol. 150,
No.3, Sept, 1996, pp. 29-31.

J. B. Conway, Numerical Methods for Creep and Rupture Analysis, Gordon and
Breach, New York, 1967.

R.W. Evans and B. Wilshire, Creep of Metals and Alloys, The Institute of
Materials, 1985, pp. 197-243.

R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials,
1993,p.80.

K. Maruyama, C. Harada, and H. Oikawa, Journal of the Society of Materials
Science, Japan, 34 (1985), p. 1289.

K. Maruyama, and H. Oikawa, Scripta Metall., vol. 21, 1987, p. 233.

CH. Wu, Master Thesis in Michgan State University, 1995, p. 30.

D. Y. Seo, T.R. Bieler and DE. Larsen, Proceedings of 7th International
Conference on Creep and Fracture of Engineering Materials and Structures in
Irvine, CA, Aug. 1997, p. 577.

J. Beddoes, W. Wallace, and L. Zhao, Intl. Mater. Rev., 40, 1995, p. 197.

S. Tsuyama, S. Mitao and K. Minakawa, Mater. Sci. Eng, A153, 1992, p. 427.

SC. Huang, D.W. McKee, D.S. Shih and J .C. Chestnutt, Intermetallic Compounds-
Structure and Mechanical Properties, ed. 0. Izumi, (The Japan Institute of Metals,
Sendai, 1991), p. 363.

T. Maeda, M. Okada, Y. Shida and M. Nakanishi, Proc. 1989 Sapporo Meeting,
(The Japan Institute of Metals, Sendai, 1989), p. 238.

T. Maeda, M. Okada and Y. Shida, MRS Symp. Proc., vol 213, (MRS Pittsburgh,
PA, 1991), p. 555.

PR. Bhowal, H.F. Merrick and DE. Larsen, Mater. Sci. Eng, A192/ 193, 1995,

pp. 685-690.

298

212.

213.
214.

215.
216.
217.

218.

219.

220.

221.

222.

223.
224.
225.
226.
227.
228.
229.
230.

231.
232.

D.Y. Seo, S.U. An, T.R. Bieler, DE. Larsen, P. Bhowal, H. Merrick, Gamma
Titanium Aluminides, ed. Y-W. Kim, et al. (TMS, Las Vegas, NV 1995), p. 745.

Z. Jin and T.R. Bieler, Phil Mag, 70,(1995), pp. 819-836.

T.S. Kallend, U.F. Kocks, A.D. Rollett, and HR. Wenk, Mater. Sci. Eng, A132,
1991, pp. 1-11.

Y. Zheng, L. Zhao and K. Tangri, Scripta Metall., 26, (1992), p. 291.

SC. Huang and EL. Hall, MRS Symp. Proc., 133, 1989, p. 373.

V.K. Vasudevan, S.A. Court, P. Kurath and H.L. Fraser, Scripta Metall., 23(6),
1989.p.907.

M. Aindow, K. Chaudkuri, S. Das and H.L. Fraser, Scripta Metall., 24, 1990,
p.1105.

S. Sriram, V.K. Vasudevan and D.M. Dimiduk, MRS Symp. Proc., 213, 1991,
p. 375.

W.H. Tian and M. Nemoto, Gamma Titanium Aluminides, ed. Y-W. Kim,, R.
Wagner, and M. Yamaguchi, (TMS, Warrendale, PA, 1995), p. 689.

B.D. Worth, J .W. Jones and J .E. Allison, Gamma Titanium Aluminides, eds. Y.W.
Kim, R. Wagner, and M. Yamaguchi, (TMS, Warrendale, PA, 1995), pp. 931-938.
K. Muraleedharan and TM. Pollock, unpublished research, PRET home page,
http://titan.mems.cmu.edu/etch.html.

S. Yarnauchi and H. Shiraishi, Mater. Sci.Eng., A152, 1992, p. 283.

DE. Larsen, S.L. Kampe, Christodoulou, MRS Symp. Proc., 194, 1990, p. 285.
SC. Huang and EL. Hall, Metall. Trans. A, 22A 1991, p. 1619.

T. Kawabata, T. Tamura, and O. Izumi, Metall. Trans. A, 24A, 1993, p. 141.

M.J. Blackburn and MP. Smith, US. Patent 4294615, 1981.

SC. Huang and EL. Hall, Acta Metall. Mater., 6, 1991, p. 1053.

T. Tsujimoto and K. Hashimoto, Mat. Res. Soc. Symp. Proc., 133, 1989, p. 391.
G.H. Meier, D. Appalonia, R.A. Perkins, and KT. Chiang, in Oxidation of High-
Temperature Intermetallics, ed. T. Grobstein and J. Doychak (TMS, Warrendale,
PA 1988), p. 185.

D.W. McKee and SC. Huang, Corrosion Sci., 33, 1992, p. 1899.

J. B. McAndrew and H.D. Kessler, J. Metals, 8, 1956, p. 1348.

299

233.
234.
235.
236.

237.

238.
239.

240.
241.

242.

243.

244.

245.

246.

247.
248.

249.
250.

251.
252.

253.

W. Zhang, G. Chen, and Z. Sun, Scripta Metall., 28, 1993, p. 563.

S. Tsuyama, S. Mitao and K. Minakawa, Mater. Sci. Eng, A153, 1992, p. 427.

S. Tsuyama, S. Mitao, and K. Minakawa, Mater. Sci. Eng, A153, 1992, p. 451.

K. Kasahara, K. Hashimoto, H. Doi, and T. Tsujimoto, J. Japan Inst. Metals, 53,
1990,p.948

Y. G. Nakagawa, S. Yokoshima, and K. Matsuda, Mater. Sci. Eng, A153, 1992,
p. 722.

RA. Perkins, K.T. Chiang, and G.H. Meier, Scripta Metal, 21, 1987, p. 1505.

D. Y. Seo, T.R. Bieler and DE. Larsen, Mater. Res. Soc. Symp. Proc. Vol. 460
(Pittsburgh, PA, 1997) p. 281.

D.Y. Seo, CH. Wu, and T.R. Bieler, Mater Sci. and Eng. A239-40, 1997, p. 450.
R.W. Evans and B. Wilshire, Introduction to Creep, The Institute of Materials,
London, 1993.

M. Yamaguchi and Y. Umakoshi, Progress in Materials Science, 34, (1990), p. 71.
S. Farenc, A. Coujou and A. Couret, Mater. Sci. Eng, A164,l993, p. 438.

T. Hanamura and M. Tanino, J. Mater. Sci. Lett. 8, 1989, p. 24

F. Appel and R. Wagner, Gamma Titanium Aluminides, (TMS, Warrendale, PA,
1995), pp. 231-244.

S. Kroll, H. Mehrer, N.N. Stolwijk, C. Herzig, R. Rosenkranz, G. Frommeyer, Z.
Metallkde, 83, (1992), pp. 591-595.

L.M. Hsiung and T.G. Nieh, Mater Sci and Eng. in press.

T. Maeda, M. Okada and Y. Shida, in Proc. 6th Int. Conf. on ‘Mechanical behavior
of Materials VI’, Oxford, Pergamon Press, 1992, p. 199.

H. Oikawa, Mater. Sci. and Eng, A153 (1992), p. 427.

F. Garofalo, Fundamentals of Creep and Creep-Rupture in Metals, The MacMillan
Co., New York, Chap. 2.

Chih-Huei Wu, Master Thesis, Michigan state University, 1995, p. 31

K. Ouchi, Y. Iijima, and K. Hirano, Titanium ‘80-Science and Technology, (TMS
Warrendale, PA, 1981), p. 559.

D.Y. Seo, T.R. Bieler, S.U. An, and DE. Larsen, Mater. Trans. A, vol. 29A (1998),
pp. 89-98.

300

254

255.

256.

257.
258.

259.

260.
261.
262.
263.
264.
265.

266.
267.

268.

269.

T. Maeda, M. Okada, Y. Shida and M. Nakanishi, Proc. 1989 Sapporo Meeting,
(The Japan Institute of Metals, Sendai, 1989), p. 238.

Z. Jin and GT. Gray III, TMS symposium on Fundamentals of Gamma Titanium
Aluminides, Feb. 1997, eds. K.S. Chan, V.K. Vasudevan, Y.-W. Kim, to be
published in Mater. Trans.

R. Mahapatra, A. Girshick, D.P. Pope, V. Vitek, Scripta Metall. et Mater. 33, 1995,
pp. 1921-1927).

R. A. MacKay and I. E. Locci, unpublished research, NASA Lewis.

Wang and Nieh, TMS symposium on Fundamentals of Gamma Titanium
Aluminides, Feb. 1997, eds. K.S. Chan, V.K. Vasudevan, Y.-W. Kim, to be
published in Mater. Trans.

P.I. Gouma, S.J. Davey, and M.H. Loretto, Materials Science and Engineering,
A241(1998),pp.151-158.

H.S. Cho et al., submitted to Materials Science and Engineering (1998).

S. Chen, P.A. Beaven, and R. Wagner, Scripta Metall., 26 (1992), pp.1205-1210.

J .D. Whittenberger and Ranjan Ray, Scripta Metall., 33 (1992), pp.1505-1512.

H.S. Cho and SW. Nam, privcate communication, 1998.

Denquin et al., Scripta Metall., 28 (1993), pp.1 131-1136.

Menand, A. Huguet, and A. Nerac-Partaix, Acta Metall. Mater., 44 (1996),
pp. 4729-4737.

H.S. Cho, and D.Y. Seo, unpublished research, 1998.

R.W. Evans and B. Wilshire, Introduction to Creep, The institute of Materials,
London, 1993.

Kad, B.K., P.M. Hazzledine, H.L. Fraser, Defect-Interface Interactions , MRS
Symp. Proc. Vol. 319, (MRS, Pittsburgh, PA 1994), p. 311.

D.Y. Seo and T.R. Bieler and PA. McQuay, Interstitial and Substitutional Solute
Effects in Intermetallics, eds. 1. Baker, R.D. Noebe, and E.P. George, (TMS,
Warrendale, PA, 1998) p. 227

301