._, ..
. kw

hm

QLJQ

. 9..

, ,
,....,.r..mm.,
Iivd 5‘-
..&...§¥
n, 4.. QB.

m. .

 

_ «mmwmwgm» Ifé ..

‘1.

2

. 3
$331

hlfl

 

‘ . LIBRARY
2000‘ Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

REACTIVE EUTECTIC BRAZING OF NITINOL

presented by

KE-BIN LOW

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in ‘ ERIALS SCIENCE AND
MINEERING

 

 

QAQ, ,, ’4 Q7

 

VMajor Professor’s Signature

APRIL 27, 2009

 

Date

MSU is an Affinnative Action/Equal Opportunity Employer

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5I08 K:lProj/Acc&Pres/ClRC/DateDuerindd

 

 

REACTIVE EUTECTIC BRAZING OF NITINOL
By

Ke-Bin Low

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Materials Science and Engineering

2009

 

 

 

ABSTRACT
REACTIVE EUTECTIC BRAZING OF NITINOL
By
Ke-Bin Low
Although NiTiNb alloys are well known as wide-hysteresis shape-memory alloys
with important applications as coupling materials, the significance of one aspect of the
Ni-Ti-Nb ternary system has not been fully appreciated. Based on the existence of a
quasibinary NiTi-Nb eutectic isopleth in this ternary system, a novel braze method has
been devised to fabricate metallurgical bonds between functional nitinol (N iTi) sections.
When NiTi and pure Nb are brought into contact at temperatures above 1170 °C,
spontaneous melting occurs, forming a liquid that is extremely reactive and not only wets
NiTi surfaces, but also apparently dissolves oxide scales, obviating the need for fluxes
and providing for efficient capillary flow into joint crevices. The melting process is
diffusion-controlled and rate-limited by the diffusivity of Nb in the liquid. The braze
liquid will subsequently solidify into rnicrostructures containing predominantly ordered
NiTi and disordered bcc-Nb. Mechanical tests revealed that the braze joints are strong,
ductile, and biocompatible. With appropriate post-braze aging, the functional
performance of the parent NiTi alloy can be restored. Micro-alloying the Nb filler metal
with Zr or tungsten showed great potential for solution-strengthening of the braze joints.
For applications where biocompatibility is not an issue, Nb metal can be substituted by
Pure vanadium as the. braze filler, which is demonstrated to possess tensile strengths that

can be potentially superior to the Nb counterparts.

 

This dissertation is dedicated
to my dear wife, Ting

and to my beloved daughter, Sophie

iii

 

ACKNOWLEDGEMENTS

I would like to thank my academic advisorth committee chairman, Professor
David S. Grummon, for his guidance in not just research, but also in teaching. Professor
Grummon has taught me how to think out-of-the-box and has given me a lot of
opportunities to develop my interpersonal and leadership skills through teaching and
public-speaking at professional conferences. He is both a great mentor and a great friend.
I would also wish to express my gratitude towards Nitinol Devices and Components for
supplying the materials and generously funding the brazing project. I am also grateful
towards my PhD committee members, Professor Martin A. Crimp, Associate Professor
Thomas R. Bieler, from the Department of Chemical Engineering and Materials Science,
and Associate Professor Farhang Pourboghrat, from the Department of Mechanical
Engineering, for giving valuable comments and suggestions during my comprehensive
examination. I am also grateful towards Dr. Xudong Fan, from the Center for Advanced
Microscopy, for the scientific discussions on specimen preparations and transmission
electron microscopy. His knowledge and experience in this field has helped contribute
significantly towards the success of this dissertation. Lastly, I would like to thank my
colleagues, Xueling Fei, Jason W. Carter and John W. Foltz for all the help and laughter

in the laboratory. Without them, a lot of work will not be accomplished.

iv

PREFACE

The shape-memory/superelastic properties and excellent biocompatibility of
nitinol (N iTi) have made it an ideal candidate material for biomedical, aerospace and
micro-actuation applications. Many of these applications require nitinol to function in
tandem with another nitinol component or with dissimilar alloy systems. For example,
the nitinol core of a surgical guidewire has to be bonded to a platinum-tungsten alloy
outer coil for proper function; radio-opaque markers, such as tantalum, has to be attached
to nitinol stents reliably in order to improve radio-opacity; porous nitinol needs to be

bonded onto monolithic nitinol bone implant surfaces to improve osteo-conductivity.

However, joining nitinol to itself or to dissimilar alloy systems is very difficult
due to wettability issues and strong tendency for brittle intermetallics and oxides to form
within the joint. Many joining techniques, such as trmgsten-inert gas (TIG), laser,
resistance welding under Ar or He environments and soldering, have been explored.
These methods have not gained significant popularity, either due to the need for
aggressive fluxes, poor wetting, or reproducibility issues. In addition, most filler
materials for brazing or soldering are not biocompatible and hence cannot be employed in

biomedical applications.

Recently, a reactive eutectic braze method has been successfully devised for

joining separate nitinol components. It relies on the existence of a quasi-binary NiTi-Nb

eutectic isopleth in the ternary system. The joining process involves vacuum brazing
using pure Nb as the filler material. Preliminary work shows that brazing occurs
promptly above 1170°C; and upon cooling, the joints solidify mainly into a pro-eutectic
phase based on NiTi, and eutectic microstructure consisting of Nb plates or rods in a NiTi
matrix. Nitinol butt-joints, mechanically-tested in tension, strongly suggests that the
brazed joints are strong and ductile as long as braze duration does not extend beyond a
few minutes. With appropriate post-braze annealing, the NiTi base-alloy is still capable

of exhibiting shape-memory and transformational superelasticity.

This joining method demonstrates definitive potential for cost-effective industrial
applications because (1) the braze process does not destroy the shape-memory or
transformational superelastic properties of the base-alloy; (2) joints made by this process
have excellent mechanical strength and toughness; (3) the braze filler metal is
biocompatible, allowing the joining approach to be extended to medical device
applications. But to fully realize its potential, process optimization is of paramount
importance, which can only be achieved through thorough understanding of the relevant
phase equilibria, contact melting, and factors influencing the microstructure of the

solidified braze.

It is commonly known that metallurgical joining of Ti-based alloys tend to yield
brittle interrnetallic compounds that have detrimental effects on the mechanical
performance of the joints. Besides mechanical and microstructural aspects,

understanding the driving force for contact melting between NiTi and Nb during braze

 

processing, the melting kinetics and isothermal solidification are also of critical
importance for optimizing braze parameters. For example, the knowledge of the melting
kinetics will allow one to predict, based on simulations, the braze duration required for

complete dissolution of various thicknesses of the Nb filler metal.

The objectives of this research are: (1) to address the thermodynamic and kinetic
aspects of contact melting between NiTi and Nb during braze processing; (2) to study
contact melting kinetics and microstructural evolution of the joints as a function of braze
time; (3) to investigate the effects of post-braze thermal processing on mechanical
properties of the braze joints and base-alloys; and (4) to explore micro-alloying of braze

joints for possible improvements in mechanical performance.

The dissertation is organized into 6 chapters. Chapter 1 provides background and
literature review on phase equilibria, physical metallurgy, and joining of NiTi. In
addition, some theories on ternary phase equilibria, brazing, liquid phase diffusion, and
alloy solidification are introduced. Chapter 2 documents the methodologies associated
with the experimental and numerical work. Chapter 3 presents results on the kinetic
studies of contact melting and numerical simulations of this process. Solidification
products from brazing, and the evolution of microstructures as a function of time and
temperature are discussed. Chapter 3 also provides a qualitative analysis of the phase
field evolution in the Ni-Ti-Nb system that is based on current experimental findings.
Chapter 4 focuses on the mechanical properties of NiTi braze joints. Results were

derived from nanoindentation and tensile experiments. The effects of post-braze heat

vii

treatments on mechanical properties of these joints, and preliminary attempts to join NiTi
to Ta and alumina, are also discussed. Chapter 5 presents results of indentation and
tensile experiments on braze joints alloyed with Zr and tungsten. In addition, vanadium
was explored as an alternative braze filler metal and as a possible alloying element for Nb.
The microstructure, phase equilibria and mechanical properties of pure V-based joints
and V-alloyed joints will be discussed in this chapter. Finally, Chapter 6 provides
concluding remarks for the dissertation and lists possible future work pertaining to the

reactive eutectic braze process.

viii

TABLE OF CONTENTS

LIST OF TABLES ........................................................................... xii
LIST OF FIGURES .......................................................................... xiv
CHAPTER 1
BACKGROUND AND LITERATURE REVIEW ...................................... 1
1.1 Introduction .......................................................................... 1
1.2 The NiTi System .................................................................... 1
1.2.1 Transformation Properties ................................................ 1
1.2.2 Mechanical Properties ..................................................... 5
1.2.3 Phase Equilibria in the Ni-Ti Binary System ........................... 7
1.2.4 Effects of Composition on Transformation Temperature ............ 9
1.2.5 Effects of Therrno-mechanical Treatment on Shape-Memory
and Superelastic Characteristics .......................................... 10
1.2.6 Effects of Ternary Alloying .............................................. 14
1.2.7 Phase Equilibria of the Ni-Ti-Nb Ternary System ..................... 17
1.3 Joining of NiTi Components ...................................................... 20
1.3.1 Welding ..................................................................... 20
1 .3 .2 Soldering .................................................................... 24
1.3.3 Mechanical Joining ........................................................ 25
1.3.4 Brazing ...................................................................... 25
1.3.5 Transient Liquid Phase Bonding ......................................... 28
1.3.6 Preliminary Work on Reactive Eutectic Brazing of NiTi ............ 30
1.4 Brazing ............................................................................... 32
1.5 Transient Liquid Phase Bonding .................................................. 35
1.6 Liquid-Phase Diffusion ............................................................ 37
1.7 Interpretation of Ternary Phase Diagrams ....................................... 42
1.7.1 Basic Characteristics and Rules .......................................... 42
1.7.2 Classes of Ternary F our-Phase Equilibrium ........................... 45
1.7.3 Isothermal Sections and Isopleths ....................................... 49
1.7.4 Alkemades Theorem ...................................................... 52

ix

 

CHAPTER 2

EXPERIMENTAL METHODOLOGY ................................................... 55
2.1 Introduction .......................................................................... 55
2.2 Surface Cleaning .................................................................... 55
2.3 Braze and Post-braze Processing .................................................. 56
2.4 Micro-alloying with Tungsten in Nb-based Braze Joints ...................... 57
2.5 Sample Preparation and Setup for SEM and TEM Studies .................... 59
2.6 Tensile Testing ....................................................................... 61
2.7 Micro- and Nanoindentation ....................................................... 63
CHAPTER 3
MICROSTRUCTURAL EVOLUTION IN REACTIVE EUTECTIC BRAZIN G .. 70
3.1 Introduction ........................................................................... 70
3.2 Nomenclature ........................................................................ 70
3.3 Contact Melting Mechanisms ...................................................... 73
3.3.1 Initial Melting ............................................................... 74
3.3.2 Braze Liquid Propagation ................................................. 81
3.3.3 Termination ................................................................. 83
3.4 Contact Melting Kinetics ........................................................... 90
3.5 Solidification ........................................................................ 105
3.5.1 As-Brazed Microstructures ................................................ 105
3.5.2 Annealed Microstructures ................................................. 115
3.5.3 Effects of Surface Preparation and Vacuum Contaminants
on Joint Microstructures ................................................. 120
3.6 The Origin of the Perfect NiTi-Nb Quasibinary Eutectic Isopleth 122
3.7 Crystallography and Phase Equilibria of TizNi .................................. 128
3.8 Summary ............................................................................. 136
CHAPTER 4
MECHNICAL PROPERTIES OF NIOBIUM- EUTECTIC BRAZE JOINTS ...... 138
4.1 Introduction .......................................................................... 138
4.2 Nanoindentation Studies ........................................................... 138
4.3 Base-NiTi Properties and Annealing Response ................................. 143
4.4 Tensile Tests on Eutectic Braze Butt-Joints ..................................... 148
4.5 Tensile Tests on Tungsten Inert Gas-Welded Butt-Joints ..................... 156
4.6 Eutectic Braze Joints of Dissimilar Materials ................................... 160
4.6.1 NiTi-Tantalum Joints ...................................................... 161
4.6.2 NiTi-Alumina Joints ....................................................... 162
4.7 Summary .............................................................................. 166

CHAPTER 5
MICRO-ALLOYING OF NIOBIUM-EUTECTIC BRAZE FOILS AND
ALTERNATIVE EUTECTIC BRAZE SYSTEM ....................................... 168
5.1 Introduction .......................................................................... 168
5.2 Micro-Alloying of the Niobium-Eutectic Micro-constituents with
Zirconium and Tungsten ............................................................ 168
5.2.1 Effects on Microstructures ................................................ 169
5.2.2 Effects on Mechanical Properties ........................................ 172
5.3 Vanadium-Based Eutectic: An Alternative Braze System ..................... 180
5.3.1 Joint Microstructural and Mechanical Properties ...................... 181
5.3.2 Ternary Phase Equilibria in the Ni-Ti-V ................................ 188
5.4 Vanadium-Alloyed Niobium Braze System ..................................... 195
5.4.1 Joint Microstructural and Mechanical Properties ...................... 195
5.4.2 Quaternary Phase Equilibria in the Ni-Ti-Nb-V System .............. 203
5.5 Summary .............................................................................. 207
CHAPTER 6
CONCLUSION AND FUTURE WORK ................................................. 209
6.1 Conclusion ........................................................................... 209
6.2 Future Work .......................................................................... 215
APPENDIX A ................................................................................ 217
APPENDIX B ................................................................................. 221
APPENDIX C ................................................................................. 226
APPENDIX D ................................................................................ 231
BIBLIOGRAPHY ............................................................................ 235

xi

 

 

Table 3.1

Table 3.2

Table 3.3

Table 3.4

Table 3.5

Table 4.1

Table 4.2

Table 5 .1

Table A-1

Table A-2

Table B-1

Table B-2

Table 01

Table 02

Table D-l

LIST OF TABLES

Surface atomic concentrations of NiTi and Nb on low-index
planes. ........................................................................ 81

Localized concentration at diffusion couple interface of various
low-index plane combinations (First monolayer approximation). ...81

X-ray fluorescence analysis on selected spots in Figure 3.18. ....... 107
X-ray fluorescence analysis on selected spots in Figure 3.28. ....... 118
X-ray fluorescence analysis on selected spots in Figure 3.35. ....... 128
X-ray fluorescence analysis on selected spots in Figure 4.9. . 151
X-ray fluorescence analysis on selected spots in Figure 4.14. ....... 157
X-ray fluorescence analysis on selected spots in Figure 5.13. ....... 183
Input parameters for calculating incubation time for initial

melting. ...................................................................... 220
Results from the incubation time calculation. .......................... 220

Input parameters for simulating the melting kinetics before the
complete melting of Nb (assume rate-limited by liquid phase
diffusion). .. ...................................................................................... 225

Results from the melting kinetics calculation (before the
complete melting of Nb, assume rate-limited by liquid phase
diffusion). .................................................................... 226

Input parameters for simulating the melting kinetics after the
complete melting of Nb. ................................................... 230

Results from the melting kinetics calculation (after the complete
melting of Nb). .............................................................. 230

Input parameters for simulating the melting kinetics before the
complete melting of Nb (assume rate-limited by heat transfer). ..... 235

xii

 

Table D-2 Results from the melting kinetics calculation (after the complete
melting of Nb, assume rate-limited by heat transfer). ................. 235

xiii

 

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4
Figure 1.5

Figure 1.6

Figure 1.7

Figure 1.8

Figure 1.9

Figure 1.10

Figure 1.11

Figure 1.12

Figure 1.13

LIST OF FIGURES

Hysteresis loop involved during Martensite-Austenite

transformation with Ms, Mf, A, and Af indicated. ....................... 2
Graphical representation of modified Clausius-Clayperon equation
showing change in M, as a function of applied stress. ................ 5
Shape-memory effect and transformational superelasticity of NiTi
alloys. ........................................................................ 6
Ni-Ti binary phase diagram [Massalski 1990]. ........................ 8
‘I'I'I' diagram of 52 at.% Ni-Ti alloy [Nishida et a1. 1986]. .......... 9

Effect of Ni composition in NiTi alloys on martensite start
temperature. ................................................................. 9

Effect of annealing temperature on room-temperature yield
strength and martensite start temperature after 40% cold work in
the martensite phase of a 50.6 at.% Ni alloy [Melton 1990]. 12

900°C — Isothermal section and liquidus projections of the

Ni-Ti-Nb ternary system reported by [Gupta 1991] and

[Prima et a1. 1995]. (NiTi + Nb(Ni,Ti) tie-line at 900°C, close to

the quasibinary eutectic point, is also indicated.) ...................... 18

Summarized findings on the 900°C — isothermal section
[Yang et a1. 2000] and liquidus projection [Matsumoto et al. 2005]
of the Ni-Ti-Nb ternary system. .......................................... 19

In-plane compression test results showing (a) Strain — Time and
(b) Stress — Strain plots with (0) digital images at selected test
stages [Grummon et al. 2006, Shaw et a1. 2007]. ....................... 31

Collage of SEM micrographs showing the braze area of a room-
temperature shape-memory lap joint (inset: Higher-magnification

image of the braze microstructure). ..................................... 32
Sessile drop of liquid on a solid surface. ................................ 33
Schematic diagram showing a simple flow scenario. .................. 34

xiv

Figure 1.14

Figure 1.15

Figure 1.16

Figure 1.17

Figure 1.18

Figure 1.19

Figure 1.20

Figure 1.21

Figure 1.22

Figure 1.23

Figure 2.1

Figure 2.2

Gibbs triangle and a ternary alloy with an overall composition P. .. 43
Examples of tie-lines and tie-triangles in a ternary phase diagram. .45

Schematic of a ternary 4-phase Class-I reaction plane. (a + ,B + L),
(a + y + L) and (,8 + y + L) tie-triangles are above the reaction
plane. (a + ,8 + y) tie-triangle is below the reaction plane. ........... 46

Four-phase Class-I (NiTi, X3, X5 and Liquid) and Class-II (NiTi,
TizNi, Nb(Ni,Ti) and Liquid) reaction planes in the Ni-Ti-Nb
ternary phase diagram illustrated in 3D. ................................ 47

Schematic of a ternary 4-phase Class-II reaction plane.
(a + ,B + L) and (a + y + L) are above the reaction plane.
(a + ,B + y) and (,8 + y + L) are below the reaction plane. ............. 48

Schematic of a ternary 4-phase Class-III reaction plane.

(a + ,6 + L) tie-triangle is above reaction plane. (a + y + L),

(3 + y + L) and (a + fl + y) tie-triangles are below the

reaction plane. .............................................................. 49

Three-dimensional view of a 900°C — isothermal section in a
Ni-Ti-Nb ternary phase diagram. ......................................... 50

Three-dimensional view of a split Ni-Ti-Nb ternary phase
diagram, exposing the NiTi-Nb quasibinary section or isopleth. .. .. 51

Schematic diagram of a ternary system with a congruently-melting
phase, D, showing (a) the construction of Alkarnades line CD and

the directions of decreasing temperatures on the liquidus and
Alkarnades lines of concern; (b) the directions of decreasing
temperatures on the rest of the liquidus projection, based on the
Alkarnades theorem. ....................................................... 53

Schematic diagram of a ternary system with an
incongruently-melting phase, E, showing the construction of
Alkarnades line CE and the directions of decreasing temperatures

on the liquidus and Alkarnades lines of concern. ...................... 54

Typical temperature-time and pressure-time profiles of a braze
process. ...................................................................... 57

Plot of thickness vs. deposition time of tungsten at a cathode

power of 200W, argon working pressure of 5 mTorr and working
distance of 50 mm. ......................................................... 58

XV

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Schematic diagram showing top-view of the cross-sectional slice

of a eutectic joint that would eventually be prepared as a TEM
specimen. (The dotted circle indicates the outline of the
open-aperture copper grid that will be attached to the specimen.) .. 61

Schematic diagrams showing the procedures for fabricating NiTi
butt-joint specimens. ...................................................... 62

A typical load vs. displacement curve associated with loading and
unloading during nanoindentation experiments. ....................... 63

Hardness and elastic modulus of alumina derived from various
indentation holding depths. ............................................... 67

Hardness and elastic modulus of Nb derived from various
indentation holding depths. ............................................... 68

SEM micrograph showing an array of indents on pure Nb.
(Indent depth of 500 nm.) ................................................. 68

The ‘perfect’ NiTi-Nb quasibinary eutectic isopleth with key
compositions labeled for 1185°C. ........................................ 71

Schematic diagram of a half-joint configuration with frequently
used terms indicated. ....................................................... 73

Schematic diagram showing the temperature profile across a
half-joint as a frmction of time during the incubation period. .. . 75

Predicted NiTi-Nb interface temperature as a function of time
during the incubation period. ............................................. 77

SEM micrographs of the braze region showing NiTi-Nb contact
melting at braze duration of (a) 1 second; (b) 8 seconds;
(c) 21 seconds; (d) 31 seconds. .......................................... 82

SEM micrographs of the braze region at braze durations of
(a) 31 seconds; (b) 120 seconds; (c) 1200 seconds;
(d) 3600 seconds. ........................................................... 84

X-ray fluorescence line profiles of Nb La x-ray intensities across
pro-eutectic NiTi/pure NiTi interfaces of specimens subjected to
braze durations of (a) 31 seconds; (b) 3600 seconds. ................. 85

Microstructures of joints fabricated using 6 pm Nb film with braze
durations of (a) 30 seconds; and (b) 900 seconds. ..................... 87

xvi

Figure 3.9

Figure 3.10

Figure 3.11

Figure 3.12

Figure 3.13

Figure 3.14

Figure 3.15

Figure 3.16

Figure 3.17

Figure 3.18

Figure 3.19

Figure 3.20

Schematic diagram illustrating the establishment and evolution of a
concentration gradient in the braze liquid as a function of time,

based on the assumed perfect NiTi-Nb quasibinary eutectic

isopleth. ...................................................................... 92

Experimental and predicted solid/liquid interface positions during
dissolution at 1185°C. (Experimental interface positions were
measured and defined with respect to the initial NiTi-Nb mating
interface.) .................................................................... 95

Estimated isothermal section of the NiTi'NiéNb7'Nb‘Ti phase
diagram at 1185°C. Inset: Magnified view of the liquid phase at

the NiTi-Nb isopleth revealing compositions across the liquid

when Ni and Ti possess similar diffusivity (Line 1); Ni diffuses
slower than Ti (Line 2); and Ti diffuses slower than Ni (Line 3).

The average liquid composition is indicated by C1,. ................... 96

Predicted melt interface velocities for NiTi and Nb at 1185°C. 99

Schematic diagram illustrating the evolution of a concentration
gradient in the braze liquid as a function of time after Nb
exhaustion. ................................................................... 100

Experimental and predicted liquid half-width as a function of time
at braze temperature of 1 185°C. .......................................... 101

Schematic diagram showing the temperature profile across a
half-joint at initial condition and after some time, t, if melting is
rate-controlled by heat transfer. ........................................... 102

Predicted thickness of melted NiTi alloy as a function of time after
the incubation period. ...................................................... 104

Typical braze microstructure solidified before exhaustion of solid
Nb (contact melting at 1185°C). ........................................... 106

Elemental XRF maps showing (a) secondary-electron, (b) Nb La,
(c) Ti Ka and ((1) Ni Ka intensities originating from base-NiTi,
pro-eutectic NiTi and eutectic phases. ................................... 107

SEM micrographs showing (a) facetted TizNi-type precipitates
(indicated by ‘A’), (b) fine Nb-rich precipitates decorating sub-grain
boundaries of a pro-eutectic NiTi dendrite (indicated by ‘B’). . .. 108

SEM micrographs of a pro-eutectic NiTi dendrite observed under
(b) SE and (c) BSE modes. ................................................ 109

xvii

Figure 3.21

Figure 3.22

Figure 3.23

Figure 3.24

Figure 3.25

Figure 3.26

Figure 3.27

Figure 3.28

Figure 3.29

Figure 3.30

X-ray fluorescence line profile of Nb La across the perceived cored
pro-eutectic NiTi phase and the base-NiTi alloy. ...................... 1 10

Qualitative model accounting for the pro-eutectic NiTi
solidification morphology at different stages of brazing. . . . . . . . 11 1

Eutectic-Nb phase possessing larnellar and rod-like morphologies. 113

TEM micrograph of braze joints processed at 1185°C for 360

seconds showing in-plane eutectic-Nb rods and SAD pattern of the
eutectic-NiTi. (Diffraction originating from bottom-right of
micrograph.) ................................................................. l 14

TEM micrograph of braze joints processed at 1185°C for 360

seconds showing eutectic phase and SAD pattern of the

out-of-plane eutectic-Nb rod. (Diffraction originating from

indicated eutectic-Nb rod.) ................................................ 115

Microstructures of joints subjected to post-braze anneal at 800°C
for (a) 29 hours; and (b) 104 hours. ...................................... 116

SEM micrographs showing (a) eutectic-Nb rods undergoing
spheroidization and (b) completely spheroidized eutectic-Nb in
braze joints subjected to post-braze anneal at 920°C for 20 hours. .. 117

TEM micrograph from a braze joint subjected to post-braze anneal

at 920°C for 20 hours and water-quenched showing a spheroidized
eutectic-Nb particle in the eutectic-NiTi matrix, together with two
SAD patterns from the eutectic-NiTi. (SAD 2 was taken from
eutectic-NiTi at another location within the same specimen.)

Positions A and B were sites for XRF analysis. ........................ 118

High-resolution TEM micrograph of eutectic-NiTi revealing lattice
fiinges oriented on the B2 [113]-zone axis. Top-right image
schematically shows the real-space lattice arrangement (bright

spheres: Ni atoms; dark spheres: Ti atoms). Bottom-1i ght image is

the FFT pattern of the selected area showing periodicity in

reciprocal space. (F FT pattern originated from the indicated area

of the micrograph.) ......................................................... 1 19

SEM micrographs of braze joints processed at 1185°C for 120

seconds under (a) scenario 1; (b) scenario 2; and (c) scenario 3.
(Scenario 1: Good vacuum, Clean NiTi; Scenario 2: Poor vacuum,
Clean NiTi; Scenario 3: Poor vacuum, Oxidized NiTi) ................ 121

xviii

Figure 3.31

Figure 3.32

Figure 3.33

Figure 3.34

Figure 3.35

Figure 3.36

Figure 3.37

Figure 3.38

Figure 3.39

Figure 3.40

Figure 3.41

Proposed liquidus projections of the Ni-Ti-Nb ternary system with

the reported 900°C isothermal section. (The literature-reported
NiTi-Nb quasibinary eutectic point and NiTi + Nb(Ni,Ti) tie-line

at 900°C (close to the quasibinary eutectic point)

[Prima et a1. 1995]; and experimentally-determined pro-eutectic

NiTi and Nb compositions (Table 3.4) are indicated.) ................ 123

F our-phase Class-II reaction planes (U 1 and U2) in close proximity
to the NiTi-Nb quasibinary eutectic point. .............................. 124

NiTi + Nb(Ni,Ti) + Liquid phase fields from U; and U2

degenerating into a single NiTi + Nb(Ni,Ti) tie-line at 1170°C,
forming the reaction horizontal for the perfect NiTi-Nb quasibinary
isopleth. ...................................................................... 125

The proposed NiTi-Nb isopleth of the Ni-Ti-Nb ternary system. 126

SEM micrographs of joints brazed at 1155°C showing coexistence
of pro-eutectic NiTi, eutectic micro-constituents and TizNi-type
phase. ........................................................................ 128

TEM micrograph of braze joints processed at 1185°C for 360
seconds showing a (Ti,Nb)2Ni precipitate and its SAD pattern. . . 129

The 900°C isothermal tetrahedron of the quaternary Ni-Ti-Nb-O
system [Nevitt 1960, Gupta 1991, Prima et a1. 1995,
Yang et a1. 2000, Chuprina et a1. 2002]. ................................. 130

Unit cells of TizNi and Ti4Ni20 [Mueller et a1. 1963], and their
respective calculated power diffraction spectra. (Both phases

share similar strongest peaks, which are labeled according to their

hkl planes.) ................................................................... 1 3 1

NiTi + Nb(Ni,Ti) + Liquid phase field evolution as temperature
decreases fi'om 1165°C to 1155°C, resulting in Ti-enrichment in
the liquid phase. ............................................................ 134

Hypothesized NiTi + Nb(Ni,Ti) + (Ti,Nb)2Ni phase field evolution

as temperature decreases from 950°C. Gently-sloping NiTi +
Nb(Ni,Ti) + (Ti,Nb)2Ni / NiTi + Nb(Ni,Ti) phase boundary

providing re-solutionizing possibility for (Ti,Nb)2Ni. (Color tone

of phase fields corresponds to temperatures: Darker - Higher
temperature; Lighter - Lower temperature.) ............................ 135

The proposed 20°C isothermal section of the Ni-Ti-Nb system. .... 135

xix

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Load vs. Displacement curves of various phases. Indentation
holding depth: 100 nm. Strain rate: 0.001 per second.) ............... 139

Histograms of (a) hardness and (b) elastic modulus of as-brazed
eutectic joints showing the typical sampling distributions.

Continuous lines are Gaussian distribution curves fitted over the
histograms. (Data derived from indentation holding depth of

50mm.) ........................................................................ 141

Plots of (a) hardness and (b) elastic modulus versus post-braze

anneal temperatures of 350°C, 800°C and 920°C of eutectic joints

and base-NiTi. (Data derived from indentation holding depth of
50m.) ........................................................................ 142

Stress vs. strain plot of as-received room-temperature superelastic
base-NiTi control specimen. .............................................. 144

Stress vs. strain plot of as-received room-temperature superelastic
base-NiTi control specimen that had been subjected to aging

treatment at 350°C for 1.5 hours (then water-quenched) prior to
testing. ........................................................................ 145

Stress vs. strain plot of as-received room-temperature superelastic
base-NiTi control specimen that had been subjected to the typical
braze process conditions prior to testing. ................................ 146

Stress vs. strain plot of as-received room-temperature superelastic
base-NiTi control specimen that had been subjected to the typical
braze process followed by aging at 350°C for 1.5 hours (then
water-quenched) prior to testing. ......................................... 147

Stress vs. strain plot of as-brazed room-temperature superelastic
butt-joint specimen. (Initial joint clearance: Approximately 400
um.) ........................................................................... 149

Photograph of a room-temperature shape-memory single-lap joint

that had experienced onset of failure under shear loading. SEM
micrograph of cracks traversing facetted Ti-rich precipitates in the
joint region. Elemental XRF maps of the same area are also

included. ...................................................................... 150

Fracture surfaces of the (a) pro-eutectic NiTi; and (b) eutectic

phases in a typical superelastic butt-joint specimen that failed
under tension. Elemental XRF maps of (a) are also shown. ......... 151

XX

Figure 4.11

Figure 4.12

Figure 4.13

Figure 4.14

Figure 4.15

Figure 4.16

Figure 4.17

Figure 4.18

Figure 4.19

Figure 4.20

Stress vs. strain plots of as-brazed and aged superelastic butt-joint
specimens. (Aging regime: 350°C for 1.5 hours; 500°C for 20
minutes; 750°C for 10 minutes, followed by 350°C for 1.5 hours.

All anneals performed in air and water-quenched. Initial joint
clearance: Approximately 400 pm.) ...................................... 153

Stress vs. strain plots of an aged superelastic butt-joint undergoing

10 tensile loading/unloading cycles. (Aging regime: 300°C in air

for 8 hours followed by a water-quench. Initial joint clearance:
Approximately 400 um.) ................................................... 154

Stress vs. strain plots of as-brazed and aged superelastic butt-joints
with initial joint clearances of approximately 300 pm (narrow) and
400 um (wide). (Aging regime: 350°C for 1.5 hours in air,

followed by a water-quench.) ............................................. 155

SEM micrographs of (a) general as-welded joint microstructure,

and (b) high-magnification of the indicated area of (a). (XRF line
profiles showing composition variations across X-Y are also
provided.) .................................................................... 157

Stress vs. strain plots of as-welded and aged superelastic

butt-joints. (Aging regimes: 350°C for 1.5 hours; 500°C for 20
minutes. All anneals performed in air and cooling by

water-quench. Initial joint clearance: Approximately 400 um.) . .159

SEM rrricrographs of the fracture surface of a welded
room-temperature superelastic butt-joint that failed under tension.

(‘A’ indicates ‘caterpillar track’ features; ‘3’ indicates

brittle-mode cracks associated with the presence of ‘A’.) .. . . . . . . . 160

Stress vs. strain plot of as-brazed superelastic NiTi-Ta butt-joint.
(Inset shows necking occurred in the Ta section of the specimen.) .. 162

Schematic diagram showing the setup of NiTi-A1203 lap shear
specimen. .................................................................... 163

SEM micrograph showing (a) the typical cross-section of a
NiTi-A1203 braze joint. (b) is the contrast-enhanced duplicate of
(a). ............................................................................ 164

Tensile stress vs. strain plots of as-brazed superelastic NiTi-A1203
shear specimen. .............................................................. 165

xxi

Figure 4.21

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

Figure 5.6

Figure 5.7

Figure 5.8

Figure 5.9

Bar-chart (left) illustrating the hardness of various phases in a
NiTi-A1203 joint obtained from nanoindentations performed at the

area shown in the optical micrograph (right). Array of indents

across the joint can be seen (except the indents in the

base-A1203). ................................................................. 1 66

SEM micrographs showing (a) global joint morphology; and
(b) eutectic microstructure of a Zr-alloyed eutectic joint. .. . . . . . . . 170

Microstructures of braze joints fabricated using (a) pure Nb foil;
and (b) Nb-er foil. (Mating surfaces were deliberately
pre-oxidized before brazing under ‘poor’ vacuum conditions.) .. 171

SEM micrographs showing (a) global joint morphology; and

(b) eutectic micro-constituents of a W-alloyed eutectic joint. (Inset

of (b) shows the eutectic micro-constituents of a pure Nb-eutectic

joint under similar magnification as (b).) ............................... 172

Load vs. displacement plots of pure Nb-eutectic, Zr-alloyed
eutectic and W-alloyed eutectic. (Indentation holding depth:
50 nm. Strain rate: 0.001 per second.) ................................... 173

Histograms of (a) hardness; and (b) elastic modulus of pure Nb,

Zr- and W-alloyed eutectic braze joints. Continuous lines are
Gaussian distribution curves fitted over the histograms of the three
eutectic braze joints. (All tests performed on as-brazed joints.

Data derived fi'om indentation holding depth of 50nm.) .............. 174

(a) Optical micrograph; and (b) SEM micrograph showing Vickers
indents within the eutectic phase. ........................................ 175

Bar chart showing the Vickers hardness of pure Nb, Zr- and
W-alloyed eutectic braze joints. (All tests performed on as-brazed
joints at 300 grams.) ........................................................ 175

Stress vs. strain plots of Zr-alloyed as-brazed and aged superelastic
butt-joint specimens. (Aging regime: 300°C for 4 hours; 350°C

for 1.5 hours; 400°C for 4 hours; 500°C for 1 hour. All anneals
performed in air and cooling by water-quench. Initial joint

clearance: Approximately 400 um.) ...................................... 177

SEM micrographs showing (a) global fracture morphology; and

(b) highly-dirnpled fracture surface of a Zr-alloyed eutectic
butt-joint that failed under tensile loading. .............................. 178

xxii

Figure 5.10

Figure 5.11

Figure 5.12

Figure 5.13

Figure 5.14

Figure 5.15

Figure 5.16

Figure 5.17

Figure 5.18

Figure 5.19

Stress vs. strain plots of W-alloyed as-brazed and aged superelastic
butt-joint specimens. (Aging regime: 350°C for 1.5 hours. Anneal
performed in air and cooling by water-quench. Initial joint

clearance: Approximately 400 um.) ..................................... 179

SEM micrographs showing (a) global fi'acture morphology; and

(b) brittle cracks (indicated by ‘A’) among dirnpled fracture

surface of a W-alloyed eutectic butt-joint that failed under tensile
loading. ....................................................................... 180

SEM micrographs showing (a) global joint rrricrostructures; and
(b) eutectic phase of a V-eutectic braze joint. ........................... 182

Elemental XRF mapping of a V-eutectic joint. ........................ 182

Load vs. displacement plots of pure Nb- and V-eutectic
micro-constituents. (Indentation holding depth: 50 nm. Strain
rate: 0.001 per second.) ................................................... 183

Histograms of (a) hardness; and (b) elastic modulus of as-brazed

pure Nb- and V-eutectic braze joints. Continuous lines are

Gaussian distribution curves fitted over the histograms of the two
eutectic braze joints. (Data derived fiom indentation holding

depth of 50nm.) .............................................................. 184

Histograms of (a) hardness; and (b) elastic modulus of

thermally-aged pure Nb- and V-eutectic braze joints. Continuous

lines are Gaussian distribution curves fitted over the histograms of

the two eutectic braze joints. (Aging regime: 350°C in air for

1.5 hours, followed by water-quench. Data derived from

indentation holding depth of 50nm.) ..................................... 185

Stress vs. strain plots of as-brazed and aged room-temperature
superelastic butt-joints fabricated using V-based eutectic. (Aging
regime: 350°C for 1.5 hours. Anneal performed in air and cooling

by water-quench. Initial joint clearance: Approximately 400 pm.) .186

SEM micrographs showing (a) global fracture morphology; and

(b) ‘web’-likc ductile tearing features associated with the eutectic
phase of an as-brazed V-based eutectic butt-joint that failed under
tensile loading. .............................................................. 187

SEM micrographs showing (a) global fracture morphology; and

(b) ‘web’-like ductile tearing features associated with the eutectic
phase of a 350°C-aged V-based eutectic butt-joint that failed under
tensile loading. .............................................................. 1 87

xxiii

Figure 5.20

Figure 5.21

Figure 5.22

Figure 5.23

Figure 5.24

Figure 5.25

Figure 5.26

Figure 5.27

Figure 5.28

Figure 5.29

Figure 5.30

Estimated 920°C-isothermal section of the Ni-Ti-V ternary system
[Prima 1992]. ............................................................... 189

Stereology on NiTi-V eutectic micro-constituents (performed using
WSXM software from NanotecElectronica). ........................... 190

Schematic diagram showing a 3-dimensional network of a phase.

(a) Under-estimation of volume fi'action when stereology is

performed on out-of-plane rods of the network.

(b) Over-estimation of volume fraction when stereology is

performed on in-plane rods of the network. ............................. 191

Two possible liquidus projections for the Ni-Ti-V system: (a) two
class-I (E 2 and E2), two class-II (P2 and P2) reactions; (b) one

class-I (E 1), three class-II (P1, P2 and P3) reactions. (Quasibinary
eutectic points are also indicated.) ....................................... 193

A four-phase class-II reaction plane (P1 and P2) on either sides of

the NiTi-V isopleth in the ternary system. Both possess common

NiTi + V(Ni,Ti) + Liquid phase fields. (The reaction plane
temperatures are speculative.) ............................................ 194

A perfect NiTi-V quasibinary eutectic isopleth in the ternary
system. ....................................................................... 194

SEM micrographs showing (a) NiTi-x eutectic micro-constituents;

and (b) coexistence of NiTi-x eutectic (indicated by ‘A’) and

NiTi-Nb eutectic (indicated by ‘B’) micro-constituents in joints
fabricated using 54Nb-46V braze foils. ................................. 196

SEM micrographs showing (a) elongated grains of x phase

(indicated by ‘A’) among dense colonies of NiTi-Nb eutectic; and

(b) higher-magnification of the elongated grains of x phase in joints
fabricated using 77Nb-23V braze foils. .................................. 197

Elemental XRF mapping of a joint fabricated using 77Nb-23V
braze foil. .................................................................... 198

Histograms of (a) hardness; and (b) elastic modulus of as-brazed
54Nb-46V joints. ........................................................... 199

Stress vs. strain plots of aged room-temperature superelastic
butt-joints fabricated using 54Nb-46V and 77Nb-23V braze foils.
(Aging regime: 350°C for 1.5 hours. Anneal performed in air and
cooling by water-quench. Initial joint clearance: Approximately

400 um.) ..................................................................... 200

xxiv

Figure 5.31

Figure 5.32

Figure 5.33

Figure 5.34

Figure 5.35

Figure C-l

SEM micrographs showing (a) global fracture morphology; and

(b) mixed ductile/brittle fracture features in a 350°C-aged
room-temperature superelastic butt-j oint, fabricated using

54Nb-46V foils, which failed under tensile loading. (Ductile

features are indicated by ‘A’, while brittle features are indicated

by ‘B’.) ....................................................................... 201

Ultimate tensile strength of Nb braze joints as a function of
vanadium composition (experimental and fitted data shown). ....... 202

920°C-isotherm of the Ni-Ti-Nb-V quaternary phase diagram
showing only the NiTi, BCC-(Nb,V,Ti,Ni), Ti2Ni, X5 and 0’
phases and their tie-triangles. ............................................. 204

Possible 920°C-isotherm of the Ni-Ti-Nb-V quaternary phase

diagram showing how the x phase could establish equilibrium with

(a) NiTi, BCC-(Nb,V,Ti,Ni), Ti2Ni and 0’ phases in the Ni-Ti-V
system; or (b) NiTi, BCC-(Nb,V,Ti,Ni), TizNi and X5 in the

Ni-Ti-Nb system. ........................................................... 205

(a) Possible 920°C-isotherm of the Ni-Ti-Nb-V quaternary phase
diagram showing common “NiTi + BCC-(Nb,V,Ti,Ni) + x”
tie-triangles as one of their faces of the tie-tetrahedrons illustrated

in the bottom of the phase diagram in Figure 5.31. Space
configurations of the (b) “NiTi + BCC-(Nb,V,Ti,Ni) + x” phase

field; and (c) “NiTi + BCC-(Nb,V,Ti,Ni)” phase field. ............... 206

Predicted liquid half-width as a function of braze time. .............. 231

XXV

CHAPTER 1: BACKGROUND AND LITERATURE REVIEW

1.1 Introduction

In this chapter, the NiTi system will be reviewed with emphasis being placed on
its transformation properties, mechanical properties, phase equilibria, composition
variations, thermo-mechanical treatments and ternary alloying possibilities. Problems
and recent development associated with joining NiTi components will also be reviewed.
A brief overview of the principles of brazing, transient liquid phase bonding, liquid-phase

diffusion, and principles of ternary phase equilibria are also provided.

1.2 The NiTi System

Nickel-titanium is an ordered intermetallic compound. It was discovered by the
Naval Ordnance Laboratory in the 19603 to exhibit shape-memory effect, where a
deformed or strained body is able to revert to its pre-programmed geometry and
dimension upon exposure to a thermal stimulus. Given its discovery by the Naval

Ordnance Laboratory, NiTi is also commonly referred to as NiTi-NOL or nitinol.

1.2.1 Transformation Properties
At elevated temperature, NiTi exists in the form of an ordered cubic B2 (CsCl —
type) superlattice, which is commonly known as the ‘austenite’ phase. It has a

superlattice parameter of 0.301 5 nm [Plietsch et a1. 1997]. However, as temperature

decreases below the martensite start temperature, M S, the austenite phase begins to

transform into martensite phase, which has a 819’ monoclinic crystal structure. The

martensite transformation proceeds as temperature decreases. At Mf, which is the

martensite finish temperature, all 32 phase become fully transformed into the BI 9 ’ phase.
Such transformation is atherrnal, diffusionless and does not involve significant change in

overall volume in NiTi. Similarly, the martensite phase will begin to transform back into

the austenite phase when temperature increases beyond the austenite start temperature, A s.

The transformation will only be complete beyond the austenite finish temperature, A f. It

is important to highlight that phase transformation in NiT i does not always involve a
direct BZ—to-BI 9 ’ pathway. In some cases, R-phase transformation may occur preceding
the BI9 ' transformation in therrnomechanically-treated NiTi alloys [Miyazaki et a1. 1982,
1986]. As the properties of R-phase and the mechanism of transformation are beyond the
scope of this dissertation work, they will not be reviewed. It is the intrinsic nature of

shape-memory alloys to exhibit thermal hysteresis in martensite-austenite phase

transformations. Specifically, Ms (My) is lower than Mf(AS) and the width of the

hysteresis loop is typically about 30°C for NiTi. The hysteretic behavior of shape-

memory alloy is shown in Figure 1.1. Note that it is possible for MS to be lower than A s.

% Austenite
A Ms Af
100 <

 

 

 

 

 

 

 

Figure 1.1 Hysteresis loop involved during Martensite-Austenite transformation with
Ms, Mf, A S and Afindicated.

Given the lower symmetry of the BI9 ' structure, it may first be expected that the
austenite to martensite transformation will result in volume or shape changes. However,
based on the Wechsler—Lieberman—Read theory [Wechsler et al., 1953, Lieberman et a1.
1955], it was predicted that up to 24 martensite habit plane variants can form in one
austenite crystal [Otsuka et a1. 1998]. The internal twinning of martensite variants will
eliminate the macroscopic shear strain associated with the transformation. Hence, the
self-accommodating capability of these martensite variants will not result in shape

change during transformation from austenite to martensite.

On the other hand, when NiTi is deformed below Mf, biased martensite variant

distributions will be created in response to the macroscopic strain through twinning. The
material appears to suffer permanent deformation due to the reorientation of martensite
variants. Slip mechanisms associated with typical plastic deformation do not play a role

in this case, unless the external stress applied exceeds the critical slip stress of the

martensite phase. Shape recovery begins to occur when the material is heated above A S,

and is completed at A f The shape recovery capability stems from the inherent ordering

across the martensite-austenite interface resulting in the detwinning of martensite variants

through an exactly reversed pathway taken during twinning.

Besides the shape-memory effect, NiTi also possesses a transformational
superelasticity or pseudoelasticity characteristic, which is an isothermal form of shape-

memory. While decreasing temperature can cause the austenite-to-martensite

transformation to occur, externally applied stresses can also drive the austenite-to-

martensite transformation even when temperature of the material is above A f. This is

known as the stress-induced martensite (SIM) transformation and it is the underlying
principle for transformational superelasticity. The SIM transformation is a
thermodynamic phenomenon and is governed by the modified Clausius-Clayperon

equation expressed in Eq. (1.1).

thrans = T135

E. 1.1
do —AH q( )

where T trans is the stress-induced transformation temperature, T is the stress-flee BZ-to-

BI9 ’ transformation temperature, a is the extemally—applied stress, Ar: is the stress-

induced strain and AH is the martensite transformation enthalpy.

Figure 1.2 qualitatively illustrates how the Clausius-Clayperon relation [Wollants

et a1. 1979] governs the dependence of A S, A f, M, and Mfon extemally-applied stress. It

is important to highlight that the plots in Figure 1.2 have positive slopes because the

martensite transformation is an exothermic process. If NiTi is subjected to sufficient
external stress at T > A f, the austenite phase becomes thermodynamically unstable and
transforms into martensite phase when magnitude of stress crosses into the MS boundary

(Vertical upward line in Figure 1.2). The martensite variants will have crystallographic

Orientations dictated by the stress states and, hence, give rise to macroscopic strain in a

similar fashion described in the previous section. However, given that T > A f, the

material will spontaneously revert to austenite phase upon removal of the applied stress

(vertical downward line in Figure 1.5). If the applied stress magnitude remained below

the critical stress for slip, Md, full strain recovery will be possible. But if the applied

stress magnitude exceeds Md, conventional slip mechanisms will be initiated, resulting in

permanent deformation with only partial strain recovery.

 

 

 

 

 

Figure 1.2 Graphical representation of modified Clausius-Clayperon equation
showing change in MS as a function of applied stress.

1.2.2 Mechanical Properties

The shape-memory effect and transformational superelasticity of MT i are
graphically srmrmarized in Figure 1.3, which show the typical stress-strain curves of NiTi
when it is deformed in the martensite and austenite state. The stress-strain behavior of

martensitic shape-memory NiTi mainly appears to be analogous to conventional metallic

systems. But when the martensitic NiTi is heated beyond A f, prior deformation strain is

recovered. For superelastic or austenitic NiTi, stress plateaus appear in each of the stress-

 

 

strain curves for mechanical loading and unloading. The upper-plateau is attributed to
SIM transformation during loading, while the lower-plateau is associated with martensite
to austenite transformation during unloading. The upper-plateau stress of NiTi alloys
varies greatly, but most alloys with useful superelastic performance tend to possess

upper-plateau stresses around 400 MPa.

 

 

 

 

 

 

Figure 1.3 Shape-memory effect and transformational superelasticity of NiTi alloys.

Typically, full strain recovery can be achieved in NiTi if it is not strained more
than 6%. However, Miyazaki et a]. has shown that recoverable strain of up to 10% can

be achieved through thermomechanical treatment of NiTi [Miyazaki et a1. 1982, 1986].

Determination of elastic modulus of NiTi has been a challenging problem. This is
due to the highly non-linear slope of the unloading stress-strain curve. Furthermore, the
elastic modulus of NiTi is also dependent on temperature [Melton 1990]. Experimental
work has determined that NiTi can have elastic modulus ranging fi'om about 30 to 100
GPa [Melton 1990, Zadno et a1. 1990]. Ultimate tensile strengths of NiTi alloys vary
significantly and are highly-dependent on compositions and therrno-mechanical

treatments. However, most alloys tend to have strengths in excess of 1 GPa.

1.2.3 Phase Equilibria in the Ni-T i Binary System
The binary phase diagram of Ni-Ti is shown in Figure 1.4. The phase diagram

indicates that NiTi melts congruently at 1310°C. Besides NiTi, there are two other stable

intermediate compounds in the binary system, namely, the TizNi and Ni3Ti phases. The

former melts peritectically near 984°C and the latter melts congruently around 1380°C.
The Ni4Ti3 and the Ni3Ti2 phases are two metastable compounds that exist in the Ni-rich
side of the NiTi — phase field [Nishida et a1. 1986, Otsuka et a1. 2005]. When subjected
to elevated temperature anneals for prolong period of time, these metastable phases will
eventually transform into the stable Ni3Ti phase. As discussed in Section 1.2.1, NiTi is
an ordered intermetallic compound in the Ni-Ti binary system. However, it is not a line-
compound of fixed composition at elevated temperatures. The binary Ni-Ti phase
diagram in Figure 1.4 shows that excess Ni of about 6 at.% is soluble in NiTi above

1000°C. On the other hand, the steep NiTi phase boundary on the Ti-rich side does not

permit excess Ti to be soluble in NiTi at any temperature. At temperatures below 630°C,

 

the phase diagram indicates that the NiTi — phase field degenerates into an equi-atomic

line-compound.

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

2500
: Ni-TI Binary
o 2000‘
o 1
a; .
B ‘ 1670 Liquid
9 .1
8. 15004 1380 1455
E ‘ 1310 ,
,2 q 1304 .
, bcc- N1
1000] TI 942 984
«883/ e
\ 1— .
hop-TI . 765 g }
i F” stable
.g,
500 . , “2N3 , . . , .
Ti 20 4o 60 NI3TI 80 Ni
Atom % Ni

Figure 1.4 Ni-Ti binary phase diagram [Massalski 1990].

With such limited solubility of Ni and Ti in the NiTi phase at low temperatures,

any alloy composition that deviates from 50Ni-50Ti (at.%) will result in precipitations of

either TizNi or Ni3Ti in NiTi. It has been reported that TizNi precipitates exist within the

matrix or at the grain boundaries of NiTi when initial composition of Ti exceeds 50.4 at%

[Surbled et a1. 2001 , Lehnert et al. 2002]. On the other hand, Nishida and coworkers

[Nishida et a1. 1986] claimed that their 52 at.% Ni alloy decomposed into metastable

phases (first Ni4Ti3 then Ni3Ti2), then finally into the thermodynamically-stable Ni3Ti

upon thermal aging. The reaction path taken depends on the aging time and temperature,

which is illustrated in the I l l diagram of Figure 1.5.

Aging Temperature (°C)

 

 

 

     

 

800 NiTi ( NiTi +Ni3Ti
700 \

,—--
600 ' NiTi + - _ -NT

+
Ni4Ti3 .1 1.
500 N13T12
l L l l l
0.1 1 10 100 1000
Aging Time (Hr)

Figure 1.5

‘I'I'I' diagram of 52 at.% Ni-Ti alloy [Nishida et a1. 1986].

1.2.4 Effects of Composition on Transformation Temperature

Transformation temperatures associated with NiTi are important parameters to

control as they dictate the temperature range for shape-memory alloys to carry out their

proper fimctions. The martensite transformation temperature is strongly dependent on

composition. Varying the composition of NiTi is one of the most effective methods of

changing transformation temperatures. In Ni-rich alloys, a composition change of less

than 1 at.% can change the transformation temperature by more than 100°C.

Figure 1.6

(°C)

 

 

49 50 51 52

Ni Content (at. %)

48

Effect of Ni composition in NiTi alloys on martensite start temperature.

Figure 1.6 shows that for Ti-rich to equi-atomic alloys, MS remains rather constant
at 60°C. However, MS decreases rapidly from 60°C to -140°C when Ni composition

increases from 50 to 51 at.%. [Melton 1990, Otsuka et a1. 2005]. The insensitivity of MS

to composition changes when alloys are Ti-rich is believed to be associated with the
almost vertical NiTi phase boundary on the Ti-rich side of the phase diagram (Figure 1.4)
causing insignificant solubility of Ti in NiTi phase. Given that it is not possible to obtain
Ti-rich NiTi solid solutions, Ti-rich alloys exhibit behaviors similar as equi-atomic NiTi

[Otsuka et a1. 2005].

The strong dependence of MS on composition for Ni-rich alloys can be understood

fi'om a qualitative explanation. Recall that the microstructures of Ni—rich alloys consist of

the metastable Ni4Ti3 phase in the NiTi matrix. Besides thermal aging, fine precipitates

of the Ni4T13 phase can also form upon quenching from solution. These precipitates are
coherent with the NiTi lattice and yield strain fields that will strengthen the 32 matrix.
Hence, a larger driving force or a lower M 3 temperature is required to initiate martensite

transformation [Miyazaki et a1. 1982, Saburi et al. 1982].

1.2.5 Eflects of Thermo-mechanical Treatment on Shape-Memory and Superelastic
Characteristics

Recall in Section 1.2 that permanent deformation of NiTi will occur if the applied

stress exceeds the critical stress for slip, Md. Hence, significant efforts have been placed

10

to explore various approaches to improve Md. Strengthening methods common in

physical metallurgy, such as work hardening, precipitation hardening, and grain size

refinement, have been studied for their applicability in NiTi.

It was found that cold work alone will improve yield stress but destroy the
superelasticity of NiTi, resulting in low recoverable strains [Melton 1990]. It was
believed that cold work introduces high densities of random dislocations, which hinder
twin boundary mobility. Hence, the strengthening effect is attributed to the presence of
these dislocations, but the constraint of twin boundary movements by dislocations also
signifies a compromise of strain recovery performance. On the other hand, post cold-
work annealing will improve strain recovery but reduces yield stress. The annealing
process rearranges these dislocations to form cells or sub-grain boundaries, giving rise to
relatively dislocation-free regions. Martensite twin boundaries will have relatively higher
mobility in these regions. This translates into poorer strengthening effect but better strain
recovery properties. Therefore, one can appreciate that the choice of degree of cold work
and the annealing temperature will dictate the trade—off in the yield stress and recoverable

strain.

Figure 1.7 shows the effect of annealing temperature on room-temperature yield

strength and MS after 40% cold work in the martensite phase of a 50.6 at.% Ni alloy

[Melton 1990]. Annealing times are within the order of an hour. A rapid decrease in the

yield strength is observed in the range from 350°C to 450°C, followed by a relatively

gradual decrease from 450°C up to 850°C. On the other hand, M 5 increases promptly

ll

when annealing temperature increases from 350°C to 450°C, but becomes weakly

dependent on temperature above 450°C.

1200 -r~fi~r~r-r~r90

 

~80 93,
3.

, (D
“‘70 m

o-O-

' (D
~60 513
0)

~50 3
(D

..s
O
O
O

   

 

 
   

—D— Yield Strength

800 1 + Martensite Start

 

 

 

600 .

‘ 3
~40p

Yield Strength (MPa)

400 -

 

 

 

. , . . . . . , . . . .30 Q
300 400 500 600 700 800 900
Annealing Temperature (°C)

Figure 1.7 Effect of annealing temperature on room-temperature yield strength and
martensite start temperature after 40% cold work in the martensite phase of a 50.6 at.%
Ni alloy [Melton 1990].

Age hardening by precipitation of the Ni4Ti3 phase is also known to significantly

affect shape-memory effect and superelasticity [Miyazaki et a1. 1982, Saburi et a1. 1982].
It was observed that a 1000°C solution-treated 50.6 at.% Ni alloy wire that was
subsequently aged at 500°C exhibits poor strain recovery when mechanically-tested
between -64°C and 51°C. Complete superelasticity can only be achieved within a limited
temperature range from 31°C to 41°C, while permanent strains were introduced in the
rest of the test temperatures. The solution treatment and aging durations were both fixed
at 1 hour. When a similar solution-treated wire was aged at 400°C for 1 hour,
superelasticity was observed over a significantly wider range from 4°C to 61°C.

Furthermore, onset stresses of the upper-plateau in superelastic stress-strain curves tend

12

to be higher for the specimen aged at 400°C, which indicates strengthening of the 32

phase [Miyazaki et a1. 1982].

The strengthening of the BZ phase can be understood from the physical
metallurgy principles that the increase in critical stress for slip is most pronounced when

high-density of fine, coherent precipitates are dispersed in the matrix. A lower annealing
temperature may initiate extensive nucleation of Ni4Ti3 particles but does not encourage
significant growth of these precipitates. On the other hand, higher annealing

temperatures will enhance growth kinetics, resulting in coarser Ni4Ti3 precipitates. This

deduction was supported by TEM experiments that revealed fine and densely-populated

precipitates of Ni4Ti3 in the NiTi matrix when the specimen was annealed at 400°C,

while the specimen that was annealed at 500°C contains fewer but larger Ni4Ti3

precipitates [Otsuka et a1. 2005].

Grain size refinement has also been reported to be a viable approach to improve
superelasticity of NiTi. Saburi et al. demonstrated the effect of grain size on superelastic
performance of NiTi by employing single crystal and polycrystalline specimens of 50.5

at.% Ni alloy and subjected them to annealing at 800°C for 1 hour before tensile testing

[Saburi et a1. 1984]. Tests were performed at 40°C, which is 10°C higher than the A fof

these specimens. The authors found that single crystal specimens showed incomplete
superelastic behavior, while polycrystalline specimens of 1 mm or smaller grain size

revealed improvement in superelasticity. At a grain size of 50 um, complete

13

superelasticity was observed. Therefore, the study clearly suggests that the reduction in

grain size is also effective for improving superelastic performance of NiTi.

1.2.6 Effects of Ternary Alloying

There has been much effort to modify NiTi alloys by adding different elements to
the binary system. It has been reported that transformation temperatures, hysteresis width,
mechanical strength and corrosion properties of NiTi can be controlled by the addition of
a third element [Melton 1990]. Alloying elements commonly-employed for study are An,

Cu, Co, Cr, Fe, Hf, Nb, Pd, Pt and Zr [Otsuka et al. 2005].

Some elements exhibit preference to substitute for a particular site in the NiTi
lattice while others do not. For example, Nakata et al. found that Co, Fe and Pd have a
strong preference for entering Ni sites, while there is no preference for Cr, Mn and V to
occupy either Ni or Ti sites [Nakata et al. 1991]. From their findings, the authors
deduced that electron configuration, chemical affinity and atomic size effect play

important roles in governing the preferences for site occupancy of each alloying element.

The martensite start temperature, M s, was observed to decrease monotonically
with increasing V, Cr or Mn substitution for Ti sites [Honma 1987]. The suppression of

Ms can also be effected by substituting Ni with Co or Fe [Eckelmeyer 1976, Honma

1987]. On the other hand, significant increase in transformation temperatures can be
achieved by substitution of Ni with Pd or Pt [Lindquist et al. 1990, Winzek et al. 2001] or

Ti with Hf or Zr [Johnson et al. 1995, Hsieh et al. 1998, Winzek et al. 2001, Firstov et al.

14

2004]. Lindquist et al. reported that MS can be increased to about 500°C when more than

45 at.% Pd was alloyed, while 30 at.% Pt addition will increase MS beyond 600°C

[Lindquist et al. 1990]. Alloys of NiTin and NiTin were also observed to possess
transformation temperatures between 300°C to 500°C [Hsieh et al. 1998, Firstov et a1.
2004]. Hence, NiTiPd, NiTiPt, NiTin and NiTin alloys are considered as candidates
for high-temperature shape-memory alloys. Grummon et al. have studied the effect of
substituting Ni with Cu and observed a pronounced reduction in transformation hysteresis
[Grummon et al. 1999]. Shape-memory alloys with narrow transformation hysteresis will
be more sensitive and responsive to temperature changes, making them suitable for

sensors and actuators applications.

Special attention will now be placed on the ternary addition of Nb in NiTi.
Niobium is given particular emphasis in this review as it is the filler metal for the reactive
eutectic braze process in this work. Ternary alloys of NiTiNb, particularly compositions
close to 44Ti-47Ni-9Nb (at.%), are widely-studied because of their excellent constrained
recovery properties that can be harnessed for use as pipe couplings and mechanical
fastening devices [Duerig et al. 1990]. Details on constrained recovery can be found in

[Prof’t 1990].

Another reason why NiTiNb alloys are excellent candidates for pipe coupling and
mechanical fastening devices is because they exhibit wide transformation hysteresis

behavior when subjected to cold-working [Duerig et al. 1990, Zhang et al. 1990, Zhao et

al. 1990, Piao et al. 1992A, Siegert et al. 2003]. Specifically, NiTiNb alloys can have MS

15

as low as -140°C and A 3 about 50°C [Duerig et al. 1990]. By adjusting the

transformation temperatures such that ambient temperature is located near the center of
the wide hysteresis, these ternary alloys can (1) lower storage costs by eliminating the
need to store the coupling components under impractically low temperatures after they
have been expanded for future use; and (2) prolonged service life due to the ability to
maintain the desired coupling strength at ambient (or moderately elevated) temperatures

beyond a reasonably long period of time.

The transformation characteristics can be understood from the microstructures of
these ternary alloys, which will consist of insoluble Nb precipitates in a matrix of
martensitic NiTi phase when the alloys warm up to room temperature from storage
temperature (i.e. less than -140°C) [Duerig et al. 1990, Zhao et al. 1990, Piao et al. 1993].
Given that the flow stress of the Nb precipitates is similar to that of the martensitic NiTi
(approximately 150 — 200 MPa), Duerig et a1. and Zhao et al. proposed that the NiTi and
Nb phases undergo deformation simultaneously at relatively low stresses from cold-
working. However, the martensite deforms reversibly by twin boundary motion, while

the Nb phase deforms permanently via conventional slip. When the deformed alloy is

heated above A s of the NiTi, the martensite begins to recover but can only proceed

significantly by first plastically re-deforrning the Nb precipitates back to their original

morphologies. Hence, the irreversible component of deformation delays recovery,

causing an increase in apparent A 3. Such increase in apparent A S is exclusively observed

16

in the first heat cycle of the NiTiNb alloy [Duerig et a1. 1990, Piao et al. 1992A].

Subsequent heat cycles cause the apparent A S to revert back to the original value.

1.2. 7 Phase Equilibria of the Ni-Ti-Nl) Ternary System
To date, the most widely-studied isothermal section of the Ni-Ti-Nb ternary

system is at 900°C. Gupta [Gupta 1991] reported the presence of 4 intermediate ternary
intermetallic phases (X1, X2, X3 and X4) in the Ni-rich region, with numerous two- and

three-phase fields in the 900°C isothermal section; while the Ti-rich side has no known
ternary intermediate phases. Piao et al. [Piao et al. 1992B] observed a eutectic
microstructure in their 40Ni-40Ti-20Nb (at.%) specimens prepared by Vacuum Arc Re-
melting (VAR). Based on their experimental findings, the authors proposed a NiTi-Nb
quasibinary eutectic isopleth in the ternary system. They claimed that the eutectic point

is 4ONi-40Ti-20Nb (at.%) at 1150°C. However, an extensive study of the solidus
temperatures in the NiTi —- Ni6Nb7 — Nb — Ti region by Prima et al. [Prima et al. 1995]
revealed a quasibinary eutectic point close to 38Ni-3 6Ti-26Nb (at.%) at 1170°C
(indicated on Figure 1.8). The findings of Gupta and Prima et al. on the 900°C
isothermal section and liquidus projections are summarized in Figure 1.8. The NiTi +
Nb(Ni,Ti) tie-line at 900°C, close to the quasibinary eutectic point [Prima et al. 1995], is

also constructed in Figure 1.8.

17

I 1 ~ Phase Fields ”I

   
   
 

[:l 2 - Phase Fields

3 ~ Phase Fields

BCC ~ Nb (Ti, Ni)

Figure 1.8 900°C — Isothermal section and liquidus projections of the Ni-Ti-Nb
ternary system reported by [Gupta 1991] and [Prima et al. 1995]. (NiTi + Nb(Ni,Ti) tic-
line at 900°C, close to the quasibinary eutectic point, is also indicated.)

More recently, a fifth ternary intermetallic phase (X5) was discovered by Yang

and coworkers in the Ni-rich region [Yang et a1. 2000]. The liquidus projections of the
ternary system were predicted by Matsumoto and coworkers using CALPHAD
calculations [Matsumoto et al. 2005], taking into considerations all the reported phases,
including the X5 phase. However, the quasi-binary eutectic point was not accounted for.
Figure 1.9 graphically summarizes the findings on the liquidus projections and 900°C

isothermal section of the Ni-Ti-Nb system by Yang and Matsumoto groups.

I 1 - Phase Fields N'

 
   
  
  
 
 
  
   
 

I] 2 - Phase Fields
3 - Phase Fields X2
X1: Nb13Ni75'l'I12 Ni3Ti
x2: Nb5N175T120
X3: Nb15N556Ti29

X4: Nb15Ni30Ti5 NiTi
X5: waNi45Ti-l5

BCC - Nb (Ti, Ni)

Figure 1.9 Summarized findings on the 900°C — isothermal section [Yang et a1. 2000]
and liquidus projection [Matsumoto et al. 2005] of the Ni-Ti-Nb ternary system.

Phase equilibria data for other temperatures are very limited. Specifically, little is

known of the temperature-evolution of phase fields, such as those of NiTi + Nb(Ni,Ti) +
(T i,Nb)2Ni and the NiTi + Nb(Ni,Ti) + X5. These may be presumed to degenerate to
points at 0 K. The knowledge of the phase field evolution is critical for determining if

(T i,Nb)2Ni or X5 phases will solidify in the eutectic matrix at lower temperatures.

1.3 Joining of NiTi Components

The ability to regulate the transformation and mechanical properties by adjusting
its binary and ternary compositions makes NiTi alloys suitable for use over a broad range
of operating temperatures. However, a major shortcoming of NiTi alloys is the difficulty
of joining NiTi alloys to themselves and to other materials. The problem stems from
either wettability issues or the tendency to form brittle Ti-rich intermetallic compounds in
joints when conventional joining techniques are employed. Furthermore, shape memory
performance of the parent material is severely affected by overheating from the various
joining techniques. As a consequence, applications of NiTi alloys are currently restricted
to situations where only simple geometries are required. There have been limited
disclosed studies on joining of NiTi alloys. Within the scarce amount of published
research, there are 5 techniques that have been explored for joining NiTi components.
They are: (1) welding, (2) soldering, (3) mechanical joining, and (4) brazing, and (5)

transient liquid phase (TLP) bonding.

1.3.1 Welding
Various types of welding, such as tungsten inert gas (TIG) welding [Ika 1996] and
plasma welding [Van der Eijk et al. 2003] under Ar or He inert environment, have been

employed for joining similar NiTi components, mostly yielding poor results. Common

problems encountered were formations of TizNi and Ni3Ti, which are brittle and have no

shape-memory effect. However, laser welding appears to be a promising technique as

there have been a few successful demonstrations showing reasonably strong NiTi joints

20

with their base alloys partially retaining their functional properties [Schlossmacher et al.

1995, Hsu et al. 2001, Schetky and Wu 2004].

Schlossrnacher and coworkers [Schlossmacher et al. 1995] employed a Nd:YAG
laser for joining a superelastic Ti-51.5Ni (at.%) alloy. Tensile tests on sheet metal of 0.5
mm thickness were performed at under martensitic and austenitic conditions. Their laser
welded specimens exhibited an ultimate tensile strength of about 820 MPa and are
independent of test temperature. This value corresponds to 80% of the ultimate tensile
strength of their base material, which was also tested in both martensitic and austenitic
conditions. Fractographic observations of ruptured specimens by scanning electron
microscopy (SEM) revealed ductile characteristics of laser-welded joints. Schetky and
Wu [Schetky and Wu 2004] also reported that Nd:YAG lasers are capable of producing
welds with excellent strength retention, which they claimed to be 75% of the base metal
strength. Their study showed only 0.2% permanent deformation after a 7% strain of a

superelastic weld specimen.

In another case, Hsu and coworkers [Hsu et al. 2001] successfully demonstrated

the ability to produce strong NiTi joints using laser welding. The authors welded an
equi-atomic NiTi alloy using a C02 laser operated within 2200 to 2500 W at a travel
speed of 1.5 m/min. Helium plasma was used during welding. Vickers micro-hardness
tests showed that weldments possess similar hardness to the parent material. Tensile

strengths as high as 740 MPa were observed when their laser welded joints were

subjected to tensile tests, but strain at failure was only about 6%, which is poor.

21

Fractographic observations of the weldment showed brittle mode fracture. The authors
attributed the poor fracture strain to segregation of impurity elements at the weldments

during solidification, and increase in grain size.

Hirose and Akari [Akari et al. 1989, Hirose and Akari 1989] joined 3 mm-thick

50.7Ni-Ti (at.%) sheets using a 10 kW C02 laser, with a travel speed of 3.4 m/min. All

welded specimens were subjected to post-weld heat-treatment at 400°C for 1 hour.

Differential scanning calorimetry showed that the A f, A 5, MS and Mfof the base metal
were 23°C, -22°C, -24°C and -79°C, respectively. There were no changes in A fand MS

in the weldment, but A S and Mfwere each about 10°C lower in the weldment than the

base metal. When subjected to tensile loading and unloading cycles at strains up to 4%,
the welded specimen yielded room temperature superelastic recovery with about 0.3%
unrecoverable strain. The authors also investigated the tensile strength of the welded
joints at -40°C, 40°C and 120°C, which they found to be about 480 MPa, 670 MPa and

800 MPa, respectively.

Besides laser welding, fiiction stir welding is another promising technique that
has attracted a lot of attention in the field of joining. Recently, London and coworkers
[London et al. 2005] explored friction stir welding for joining NiTi alloys. Void-flee
friction stir processed zones were produced in 6.35 mm—thick NiTi plates using
polycrystalline cubic boron nitride and tungsten-rhenium tools that were rotating from

400 to 600 rpm at a travel speed of 12.7 mm/min. Transformation temperatures of the

22

processed zones were measured using differential scanning calorimetry, which showed a

lower A fin the processed zones (—22.5°C) compared to the parent alloy (15.0°C). Optical

and SEM revealed a slight reduction in grain diameter (33 pm) from the parent alloy (43
um). Welded sheets were hot-rolled at 850°C to 76% total reduction and then subjected
to tensile tests. Results showed that these fiiction stir welded NiTi alloys retained their

superelastic responses with tensile strength approaching 700 MPa at a strain of about

16%.

Although some positive results were reported for welding NiTi alloys to itself,
successful welding of dissimilar materials proves to be an even greater challenge. The
main reason is the undesirable formation of hard and brittle intermetallic compounds,
oxides or carbides at the weld zone. Van der Eijk and coworkers [Van der Eijk et al.
2003] had attempted to join NiTi alloys to stainless steel and Hastelloy C276 using
plasma arc welding. The joining procedures were performed between 15 and 42 A of
current under Ar plasma, and the shielding gas was also Ar. Working distance between
the torch and the sample was approximately 6 mm. The authors observed significantly
more melting of the NiTi as compared to the stainless steel, resulting in leaching of
elements (predominantly Fe, Cr, Ni, Mn and Si) from the steel into the molten NiTi to
form a mixed zone and TiC precipitates. Scanning electron microscopy revealed cracks
within the mixed zone and the carbide precipitates, indicating that they are brittle.
Welding of NiTi alloys to Hastelloy C276 also resulted in significant leaching of

elements into the weldment, producing a mixed zone. Tensile tests could not be

23

performed on the NiTi-steel joint due to serious embrittlement, while tensile tests on the

NiTi-Hastelloy joint revealed a tensile strength of only 70 MPa with negligible strain.

Chan and coworkers [Chan et al. 2006] investigated the possibility of using
Nd:YAG laser for spot-welding 0.5 mm-thick NiTi sheets to 0.1 mm-thick commercially-
pure Ti sheets. They found that welding at a beam voltage of 220V for 35 ms yielded the
highest failure load of 193N, which translates to shear strength of 109 MPa. Such

strength level does not meet the requirement of most applications.

1.3.2 Soldering

The predominant problem associated with soldering NiTi alloys is a lack of
wettability, which is attributed to the presence of Ti-rich native oxide on NiTi. Given the
chemically-inert nature of Ti-rich oxides, aggressive fluxes are usually required to
remove them prior to soldering. In a US. patent by Hall [Hall 1993], it was claimed that
an aluminum paste flux applied to NiTi surfaces will remove the Ti-rich native oxide
layer when the flux is heated to its activation temperature. Consequently, Ti leaches from
the surface of the base metal, forming a Ni-rich surface. The activated flux also coats the
Ni-rich surface to prevent further oxidation. A bonding material, such as Sn — Ag solder
is flowed onto the Ni-rich surface, which will then displace the flux. At this point,
another NiTi member will be applied to the molten solder to form a joint. Ultrasonic
soldering, developed for removing oxides from the almninum to promote solder
wettability, has also been employed successfully on NiTi alloys with a Sn—based solder

[Schetky and Wu 2004].

24

Successful soldering of NiTi alloys was also claimed by Murai [Murai 1990].
The author reported in his US. patent that plating Ni onto NiTi creates a surface that will
accept solder with mild flux can improve solderability. Other variants of this approach

are plating Cu/Ni or Cu/Ni/Au onto NiTi surfaces.

1.3.3 Mechanical Joining

Various types of mechanical joints have been explored in industry. For example,
threading and tapping can effectively couple NiTi components, but the machining process
of NiTi is cost-ineffective. Joining by interference fit or interlocking components can be
achieved by harnessing the SME and TSE properties of NiTi [Kapgan et al. 1990, Wu
2001]. However, precision machining of the joining part to ensure optimal fit poses a

huge challenge in terms of reducing production time and cost.

1.3.4 Brazing

Problems associated with brazing NiTi alloys are analogous to those involved in
welding. That is, the strong tendency to form intermetallic compounds, oxides or
carbides result in serious embrittlernent of braze joints. Brazing NiTi alloys to stainless
steel and Ti-metal/alloys using industrial-standard braze fillers have been studied by a
few groups. Seki and coworkers [Seki et al. 2000] employed a series of BAg-8
(American Welding Society standard brazing alloy with compositions of 72Ag-28Cu
(at.%)) filler alloy with increasing Ni content for brazing NiTi alloys to 304-grade
stainless steel. Brazing was performed above the flow point of the filler alloy of 790°C.

The authors found that inter-diffusion of Fe, Ni and Ti during brazing resulted in the

25

formation of (F e,Ni)2Ti at the bond line when the BAg—8 filler alloy did not contain Ni.

Tensile tests on these NiTi-steel joints yielded fracture strength of only 275 MPa.

Fracture occurred in the (Fe,Ni)2Ti phase. With the addition of Ni, Ni3Ti phase formed

in the joint instead of (Fe,Ni)2Ti, and tensile strength of 400 MPa can be achieved.

Fracture was observed to occur in the Ni3Ti phase.

In a separate study, Kunimasa’s group [Kunimasa et a1. 2001] investigated
brazing of NiTi to commercially-pure Ti using BAg-8 filler alloy at temperatures above
880°C. They reported that the predominant phase in the braze region is a Ti-Cu
intermetallic compound, which solidified isotherrnally between the commercially-pure Ti
and the BAg—8 filler during brazing. Mechanical tests revealed joint strength up to 350

MPa Fracture was observed in the Ti-Cu intermetallic compound.

Other variants of brazing to join NiTi alloys to itself and to dissimilar materials
have also been explored. Infrared brazing of NiTi was investigated by Yang’s group
[Yang et al. 2004]. Nitinol sheets of 1 mm thickness were brazed using using 25 pm-
thick pure Cu and 50 urn-thick Ti-15Cu-15Ni (at.%) braze filler foils. Brazing was
performed up to 1150°C. The authors observed CuNiTi and Ti(Ni,Cu) phases in the

joints fabricated using pure Cu filler. Bend tests on these braze joints showed shape

recovery. Extensive formation of Ti2(Ni,Cu) phase was observed in braze joints prepared

using the Ti-lSCu-15Ni filler alloy. Bend tests could not be performed on these joints

due to serious embrittlement fiom the Ti-rich phase.

26

An attempt to join NiTi alloys to Ti-6Al-4V (at.%) alloys using infrared brazing
was reported by Shiue and Wu [Shiue and Wu 2005]. The authors employed BAg—8 as
the filler alloy. Braze joints comprised mainly of hypo-eutectic Ag-Cu when brazing was
performed below 850°C. It was found that Ag did not react with both base-alloys but Cu
readily alloyed with Ti, forming various Ti-Cu intermetallic compounds. Mechanical

tests on the 850°C-brazed joints revealed shear strength of about 200MPa. When brazed

at 900°C, joint microstructure consisted mainly of TizNi phase. Due to serious

embrittlement, shear tests could not be performed on the 900°C-brazed joints.

Chiu and coworkers [Chiu et al. 2005] adopted microwave-assisted brazing for
cladding NiTi to 316L-grade stainless steel. The motivation of their work was to
improve cavitation erosion resistance of stainless steel used in liquid handling systems
with a NiTi surface layer. Brazing was performed in a commercial microwave oven
operating at 2.45 GHz and a power of 1700 W for 150 seconds. The braze filler alloy
used had a composition of 50Cu-33Zn-9Ag-8Cd (at.%). The authors found that the braze
microstructure contained predominantly Fe and Ti elements, but did not specify the
phases present. Indentation tests indicated that the brazed region had a Vickers hardness
number of about 450, which was significantly higher than both NiTi and stainless steel.
Nanoindentation performed on the post-braze NiTi surface showed superelastic recovery
that is comparable to that of the as-received NiTi alloy. Shear strength of the braze joint

was determined to be in the range of 100 to 150 MPa.

27

Li and coworkers [Li et al. 2006] employed a Nd:YAG laser for brazing NiTi
wires to stainless steel wires. The braze filler alloy used had a composition of 52Ag-
220u-18Zn—8Sn (wt.%). Laser power and brazing time were varied from 50 to 70 W and
10 to 20 seconds, respectively. The authors observed good wettability of the filler alloy
on both NiTi and stainless steel. The braze microstructure consisted of a-Ag, a’-Cu and
various types of intermetallic compounds. Braze joints were subjected to tensile loading
and unloading cycles up to 4% strain. The 50 W/ 10 seconds braze joint incurred an
unrecoverable strain of 0.37% after the loading/unloading cycle. Unrecoverable strain
increased to 2.2% when laser power and braze duration increased to 70W and 20 seconds,
respectively. Indentation tests indicated that the braze region possessed a Vickers
hardness number of about 220, which is only 60% of the base NiTi alloy. However, the

authors did not probe the tensile or shear strength of the braze joints.

1.3.5 Transient Liquid Phase Bonding

The transient liquid phase (TLP) bonding method had also been investigated for
the possibility of joining NiTi alloys to other materials. Gale’s group [Gale et al. 1997]
applied TLP bonding to join NiAl and NiTi. Pure Cu interlayers were used as the
melting point depressant. Melting at the bond interfaces were initiated at a temperature
of 1150°C. The authors reported that complete isothermal solidification or removal of

liquid phase was achievable within an hour at the process temperature, which resulted in

the formation of Ni2AlTi intermetallic compound at the bond line. Extending the

processing duration to two hours caused significant leaching of Ti from the base NiTi

28

alloy to the joint region. No mechanical characterization was performed on these TLP-

bonded joints.

The review on joining of NiTi alloys suggests that there are only limited reports
on successful joining of NiTi alloy to itself, most of which involves laser welding.
Despite claims that the functional properties of NiTi at the heat-affected zones can be
reasonably retained, the high cost and low yield associated with laser welding tend to
prevent the technique fi'om being widely adopted in industries. Furthermore, welding of
NiTi alloys to dissimilar materials, such as stainless steels, tend to produce brittle
intermetallic compounds in the joints, resulting in mechanical strengths that are

insignificant for most applications.

Although being a relatively low temperature process that will not adversely affect
the functional pr0perties of the base NiTi alloy, soldering NiTi alloys requires the use of
aggressive fluxes to promote wetting and flow of typical solder materials on the mating

surfaces. Most of these fluxes contain elements that are not biocompatible.

Brazing appears to be more suited for joining NiTi alloys to dissimilar materials,
and it is an economical, high yield method that can be readily adopted by industries. But
successful brazing tends to rely on the use of filler alloys with elements that are also not
biocompatible. Most of the braze processes also produce undesirable intermetallic

compounds in the braze region, causing embrittlement.

29

1.3. 6 Preliminary Work on Reactive Eutectic Brazing of Ni T i

Recognizing the ternary alloying ability of Nb in NiTi and the presence of a
quasi-binary eutectic point in the ternary system, Grummon and Shaw [Grummon et al.
2006, Shaw et a1. 2007] proposed using pure Nb filler metal to react and form a eutectic
liquid for joining NiTi alloys. The authors prepared honeycomb structures from
corrugated superelastic NiTi strips brazed together by Nb foils at 1 17 5°C for less than 10
minutes under high-vacuum conditions. The honeycomb was thermally-aged at 450°C

prior to mechanical tests.

Preliminary in-plane compression test results, summarized in Figure 1.10, show
that the honeycomb joint integrity is not compromised despite a progressive compressive
strain of up to 60%. Interestingly, the structure also exhibited very good transformational
superelastic properties. The recoverable strain after each strain cycle is close to 95%.
This strongly suggests that the structure possesses mechanically strong joints while the
NiTi still retains its superelasticity. However, the NiTi and joints became weak and

brittle when the braze duration was increased much beyond 10 minutes.

30

 
 
  

Compressive
Stress (MPa)

0.4 5

 

 

Figure 1.10 In-plane compression test results showing (a) Strain — Time and (b) Stress
— Strain plots with (c) digital images at selected test stages [Grummon et al. 2006, Shaw
et a1. 2007].

Scanning electron microscopy (SEM) of a ‘strong’ lap joint area revealed
coexistence of eutectic micro-constituents with a dendritic phase nucleating against the
base-metal (Figure 1.11). Menisci were observed at the edges of the lap joint, suggesting
that the braze liquid, once formed, readily wets and flows on the NiTi alloy surfaces.
However, there is no compositional and crystallographic information on the brazed joints.
Furthermore, the reactive eutectic brazing dynamics and mechanical properties (e.g. yield

strength and failure strength, etc.) of the joints remains unclear.

31

   

Figure 1.11 Collae of SEM micrographs showing the braze area of a room-
ternperature shape-memory lap joint (inset: Higher-magnification image of the braze
microstructure).
1.4 Brazing

Brazing is a joining technique that is similar to soldering but done at much higher
temperatures (typically above 450°C). It involves the use of ferrous or non-ferrous filler
metal/alloy that melts and forms a liquid within the joint. In the case of reactive brazing,
the filler metal will react with a thin layer of the base-metal of concern and form a molten
filler alloy mixture. This liquid, being ‘sandwiched’ between the base-alloys, will be
evenly distributed to the close-fitting regions of the joints by capillary action. Upon
cooling, the brazed joint becomes metallurgically linked to the base material, hence,

yielding a strong joint.

32

Given that brazing involves formation of liquid in the joints, good wettability and
flow of braze liquid become the two critical requirements for brazing processes to be
successful. For wettability, consider the scenario of a sessile drop of liquid on a solid
surface, as shown in Figure 1.12. At equilibrium, the sessile drop wets the solid and

makes an angle, a, with the surface.

Liquid YLV
W

Solid lSL

 

 

 

 

Figure 1.12 Sessile drop of liquid on a solid surface.

The forces acting at the solid-liquid-vapor triple-point at equilibrium are also

indicated in Figure 1.12 and can be mathematically described by,

75L = 75V - 7LV 0080 BL (12)

where YSL is the solid-liquid interface energy, i’SV is the solid-vapor surface energy, and
YLV is the liquid-vapor surface energy. Figure 1.12 and Equation (1.2) show that

maximum wetting of o = 0° occurs when J’SL = YSV - H, V- Wettability begins to decrease

as 0 increases. At a = 180°, a system will experience complete non-wetting condition.

Flow is another important aspect of brazing. If the braze liquid wets but does not

flow into joints, then the use of the filler metal or alloy will be greatly restricted. Figure

33

1.13 illustrates a schematic diagram of a simple flow scenario. Flow is driven by
capillary attraction, which in turn, results from surface energy effects. If a braze liquid
with viscosity, 7], is confined in a narrow joint with a gap width, w, such as between two
flat plates, the surface of the liquid must curve in order for its contact angle, a, to form on
both plates (recall contact angle in wettability). This curvature results in a pressure

difference across the surface. The liquid flows into the gap until pressure caused by the
surface energy, )1 V, is equal to the hydrostatic pressure. By applying the principles of

fluid flow under equilibrium conditions, it can be shown that the distance, x09, covered in

time, t, by the braze liquid is given by Eq. (1.3) [Schwartz 1969].

 

Figure 1.13 Schematic diagram showing a simple flow scenario.

x0) = ,lwtyLV 0050'

E . 1.3
377 q ( )
From Eq. (1.3), it becomes apparent that high liquid surface tension, low contact
angle and low viscosity are desirable in brazing. Surface tensions of liquid metals are
usually high [Iida et al. 1988], so it is usually not the deciding factor for choice of braze

filler alloys. On the other hand, viscosity of liquid metals or alloys can vary significantly.

34

Alloys with narrow fi'eezing ranges (i.e. temperature range between the solidus and
liquidus) that are close to the eutectic composition generally have lower viscosity than
those with wide freezing range [Schwartz 1969]. Hence, eutectic filler alloys are

commonly used for brazing operations.

1.5 Transient Liquid Phase Bonding

Transient liquid phase (TLP) bonding is a form of brazing that involves solid-state
diffusion. This joining method is known to produce strong and interface-free joints with
no remnant of the bonding phase. Theoretical aspects of the TLP bonding process were
extensively studied by Tuah-Poku et al. and MacDonald et al. [Tuah-Poku et al. 1988,
MacDonald et al., 1992, 1998]. The authors categorized TLP bonding into 4 main stages:
(1) melting, (2) widening, (3) isothermal solidification and (4) homogenization. Only
qualitative description of their model will be reviewed. For detailed quantitative

discussions, refer to the original literature in [MacDonald et al. 1992, 1998].

Most TLP bonding requires the use of interlayers, which can be provided by foils,
electroplating or sputter coating. The interlayer is sandwiched between the parent
materials to be joined. Upon heating to the bonding temperature, the interlayer and the
parent material undergo inter-diffusion and form a liquid phase. In other words, the
interlayer acts as a melting-point depressant to initiate melting at the interface. Melting
continues, causing the liquid or joint gap to widen. The melting process typically takes
minutes to complete, depending on thickness of the interlayer. Upon complete melting of

the interlayer, isothermal solidification dominates, resulting in the narrowing of the joint

35

through solid-state diffusion of solute into the base-material. The isothermal
solidification stage is the most critical stage in TLP bonding as it requires the greatest
amount of time (typically hundreds of hours for interlayers that are about 100 pm thick)
and is dependent on the width of the liquid and the rate of solute diffusion into the parent
material. Finally, homogenization is achieved in the joints when isothermal solidification
causes narrowing of the joint to the point where no liquid is left at the processing
temperature. Microstructures of the joint will resemble closely to that of the parent

material in this final stage, except for the effects of recrystallization and grain growth.

Tuah-Polcu and coworkers [Tuah-Poku et al. 1988] developed a model to predict
the solidification time required for homogenization and applied it to the Cu-Sn system.
Their model was based on a stationary solid-liquid interface. The solidification time, t, is

dependent on the diffusivity of the solute in the parent material, D, the initial

concentration of solute in the interlayer, C3: the solidus composition of the parent

material, cab and the initial interlayer thickness, W0. The mathematical relation among

them is expressed as,

2
W C
,=£(__0 19] Eq.(1.4)
D 4CaL

The main advantages of TLP bonding are (1) no interface remains after bonding;
(2) joints possess properties similar to those of the parent material; (3) large and complex

geometries can be bonded simultaneously. On the other hand, TLP bonding also has its

36

 

 

shortcomings: (1) long bond durations and high temperatures are required; (2) the
bonding process is only suitable for fast diffirsers; (3) post-bonding heat treatments for

age-hardenable alloys are required.

The high temperatures involved in TLP bonding (typically higher than 50% of the
parent materials’ melting points) makes it suitable for bonding high-temperature
materials such as Ti-alloys and Ni-based superalloys [MacDonald et al. 1992, Yan et al.
1993, Gale et al. 1996, Bird et al. 1998, Su et al. 1999, Campbell et a1. 2000]. lnterlayers
of pure Cu, Ni and Zn, eutectic Ag-Cu, Cu—Ni, Cu-Sn are commonly-used for joining Ti-
alloys [MacDonald et al. 1992, Bird et al. 1998], while Ni-based interlayers containing Cr

and Co are widely-employed for bonding Ni-based superalloys [Su et al. 1999].

For some material systems, joints solidified from the transient liquid may yield
better mechanical properties as compared to those that had been homogenized. Hence,
isothermal solidification and homogenization stages are deliberately avoided by
premature termination of the bonding process. Examples of such TLP variants are
reactive eutectic brazing of Zircaloy to stainless steel [Owczarski 1962], and Al-B
composites to Al [Niemann et al. 1974]. In both cases, the liquids are not transient and

are allowed to solidify into eutectic microstructures.
1.6 Liquid-Phase Diffusion

The reactive eutectic braze process discussed here involves liquid formation in the

joint via contact melting between NiTi and Nb, so it is important to understand liquid-

37

phase diffusion, which plays a critical role in the dissolution kinetics. Although there has
been no literature reporting on diffusion in liquid ternary alloys, it is still useful to review
the basic characteristics of pure liquid metals, the fundamental principles of liquid-phase

diffusion, and the key parameters affecting diffusion in these simple liquids.

The main observable difference between a solid and liquid is the inability of the
latter to support a static shear force. Other differences include higher energy of liquids,
higher compressibility, greater thermal expansion and higher difiusivity [Darken et al.
1953]. Many theories on the structure of a liquid metal have been proposed to account
for the abovementioned physical differences, but the most widely-accepted general
theory was proposed by Eyring [Eyring 193 6]. The author stated that liquid metals can
be considered analogous to their corresponding solids, except that some of the ‘lattice
sites’ are empty, resulting in lattice distortion. Another perspective is to consider a case
of a disordered close-packed crowd of atoms, with coordination numbers around 10 or 11
instead of 12. As a consequence, an atom can move into an adjacent ‘hole’ more
frequently and easily, hence, its inability to support a static shear force. Since this hole-

theory calls for only a relatively small number of vacancies in crystalline metals (~ 1 in

103 to 106 atoms even near melting point), whereas for liquids it requires one vacancy for
about every 10 atoms, it predicted that the diffusivity of liquid metals should be 102 to

5 . . . -ll -14 2 . .
10 trrnes greater than the corresponding SOlldS (10 - 10 m /s). In fact, most quurd

metals are observed to possess diffusivities of the order of 10‘9 mz/s just above their

melting temperatures [Iida et al. 19888], which is grossly consistent with the prediction.

38

The fluidic characteristics of liquid metals also suggest that the ease of flow of
atoms or the diffusive movement of atoms must be strongly dependent on the resistance
to shearing among neighboring atomic layers, which is, related the viscosity of the liquid
and is dependent on temperature. Equation (1.5) describes the Arrhenius type relation

between viscosity, 1], and temperature, T for liquid metals [Iida et al. 1988C].

Qvisc
=Aex —— E. 1.5
77 p( RT) q( )

where A is a pre-exponential constant, Qvisc is the viscosity activation energy required to

overcome the resistance of shearing among neighboring atomic layers, and R is the

universal gas constant.

In the hole-theory of Eyring [Eyring 1936], an attempt was made to develop a
mathematical expression (Eq. (1.6)) that describes the relation between self-diffusivity, D,

and 7],

D77 =k—T Eq. (1.6)
2r

where r is the atomic radius, k is Boltzmann’s constant and T is the temperature.
However, self-diffusivity data predicted by Eq. (1.6) did not satisfactorily agree with
many experimental values. Therefore, the Stokes-Einstein and Sutherland-Einstein

relations [Iida et al. 1988B], given in Eq. (1.7) and (1.8), respectively, were formulated.

39

These two expressions take into account the hydrodynamics of fluids and are widely-
accepted as one of the most robust empirical expressions for self-diffusivity in liquid
metals and solute diffusivities in very dilute metallic solutions. Values predicted using

Eq. (1.7) and (1.8) appears to agree reasonably well with experimental data, at least

within the same order of magnitude (i.e. 10.9 mZ/s).

D = 6—:77];r_ if diffusing particle is large compared Eq. (1.7)
with that of the medium.

D = g; if diffusing particle is approximately Eq. (1.8)
equal to that of the medium.

Little quantitative theoretical work has been performed on solute diffusion in
liquid alloys because knowledge of the fundamental quantities, such as the pair potentials
and the pair distribution functions of liquid alloys are insufficient. However, it still
seems reasonable to extend the qualitative hole-theory of the liquid state to metallic
solutions. For liquids originating from ordered alloys, such as ,B-brass [Darken et al.
1953], it is believed that short-range order may still persist despite absence of long-range
ordering. The difference in either electronegativity or size factor is likely to still exert

their influences in the liquid state but is expected to diminish as temperature increases.

While there are a few well-established methods for measuring self-diffusivities in
pure liquid metals [Iida et al. 1988B], experimental work on solute diffusivities in liquid
alloys, especially eutectic systems, are very limited. Getazheev’s group [Getazheev et al.

1970] attempted to address this issue by developing an experimental approach that

40

involves contact melting between two pure metals that will ultimately produce a eutectic
mixture. The authors assumed a diffusion-controlled melting process where the rate of
diffusive removal of solute from the liquid at the solid-liquid interface is slower than the
jump rate of particles fi'om the solid into the liquid. They firrther assumed steady-state

diffusion in the liquid and neglected hydrodynamic flow. The concentration profile in the

liquid, CL(x), and melting interface velocities of A and B, vA and vB, were described by,

ACE

CL(x)=—£é—"—L—x+cf Eq. (1.9)
W
dC
”(of -C§‘)=fl(DL—L—] Eq. (1.10)
PA 4"
dC
vB(Cf —C§)=££(DL—£) Eq. (1.11)
PB ‘11

where C LA and C LB are the liquidus compositions of metal A and B, respectively, CSA

and C SB are the solid compositions of metal A and B, respectively, p A, ,0]; and [2], are the

densities of A, B and the liquid phase, respectively, x is the distance in the liquid, w is the

width of the liquid, DL is the mutual diffusivity in the liquid. According to Darken’s

equation, partial-diffusivities of A and B in the liquid, D A and D3, are related to D], by,

41

DL=CLDA+(1_CL)DB Eq.(1.12)

Using the above numerical model, Getazheev et al. predicted the contact melting
rates between pure Bi and Sn at 200°C [Getazheev et al. 1970], which consequently
allowed them to derive solute diffusivities of Bi and Sn in the eutectic mixture at this
temperature. Their model was also applied to predict diffusivities in the Ga-In system

[Savintsev et al. 1971]. All predicted results produced reasonable diffusivity values

. . -9 2 . . . .
wrthm the order of 10 m /s, thus demonstrating great potentlal 1n measurrng solute

diffusivities in eutectic systems. However, there is limited literature available to verify
their findings. Furthermore, their approach has only been applied to binary eutectic
systems. The ability to extend their model to quasibinary eutectic in ternary systems has

not yet been tested.

1.7 Interpretation of Ternary Phase Diagrams
1.7.1 Basic Characteristics and Rules

Ternary systems are materials that are made up by three elements. The phases
present in a ternary alloy and their relative fractions at a particular temperature and
pressure can be described by a 4-dimensional phase diagram. However, it is visually
impossible to illustrate such a phase diagram. Hence, it is common to assume constant
pressure (atmospheric pressure in most cases) so that composition and temperature can be
adequately described using just three-dimensional space. The composition of a ternary

alloy is represented on a two-dimensional scale that takes on an equilateral triangular

42

geometry, also known as the Gibbs Triangle (see Figure 1.14), while the third dimension

(out of the page) describes temperature.

70 " 30

wAAAA 6 Am
so AWAVAVAVAA VA 10
B AVAVAVAVAVAVAVAVAVA c

102030405060708090
at.%C

Figure 1.14 Gibbs triangle and a ternary alloy with an overall composition P.
The Gibbs triangle provides the simplest means for plotting compositions, where
the three pure components are represented at the comers, A, B and C. Binary
compositions are represented along the edges AB, AC and BC. Ternary alloys are
represented within the Gibbs triangle. The composition scales are usually ruled with
lines parallel to the edges for ease of reading. For example, the overall composition of a
ternary alloy, P, in Figure 1.14 can easily be determined to be 30 at.% A, 50 at.% B and

20 at.% C.

Equilibria in phase diagrams are governed by the Gibbs phase rule, which is

mathematically expressed as,

P+F=C+2 Eq.(1.13)

43

where P is the number of phases, F is the degrees of freedom and C is the number of

components in the system.

Given that most phase diagrams are drawn with pressure being held constant, Eq.

(1.13) becomes,

P+F=C+1 EqUJQ

Therefore, in a ternary system where C is 3, equilibrium between two phases
possesses two degrees of freedom, which can either be temperature and composition of
one phase, or compositions of two phases. In other words, the knowledge of the
temperature and composition of one phase allows one to determine the composition of
the other phase, or the knowledge of the compositions of both phases will allow one to
determine the temperature at which they occur. For equilibrium involving three phases,
there is only one degree of freedom, where the equilibrium compositions of the three
phases will be fixed for a particular temperature. For four phases to coexist at
equilibrium, there are no degrees of freedom. This implies that such equilibrium can only
occur at a single point in the ternary phase diagram. Ternary 4-phase equilibrium will be

discussed in detail later.

Equilibrium compositions of two phases are described by tie-lines in the phase

field. Tie-lines are construction lines that join the composition points of the two

conjugate phases that exist at a particular temperature and pressure. Equilibrium

44

compositions of three phases are described by tie-triangles, whose edges are defined by
tie-lines among the three phases. It is important to note that no phases can have
compositions that fall along the tie-lines or within tie-triangles. Phases can assume
compositions only at the end-points of tie-lines (i.e. corners of tie-triangles). Examples
of tie-lines and tie-triangles of two- and three-phase fields are illustrated in Figure 1.15,

respectively.

   
 

Nb (Ti, Ni)

Ti
Figure 1.15 Examples of tie-lines and tie-triangles in a ternary phase diagram.
1. 7.2 Classes of Ternary Four-Phase Equilibrium
As mentioned previously, a ternary system can have up to four phases coexisting
in equilibrium at a single point in the composition-temperature space diagram. There are
three possible classes of four-phase equilibria in a ternary system, namely the Class-I,
Class-II and Class-III equilibria. Class-I equilibrium occurs at a particular composition
when a single liquid (L) phase fieezes isothermally into three different solid phases (a, ,8
and y) [Rhines 1956A]. It can be considered a ternary eutectic reaction, which is

mathematically expressed in Eq. (1.15).

45

Lflhigaa+B+y;L<—M—a+fl+y Eq.(1.15)

Class-I equilibrium is represented in a phase diagram by a reaction plane that
exists at a particular temperature. Figure 1.16 illustrates the plan-view of the reaction

plane involving a, B, y and L.

 

 

 

 

 

B 1'

Above Reaction Plane Below Reaction Plane

Figure 1.16 Schematic of a ternary 4-phase Class-I reaction plane. (a + B + L), (a + y
+ L) and (B + y + L) tie-triangles are above the reaction plane. (a + B + y) tie-triangle is
below the reaction plane.

The temperature axis is pointing out of the plane of the page. Just above the
reaction temperature, the reaction plane consists of three tie-triangles, namely the (a + B
+ L), (a + y + L) and (B + y + L). Below the reaction temperature, the reaction plane
consists only of a single tie-triangle of (a + B + y). Any initial liquid composition that
falls within this reaction plane will eventually produce a (a + B + y) — ternary eutectic

solid when cooled below the reaction temperature. An example of the four-phase Class-I
reaction is the equilibrium among NiTi, X3, X5 and liquid phases in the Ni-Ti-Nb system,

which is illustrated in Figure 1.17.

46

  

 

 

 

 

\. 7

mm W Class II @9500c

Nb(Ni,Ti)

 

 

‘0
Figure 1.17 Four-phase Class-I (NiTi, X3, X5 and Liquid) and Class-II (NiTi, TizNi,

Nb(Ni,Ti) and Liquid) reaction planes in the Ni-Ti—Nb ternary phase diagram illustrated
in 3D.

Ternary four-phase Class-II equilibrium is one that may be thought of as being
intermediate between eutectic and peritectic reaction [Rhines 1956B], which does not

exist in binary systems. The reaction, which is also represented by a plane in the ternary

phase diagram (see Figure 1.17), can be expressed as,

L+or————>C°°""‘g B+y; L+ae——He°""g B+7 Eq. (1.16)

47

During heating or cooling, two phases react to form two new phases. Figure 1.18
shows that just above the reaction temperature, the reaction plane consists of two tie-
triangles, namely the (a + B + L) and (a + y + L). Below the reaction temperature, the
reaction plane consists of two different tie-triangles, namely the (a + B + y) and (B + y +
L). For the case of cooling, four scenarios are possible depending on the overall
composition of the system. Consider alloys of 4 different compositions, A, B, C and D
that will be cooled through the Class-II reaction plane (Figure 1.18). Due to the
difference in initial compositions, each alloy will undergo different reaction pathways

that are summarized as follow:

A: a+B+L-—)a+B+7 Eq.(1.17)
B: a+7+L——>a+B+y Eq.(1.18)
C: a+B+L——-)B+7+L Eq.(1.19)
D: a+y+L—)B+7+L Eq. (1.20)

 

 

 

 

 

Above Reaction Plane Below Reaction Plane

Figure 1.18 Schematic of a ternary 4-phase Class-II reaction plane. (a + B + L) and (a
+ 1’ + L) are above the reaction plane. (a + B + y) and (B + y + L) are below the reaction
plane.

48

Ternary four-phase Class-III equilibrium is analogous to a pe1itectic reaction in
the binary system [Rhines 1956C]. It can also be thought of as an inverse of the Class-I
equilibrium, where upon cooling, three phases react isothermally to form a single new

phase (Figure 1.19). The reaction pathway is described as follows:

L+a+B—Mg——)7;L+a+B<—M§—y Eq.(l.21)

 

 

 

 

Above Reaction Plane Below Reaction Plane

Figure 1.19 Schematic of a ternary 4-phase Class-HI reaction plane. (a + B + L) tie-
triangle is above reaction plane. (a + y + L), (B + y + L) and (a + B + y) tie-triangles are
below the reaction plane.
I. 7.3 Isothermal Sections and Isopleths

The illustrations of three-dimensional ternary phase diagrams are often simplified
further through constructions of two-dimensional horizontal and vertical sections. The
former is known as an isothermal section, while the latter is termed an isopleth. Both
forms will be widely-used in this dissertation. An isothermal section, which is illustrated

in 3D in Figure 1.20, allows the location of phase fields at a particular temperature to be

visualized more easily.

49

Ni-Tl-Nb Ternary Phase Equilibria
2°00 (900°C - Isothermal Section)

 

Figure 1.20 Three-dimensional view of a 900°C — isothermal section in a Ni-Ti-Nb
ternary phase diagram.

Because of their resemblance to binary phase diagrams, isopleths are also widely-
used to illustrate phase evolution at a particular composition as a function of temperature.
Liquidus and solidus temperatures for any alloys represented can also be read from these
vertical sections. However, it is not usually possible to represent tie-lines in an isopleth
because tie-lines will not generally lie in the plane of the isopleth, with the exception of
quasibinary sections. Isopleths selected for study are usually (1) those that originate fi'om
one comer of the Gibbs triangle and end at mid-point of the opposite edge, therefore

representing a fixed ratio of two of the components (Figure 1.21), or (2) those that are

50

parallel to one edge of the phase diagram, representing a constant fraction of one of the

components, or (3) those that extend from one binary line compound to another (e. g.

Ni3Ti to Ni3Nb).

 

.{\ .

 

 

 

 

 

Figure 1.21 Three-dimensional view of a split Ni-Ti-Nb ternary phase diagram,
exposing the NiTi-Nb quasibinary section or isopleth.

When a congruently-melting intermediate phase occurs in a ternary system, it
sometimes happens that this phase forms a quasibinary system with one of the other
phases. An example is shown in Figure 1.21, where a quasibinary system exists between
the intermediate NiTi phase of the binary system Ni-Ti and the Nb component of the
ternary system. A vertical section taken along the line joining these two phases will be,
in all respects, equivalent to a binary phase diagram. The line that joins these two phases
is also known as the Alkemades line (see below) [Hillert 1998]. All tie-lines in two-

phase fields will lie in the plane of the section, while three-phase equilibrium will be

51

represented by a single horizontal line. No four-phase equilibrium can occur in the
quasibinary section. All alloys behave as though binary in their structural response to

temperature changes.

I. 7.4 Alkemades Theorem

Another important feature of ternary phase diagram is its liquidus projections. It
provides information on the change in liquid composition and the type of reaction
involved (e. g. eutectic or peritectic, class-I, H or III) as an alloy is freezing. The
directions of decreasing temperature of liquidus lines can be predicted using the
Alkemades theorem [Hillert 1998], which states that the intersection of a liquidus line (or
extrapolated liquidus line) with its corresponding Alkemades line (or extrapolated
Alkemades line) is a temperature-maximum on the liquidus line and a temperature-
minimum on the Alkemades line [Hillert 1998]. Alkemades lines are those that serve to

connect two solid phases that share a common liquidus boundary.

An example is shown in Figure 1.22a. It shows a ternary system with a
congruently-melting binary intermediate phase, D, and three other primary phases, A, B
and C. The liquidus projection of the ternary system is also illustrated. In order to know
the direction of decreasing temperature for the liquidus line separating C and D, an
Alkemades line CD is drawn to connect C and D. Figure 1.223 shows that the liquidus
line intersections the Alkemades line. Hence, a temperature-maximum exists on the
liquidus line at the intersection and temperature decreases on either side as illustrated in

Figure 1.22a. On the other hand, temperature decreases from either end. of the

52

Alkemades line and a temperature-minimum occurs at the intersection with the liquidus
line. Based on the Alkemade’s theorem, directions can be assigned to the rest of the

liquidus projections (Figure 1.22b).

Another example shown in Figure 1.23 describes a ternary system with an
incongruently-melting binary intermediate phase, E, and the same primary phases, A, B
and C. Given that the liquidus line between C and E does not intersect the corresponding
Alkemades line, the liquidus line is extrapolated to facilitate the intersection (dotted-line
in Figure 1.23). Therefore, direction can now be assigned to the liquidus line as shown in

Figure 1.23.

C
(a) ........... Alkamades Line

Liquidus Line

  
   
   

 

Arrows indicate
direction of decreasing
temperature.

 

O B

 

A

Figure 1.22 Schematic diagram of a ternary system with a congruently-melting phase,
D, showing (a) the construction of Alkarnades line CD and the directions of decreasing
temperatures on the liquidus and Alkarnades lines of concern; (b) the directions of
decreasing temperatures on the rest of the liquidus projection, based on the Alkarnades
theorem.

53

(b)

 

Liquidus Line

Arrows indicate
direction of decreasing

 

 

temperature.
C
A D “ B
A > < 0 e l < B
D

Figure 1.22 (Continued).

  
  
 
 

----------- Alkarnades Line

Liquidus Line

 

Arrows indicate
direction of decreasing
temperature.

 

A
Figure 1.23 Schematic diagram of a ternary system with an incongruently-melting

phase, E, showing the construction of Alkarnades line CE and the directions of
decreasing temperatures on the liquidus and Alkarnades lines of concern.

54

CHAPTER 2: EXPERIMENTAL METHODOLOGY

2.1 Introduction

Details on experimental procedures are discussed in this chapter. Specifically,
methodologies for specimen cleaning, brazing, annealing and alloying are provided,
followed by details of sample preparation for electron microscopy. Mechanical testing

procedures will be covered at the end of the chapter.

2.2 Surface Cleaning

Two types of base-NiTi alloys were employed in this work. They were room-
temperature shape-memory (50 at.%Ni-Ti) and superelastic (51 at.%Ni-Ti) alloys with
nominal thicknesses of 3.175 mm and 1.588 mm, respectively. Both alloys were supplied
by Nitinol Devices and Components Inc. The room-temperature shape-memory alloy
was mainly used for microstructural and indentation studies, while the superelastic form
was extensively employed for preparing single-lap and butt-joints. The lateral

dimensions for each purpose will be discussed in subsequent sections.

Prior to braze processing, native oxides on NiTi surfaces were chemically-

removed. Surface cleaning was similar for both NiTi alloys used. The base-alloys were

immersed in a pickling solution, consisting of 1HF:20HNO3:10H20 (by volume), for

about 5 minutes. The oxide-removal process was then terminated by rinsing the NiTi
sections in water. Base-NiTi and Nb braze foils were ultrasonically-cleaned in acetone,

followed by methanol in order to remove organic contaminants.

55

2.3 Braze and Post-braze Processing
All braze processes were performed in a Centorr M60 vacuum annealing firrnace.

The specimen chamber is evacuated by a Varian VHS-4 diffusion pump and is capable of

maintaining a base pressure better than 4 x 10.6 Torr. The hot zone is cylindrical, 2.5” in

diameter and 6” in length. Specimens and an Omega Type-K thermocouple probe were
positioned close to the center of the hot zone to ensure uniformity and accuracy in

temperature measurements.

Braze specimens were in the form of an NiTi/Nb/NiTi stacking configuration.
The Nb foils, which were acquired from ESPI Metals Inc., were held between the base-
NiTi via spot-welding at their comers. No external pressure was applied to the samples
during brazing. The fabrication of butt-joint specimens will be elaborated in the Tensile

Testing section.

After loading of specimens into the vacuum chamber, degassing was performed at

250°C for at least 3 hours. Brazing was allowed to commence when base pressure was in

the range of 10.6 Torr. Approximately 2800W of power was consistently applied to the

heating elements to achieve a constant temperature ramp-up profile for all braze
processes. Upon attaining the desired braze temperature, power to the heating elements
was manually controlled to maintain the desired braze temperature, which in most cases

were 1185°C. The manual control of braze temperature involves an error of +/- 5°C.

During brazing, chamber pressure occasionally crept up to about 2 x 10.5 Torr. However,

56

pressure remained in the range of 10'6 Torr throughout most stages of the braze process.

Termination of brazing was effected by shutting off power to the heating elements.
Specimens were cooled naturally in the vacuum fumace before removing them from the
chamber. Representative temperature-time and pressure-time profiles for the braze

process are shown in Figure 2.1.

 

 

 

 

 

 

 

1400 ~ . ~ . ~ . . 10
1200- l
A - \ “8 1’
01000- D o
e ...- -6 .c,
3 800- (D
g . 0 Pen 3?
0’ - . E3
g 500‘ \mfi \x 4 .3”
'— 400‘I +Temperature -U-Uflu’u’” :21
~ —D—Pressure '2
200 ~ r - . . . . , -
0 5 10 15 20 25

 

Time (Minutes)
Figure 2.1 Typical temperature-time and pressure-time profiles of a braze process.
Post-braze heat treatments of braze joints were performed in air, using a Lindberg
resistance-heated furnace. Temperature in the furnace was monitored by a K-type
thermocouple. Heat-treated specimens were cooled by quenching in room-temperature

water promptly upon removal from the furnace.

2.4 Micro-alloying with Tungsten in Nb-hased Braze Joints
Tungsten was alloyed into Nb-based braze joints by coating Nb braze foils. Films

of tungsten were deposited on Nb foils by DC magnetron sputtering at room-temperature

57

under a base vacuum of 2 x 10.6 Torr using a 3”-diameter Kurt J. Lesker Torus gun and

MDXIK power supply. Argon working pressure during sputtering was 5 mTorr. Power
applied to the cathode was maintained at 200W, while working distance from the
tungsten sputter target to the substrate was 50 rmn. The deposition rate was first
determined before the actual film deposition runs. Tungsten films were deposited on
glass microscope slides separately with increasing sputtering durations. The thicknesses
of the films were measured using a Wyko optical profilometer. A plot of thickness vs.
deposition time is provided in Figure 2.2. Linear regression of the data points estimated

the deposition rate to be 0.1 pm per minute.

Calibration of W Deposition Rate

 

.5 _s ..s
O N -§
1 I

oo
1

Thickness (microns)
O

O

 

 

 

0 l l T T i i T T I l T T

0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time(min)

Figure 2.2 Plot of thickness vs. deposition time of tungsten at a cathode power of
200W, argon working pressure of 5 mTorr and working distance of 50 mm.

58

For the W-alloying studies discussed in Chapter 5, 1 pm of tungsten was
deposited on 51 rim-thick Nb foils, which yield a composition of approximately 2 vol.%.
Assuming theoretical density, this thickness will coincidentally also produce an overall

braze foil composition of about 2 at.% tungsten.

The procedures and experimental conditions for the Nb film depositions on NiTi
alloys were similar to those described for tungsten deposition. Niobium film depositions
were required in an experiment that studied the effects of braze width on isothermal
solidification time, which is discussed in Chapter 3. Niobium films of approximately 6
pm thick were deposited onto 300 pm thick superelastic NiTi strips as the braze filler for

the experiment. The deposition rate for Nb was also found to be about 0.1 pm per minute.

2.5 Sample Preparation and Setup for SEM and TEM Studies
For scanning electron microscopy (SEM) studies, cross-sections of as-brazed

specimens were first ground using progressively finer grit SiC abrasive papers (from 240-

grit to 600-g1it), followed by polishing with progressively finer A1203 powders (from 5

pm to 0.05 pm). Polishing was terminated only after mirror-like finishes of specimen

surfaces were achieved. Finally, specimen surfaces were chemically-etched prior to SEM

observations using a solution of 1HF:2HNO3:10H20 (by volume). Etching time was

restricted to less than 10 seconds to limit formation of etch pits.

Most SEM observations were performed using a Hitachi S2500 SEM under an

accelerating voltage of 25 KV at a working distance of 10 mm. All images shown were

59

taken in secondary electron (SE) mode, unless otherwise specified. X-ray fluorescence
experiments were conducted using an Oxford Instruments detector in the SEM under an
accelerating voltage of 20KV. All energy spectra were collected under the same
acquisition duration of 120 seconds with spectrometer dead-time less than 30%.
Calibration was performed on Ni, Ti and Nb standards and the ZAF correction was
applied to all XRF quantitative results. The ZAF scheme corrects for matrix effects from
the specimen, which are affected by atomic number, absorption and fluorescence. Details

on ZAF corrections are described in [Flegler et al. 1993].

For Transmission Electron Microscopy (TEM) studies on the eutectic joints,
approximately 500 urn-thick cross-sectional slices of the joint area were out (see Figure
2.3). These bulk slices were mechanically-thinned down to less than 20 pm using SiC
abrasive papers before being mounted onto Open-aperture copper grids of 3 mm in
diameter using vacuum-compatible Gatan GI epoxy. Excess material outside the copper
grid was carefully trimmed away using fine surgical scissors. Next, eutectic joint areas of
the samples were ion-milled down to electron transparency using the Gatan Precision Ion
Polishing instrument. A milling angle of 4° and Ar ion energy of 4.0 KeV were
employed throughout the thinning process. Penetration of light through the thinned area
would indicate perforation and possible electron transparency along the perimeter of the
hole(s), which were achieved within 4 to 6 hours of milling time. All samples were

plasma-cleaned for 2 minutes before loading into the TEM specimen airlock.

60

 

Top-View of Cross-Sectional Slice of a Braze Joint

 

Outline of Open-
Aperture Copper Grid

~3 mm ~0.5 mm

 

Figure 2.3 Schematic diagram showing top-view of the cross-sectional slice of a
eutectic joint that would eventually be prepared as a TEM specimen. (The dotted circle
indicates the outline of the open-aperture copper grid that will be attached to the
specimen.)

Transmission electron microscopy studies were mainly conducted using a J EOL
2200FS field-emission TEM, which is also equipped with a similar XRF system to the
SEM. Imaging was performed under bright-field conditions under an accelerating
voltage of 200KV, while diffraction patterns were obtained using selected-area
diffraction apertures. Quantitative results from XRF experiments in the TEM were

corrected by the Cliff-Lorimer approach for thin-foils, which is described in [Flegler et al.

1993]. This correction scheme requires knowledge of the mass-density and thickness of
the specimen foil. In this case, mass-density of NiTi is 6.45 g/cm3, while the thickness

was assumed to be 100 mn. Spectrum acquisition duration was 120 seconds and dead-

time never exceeded 30%.

2.6 Tensile Testing
Butt-joint specimens were prepared using 50.8 mm (L) x 25.4 mm (W) NiTi
coupons with slits cut almost across the widths (see Figure 2.4). The nominal coupon

thickness for room-temperature shape-memory NiTi was 3.175 mm, while thickness for

61

room-temperature superelastic NiTi was 1.588 mm. The slits were electro-discharge
machine (EDM)-cut with either 254 m- or 356 mm -diameter wires, resulting in initial
joint clearances of approximately 300 pm or 400 pm, respectively (some material loss

occurred close to the wire as it traverses through the specimen).

Feeding Nb braze foil Eutectic Brazing Sectioning of Brazed Coupc
between mating NiTi surfaces under High-Vacuum into Butt-Joint Specimens

   

 

Nb Foil
(oversized)

    

Initial Gap:

. mm
~ 300 or 400 um

Figure 2.4 Schematic diagrams showing the procedures for fabricating NiTi butt-joint
specrmens.

After removing native oxides from the NiTi coupons, sufficient Nb braze foils
were fed into the slits until they fit snugly between the mating NiTi surfaces. The Nb
foils were deliberately oversized (i.e. wider than the thicknesses of the NiTi coupons) by
about 50% so that sufficient braze liquid would form to completely fill up the gap. The
Nb-filled NiTi coupons were individually loaded into the vacuum firrnace for braze
processing. Finally, brazed coupons were EDM-sectioned into approximately 5 mm-
wide butt-joined bars for tensile testing. End pieces from each coupon were discarded.
Tensile tests were performed using an Instron screw-driven universal testing instrument
at room temperature. The gage length of butt-joints was fixed at 30 mm and tests were

conducted at an elongation rate of 0.1 mm/s (measured at cross-head).

62

2.7 Micro- and N anoindentation

Nanoindentation was carried out using the MTS Nanoindenter XP with a diamond
Berkovich indenter. Hardness and elastic modulus were calculated from the basis of
Oliver and Pharr’s model [Oliver et al. 1992]. This model states that as the indenter tip is
pressed into the specimen, elastic and plastic deformations occur, forming an impression
related to the geometry of the indenter and to the plastic deformation mechanisms that
occurred in the material. Figure 2.5 shows a typical load vs. displacement curve

associated with loading and unloading during nanoindentation experiments.

 

 

hf hc hmax
Displacement

Figure 2.5 A typical load vs. displacement curve associated with loading and
unloading during nanoindentation experiments.

In Figure 2.5, hmax represents the displacement at peak load, Pmax, while he is the

indenter displacement projected from the tangent of the initial tmloading slope and hfis

the final displacement upon complete unloading. Given that only the elastic component

of the displacement is recovered upon retracting the indenter from the specimen (see

63

Figure 2.5), elastic modulus of the specimen can be calculated from the initial
unloading contact stiffness, S. The unloading curve in Figure 2.5 can be described

as a power law function [Oliver et al. 1992]. Its mathematical form is shown in Eq.

(2-1),
P=B(h—hf)"’ Eq.(2.1)

where B and m are empirically-determined fitting parameters. By differentiating Eq.

(2.1) at hmax, S can be determined.

dP
s = — = B h — h "1"
( dh Lima. m( 2) Eq. (2.2)

An expression was established by Sneddon [Sneddon 1965] involving contact

stiffness, S, contact area, AC, and reduced modulus, E,, which is shown in Eq. (2.3),

A
S = ZBErJ—{fj Eq. (2.3)

where B is a constant that depends on the geometry of the indenter (B = 1.034 for a

Berkovich indenter). E r is a modulus that accounts for elastic deformations in both

specimen and indenter.

64

The mathematical form of E, is given in Eq. (2.4),

 

E, = +1’UIMM Eq. (2.4)
E Eindenter

where E and v are the elastic modulus and Poisson’s ratio for the specimen, respectively.

In this study, a diamond Berkovich indenter was used where E indenter = 1141 GPa and

”indenter = 0-07-

The contact area, Ac, can be obtained from the tip area function, as shown in Eq.
(2.5). The first term corresponds to a perfect Berkovich indenter, while the last 2 terms

were attributed to tip imperfections and were determined fiom calibrations performed on

fused silica.

Ac = 24.56er2 + 243.05hc —176.15J7r: Eq. (2.5)

The contact depth, he, can be approximated using the following relation,

hc = hmax —e—n—‘a-l Eq. (2.6)

65

where s is a indenter geometry-dependent parameter. In this case, a = 0.75 for a

Berkovich indenter.

. . . . . dh
The mdentatron strain rate, 4', IS defined as the descent rate of the indenter, -d— ,
t

divided by its current depth, h. That is,

dh 1
C =[EI-5) Eq. (2.7)

The nanoindentation hardness, H, is defined as the indentation load at maximum

displacement divided by the corresponding contact area. Given thath and Ac are

known, hardness can easily be calculated from Eq. (2.8).

P
H=_II& E,(2,8
I q )

Prior to conducting actual experiments, indentations on polished A1203 (99%

density) were performed at holding depths of 50 nm, 100 nm and 200 nm with a strain
rate of 0.001 per second. This was to verify accuracy of the indenter tip shape function

described in Eq. (2.5). At least 10 indents were made at each depth. Using a Poisson’s

ratio of 0.18 for A1203, the Oliver and Pharr [Oliver et al. 1992] model was employed for

calculating the hardness and elastic modulus of A1203. Results are shown in Figure 2.6.

66

The calculated hardnesses and elastic moduli were found to be reasonably constant across
these indentation holding depths. The average hardness and modulus were determined to
be 27.3 GPa and 402 GPa, respectively, which are consistent with values reported by

[Munro 1997, Krell et al. 2001].

 

 

 

 

 

 

4o - , . . . . . . ~ 440
—e—Hardness ’
—Cl—Modu|us ”420
A 35- ~
8; i L] 6400 9:1
7; 30- ~380 g-
3 _ m
C A
'o ~360 0
21:0 25- ~ 9‘3
~340
20 ' l

 

 

 

. . . . . . . 320
0 50 100 150 200 250
Indentation Holding Depth (nm)

Figure 2.6 Hardness and elastic modulus of alumina derived from various indentation
holding depths.

The indenter was then tested against pure Nb at various indentation holding

depths. The strain rate was similar to that for A1203. At least 10 indents were made at

each depth. Hardness and elastic modulus were calculated based on a Poisson’s ratio of
0.38. Again, the mechanical properties remained reasonably consistent across the studied
range of indentation depths (Figure 2.7). Mean hardness and modulus were 1.2 GPa and
99 GPa, respectively, which agree reasonably well with values reported by [Myneni et al.

2003, Wang et al. 2005, Alkorta et al. 2006]. The indents on pure Nb were observed in

67

SEM (Figure 2.8). No significant pile-up or sink-in behaviors were associated with these

indents.

 

 

 

 

3.0 I I I ' I ' I T' I120
‘ -e—Hardness ’

2.5- % —D—Modulus '110
g 2.0- \gxq ~100§
V E
g 1.5. g} :90 E,
E 1.0- \ ~80 ’63
u 'o

0.5- ~70

0.0 T I ' l ' If T I 1 I ' I60

0 200 400 600 800 1000 1200

Indentation Holding Depth (nm)
Figure 2.7 Hardness and elastic modulus of Nb derived from various indentation
holding depths.

  

Figure 2.8 SEM mcrograph showing an array of inednts on pure Nb. (Indent depth
of 500 nm.)

68

The verification studies on A1203 and pure Nb showed that reasonably accurate

results can still be achieved with indentation depths as low as 50 nm. Hence, most of the
nanoindentation work on pre-polished eutectic phase in braze joints and the base-NiTi
alloy were conducted at holding depths of 50 nm. The reason for using a low indentation
depth was to minimize the probability of probing any pro-eutectic NiTi phase that could
be present beneath the exposed eutectic phase. Poisson’s ratios of 0.30 were used for
calculating indentation results derived fi'om pure NiTi. Since there are no reported
Poisson’s ratio of the NiTi-Nb eutectic micro-constituent, a value of 0.35, intermediate of

pure NiTi and Nb, was chosen for the indentation calculations.

Microindentations were employed to determine microscopic hardness.
Experiments were conducted using Vickers indenter with a load of 300 grams. This load
was selected as it produced indents that were almost as wide as the field-of-view of the
microsc0pe’s eye-piece, hence minimizing measurement errors. Joint specimens were
polished based on procedures described previously. The widths of indents were
measured using the built-in microscope with a magnification of 50X. Only

measurements of symmetrical indents were used for calculating hardness values.

69

CHAPTER 3: MICROSTRUCTURAL EVOLUTION IN REACTIVE
EUTECTIC BRAZIN G

3.1 Introduction

In this chapter, results of experiments on the initial contact melting between base-
NiTi and pure Nb filler metal, the propagation of the liquid, and termination stages
(solidification) of the braze process will be presented. A diffusion-based numerical
model is subsequently established to simulate the melting kinetics. Solidification
products associated with the braze process will also be discussed. In particular, the
effects of braze duration, post-braze anneal, surface preparation and vacuum conditions
on microstructures and morphologies of the joints were extensively studied. In order to
understand the origin of the NiTi quasibinary eutectic isopleth in the ternary system, the

phase equilibria of Ni-Ti-Nb will be revisited. Finally, a section has been included that

addresses the crystallography and phase equilibria of TizNi and (Ti,Nb)4Ni20.

3.2 Nomenclature

The nomenclature adopted for most discussions in this chapter is introduced.
Figure 3.1 shows the ‘perfect’ NiTi-Nb quasibinary eutectic isopleth that will be
extensively used for discussions in terms of both thermodynamic and kinetic aspects of
contact melting. The isopleth is analogous to a typical binary eutectic phase diagram
with a quasibinary eutectic point at 26 at.% Nb occurring at 1170°C. Solidus
compositions of NiTi and Nb at 1170°C were estimated fi'om the literature [Gupta 1991,
Prima et. al. 1995, Yang et. al. 2000, Matsumoto et. al. 2005] to be 7 at.% Nb and 91

at.% Nb, respectively. The origin of the perfect quasibinary eutectic isopleth in the

70

ternary system will be discussed in a later section. Two solid phases exist in Figure 3.1,
which are the NiTi and Nb phases. Pure NiTi melts congruently at 1310°C, while pure
Nb melts at 2469°C. Any liquid composition beyond the solubility limits of both solid

phases will result in eutectic NiTi + Nb solidification.

2500

 

2000 - Liquid

     
    

L + bcc-Nb

  

Temperature, °C
G?
O
O

1‘52; 1185°C
..-- . —————————————————————————————

 

 

 

 

CN CN,s CNJ' CBJ 1170 CB,s bCC- CB
l
TiNi :
l TiNi + bcc-Nb
500 . . . . .
TiNi 20 25 4o 50 80 Nb
Atom % Nb

Figure 3.1 The ‘perfect’ NiTi-Nb quasibinary eutectic isopleth with key compositions
labeled for 1185°C.

Given that most contact melting experiments were performed at 1185°C, the key
compositions associated with NiTi and Nb phases at this temperature are represented by
the symbols listed below. Values associated with the compositions (indicated by
parentheses) were derived based on the assumption that the liquidus and solidus lines of

Figure 3.1 are linear.

71

CN: Pure NiTi (0.0 at.% Nb)

Cit/.31 NiTi Solidus Composition (7.0 at.% Nb)

CN. 1: NiTi Liquidus Composition (23.2 at.% Nb)

CB, 1: Nb Liquidus Composition (26.9 at.% Nb)

CBJ: Nb Solidus Composition (91.0 at.% Nb)

CB: Pure Nb (100.0 at.% Nb)

CE: Quasibinary Eutectic Composition (26.0 at.% Nb)

CL: Instantaneous Average Liquid Composition = (CNJ + CB,1)/2

All numerical modeling and related mathematical expressions were based on half-
joint configuration as illustrated in Figure 3.2. Vector quantities abide by the convention
that right-to-left direction is positive. The following lists the terms frequently used for

discussion in this chapter and have also been labeled accordingly in Figure 3.2. Note that

l . . -2 -1 . . .
JNi = JTi and JNiTi = JNi + JTi were assumed. Fluxes are 1n umt of m s , velocrtres 1n

-1 . . .
ms and posrtrons 1n m.

JNiTi—rLi Flux of Ni plus Ti from solid NiTi into liquid
JNb—rLi Flux of Nb from solid Nb into liquid

JNiTi—rNbi Flux of Ni and Ti from liquid into solid Nb

72

JNb_,NiTiZ Flux of Nb from liquid into solid NiTi

VNiTi3 NiTi-Liquid interface velocity
va: Nb-Liquid interface velocity
xM-Ti: NiTi-Liquid interface position (wrt to initial NiTi-Nb contact position)
be: Nb-Liquid interface position (wrt to initial NiTi-Nb contact position)
w: Liquid width of the half-joint configuration = xNiTi - be
Initial NiTi-Nb
Contact Position

leTi = 1% = 0

   
   
 

l‘l/Hr' T

   
  

   
 

xNiTi 'be

Time :1 _

- W_xNiTi' be

W. ,. i
- ‘H‘JI$:UI1‘I:"III'I'II

5 ’- I'I l'-:::::4':"- I: 1:, IRIS“ :h:l:l:':.-I:::'-

   
    

  
    

IIII:I II

uuuuuuuuuuuu

    

:I:1I IIIIIIIIIII
I. I:- ..I:::I:‘:'

‘ '1'-

 
 

i—J
vNiTi<—‘ E NiTi—>1.

JNb—lNiTi <—§ -> 'JNiTi—rNb

Figure 3.2 Schematic diagram of a half-joint configuration with frequently used
terms indicated.

73

3.3 Contact Melting Mechanisms

In order to study the reactive brazing process, room-temperature shape-memory
NiTi sheets (10.2 mm (L) x 7.6 mm (W) x 3.175 m (T)) and 51 um-thick Nb foils were
stacked in a NiTi/Nb/NiTi configuration and brazed at 1185°C. Details on brazing
procedures have been documented in Chapter 2, and will not be repeated here. Contact
melting between the base-NiTi and Nb filler metal was found to occur at the braze
temperature, resulting in liquid formation in the joint. The contact melting process can be
classified into 3 stages. They are (1) initial melting, (2) braze liquid propagation, and (3)
termination. Experimental observations pertaining to each of these stages of contact

melting will be presented and discussed separately.

3.3.] Initial Melting

Joint specimens were separately brazed at 1185°C for durations ranging from 5
seconds to 20 seconds. The braze temperature-time profiles were similar to that shown
in Figure 2.1. A timer was initialized when temperature reached 1185°C and braze was
terminated at progressively longer durations for each specimen with intervals of 5
seconds between them. It was found that braze liquid formed consistently only after a
duration of 15 seconds, suggesting the presence of an ‘incubation’ period for contact

melting to commence.

The incubation period might be attributed to the thermal inertia of the specimen

requiring a finite amount of time for the joint area to approach braze temperature. A

simple computation model was established to determine if thermal inertia is responsible

74

for the incubation time. The model is illustrated in Figure 3.3, which shows a half-joint

configuration. The NiTi alloy and the Nb filler metal have thicknesses of 3.175 mm and
25.5 um, respectively. The initial NiTi-Nb interface temperature, Tint, was 300°C
(degassing temperature), while the specimen surface temperature, Tswfi was held
indefinitely at 1185°C (brazing temperature). This give rise to an initial temperature
difference, AT= Tm,f- Tim, of 885°C. The model assumed that heat can only flow to

the NiTi-Nb interface through the NiTi surface.

NiTi—Nb
Inteiface
fig; .

Temperature

Tint =
— 300°C

 

3.175 mm 25.5 um
Tsurf=

1185°C - "'7

   
   

. . : Tint,t
................ 3000C

Distance
Figure 3.3 Schematic diagram showing the temperature profile across a half-joint as a
fimction of time during the incubation period.

These initial conditions present an extremum situation that should provide an

upper-bound to the time required for the NiTi-Nb interface temperature to reach 1170°C

75

(believed to be the onset temperature for contact melting) after the NiTi surface attains a
temperature of 1185°C. But in actual experiments, ATshould be smaller than 885°C

because the NiTi surface and NiTi-Nb interface temperatures were both increasing

dynamically over a finite amount of time during ramp—up such that T int will be higher

than 300°C when T swfreaches 1185°C. Furthermore, besides the NiTi surfaces, heat can

also enter through the edges. Hence, the actual duration for the NiTi-Nb interface to
reach 1170°C (i.e. for melting to start) after the NiTi surface attains a temperature of

1 185°C should be less than that predicted in the model.

Given the temperature setup, heat will flow from the specimen surface towards
the NiTi-Nb interface. Since the thickness of Nb filler metal is significantly smaller than
the thickness of the NiTi alloy, the model assumed that the Nb filler metal has a uniform
temperature across its thickness at all times. Additional heat flowing across the NiTi-Nb
interface will result in uniform temperature increment in the Nb filler metal. The model
also assumed steady-state heat flow whereby a linear thermal gradient exists between the
NiTi alloy surface and the NiTi-Nb interface. The change in heat flowing across the
NiTi-Nb interface, AQ(t), at a particular time, t, over a finite time interval, At, can be

determined using the thermal conductivity expression,

AQ(t) = kA[A—T(—Q]At Eq. (3.1)

x

76

where k is the thermal conductivity of NiTi, A is the NiTi-Nb interfacial area, which is

conveniently chosen as unity, and x is the thickness of NiTi.

The heat input from AQ(t) caused T(t),-m to increase to T(t+1),-,,,, hence decreasing

the temperature difference between the surface and NiTi-Nb interface, AT (t+1 ), at time

t+I . The change in AT(t+1) due to AQ(t) are mathematically expressed as,

 

T(t +1)int = T(t)int + A90) Eq. (3.2)
me
AT(t +1) = Tswf — T(t +1)int Eq. (3.3)

where m is the mass of the Nb filler metal and CI) is the specific heat capacity of Nb.

1200

 

1000 . ..."

O) on
8 8
L 1

0
0..

NiTi-Nb Interface Temperature (Deg. C)
.5
O
o
O 1.

A
V

 

 

 

N
8

l

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Tlme(mlcro-sec)

Figure 3.4 Predicted NiTi-Nb interface temperature as a function of time during the
incubation period.

77

Using thermal conductivity of NiTi, k = 41 W/m-K [Y ung et al. 2005], specific

heat capacity of Nb, C1, = 0.265 J/g.K, and a time interval, At = 20 us, the temperature at
the NiTi-Nb interface, T int: was predicted as a function of time using an explicit finite-
difference scheme (2nd order Runge-Kutta method) applied to Eq. (3.1) and (3.3). The

simulation was terminated when T int reached 1170°C. The program code was written in

FORTRAN 77 and is provided in Appendix A.

Results from the numerical calculations shows that T in, increases rapidly fi'om

300°C to 900°C, but the rate of increase retards significantly beyond 900°C (Figure 3.4).
The simulation predicted that the time taken for the NiTi-Nb interface to start melting is
about 1.7 us fi'om the time the NiTi surface reaches 1185°C. This predicted duration is
almost 7 orders of magnitude smaller than the experimentally-observed incubation time,
which strongly suggests that the incubation time for contact melting is not attributable to

thermal inertia.

Alternatively, the incubation time might be due to the presence of thin surface
oxide layers on NiTi and Nb serving as a temporary physical barrier for contact melting.
These oxide layers would have formed during the short atmospheric exposure between
cleaning and loading of samples into the vacuum chamber. Given that the braze chamber
was only under moderate or high-vacuum, further oxidation of the specimens might also
have occurred during the few minutes of ramping up to braze temperature. Hence,

contact melting could only have occurred after Nb or Ni and Ti diffuses across this oxide

78

barrier in a finite amount of time, which manifests as the incubation period. To test the
hypothesis of an oxide difiusion barrier, the characteristic diffusion lengths of Ni, Ti and

Nb at 1170°C within a time frame of 15 seconds were estimated using Eq. (3.4).

L = m Eq. (3.4)

where L is the characteristic diffusion length, D is the diffusion coefficient and t is time.

Given that the intrinsic diffusion coefficients of NiTi and Nb at 1170°C are in the

order of 10'12 [Bastin et al. 1974] and 10‘15 m2/s [Einziger et al. 1978], respectively, Eq.

(3.4) shows that the characteristic diffusion lengths of these elements after an incubation
time of 15 seconds range from about 200 nm to 7700 nm. Although oxide thickness
measurements were not performed on the NiTi and Nb, it is reasonable to believe that
tens to hundreds of nanometer thick oxides can form on these alloy/metal surfaces during
the ramp-up stage, where the specimens were subjected to high temperatures under
moderate or high vacuum conditions. The extent of oxidation is supported by the
observation that alloy surfaces always have a yellowish-gold tint after brazing, indicating
an oxide thickness about one-half the wavelength responsible for the color. Given that
yellow and gold colors possess wavelengths within 570 — 620 nm, the approximate oxide
thickness is about 300 nm. Since the calculated characteristic diffusion lengths fall
within similar orders of magnitude with the predicted range of oxide thicknesses, it is
reasonable to deduce that the incubation time for contact melting is attributed to the

presence of oxide barriers on the NiTi and Nb mating surfaces.

79

It is crucial to highlight that superheating is not required for onset of melting (i.e.

nucleation of liquid phase) [Porter et. al. 2001]. This can be understood from the relative

free energies of the solid-liquid (ySL), liquid-vapor (n V) and solid-vapor (ySV) interfaces,

which are given by:

YSL+YLV< l’SV Eq- (35)

Equation (3.5) implies that the wetting angle is zero, hence the absence of a
nucleation barrier for melting. This is in contrast with solidification, where undercooling

is required to overcome a nucleation barrier.

After Nb or Ni and Ti diffuse across the oxide layer, liquid forms and dissolves
the oxide. In the following discussion, however, the presence of the oxide layer will be
ignored. Experimental evidence, which will be provided in the next section, shows that
melting commenced at the mating interface (hence, the term ‘contact melting’). The
driving force can be understood through analysis of the local concentration at the mating
interface. First, surface atomic concentrations of NiTi and Nb of various low-index
planes were calculated and summarized in Table 3.1. Since the base-alloy and filler
metals used in this study were polycrystalline in nature, permutations of pairs of NiTi-Nb
surface planes were performed to simulate some possible ‘ideal interface’ scenarios. The
interface concentration, confined to the first-monolayers of NiTi and Nb, was then
determined for each case. The purpose is to develop a general sense of the local

concentrations at various NiTi-Nb interfaces. Calculations, shown in the matrix in Table

80

 

3.2, revealed that the NiTi-Nb interface generally have concentrations above 40 at.% Nb

(average concentration is 61.5 at% Nb). These compositions fall between C3,] and C3”,

or within the Nb + Liquid phase field of Figure 3.1, which are unstable against melting.

Hence, contact melting occurred at the mating interface, producing the braze liquid.

 

 

 

Table 3.1 Surface atomic concentrations of NiTi and Nb on low-index planes.
Planes Surface Concentration of NiTi Surface Concentration of Nb
l4 -2 l4 -2
(x 10 cm ) (x 10 cm )
(100) 5.56 9.14
(1 10) 7.87 12.9
(1 1 1) 3.21 5.28
Table 3.2 Localized concentration at diffusion couple interface of various low-index

plane combinations (First monolayer approximation).
Localized Concentration at Diffusion Couple Interface (at% Nb)

 

 

 

Contact Surface of Nb
(100) (110) (111)
(100) 62.2 69.9 48.7
“mm???“ M (110) 53.7 62.1 40.2
(111) 74.0 80.1 62.2
Average 61.5

 

 

3.3.2 Braze Liquid Propagation

After initial melting at the NiTi-Nb mating interface, propagation of the liquid
phase occurred in the joint. A series of ex—situ SEM micrographs, capturing changes in
the width of the liquid at 1, 8, 21 and 31 seconds of the NiTi-Nb contact melting process,
are shown in Figure 3.5a — (1, respectively. The braze durations stated in Figure 3.5 are
actually resultant durations obtained by subtracting the incubation time from the total.
Gray features at the top and bottom of the micrographs are the unmelted base-NiTi, while

the bright feature in the center of each micrograph is the unmelted Nb filler metal. The

81

predominantly planar (but occasionally bulbous) layers against the unmelted NiTi are
pro-eutectic NiTi. Both pro-eutectic NiTi and the eutectic micro-constituents are

solidification products of the braze liquid. These microstructural features will be

discussed in a later section.

 

Figure 3.5 ”SEM micrographs of the braze region showing NiTi-Nb contact melting at
braze duration of (a) 1 second; (b) 8 seconds; (c) 21 seconds; (d) 31 seconds.

82

It can be seen in Figure 3.53 that within the first second of contact melting (afier
the 15 seconds incubation time), the braze liquid had already propagated to an
approximate thickness of 30 pm on each half of the braze joint. The half-thickness of Nb
foil that had melted was measured to be 8 um. Hence, approximately 22 um of base-NiTi
must have melted on each half of the joint. This suggests an almost 3:1 NiTi-to-Nb
volume ratio or 1.421 molar ratio that contributed to braze liquid formation. This ratio

corresponds to an approximate composition of 37Ni-37Ti-26N i (at.%), which matches

the composition, C B As contact melting proceeded from 1 to 21 seconds, wider liquid

zones were formed, which are evident in Figure 3.5b and c. However, based on the time
rate of change of remaining Nb thiclmess, the rate of melting appeared to decrease with
increasing time. Significant attenuation of melting rates of NiTi and Nb was observed
between 21 and 31 seconds (Figure 3.5c and (1), resulting only in subtle widening of the

braze liquid.

3. 3.3 Termination

When specimens were subjected to braze durations beyond 31 seconds, complete
melting of Nb braze foil occurred (Figure 3.6b). After Nb exhaustion, the pro-eutectic
NiTi phase fraction increased substantially and a transition from a predominantly planar
morphology to a dendritic form was observed. The pro-eutectic Nb phase was absent
after Nb exhaustion. (Figure 3.6a is a lower-magnification duplicate of Figure 3.5d that
serves as a basis for comparison with Figures 3.6b — (1.) Liquid width continued to
increase slightly between 31 and 120 seconds (Figure 3.6a and b), which can be attributed

to continued melting of base—NiTi. The melting rate was significantly slower than those

83

observed in Figure 3.5. Figures 3.6c - d show that the liquid width maintained rather
constant total width of 250 pm from 1200 seconds up to the maximum studied braze
duration of 3600 seconds. Pronounced thickening of the pro-eutectic NiTi was not
observed. These observations indicate the negligible rate of isothermal solidification (to

be discussed in detail below).

Dcnrlritic
Pro-eutectic NiTi

 

1200 sec - - i . 3600 see

Figure 3.6 SEM micrographs of the braze region at braze durations of (a) 31 seconds; (b)
120 seconds; (c) 1200 seconds; ((1) 3600 seconds.

84

Isothermal solidification is controlled by solute diffusion in the base-material.
Once the solid Nb has been depleted, Nb diffusing from the liquid into the solid NiTi
upsets the interface equilibrium, which can only be restored by freezing out NiTi,
consequently rejecting Nb back into the liquid. Therefore, attention was focused on the
diffusion of Nb across the NiTi-Liquid interface. Figures 3.7a and b show the Nb La
XRF intensity profiles across NiTi-Liquid interfaces of specimens brazed for 31 and 3600
seconds, respectively. The line profile in Figure 3.7b was deliberately defined well
within the bulk of the pro-eutectic NiTi phase as the micrograph suggests the presence of
coring (to be discussed later) around the edges of the pro-eutectic NiTi phase, which will
also yield a change in Nb-concentration across the cored structure. Both profiles revealed
constant Nb compositions in the pro-eutectic NiTi phase (originated from the liquid).

Approximately 4 and 12 at.% Nb were determined in Figure 3.7a and b, respectively.

I
‘-‘-rr»‘4-«:rrr-‘IL .
- '1

v' |

' 0 at.% Nb
‘ W‘AA‘HIA,A’. .

12 at.% Nb

4 at.% NI)

“ at.% Nb (\me Profile
‘ ‘ Prior \
Liquid

\ NiTi
\

    

4* ’. I 31 sec 3600 SCC

Prior Liquid NiTi

 

Figure 3.7 X—ray fluorescence line profiles of Nb La x-ray intensities across pro-eutectic
NiTi/pure NiTi interfaces of specimens subjected to braze durations of (a) 31 seconds; (b)
3600 seconds.

85

However, Nb concentration decreased promptly across the NiTi-Liquid interface.
The Nb composition in Figure 3.7a was observed to attenuate from 4 at.% to zero across
a distance of approximately 5 pm from the NiTi-Liquid interface. It should be noted that
the large excitation volume inherent in XRF (a few um) is believed to generate an
overestimation of the spatial measurement. Hence, it is reasonable to assume that Nb
diffusion in NiTi is negligible at least within a braze duration of 31 seconds. For a braze
duration of 3600 seconds, the Nb composition attenuated from 12 at.% to zero over a
larger thickness of 20 um (Figure 3.7b), thus implying that solid-state diffusion of Nb

had become more pronounced.

For the case of Figure 3.7b, if one assumes that the solid NiTi adjacent to the

NiTi-Liquid interface has a constant concentration of Nb, C N, 5, throughout the braze

duration, t, and Nb diffusion in the semi-infinite base-NiTi alloy occurs across a
stationary solid-liquid interface over a distance, x, then the diffusion problem can be

expressed mathematically as shown in Eq. (3.6). Note that the concentration of Nb in the

base-NiTi alloy far away fiom the NiTi-Liquid interface, C N: is zero.

C___C__—N

CN, 3 "CN edeJljmt

 

Eq. (3.6)

The term 2 D th is the characteristic diffusion length, L, at which the diffuser

concentration has decayed to about 15 % of C N. s at time t. Given the limited spatial

86

resolution of XRF, the attenuation length, instead of L, will be considered in the

calculation of Nb diffusivity. With an attenuation length of 20 um and time of 3600

seconds, the diffusion coefficient of Nb in NiTi, DNb, was estimated to be 2.8 x 10-14

m2/s. This value is about an order of magnitude higher than the self-diffusion coefficient
of Nb at 1185°C [Einziger et al. 1978]. As a means for comparison, the intrinsic
diffusion coefficient of NiTi at the same temperature is about 2.0 x 10'12 mZ/s [Bastin et

al. 1974], which is almost two orders of magnitude higher. From the calculated
diffusivity, it is safe to deduce that low diffusivity of Nb in NiTi is responsible for the
sluggish isothermal solidification after the complete melting of Nb filler. Given that the
liquid layer was significantly thick as compared the diffusion distance of Nb in Figure
3.7b, homogenization of the joint through isothermal solidification was not observed, at
least within the studied time frame. That is, braze processes involving thick liquid layers

can be considered to have entered a ‘termination’ phase upon the exhaustion of Nb filler.

900 see
Figure 3.8 Microstructures of joints fabricated using 6 pm Nb film with braze
durations of (a) 30 seconds; and (b) 900 seconds.

 

87

Effects of isothermal solidification can only be observed if thin Nb films of 3 pm
thick were employed as braze filler instead of the 51 um foils. Pure Nb films were
deposited onto each mating NiTi surface at room-temperature using DC-magnetron
sputtering, hence, yielding a total Nb thickness of 6 pm. Figures 3.8a and b show the
microstructures of joints processed at 1185°C for 30 and 900 seconds, respectively. At
30 seconds, complete melting of the Nb film was observed (Figure 3.8a), resulting in pro-
eutectic NiTi and eutectic micro-constituents. The total liquid width (pro-eutectic NiTi +

eutectic) was measured to be about 20 pm.

In contrast, Figure 3.8b revealed near-homogenization of the joint due to
isothermal solidification when braze time was increased to 900 seconds. This is evident
fiom the significantly narrower eutectic width as compared to that observed in Figure
3.8a. Based on the calculated diffusion coefficient of Nb in NiTi, the characteristic
diffusion length is approximately 10 um for each half of the joint, which is in good
agreement with the extent of narrowing of the joint in Figure 3.8b. As a result of
isothermal solidification, the interfaces between pro-eutectic NiTi and base-NiTi were
not physically distinguishable. In addition, new solidification products, in the form of
facetted precipitates, were observed to formed in the liquid phase and extensively
decorate the centerline of the joint. These precipitates are presumed to be oxides and will

be discussed in a later section.

The results in Figure 3.8 suggest that homogenization of j oint through isothermal

solidification is viable if the maximum width of the liquid phase is in the same order of

88

magnitude as the Nb diffusion distance associated with the desired braze time. However,
the process of homogenization, which is a desirable step in typical TLP processing,
resulted in the precipitation of facetted phase along the centerline of the joint.
Termination of braze process in the homogenization stage will be undesirable if these

precipitates possesses brittle characteristics.

To summarize, the contact melting process can be categorized into 3 stages: (1)
initial melting; (2) liquid propagation; and (3) termination of melting. The initial melting
stage involves an incubation period, which is the time taken for Ni, Ti and Nb to inter-
diffuse across the native oxide barriers before melting commences at the interface. The
liquid phase widens or propagates as further melting of NiTi and Nb occurs, and is rate-
controlled by diffusion in the liquid (to be discussed further below). When the liquid
width is significantly larger than the diffusion length of Nb in NiTi associated with the
braze duration, the effect of isothermal solidification is insignificant. Hence, narrowing
of liquid width will not be observed after complete melting of Nb filler and the contact
melting process is considered to have terminated. However, if the braze liquid thickness
is within the same order of magnitude as the diffusion length of Nb, isothermal
solidification will dominate after the complete melting of Nb, resulting in narrowing of

liquid width.

89

"'3

3.4 Contact Melting Kinetics

In this section, a theoretical model of the contact melting process will be proposed
to account for the experimental observations in section 3.3. The model assumes a half-
space scenario where a NiTi-Nb couple was subjected to brazing at 1185°C. A thickness
of 25.5 um was chosen for the Nb filler, which corresponds to the thickness of those
specimens employed in Section 3.3 in full-space. For ease of calculation, unit areas were
assumed for both NiTi and Nb. The establishment of this model will also form the basis
for numerical calculations of the contact melting kinetics, where the aim is to determine
the diffusivities of Ni, Ti and Nb in the liquid. The contact melting kinetics will be

presented later in the section.

Recall that contact melting originates from the NiTi-Nb mating interface. Contact

melting occurs because the local concentration at the NiTi-Nb solid mating interface falls

between C B, 1 and C33, which is thermodynamically-forbidden at 1185°C. Upon the first
instance of melting, it is postulated that braze liquids close to the solid NiTi and Nb
immediately attempts to maintain equilibrium solid-liquid interface compositions at C N. 1

and CB, 1, respectively. The liquid close to the base-NiTi can establish equilibrium only if

the base-NiTi adjacent to the liquid builds up a thin solid layer with composition, CN,s-
Likewise, the liquid next to the unmelted Nb will build up a thin layer of solid of

composition, C B, 5. These layers are perhaps within the order of a few monolayers and

their formation should, therefore, not control the rate of melting.

90

The discrepancy (AC) of composition between C N. 1 and C B, 1 across the liquid

(w(t)) at a particular time (t) sets up a concentration gradient (A C/w(t)) in the liquid phase.
Flux exchanges between Ni, Ti and Nb must therefore occur in the presence of this
concentration gradient, whereby Ni and Ti atoms in the liquid diffuse towards solid Nb,
while Nb atoms in the liquid diffuse towards solid NiTi. These fluxes in turn upset the
local interface equilibria at both solid-liquid interfaces. The system can re-establish local
equilibrium only by further melting of the respective solids. Therefore, it is the deviation
from local equilibrium at the NiTi-Liquid and Nb-Liquid interfaces that forms the driving
force for continuous melting. Melting is thus hypothesized to be rate-controlled by solute

diffusion in the liquid.

Assuming steady-state liquid diffusion conditions (linear concentration gradient),

and given that the local concentrations of liquids at the two solid-liquid interfaces are

fixed at C N. 1 and C B, 1 for all liquid widths, continued melting flattens the concentration

gradient in the liquid as w increases with time. Fluxes of Ni, Ti and Nb, which are
directly proportional to concentration gradient (F ick’s first law), will therefore also
decrease. The reduction in flux associated with increasing liquid width slows down
subsequent melting of the base-NiTi and Nb filler. A schematic diagram showing the
establishment and change of a liquid concentration gradient across the two solid-liquid

interfaces as a function of time before exhaustion of Nb filler is illustrated in Figure 3.9.

91

 

 

 

 

 

 

 

 

Braze Temperature: 1185°C @- 1,
11 < t2 < t3 '2‘2
’59
NiTi , L Nb NiTi L Nb NiTi Nb. L g
«— .2 ------ £ ----- C B” E
- may, a”
w(t

 

Temp., °C
Figure 3.9 Schematic diagram illustrating the establishment and evolution of a
concentration gradient in the braze liquid as a function of time, based on the assumed
perfect NiTi-Nb quasibinary eutectic isopleth.

Based on the proposed model, the diffusion-controlled process of melting can be

represented by the mass-balance equation, shown in Eq. (3.7) and (3.8).

VNiTi (CNJ — CN,s) = -JNiTi—>L + JNb—eNiTi BL (37)

VNb(CB,s “CB,” = JNb—)L “Jilin—4M; E(1-(3-8)

Since solid-state diffusion is slow compared to liquid-phase diffusion, Eq. (3.7)

and (3.8) can be simplified into:

AC
“C S~D ,, Eq.(3.9)
w(lNT N, N,T[w(-_t))L

92

AC
v(t)Nb C3,, 4 —D§,b[—] Eq. (3.10)
W) L
where D110}, is the diffusion coefficient of Nb in the braze liquid. Dix-J,- is the collective
diffusion coefficient of Ni and Ti, or the smn of the diffusivities of Ni (DIIQi) and Ti

(0713-). Note that Ni and Ti are assumed to possess similar transport behavior in this

model in order to facilitate the use of a common concentration gradient, (A C/iv(t)). The

mathematical relation among Di}: , Di}- and Dix-J,- are summarized in Eq. (3.11) and

(3.12).
01%,,- = 0%,- Eq. (3.11)
and,
016,1,- = 131%,,- + 19%,. Eq. (3.12)

An explicit finite-difference scheme (2nd order Runge-Kutta method) was applied

to Eq. (3.13) — (3.17) in order to numerically-model the contact melting as a function of
time prior to the complete melting of Nb filler. The program code was written in

FORTRAN 77 and the input parameters are provided in the Appendix B.

 

DL. . AC
6x0) . -=v(t) . .At= N"T'( ] At Eq.(3.13)
NrTr NiTi CN,s w(t) L

93

 

L
ax(z)M, = v(t)Nb At = ——’fl[-’9£) At Eq. (3.14)

C3,: w(t) L
x(’ +1)NiTi — xl )NiTi +5xl )NiTi Eq- (3-15)
x(+1Nb) -xtM,() -dx(tM,) 1240.16)
w(t+l)= x(t+1)N,T, —x(t+l)Nb Eq. (3.17)

Due to the nature of the mathematical expressions, an initial braze liquid width of
zero cannot be used for the numerical calculations. Hence, a finite braze liquid width of
10 nm, a legitimately small value, was employed to serve as the initial condition. The
braze temperature for calculation was set at 1185°C and the initial half Nb braze foil

thickness of 25.5 um was used, both consistent with the experimental setup employed for

the time-evolution studies in section 3.3. The calculation employed values of C N. s, C N. 1,

C3,] and CB”, described in section 3.2. Given the high rate of change of interface

velocities during initial melting, small time steps had to be used to reduce computation
error. It was found that a computation time interval, At, of 0.002 seconds is suitable for

the nature of this numerical work. While the abovementioned parameters were fixed,
diffusion coefficients Dfihn and D119}, were adjusted to obtain an optimum fit between

predicted and experimental data. Hence, diffusivities of Ni, Ti and Nb can be inferred

from the numerical simulation.

Figure 3.10 illustrates the experimental and predicted solid-liquid interface

positions, me- and be, as a function of braze time prior to complete melting of Nb filler.

94

When DIIQi’Ti and Dillib were assigned values of 4.10 x 10'9 m2/s and 1.55 x 10-9 mZ/s,

respectively, simulation results were able to correctly replicate the trends of experimental
data. The predicted duration for complete melting of a 25.5 um-thick Nb foil is about 34
seconds, which is in excellent agreement with experimental observations. It is worth
mentioning that such similar contact melting model has been successfully utilized in the
past by [Getazheev et al. 1970] to predict the liquid diffusion coefficients in the Bi-Sn
binary eutectic system (see Chapter 1). The predicted diffusivities for Bi-Sn in
[Getazheev et al. 1970] and NiTi-Nb in this work are consistent with many pure liquid
metals studied close to their respective melting points [Vadovic et al. 1970, Roy et a1.

1988, Mei et al. 1990, Alemany et al. 1998] in terms of order of magnitude.

 

 
    
  
 
 
   
 

 

 

 

 

1.0E—04 .
Calculations based on
DNbL = 1.55 x 10'9 m2/s
8.0E-05 - L _9 2
DM'J',‘ =4.10X10 m [S-
.I SLEBOS-
Ev — Predicted NiTi-L Position
é 4.0E-05 _ —Predlcted Nb-L POSlthfl
g I Experimental NiTi-L Position
§ 2.0505 . 0 Experimental Nb-L Position
‘1: . . . . .
.3 Initial NiTi-Nb Mating Interface

0.0E+00

 

-2.0E-05 .

 

 

-4.0E-05

 

0 5 10 15 20 25 30 35
Braze Time (second)

Figure 3.10 Experimental and predicted solid/liquid interface positions during
dissolution at 1185°C. (Experimental interface positions were measured and defined
with respect to the initial NiTi-Nb mating interface.)

95

Although the simulation is able to determine D fqm , it is unable to quantify the
individual contributions of Ni and Ti diffusivities to D 11(7in . Recall that the calculation

assumed D110,- = 0114,- (see Eq. (3.11)). In such an ideal scenario, Ni and Ti will be
transported away fiom the unmelted NiTi at the same rate. Hence, the composition

across the liquid can be described by Line 1 in the 1185°C - isothermal section of the

NiTi-Ni6Nb7-Nb-Ti phase diagram (Figure 3.11). Note that Line 1 lies in the plane of

the NiTi-Nb isopleth (indicated by the dotted lines in Figure 3.11 and the inset), which
implies that the ratio of Ni to Ti will always be unity in the liquid at any position and

time.

 

NiTi

 

bcc-Nb(Ni,Ti) .
-nl ' 1 WT I ' l I

 

 

 

 

Figure 3.11 Estimated isothermal section of the NiTi-Ning7-Nb-Ti phase diagram at
1185°C. Inset: Magnified view of the liquid phase at the NiTi-Nb isopleth revealing
compositions across the liquid when Ni and Ti possess similar diffusivity (Line 1); Ni
diffuses slower than Ti (Line 2); and Ti diffuses slower than Ni (Line 3). The average

liquid composition is indicated by C L-

If D119,- qé D711,- , the flux of the slow-diffuser will be smaller than that of the fast-

diffuser and a build-up of the slower-diffusing solute near the NiTi-Liquid interface will

happen momentarily when contact melting first occurs. Assume a case where Ni is

96

diffusing more slowly than Ti. The liquid adjacent to the unmelted NiTi will build up
more Ni than Ti initially, while the liquid close to the unmelted Nb possesses less Ni.
Consequently, the composition across the liquid during this transient stage can be
described by Line 2 in Figure 3.11. The build-up of Ni near the unmelted NiTi end
results in an increase in Ni concentration gradient across the liquid. This, in turn,

increases the Ni flux, J 11",,- , towards the unmelted Nb. In other words, increase in J 11“,,-

will compensate for its lower diffusivity. Concurrently, the faster diffusing Ti will

responsd in an antagonistic manner by reducing its concentration gradient and flux, J7L~, ,

across the liquid. Assuming the solid NiTi to be perfectly equi-atomic, steady-state

occurs when J 119,. = J11;- and the composition across the liquid will be driven back to Line

1 in Figure 3.11 (this statement additionally assumes negligible solid-state diffusion of Ni

and Ti).

In contrast, if Ti is diffusing slower than Ni, the situation reverses and the
composition across the liquid during the transient stage can be described by Line 3 in

Figure 3.11. The system will then drive the liquid composition back to Line 1 by
simultaneously decreasing .1119,- and increasing J 711, to the point when J {9,- = J 71’} .
Therefore, a self-moderating mechanism exists in the contact melting process, which
prevents deviation of liquid composition from the NiTi-Nb isopleth even if D119,- ¢ Dfi.
F urtherrnore, mass-conservation dictates that the average liquid composition, C L, in the
joint (indicated by the black dot in the inset of Figure 3.11) is invariant whether

discrepancies between Ni and Ti diffusivities exist or not.

97

The melting kinetics simulation also predicted the interface velocities (the rate of
melting) for NiTi and Nb, which are shown in Figure 3.12. The predicted trends in
Figure 3.12 show that melting proceeds rapidly for both NiTi (> 20 rim/s) and Nb (> 5
um/s) during initial melting. However, the velocities attenuate rapidly to less than 5 um/s

within first 5 seconds of contact melting (or as the braze width increases).

Simulation results suggest that Nb possesses a lower rate of melting compared to
NiTi, which is attributed to its lower diffusivity with respect to the NiTi counterpart. The
discrepancy in diffusivities can be qualitatively explained by the Einstein theory on self-
diffusivity of liquid metals, which was discussed in Chapter 1. Equations (1.6) and (1.7)
state that diffusion coefficient is dependent on temperature, viscosity and atomic radius
of the diffusing species. Since diffusion of Ni, Ti and Nb occurred in the same liquid at
the sarrre temperature, the viscosity and temperature terms in Eq. (1.6) and (1.7) should
be constant. Therefore, atomic radius is the only parameter that should affect diffusivity.

Given that diffusivity is inversely proportional to atomic radius and Nb possesses larger

atomic radius than Ni or Ti [Hume-Rothery et al. 1954], 011%,],- will be greater than D1191) .

Furthermore, the Einstein theory also predicts Ni as a faster diffuser than Ti in the liquid.
Therefore, it is believed that the transient stage indicated by Line 2 in Figure 3.11 is the
likely scenario that precedes steady—state. Given that Nb is the slowest diffuser in the

liquid, it should be the rate-limiting parameter for the contact melting process.

98

Interface Velocity (mls)

 

3.0E-05

 

2.5E-05 _ —Ni'|"r Interface Velocity
--— Nb Interface Velocity

 

 

2.0E-05 -

 

 

1 .5E-05
1.0E-05 l

5.0E-06 .

 

 

0.0E+00

 

-5.0E-06 ~

 

 

FL

 

-1 .0E-05

l r l l l

0 5 10 15 . 20 25 30 35
Braze Time (second)

Figure 3.12 Predicted melt interface velocities for NiTi and Nb at 1185°C.

Upon complete melting of solid Nb, additional Nb flux will no longer be supplied

into the liquid phase and the liquid does not need to establish local equilibrium

composition at C B, 1. All liquid in the joint will now attempt to establish a uniform

equilibrium composition only at C N. 1. In other words, the liquid will attempt to achieve

zero concentration gradient. A final liquid composition of C N. 1, or zero concentration

gradient, can only be attained by continued melting of the base-NiTi, which will cease

only when the overall liquid composition reaches C N. 1. Figure 3.13 illustrates the

proposed time-evolution of the concentration profile in the liquid after the exhaustion of

Nb filler, where linear concentration gradients are also assumed.

99

     

Braze Temperature: 1185°C

t4 < t5 <

NiTi L Nb NiTi L NiTi

w(t4) ' ‘ w(ts) Wilts) Temp.,°C

Figure 3.13 Schematic diagram illustrating the evolution of a concentration gradient in
the braze liquid as a function of time after Nb exhaustion.

  

    

Atom % Nb —>

  

 

     

 

 

The numerical scheme was allowed to continue upon the complete melting of Nb
filler. Since the liquid can no longer support a local composition at C3,], Eq. (3.13) has
to be replaced by Eq. (3.18) and (3. 19) to allow the liquid to eventually achieve a uniform
composition C N. 1. Equations (3.18) and (3.19) describe the enrichment of liquid with Ni

and Ti through further melting of NiTi, which causes an attenuation of the concentration

gradient as a function of time. When the concentration gradient approaches zero (that is,

the overall liquid attains C N. 1 composition), there is no driving force for melting of NiTi.

Hence, the numerical scheme would be terminated at this point. The program code is

 

 

provided in Appendix C.
L AC—C(t) r
C(t+1)=DNi,Ti 7T At m Eq..(318)
L
D110: Ti AC—C(t+l)
&tl.-.= ’ A E.3.l9
( + )NlTl CN,s W(t+l) L t q ( )

100

 

1 .4E-04 l

5 + 4

1.0E-04 ~

 

    
    

 

Ta

— Predicted
I Experimental

 

 

 

Eutectic Liquid Half-Width (m)

Before Nb
Exhaustion

0.0E-I-00 i r
0 10 20 30 40 50 60 70 80 90 100 110 120
Braze Time (second)

After Nb Exhaustion

_V

 

A

 

r

 

 

 

q
—:

l l l l r T I

Figure 3.14 Experimental and predicted liquid half-width as a function of time at braze
temperature of 1185°C.

Figure 3.14 shows the experimental and predicted evolution of liquid half-width
from 0 to 115 seconds. The predicted trend in Figure 3.14 suggests that the liquid

continued to widen by approximately 5 pm after complete melting of Nb filler in order

for the liquid to achieve uniform composition at C N, 1. According to the numerical

scheme, this process took less than 15 seconds to complete, which is reasonably
consistent with experimental results. The final liquid half-width of 126 run was achieved

within 50 seconds from the point of initial melting.

Recall that the contact melting model hypothesized that melting is rate-controlled

only by solute diffusion in the liquid. The model did not address possible contributions

101

of latent heat effects to the melting rate. Hence, heat conduction calculations, based on
the half-space specimen configuration, were performed to determine if heat transfer plays
a rate-lirniting role in contact melting. Initial conditions for the calculations were such
that the NiTi surface is constantly at 1 185°C, while the NiTi-Nb interface had attained a
temperature of 1170°C and the system had just passed the incubation period for melting.
That is, any additional heat flux across the NiTi-Nb interface will initiate and cause
subsequent melting without raising the liquid and the solid Nb temperature. The
calculations assumed steady-state heat flow where a linear thermal gradient existed in the
NiTi and perfect mixing of solutes in the liquid at all times. Figure 3.15 illustrates the
temperature profile across the half-space configuration of a braze joint at initial condition

and afier some time, 1.

Temperature - -_
NrTr+ Nb Interface

    

Tswf= v
1185°C
Tint _
f 1170°C
3.175 mm 255 pm
Tswf= J
1185°C

 

Distance

Figure 3.15 Schematic diagram showing the temperature profile across a half-joint at
initial condition and after some time, t, if melting is rate-controlled by heat transfer.

102

The heat conduction problem can be mathematically expressed by Eq. (3.20),
AQ(t) = kA[£]At Eq. (3.20)
x(t)

where k is the thermal conductivity of NiTi, AQ(t) is the change in heat at time t, At is the
time interval, x(t) is the remaining thickness of base-NiTi alloy at time t, A is the NiTi-
Liquid interfacial area and AT is the temperature difference between exposed surface of

the base-NiTi alloy and the NiTi-Nb interface.

The remaining thickness of NiTi after At can be determined by Eq. (3.21),

_x _AQ(t)
x(t+1)— (z) pAL Eq.(3.21)

where p is the mass density of NiTi and L is the latent heat of fusion of the braze liquid.

The thickness of melted NiTi was predicted as a function of time (after the
incubation period) using an explicit finite-difference scheme (2nd order Runge-Kutta

method) applied to Eq. (3.20) and (3.21). Since there are no data available on the latent
heat of fusion for the braze liquid, the calculation was based on the latent heat of fusion
of NiTi, L = 357 J/g [Yung et al. 2005]. Results derived from the latent heats of fusion of
500 J/ g and 1000 J/ g were also presented as reference. The latter value represents an

exceptionally high latent heat of fusion. Therefore, results predicted from such a high

103

latent heat of fusion will serve as a lower-bound to the NiTi melting rates. The
simulation was terminated when approximately 96 pm of NiTi had melted. This
thickness was deliberately chosen because it corresponds to the calculated thickness of
melted NiTi in the diffusion-controlled model upon complete melting of a 25.5 pm thick
Nb foil, and is also reasonably consistent with experimental observation. Time interval,

At = 2 us, was used. The program code was written in FORTRAN 77 and is provided in

 

 

 

 

 

 

 

 

 

Appendix D.
1.0E-04
------ -:--------------- ---------------------- --.r---
O

8.0E-05 ~ g: ‘
e s
E f
z 6 0305 _ ’ Thickness of melted NlT i
E ' 5 associated with complete
it melting of Nb
2 3
3 405-05 - 3:
0
I:
‘6 f
E o Latent Heat = 357 Jlg
.—

2.0E-05 ~ I Latent Heat = 500 J/g

A Latent Heat = 1000 J/g
o.oe+00 . . . . . . .
0 100 200 300 400 500 600 700 800

Time (micro-sec)

Figure 3.16 Predicted thickness of melted NiTi alloy as a function of time after the
incubation period.

Figure 3.16 shows the thickness of melted NiTi alloy as a function of time after
incubation period. The predicted results revealed that NiTi melts at an almost linear rate

for all latent heats of fusion. When the latent heat of fusion of NiTi was employed for

104

calculations, the time taken to melt 96 pm of NiTi was predicted to be 112 us. Figure
3.16 shows that despite a longer required duration when a high latent heat of fusion of
1000 J/g was assumed, the time taken to melt 96 pm of NiTi is still less than 1 ms, which
is about 5 orders of magnitude smaller than what was experimentally observed. It can

safely be concluded therefore, that latent heat effects are negligible in the present context.

3.5 Solidification

The discussion will now focus on the solidification products originating from the
braze process. Specifically, the as-brazed and annealed microstructure of braze joints
will be discussed in detail. Before the end of this section, the effects of surface

preparation and vacuum conditions on joint morphology will also be considered.

3.5.1 As-Brazed Microstructures

A higher-magnification SEM micrograph of Figure 3.5a is shown in Figure 3.17.
It revealed microstructural details of a typical joint that was solidified before complete
melting of the Nb filler. Figure 3.17 reveals the presence of pro-eutectic Nb, which has
nucleated against the unmelted Nb foil during cooling. The pro-eutectic Nb phase
possessed a nearly planar solidification morphology. Likewise, the pro-eutectic NiTi
phase was also observed to nucleate against the base-NiTi, and also possessed a
predominantly planar solidification front (Figures 3.5b — (1). However, it occasionally
assumed a cellular morphology similar to those observed in Figure 3.17. The eutectic

mixture forms the main constituent in the joint and well-defined eutectic impingement

105

boundaries only developed beyond 8 seconds from initial melting (Figures 3.5c — d). The

average eutectic grain size is approximately 60 um.

Pure NiTi
Pro-eutectic NiTi

NiTi-Nb Eutectic
Pro-eutectic Nb

Unmelted Nb

 

Figure 3.17 Typical braze microstructure solidified before exhaustion of solid Nb
(contact melting at 1 185°C).

X-ray fluorescence studies revealed that the pro-eutectic NiTi phase contains
about 4 at.% Nb with a corresponding decrease in Ti composition (Table 3.3). This
suggests that Nb tends to substitute for Ti in the NiTi phase. Elemental XRF mapping of
the braze area, shown in Figure 3.18, indicates presence of Nb and reduced Ti Ka
intensity in the pro-eutectic NiTi while Ni [(01 intensity remains constant across the base-
NiTi and pro-eutectic NiTi. This reinforces the previous deduction that solutionizing of

Nb in pro-eutectic NiTi occurs by preferential substitution for Ti lattice sites in NiTi.

106

Due to spatial resolution limits of XRF, the composition of pro-eutectic Nb could not be

determined.

 

Figure 3.1
Ka and (d) Ni Ka intensities originating from base-NiTi, pro-eutectic NiTi and eutectic
phases.

Table 3.3 X-ray fluorescence analysis on selected spots in Figure 3.18.
Spot Nb 41t%) Ni (at%) Ti (at%)
A 0

 

51 49
B 5 52 43
C 4 51 45

 

107

Upon the complete melting of Nb filler, the pro-eutectic NiTi phase appeared to

change from a planar/cellular to a dendritic morphology (Figure 3.6). Occasional
facetted precipitates, later identified as the TizNi-type phase (or its oxide, Ti4Ni20), were

observed within the eutectic micro-constituents in experiments involving complete
melting of the Nb filler. Their grain size ranged from 5 to 20 um, although there were
noticeably more grains near 20 um in size when joints were subjected to longer braze
durations. Figure 3.19a shows the typical morphology and size of these facetted

precipitates in joints brazed for 3600 seconds.

6.; {1 (hi

fe— \‘ 1 1 r, ‘3‘
. ' 0» ~.._
I . "

   

   

Figure 3.19 SEM micrographs showing (a) facetted TizNi-type precipitates (indicated
by ‘A’), (b) fine Nb—rich precipitates decorating sub—grain boundaries of a pro-eutectic
NiTi dendrite (indicated by ‘B’).

\

Bright, spheroidal precipitates can also be occasionally observed forming
networks within the pro-eutectic NiTi dendrites (Figure 3.1%). Given that their SE
intensities closely resemble those of the eutectic-Nb phase, these fine precipitates are

likely to be Nb-rich and appear to have formed along sub-grain boundaries in the pro-

108

eutectic NiTi dendrites during cool-down from braze temperature. Figure 3.20a and b
shows SE and back-scatter electron (BSE) micrographs of sub-grain boundaries of the
same pro-eutectic NiTi dendrite. The sub-grain boundaries are visible possibly due to
etching effects. These sub-grain boundaries appeared dark in both micrographs either
because the Nb-rich precipitates may be too fine to be resolved by SEM, or because these
sub-grain boundaries were not decorated with Nb precipitates. The preferential
precipitation of Nb-rich phase along the sub-grain boundaries of the dendrites can be
attributed to Nb supersaturation in the pro-eutectic NiTi phase at lower temperatures.
Figure 3.1% and Figure 3.20 show abrupt changes in SE intensities near the dendrite
borders (bright near the perimeter, but dark in the bulk), suggesting higher Nb-
concentrations at the boundaries of these dendrites. Such phenomenon is termed coring,

which is due to slow solid-state diffusion associated with non-equilibrium solidification.

  

‘ .
Figure 3.20 SEM micrographs of a
and (c) BSE modes.

However, the XRF line profile in Figure 3.21 revealed that the change in Nb

composition across a perceived cored dendrite is rather gradual. The perceived cored

109

region possessed a maximum Nb concentration of 13 at.% Nb, which is comparable to
the Nb content in the bulk of the dendrite in Figure 3.7b. The prominent decrease in Nb
content actually occurred away from the perceived coring. Hence, the abrupt change in

SE intensities may, in fact, also be an artifact of the etching process.

13 at.% Nb

, ..I '1 -'.,,‘ ‘

q 0.11% Nb

XR F Linc Profile

 

Figure 3.21 X-ray fluorescence line profile of Nb La across the perceived cored pro-
eutectic NiTi phase and the base—NiTi alloy.

The transition of the character of the pro-eutectic NiTi solidification fiont from

planar to dendritic can be explained qualitatively with the aid of Figure 3.22. During the
melting process, the liquid possesses an average composition, CL, of 25 at.% Nb. If the

braze joint was cooled from 1185°C at this juncture, the pro-eutectic NiTi would solidify

against the base-NiTi, rejecting excess Nb into the liquid immediately ahead. The liquid

ahead of the solid-liquid interface may build up sufficient Nb to yield a composition C5,

110

while liquid far away from the solid-liquid interface will have composition of 25 at.% Nb.

Consequently, the liquid must have a composition-distance profile indicated as Cbefore in

Figure 3.22. The corresponding liquidus temperature-distance profile is shown as T before

in Figure 3.22. The reason for the planar solidification morphology may be attributed to

a thermal gradient in the liquid phase being tangential to the slope of Tbefore at the solid-

liquid interface. Hence, the liquid does not experience constitutional supercooling and

solidification can proceed in a planar fashion.

 

 
  
  
 

L + bcc-Nb

/ Liquid

 

 

3.,"duraj

 

 

 

 

 

 

/ -. ------------- 1185°C
W - - TE 1170°C ‘- + T'N' 1170°C
‘E‘ WWW W WW3” l 1 TiNi + bcc-Nb
I
i CE 126 at% Nb \ After Nb Before Nb
Atom % Nb -—) Exhaustion Exhaustion
H
u, 5.. CE
'0 ‘33 Cbefore
8 .5 .\°
.: N ""
< g3 Cafter
"' Constitutional

,Thermal Gradient
Tafter

Supercooling

 

 

Tbefore

 

Liquidus
Temperature
Profile (°C)

 

Corresponding Composition Profile

l Distance from SIL Interface

Figure 3.22 Qualitative model accounting for the pro-eutectic NiTi solidification
mm‘l3hology at different stages of brazing.

111

Upon complete melting of the Nb filler, the liquid now maintains equilibrium

composition at C N. 1 of Figure 3.1, which is leaner in Nb content than the previous case.

When the liquid is allowed to cool, the liquid ahead of the solidified pro-eutectic NiTi

will also build up Nb. Again, the liquid ahead of the solid-liquid interface may have as

high of a Nb concentration as C 5, while liquid far away from the solid-liquid interface
will have composition of C N. 1. The resulting composition- and liquidus temperature-
distance profiles are indicated by Cafte, and Tafle, in Figure 3.22, respectively.
Comparison between T before and T after shows that the latter has a steeper slope at the

solid-liquid interface. If the thermal gradient in T before is also imposed on T after (similar

cooling regime in both cases), a finite distance of the liquid ahead of the solid-liquid
interface will have lower temperature than the liquidus temperature of those
corresponding liquid. Hence, constitutional supercooling occurs, causing instability in
the pro-eutectic NiTi solidification front and eventual dendrite formation [Porter et. al.

2001].

The eutectic-Nb phase (appears bright in Figure 3.23) was observed to possess a
predominantly lamellar morphology. However, rod-like eutectic-Nb was also
occasionally encountered. The average eutectic spacings of lamellae and rods were
determined to be about 0.4 and 0.6 pm, respectively. Stereology performed on Figure
3.23 and other similar micrographs indicates an approximate 30 vol.% occupancy by

eutectic-Nb in the eutectic micro-constituents. If one assumes the solubility limits of

112

NiTi and Nb at eutectic solidification temperature are approximately CN,s and CB,s:

respectively, Lever-rule suggests that this volume fraction of eutectic-Nb should
correspond to a quasibinary eutectic point at 23 at.% Nb. This value puts it close to the
literature-reported value of 26 at.% Nb [Prima et al. 1995]. The morphology of the

eutectic micro-constituents will be discussed in a later section.

 

  
 
 
   
  
   

3..

 
 
 

4. ,1:
l‘mvt

   

 
      
   

l 'l
r.
’1,

  

            
   
 
        

... \ 9.." :T:\§§“" _
. .‘T "a? '._~.\¢=\\ ~\ ..
, . - - ...,
l 7‘ ..(FL'.-‘-W.~:‘i~§“3‘ \\ a)", H at
W: 4.“. '-.\~' .I‘ . .“ ‘~“‘. I. I N N\\
J/ (“'1 W‘ ‘\“'\“‘\Kl' MW e
\ ff." o"\-"’h“‘< \ “‘8 ’4 I“
‘;~ ‘..//'r."."»-r,\..;~‘>\1:'°~.\ .4.- llr )4 , ..
‘-"""r,r\-V“\~..\‘\\3\: \ ~¢>~..$. l’ 'r *

 

Figure 3.23 Eutectrc-Nb phase possessing lamellar and rod-like morphologies.

Braze joints prepared at 1185°C for 360 seconds were investigated by TEM and
results are shown in Figure 3.24 and 3.25. In-plane eutectic-Nb rods (darker contrast) in
the matrix of eutectic-NiTi (lighter contrast) can be clearly seen in Figure 3.24. Selected-
area diffraction of the eutectic-NiTi showed diffraction spots with a diffuse ring as

background. The ring indicates amorphous or low crystallinity of the eutectic-NiTi from

113

ion-milling. Indexing the diffraction spots revealed BC C-type crystal structure with two
zone-axes, possibly due to sampling of adjacent grains simultaneously by the aperture.

The first zone-axis was determined to be [011] with weak ordered-32 superlattice

reflections, while the second zone-axis was identified to be [133]. Again, the weak
superlattice reflections are indications of poor ordering, attributed to the moderately fast
cooling upon termination of the braze process. No martensite phase was observed in the
eutectic-NiTi. Note that these joint specimens were never subjected to temperatures
lower than room temperature after they were cooled fi'om 1185°C. Hence, the absence of
martensite suggests that the martensite start temperature may be lower than room

temperature.

Ordered—[1‘3

1

  
 

e [011] - Zone

0 Superlattice
Spots of [011] -
Zone

Figure 3.24 TEM micrograph of braze joints processed at 1185°C for 360 seconds
showing in—plane eutectic-Nb rods and SAD pattern of the eutectic-NiTi. (Diffraction
originating from bottom-right of micrograph.)

114

The selected-area diffraction from the out-of-plane eutectic-Nb rod in Figure 3.25
showed strong diffraction spots that are associated with disordered-BCC [012]-zone axis.
No amorphous ring was observed. Hence, the degree of crystallinity in the eutectic-Nb
phase was high despite fast cooling, which is consistent with its purity and disordered-

BCC equilibrium state.

Disordered-1366

 

.0. -o
121 200
9 o e-
121 000 121
. -.-
200 121

E i . [012] -Zone

Figure 3.25 TEM micrograph of braze joints processed at 1185°C for 360 seconds
showing eutectic phase and SAD pattern of the out-of-plane eutectic-Nb rod.
(Diffraction originating from indicated eutectic-Nb rod.)

  

3.5.2 Annealed Microstructures

Brazed joints processed at 1185°C for 1200 seconds were subjected to annealing
temperatures from 300°C to 920°C in air followed by water-quench. At lower
temperatures, SEM did not reveal any distinctive changes in joint microstructures.
Notable changes only occurred when annealing temperatures were increased to 800°C

and beyond. When annealed at 800°C for 29 hours, Nb-rich precipitate networks in pro-

115

eutectic NiTi dendrites became bigger and more extensive than those observed in the as-
brazed specimens (not shown in Figure 3.26a but are similar to those in Figure 3.26b).
However, there were no morphological changes to the eutectic micro-constituents and
facetted grains (Figure 3.26a). The eutectic mixture still remains morphologically stable
when subjected to annealing at 800°C for 104 hours, but the TizNi-type precipitates
appeared to be in the process of coalescence to form larger grains (Figure 3.26b).
Experimental results suggest that the coalescence has to be a thermally-activated process

and one can expect to observe larger TizNi-type grains if either annealing duration was

increased further or higher annealing temperatures were implemented.

M

‘2

' \ 3""
' ;““‘-.\\\'\‘i§\\.\.l

: 2 .\)\r;\\\“il
: {Wat \l.\\\\\\ll
. _ , : 800°C 29 Hr ,1 _
Figure 3.26 Microstructures of joints subjected to post-braze anneal at 800°C for (a)
29 hours; and (b) 104 hours.

     

, ‘i

Figures 3.27a and b show the microstructures of a braze joint subjected to
annealing at 920°C for 20 hours. Significant morphological change was evident in the

eutectic micro-constituents. The eutectic-Nb phase now appears to transit from lamellar

116

and rod-like morphology (Figure 3.27a) into final spheroidal form (Figure 3.27b). Recall
that in as-brazed joints that the eutectic-Nb lamellae and rods had average spacings of 0.4
and 0.6 pm, respectively. However, the spheroidized eutectic has an average spacing of
0.9 pm. Similar to the coalescence of TizNi-type grains at 800°C, the spheroidization of
the eutectic phase has to be a thermally-activated process and the driving force for such
morphological transition is the reduction of eutectic-NiTi/eutectic-Nb interfacial

boundary energy (assuming isotropic surface energies for both phases in the eutectic)

[Flemings 1974].

5; .L l cl .
. (a)
VH/' ”:l\u.|I‘: “'3. '5! a.
' I .”

lliiilll,

if!

j. "I”, ,’",n
(I
,3 A, ., //,

'l

 

‘ - l - I ..‘ q 5- 1 7‘ . I I , ., i i i. -\
Figure 3. 27 SEM micrographs showing (a) eutectic-Nb rods undergoing
spheroidization and (b) completely spheroidized eutectic-Nb in braze joints subjected to
post-braze anneal at 920°C for 20 hours.

Besides morphological changes in the eutectic micro-constituents, bigger and
more extensive networks of Nb-rich precipitates were also observed in the pro-eutectic

NiTi dendrites in Figure 3.27b than those seen in Figure 3.26b. This is not surprising as

the microstructure approaches equilibrium from the annealing process. The TizNi-type

117

grains, originally seen in prolonged braze specimens, were no longer present when

annealed at 900°C and water-quenched. This may be attributed to thermal decomposition

and solutionizing of the TigNi-type precipitates into the eutectic matrix.

Superlattice
Reflections

[011] — Zone

 

Superlattice
Reflections

   

[001] — Zone
Figure 3.28 TEM micrograph from a braze joint subjected to post-braze anneal at
920°C for 20 hours and water-quenched showing a spheroidized eutectic-Nb particle in
the eutectic-NiTi matrix, together with two SAD patterns fi'om the eutectic-NiTi. (SAD 2
was taken from eutectic—NiTi at another location within the same specimen.) Positions A
and B were sites for XRF analysis.

Table 3.4 X—ray fluorescence analysis on selected spots in Figure 3.28.
Spot Nb (at%) Ni (at%) Ti (at%)
A 9 49 42
B 76 11 13

 

 

Selected-area diffraction obtained from the 920°C-annealed eutectic-NiTi phase

in the TEM micrograph (SAD l of Figure 3.28) showed strong diffraction spots and

118

ordered-82 superlattice reflections from the [011]-zone. Diffraction obtained from
eutectic-NiTi at a different location (SAD 2) also revealed strong superlattice ordering in
the [001]-zone. Furthermore, distinct lattice fringes were observed in the high-resolution
TEM micrograph of the eutectic-NiTi in the same specimen (Figure 3.29). Fast-Fourier
Transform analysis shows that the grain was oriented along the 82 [l22]-zone axis.
Martensite was again not observed in the eutectic-NiTi phase. From Figure 3.28 and 3.29,

it can be deduced that post-braze annealing is necessary to fully restore BZ ordering in the

eutectic-NiTi phase in braze joints.

     

, , . , _ 4 82 [T22] - Zone

Figure 3.29 High-resolution TEM micrograph of eutectic-NiTi revealing lattice fringes
oriented on the B2 [l22]—zone axis. Top-right image schematically shows the real-space
lattice arrangement (bright spheres: Ni atoms; dark spheres: Ti atoms). Bottom-right
image is the FFT pattern of the selected area showing periodicity in reciprocal space.
(F FT pattern originated from the indicated area of the micrograph.)

119

X-ray fluorescence was also performed on the phases in Figure 3.28 and results
were summarized in Table 3.4. The higher spatial resolution associated with XRF in
TEM allows the composition of eutectic-Nb to be determined and was found to possess
approximately 10 at. % Ni and Ti solutionized in Nb. The TEM-XRF revealed eutectic-

NiTi compositions in close agreement with the SEM-XRF results in Table 3.3.

3.5.3 Effects of Surface Preparation and Vacuum Contaminants on Joint

M icrostructures

Besides post-braze anneal temperatures, contamination is also a critical factor that
is known to affect braze joint properties (mechanical properties will be discussed in detail
in Chapter 4). Contaminations usually originate from inadequate surface cleaning and
non-ideal braze ambient conditions. Hence, this section will specifically address the
effects of surface preparations and vacuum conditions on joint microstructures and
morphology. This study involved 3 controlled scenarios: Processing braze joints at
1185°C for 120 seconds under (1) good vacuum and using clean NiTi surfaces; (2) poor
vacuum and using clean NiTi surfaces; and (3) poor vacuum and oxidized NiTi surfaces.

Niobium foils were identically cleaned in each case. The definition of ‘good vacuum’ in

this study implies vacuum in the range of 10'6 Torr, while ‘poor vacuum’ implies

vacuum in the range of 10-4 Torr. These selected states of vacuum will provide

distinctively different oxygen partial pressures in the braze chamber. The oxidized NiTi
surfaces were obtained by first chemically removing native oxides from the NiTi
specimens as described in Chapter 2, followed by heating them in air at 350°C for 1 hour.

In this way, the mating surfaces were oxidized in a consistent and controlled fashion.

120

Figures 3.30a — c shows the SEM micrographs of braze joints processed under the 3

described scenarios, respectively.

Scenario 1

\

._;2

 

Scenario 3

Figure 3.30 SEM micrographs of braze joints processed at 1185°C for 120 seconds
under (a) scenario 1; (b) scenario 2; and (c) scenario 3. (Scenario 1: Good vacuum, Clean
NiTi; Scenario 2: Poor vacuum, Clean NiTi; Scenario 3: Poor vacuum, Oxidized NiTi)

Experimental results show that brazing under good vacuum with clean mating

NiTi surfaces produce joints with no TizNi-type precipitates and porosity (Figure 3.30a).

121

Despite clean surfaces, the number density of TizNi-type precipitates appears to increase

when poor vacuum conditions were employed (Figure 3.30b). When brazing under poor

vacuum with oxidized NiTi surfaces (Figure 3.30c), TizNi-type grains and pores form

even more extensively in the joint. The significant increase in number density of TizNi-

type precipitates as oxygen contamination worsens is consistent with the known ability of
oxygen to stabilize their formation [Nevitt 1960, Melton 1990]. This comparative study
highlights the importance of performing reactive eutectic brazing of NiTi alloys using

clean mating surfaces under reasonably high vacuum conditions.

3.6 The Origin of the Perfect NiTi-Nb Quasibinary Eutectic Isopleth

At the beginning of this chapter, discussions on contact melting between NiTi and
Nb were based on the assumption of a perfect NiTi-Nb quasibinary eutectic isopleth in
the ternary system. However, the perfect quasibinary eutectic iSOpleth may not be an
exact representation of the NiTi-Nb pseudobinary section. The literature suggests that
the actual NiTi-Nb isopleth is more complex and may contain three-phase fields.
Therefore, this section aims to address the origins of both possibilities for the isopleth
through a qualitative analysis of the ternary phase equilibria. The literature strongly
indicates that the Ni-Ti-Nb ternary phase diagram contains a NiTi-Nb quasibinary
eutectic point close to 38Ni—36Ti-26Nb (at.%) at 1170°C and a NiTi + Nb(Ni,Ti) tie-line
at 900°C has also been proposed [Prima et al. 1995] (see Figure 3.31). Although
majority of the liquidus projections recently calculated by Matsumoto and coworkers

[Matsumoto et al. 2005] is consistent with existing work, they did not take into account

122

the existence of the quasibinary eutectic point. In fact, Matsumoto’s calculated liquidus
projection at the quasibinary eutectic point violates the Alkarnade’s principle [Hillert
1998], which states that a temperature-maximum should occur in the liquidus line that
demarcates two congruently-melting phases. In this case, the two congruently-melting
phases are NiTi and Nb. Hence, a new liquidus projection of the ternary system is
proposed to accommodate the existence of the quasibinary eutectic point as well as the
Alkarnade’s principle (Figure 3.31). Therefore, the quasibinary eutectic point, which
should lie close to the tie-line in Figure 3.31, will be at the temperature-maximum of the

liquidus line.

. NiTi-Nb Quasibinary
eutectic point

    
   
  
  

- - NITi-Nb Tie-Line at 900°C

I Experimentally— _ X?
determined eutectic- N'3TI
NiTi and Nb
compositions

Ti

Figure 3.31 Proposed liquidus projections of the Ni-Ti-Nb ternary system with the
reported 900°C isothermal section. (The literature-reported NiTi-Nb quasibinary eutectic
point and NiTi + Nb(Ni,Ti) tie-line at 900°C (close to the quasibinary eutectic point)
[Prima et al. 1995]; and experimentally-detennined pro-eutectic NiTi and Nb
compositions (Table 3.4) are indicated.)

123

Compositions of eutectic NiTi and Nb equilibrated at 920°C, as determined by
TEM-XRF (summarized in Table 3.2), are also indicated in Figure 3.31. The TEM-XRF
results are in fairly good agreement with the literature-reported NiTi + Nb(Ni,Ti) tie-line
at 900°C [Prima et a1. 1995], which confirms the validity of phase field positions close to
the NiTi-Nb isopleth and will, later, aid in the determination of the true NiTi-Nb isopleth.

Data fi'om [Prima et al. 1995, Yang et al. 2000, Matsumoto et al. 2005] suggest that there

are two four-phase Class-II reaction planes, U1 and U 2, close to the NiTi-Nb quasibinary

eutectic point (Figure 3.32). U] exist at about 950°C and involves the reaction:

NiTi + Lia—33% Nb(Ni, Ti) + TizNi ; while U2 occurs close to 1140°C, and

involves the reaction: Nb(Ni,T i) + Lia—395$) NiT i + X5 .

Centigrade .
2500 Four-Phase Class II reactions planes

with common NiTi + Nb(Ni,Ti) + Liquid

2°00 phase fields.

  
    

1500

  
 
    

L

Kfir U2: 9500c

Nb (Ni,TI)

 

'leNl

Figure 3.32 Four-phase Class-H reaction planes (U 1 and U 2) in close proximity to the
NiTi-Nb quasibinary eutectic point.

124

There are NiTi + Nb(Ni,Ti) + Liquid phase fields in each reaction plane. These
will converge and degenerate into the NiTi + Nb(Ni,Ti) tie-line as the temperature
approaches 1170°C. For clarity, Figure 3.33 only illustrates how the NiTi + Nb(Ni,Ti) +
Liquid phase triangle of the Ti-rich reaction plane degenerates into the NiTi + Nb(Ni,Ti)
tie-line. The Ni-rich counterpart follows likewise. The NiTi + Nb(Ni,Ti) tie-line at

1170°C forms the reaction horizontal for the perfect quasibinary eutectic isopleth.

NiTi + Nb(Ni,Ti) + Liquid
phase fields

'U2: 1140 °C

 

/' ' '
' /
Ll " ~
g ." _ . ..u-" , ”4’ I
- , a 1;... .{s

NiTI

U1:950 Oc/f

 

 

 

 

 

Ti
Figure 3.33 NiTi + Nb(Ni,Ti) + Liquid phase fields from U] and U 2 degenerating into

a single NiTi + Nb(Ni,Ti) tie-line at 1170°C, forming the reaction horizontal for the
perfect NiTi-Nb quasibinary isopleth.

125

The NiTi + Nb(Ni,Ti) tie-line at 900°C has been reported by Prima and coworkers
[Prima et al. 1995] to be off the plane of the NiTi-Nb isopleth (see Figure 3.31). If this is

the case, a perfect quasibinary eutectic diagram would not be a true representation of the

NiTi-Nb isopleth. Furthermore, the NiTi + Nb(Ni,Ti) + X5 phase field appears to

intersect the NiTi-Nb isopleth. Therefore, the NiTi-Nb isopleth, after taking past work

into consideration, is proposed to take on the form shown in Figure 3.34. Note that the

X5 + Ni7Nb6 + Nb(Ni,Ti) phase field was assumed not to cross the isopleth.

 

2500

Liquid

  

2000-
‘ bcc-Nb+L

 

1500-

Temperature, °C

Nm :
\

 

 

1000«

 

. Ni‘l”:+bcc-Nb
500 . . . . . . . . .
NiTi 20 40 60 80 Nb
Atom % Nb
Figure 3.34 The proposed NiTi-Nb isopleth of the Ni-Ti-Nb ternary system.

 

 

 

Given its close proximity with the NiTi + Nb(Ni,Ti) tie-line, the NiTi-Nb isopleth

possesses a high degree of similarity with the perfect quasibinary eutectic isopleth with

126

exceptions at the Nb-rich end. It is because of its similarity with the true isopleth and its
simplicity that the perfect quasibinary eutectic iSOpleth was selected as the basis for
previous contact melting discussions. A noteworthy feature to highlight in Figure 3.34 is
the presence of a narrow NiTi + Nb(Ni,Ti) + Liquid phase field between the NiTi +
Liquid and NiTi + Nb(Ni,Ti) phase fields, and it is expected to exist within an estimated
range from 1130°C to 1170°C. If this is true, the presence of this NiTi + Nb(Ni,Ti) +
Liquid phase field on the true NiTi-Nb isopleth suggests that contact melting between

NiTi and Nb can actually occur between 1130°C to 1170°C.

Indeed, contact melting has been observed in this study at temperatures as low as

1155°C. Figure 3.35 shows the microstructure of a joint processed at 1155°C for 120

seconds, revealing pro-eutectic NiTi, eutectic and a high number density of TizNi-type

precipitates. Some of these precipitates (Sites A, B and C in Figure 3.35) were analyzed

by XRF and their compositions are tabulated in Table 3.5. The reason for the

precipitation of these TizNi-type grains will be discussed in the next section.

127

 

Figure 3.35 SEM micrographs of joints brazed at 1155°C showing coexistence of pro-
eutectic NiTi, eutectic micro-constituents and TizNi-type phase.

Table 3.5 X-ray fluorescence analysis on selected spots in Figure 3.35.
Spot Nb (at%) Ni (at%) Ti (at%)
37 52

 

A 11
B 13 36 51
C 12 36 52

 

3.7 Crystallography and Phase Equilibria of TizNi

Throughout the chapter, facetted precipitates were fiequently observed in joints
subjected to (l) prolong braze durations; (2) high degree of oxygen contamination; and (3)
low braze temperature (1155°C). In section 3.4, these grains were shown to coarsen at
800°C, but dissolved at 920°C. Hence, this section is dedicated to discussion on the

crystallography and phase equilibria of these facetted precipitates.

128

X-ray fluorescence was performed on the facetted precipitates in Figure 3.35.

Results revealed that they contain almost 50 at.% Ti, but only have about 35 at.% Ni

while the remaining composition is made up of Nb (Table 3.5). Since it has been

demonstrated (Table 3.3) that Nb tends to substitute for Ti, the facetted precipitates

should be of the TizNi-type phase with a more precise chemical formula of (Ti,Nb)2Ni.

The (Ti,Nb)2Ni precipitates were also studied by TEM. A micrograph and selected-area

diffraction of a Ti-rich grain are given in Figure 3.36, which revealed the F CC-type

crystal structure and the diffraction spots were associated with the [011]-zone. The

chemical composition of the (Ti,Nb)2Ni precipitate was determined by TEM-XRF to be

39Ni-58Ti-3Nb, which is in reasonably close agreement with results from SEM-XRF

(Table 3.5).

(TLN b);Ni
Precipitate

 

 

FCC-Tpr

- 11
111
. O

1
oéoo

O
2000 - -
- 000 0111
1110

[011] - Zone

Figure 3.36 TEM micrograph of braze joints processed at 1185°C for 360 seconds

showing a (Ti,Nb)2Ni precipitate and its SAD pattern.

129

Recall that oxygen contamination tends to encourage precipitation of the TizNi-
type phase (see Figure 3.30). The catalyzing role of oxygen can be attributed to the
solubility of oxygen in (Ti,Nb)2Ni. According to data from [Nevitt 1960, Gupta 1991 ,

Prima et al. 1995, Yang et al. 2000, Chuprina et al. 2002], the 900°C isothermal

tetrahedron of the quaternary Ni-Ti-Nb-O system was constructed (Figure 3.37) and it

indicates that up to approximately 15 at.% oxygen should be soluble in (T i,Nb)2Ni,

giving rise to (Ti,Nb)4Ni20 formation.

  
    
  

A g. x
I...

‘
/ J

\
.x.
7 ,/

\\
/
'
Z
W
Z
U'

1.. ‘-‘.

“ '
_.5_. :'\
. ’4’ ~<-..___‘< | , t
I ', r ‘7{ \ ‘.
1 . 4 a. ‘ .
f \ ‘ yr m 1:
' \
, ,IJ’ J \
IZ
—
M”,
” ”'7

/: " .‘
_' I
I ‘ '
I
.
n/ x/ '
. I
. , I’.-'
’-,I It.
I ’ r ’
av‘

_:i

:5;
i“ ‘F‘

 

" /‘ \j

 

     

 

, I “in—--...L
/ bcc-Nb(Ni,TI) M"
T I l r l I I I 1 I r I I I f r I l I I
at% Nb
Figure 3.37 The 900°C isothermal tetrahedron of the quaternary Ni-Ti-Nb-O system
[Nevitt 1960, Gupta 1991, Prima et al. 1995, Yang et al. 2000, Chuprina et al. 2002].

 

.
X ‘
__—.__ . _—__ \
..__ --.—.... ._-_._-__ __ ____
._ ,

 

 

n’

130

The ability of oxygen to be incorporated into the TizNi lattice can also be
appreciated through the crystal structures of TizNi and Ti4Ni20. Figure 3.38 illustrates

the unit cells of TizNi and Ti4Ni20. The former has 96 atoms per unit cell, while the

latter has 112 atoms per unit cell. Both phases have cubic structures and share the same
space group of Fd3m [Mueller et al. 1963], making them crystallographically very similar,
and almost indistinguishable by diffraction techniques. The calculated powder

diffraction spectra of both species in Figure 3.38 revealed that the three strongest

 

reflections of both phases are identical. TizNi can be transformed into Ti4Ni20 by

addition of 16 oxygen atoms to all of its unoccupied tetrahedral interstitial sites of

alternate sub-cells. However, non-stoichiometric Ti4Ni20 or sub-oxides can also form.

In other words, it is not necessary for each unit cell to have full occupancy of 16 oxygen

atoms.

 

 

 

 

 

 

 

    
  

 

O .
' 96 Moms per Unit Cell
4 Ni -Tetrahedrons (Sub-Cell Body-Center)

' Ni ~Tetrahedrons (Sub-Cell Comer)

9 9 T1 - Octahedrons (Sub-Cell Body-Center)
' . Ti - Octahedrons (Sub-Cell Comer)

. T1 Atoms at Tetrahedral Interstitial Sites

04'

\Nfi

Figure 3.38 Unit cells of TizNi and Ti4Ni20 [Mueller et al. 1963], and their respective
calculated power diffraction spectra. (Both phases share similar strongest peaks, which
are labeled according to their hkl planes.)

 

 

 

 

 

 

 

 

 

 

 

131

 

 

Riki}; mNizo

1 1 2 Atoms per Unit Cell

4 Ni - Tetrahedrons (Sub-Cell Body-Center)
' Ni -Tetrahedrons (Sub-Cell Comer)

Q T1 - Octahedrons (Sub—Cell Body-Center)
. T1 - Octahedrons (Sub-Cell Comer)

. T1 Atoms at Tetrahedral Interstitial Sites
. O Atoms at Tetrahedral Interstitial Sites

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5 (511)

— TizNi
— Ti4NiZO

 

 

 

~ Calculations based
' (422) on Cu Ka Radiation.

(220)

Intensity (A.U.)

 

5 . .I. -.ll .r. .- ._
i - .111

 

 

 

 

 

 

l T I ' I ' l ' l

0120 40 60 80 100

20 (Degrees)
Figure 3.38 (Continued).

Apparently, an oxygen-contaminated braze liquid (originating either fi'om high

oxygen partial pressures or from dissolution of native oxides) will attempt to incorporate

the oxygen chemically into the joint by precipitating sub-oxides of (Ti,Nb)4Ni20 during

cooling. The reason stems from the fact that (Ti,Nb)2Ni has a moderately high oxygen

solubility limit, perhaps much higher than eutectic-NiTi or eutectic Nb. This deduction is

132

also consistent with the fiequent observation of TizNi-type precipitates in joints subjected

to prolonged braze durations, where there is higher tendency for braze liquid to absorb

oxygen from vacuum contamination. Note that SEM-XRF and TEM-XRF study did not

reveal any oxygen in the TigNi-type precipitates because the O Ka x-ray line (0.5249

KeV) is close to the Ti La line (0.4522 KeV) and the energy resolution of XRF

(approximately 100 eV) is not good enough to resolve the two peaks.

Besides oxygen contamination, brazing at 1155°C for short durations (< 120

seconds) was also observed to result in extensive precipitation of the TizNi-type phase.

Their formation at a braze temperature of 1155°C can be explained with the aid of Figure
3.39, which illustrates the NiTi + Nb(Ni,Ti) + Liquid phase field evolution as
temperature decreases from 1165°C to 1155°C. Note that liquid compositions are

governed by the liquidus projection. At 1155°C, the braze liquid establishes equilibrium

at composition C 7; 2, making it higher in Ti content than those at C 7; 1 and C E-

For (Ti,Nb)2Ni to precipitate from liquid composition C T, 2, the temperature-

evolution of the NiTi + Nb(Ni,Ti) + (Ti,Nb)2Ni phase field should take on the form as

shown in Figure 3.40, where the phase field is shown to swing rapidly towards the NiTi-
Nb isopleth as temperature decreases from 950°C. At 20°C, the isothermal section is

estimated to resemble the one shown in Figure 3.41. If the hypothesized phase field

evolution is true, liquid composition C132 will first enter the NiTi + Nb(Ni,Ti) phase field

133

during initial cooling, but abruptly falls into the NiTi + Nb(Ni,Ti) + (Ti,Nb)2Ni three-
phase field below 950°C. Because of the gentle-sloping solidus boundary of the NiTi +

Nb(Ni,Ti) + (Ti,Nb)2Ni phase field, the hypothesized phase field evolution predicts the

ability to re-solutionize (Ti,Nb)2Ni phase at temperatures around 950°C. This prediction

agrees well with experimental observations of the TizNi - type precipitates re-

solutionizing into the eutectic phase at 920°C (Figure 3.27).

Centigrade
2500

'3’

—-I
8
O

2

 

 

§

IIIIIIITIIIIIIIIIIIT

 

 

 

 

'43

 

 

j «i .
‘ V Liq. (1): Liquid at 1165°C
1

with composition CT, 1

‘ Liq. (2): Liquid at 1155°C
with composition 67,2

05. Quasibinary eutectic
‘° composition

 

 

 

Figure 3.39 NiTi + Nb(Ni,Ti) + Liquid phase field evolution as temperature decreases
fiom 1165°C to 1155°C, resulting in Ti-enrichment in the liquid phase.

134

 

n: .....

T'ZNI‘_‘ , 5 . .

ecc- .
Nb (Ti, Nj) LowerTempera‘tures

 

9500C

 

Ti
Figure 3.40 Hypothesized NiTi + Nb(Ni,Ti) + (Ti,Nb)2Ni phase field evolution as
temperature decreases from 950°C. Gently-sloping NiTi + Nb(Ni,Ti) + (Ti,Nb)2Ni /
NiTi + Nb(Ni,Ti) phase boundary providing re-solutionizing possibility for (Ti,Nb)2Ni.
(Color tone of phase fields corresponds to temperatures: Darker — Higher temperature;
Lighter — Lower temperature.)

20 0C isotherm,
(Assuming only the
solubility ranges
associated with
known binary
equilibria.)

 
  
 
 
  

 

 
 

 

 

I ' U I I I I l I l I. l I I I I I
at% Nb
Figure 3.41 The proposed 20°C isothermal section of the Ni-Ti-Nb system.

135

It is important to highlight that TizNi phase cannot be solutionized in the binary
system due to its steep solidus boundaries (see Figure 1.4). Hence, the ability to re-

solutionize the TizNi - type precipitates is unique to the ternary system and can serve as a

means to control their precipitation. The need to control the precipitation of these TizNi

— type precipitates stems from the fact that, when coarse, they possess undesirable
mechanical properties that will compromise the ultimate strength of j oints. But when
developed as fine, coherent precipitates, or even as GP zones, they may be able to
strengthen the NiTi phase substantially. (Details on their mechanical properties will be

discussed in Chapter 4.)

3.8 Summary

Experimental and simulation results show that the driving force for NiTi—Nb
contact melting stems from the deviation from local equilibria at the respective melt
interfaces, while the melting process is rate-controlled by liquid diffusion rates. For a Nb
foil thickness of 51 um, melting was found to occur rapidly during the first few seconds.
However, the rate of melting decreased as the liquid width increased at longer braze
times. Due to sluggish solid-state diffusion of Nb, isothermal solidification plays an
insignificant role in the reactive eutectic braze process. On the other hand, isothermal

solidification was feasible only when thin Nb films were employed as the braze filler.

Owing to the presence of a NiTi-Nb quasibinary eutectic isopleth in the ternary

system, braze liquid solidifies with predominantly eutectic micro-constituents. However,

136

oxygen contamination arising from poor surface preparations, poor vacuum conditions or

prolonged braze durations tends to encourage the precipitation of (Ti,Nb)4Ni20 sub-

oxides. Phase equilibria considerations and experimental evidence have shown that these

TizNi — type precipitates can be re-solutionized at 920°C. The re-solutionizing treatment

will concurrently spheroidize the eutectic-Nb phase.

137

CHAPTER 4: MECHANICAL PROPERTIES OF N IOBIUM-EUTECTIC

BRAZE JOINTS

4.1 Introduction

Results of experiments on the mechanical performance of Nb-eutectic braze joints
will be presented in this chapter. In particular, experimental techniques, such as nano-
and microindentation, shear and tensile tests, were employed to probe the mechanical
properties of the eutectic and its constitutive phases. Fractography will also be discussed.

Besides NiTi-NiTi homo-joints, the reactive eutectic braze process was also applied to

hetero-joints of NiTi-Ta and NiTi-A1203, whose mechanical properties will be presented

at the end of this chapter.

4.2 Nanoindentation Studies

Nanoindentation experiments were performed on near-equiatomic NiTi joint
specimens that were brazed using 51 urn-thick Nb foils. Braze temperature and duration
used in this series of experiments were 1185°C and 120 seconds, respectively. The
details of the nanoindentation setup, calibration and experimental procedures were
described in Chapter 2 and will not be reiterated here. Most of the results were derived
from indentation holding depth of 50 nm and a strain rate of 0.001 per second unless

otherwise specified.

For the purpose of comparison, a control experiment was first performed on pure

Nb and on the base-NiTi material. At least 10 indents were made on each material in

138

 

order to obtain meaningful average values of hardness and elastic modulus from the
results. The typical load vs. displacement curves of pure Nb and base-NiTi at an
indentation depth of 100 nm are given in Figure 4.1. Using the Oliver-th model
[Oliver et al. 1992], the mean hardness and elastic modulus of pure Nb derived at this
indentation depth were found to be 1.1 GPa and 104 GPa, respectively. As mentioned in
Chapter 2, these values are reasonably consistent with those reported in literature
[Myneni et al. 2003, Wang et al. 2005, Alkorta et a1. 2006]. For the base-NiTi, the mean

hardness and elastic modulus are 4.3 GPa and 101 GPa, respectively.

 

1.8
5 (Ti,Nb)2Ni or (T i,Nb)4Nl20 m,Nb)2Ni or

Modulus: 175 GPa
1'6 Hardness: 12.8 GPa (Ti,Nb)4Ni20

 
   
 
 
 
 
  

1.4 - Eutectic
Modulus: 87 GPa
1.2 , Hardness: 2.0 GPa

A Nb
NiTi
E 1 ‘ Modulus: 104 GPa
; Hardness: 1.1 GPa
3 0.8 —
-l NiTi Eutectic
Modulus: 101 GPa
0-6 i Hardness: 4.3 GPa
0.4 ‘ Nb
0.2 ~

 

 

 

0 ‘ l 1 fi l l 1 i I

0 1O 20 30 4O 50 60 70 80 90 100 110 120
Displacement (nm)

Figure 4.1 Load vs. Displacement curves of various phases. Indentation holding
depth: 100 nm. Strain rate: 0.001 per second.)

Next, the effect of post-braze anneals on mechanical properties of the eutectic

joint and base-NiTi were investigated. Specifically, separate specimens were subjected

139

to post-braze annealing regimes of 350°C (1.5 hours), 800°C (104 hours) and 920°C (29
hours). All specimens were annealed in air followed by a water-quench. For this series
of experiments, at least 50 indents were made on each studied area. Statistical data for
measured hardness and elastic modulus, of an as-brazed eutectic joint, is plotted in the

form of histograms in Figure 4.23 and b to illustrate the typical sampling distributions.

Since the joint consists mainly of eutectic-NiTi and eutectic-Nb phases, which are

expected to have differing mechanical properties], one would expect the hardness and/or

F4:—

the modulus distribution to have bi-modal characteristic. However, the histograms in
Figure 4.2a and b are single-mode distributions. The mean indentation hardness and

elastic modulus were calculated to be 2.0 GPa and 87 GPa, respectively.

Recall that the typical spacing in the eutectic phase is about 0.5 um, while the
indentation holding depth employed for this experiment is 50 nm, which yield a projected
indentation width of about 0.3 pm. Given the similarity in eutectic spacing and indenter
lateral probing resolution, there is strong tendency that the indenter was actually probing
both phases simultaneously during each indentation. It is important to stress that
although the indentation holding depth can be further reduced to improve the lateral
probing resolution, it will increase measurement noise by doing so and hence

compromise measurement resolution.

 

1
At least in terms of hardness as the control tests showed that pure Nb and the base NiTi alloy have rather
similar elastic modulus.

140

12-

    

     

 

 

10; (a)
a 8‘- \ 6
5 sq § 8
3 § 3
g 1 § <3
u. 4‘ § “-
§
§

  

.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 20 40 60 80 00 120 140 160 180
Hardness (GPa) Elastic Modulus (GPa)

Figure 4.2 Histograms of (a) hardness and (b) elastic modulus of as-brazed eutectic

joints showing the typical sampling distributions. Continuous lines are Gaussian

distribution curves fitted over the histograms. (Data derived from indentation holding

depth of 50nm.)

 

o°.'.°.
d

The typical load vs. displacement curve of the eutectic micro-constituents at an
indentation depth of 100 nm is illustrated in Figure 4.1, which clearly shows that the
eutectic micro-constituents required an indentation load that is intermediate of those
required for pure NiTi and Nb in order to achieve similar indentation displacement. This
is consistent with the calculated hardness of pure NiTi, Nb and the eutectic mixture.
Recall in Chapter 3 that stereology suggested an estimated 30 vol.% occupancy by the
eutectic-Nb phase. By applying the rule of mixture to the indentation hardness of pure
NiTi and Nb, the eutectic micro-constituents was expected to possess a hardness of about
2.7 GPa, which is in fair agreement with the experimentally-determined hardness of 2.0

GPa.

The mean hardness and elastic modulus, for both eutectic joints and base-NiTi, at
various annealing temperatures are summarized and plotted in Figure 4.3a and b,
respectively. The as-brazed eutectic joint was found to possess an average hardness of

2.0 GPa, which is about half of the as-brazed base-NiTi’s hardness. A mean elastic

141

modulus of 87 GPa for the as-brazed eutectic joint is also lower than those of the base-
NiTi and the pure Nb. The joint hardness increases from as-brazed condition to a value
comparable to the base-NiTi’s hardness when subjected to annealing temperature of
350°C, but hardness decreases gradually to as-brazed state when the annealing
temperature was increased to 920°C. Annealing at 350°C seems to lower joint stiffness

while annealing beyond 800°C leads to a slightly higher modulus than as-brazed joints.

The base-NiTi hardness appears constant from the as-brazed state to an annealing
temperature of 350°C, but increases significantly after annealing at 920°C. Elastic
modulus of base-NiTi was observed to decrease gradually as the annealing temperature
was increased. To summarize, temperature studies on the joints and base-NiTi, annealing

at 350°C yields the maximum joint hardness, which becomes comparable to the hardness

 

 

 

of the base-NiTi.
(a) 7 @120
- Base-NiTi - Base-NiTi
6 ' 1:] Eutectic A :1 Eutectic
A 5_ 53 100
g 2?.
o 4 g
3’ 5 80«
8 3 E
C E
'E 2 o
(u '5 60.
I 8
1. m
o- 40«
As— 350°C 800°C 920°C AS- 350°C 800°C 920°C
Brazed Brazed

Figure 4.3 Plots of (a) hardness and (b) elastic modulus versus post-braze anneal
temperatures of 350°C, 800°C and 920°C of eutectic joints and base-NiTi. As-brazed
data included. (Data derived from indentation holding depth of 50nm.)

142

By performing nanoindentation in areas with high number-density of the facetted

(Ti,Nb)2Ni or (Ti,Nb)4Ni20 precipitates, their mechanical properties can also be

determined. Figure 4.1 shows the load vs. displacement curve of a TizNi-type precipitate

at an indentation depth of 100 nm. Given that poisson’s ratio of the Ti-rich phase is
unknown, an arbitrary poisson’s ratio has to be assigned in order to calculate the hardness
and elastic modulus. Since most metallic and intermetallic materials have values that
fall between 0.3 and 0.4, a value of 0.35 was assumed. Calculations yield hardness of
12.8 GPa and an elastic modulus of 175 GPa, which are significantly higher than those of

the base-NiTi alloy and the eutectic micro-constituents. The indentation characteristics

of TizNi-type precipitate suggest that it is likely to be brittle in nature.

4.3 Base-NiTi Properties and Annealing Response

Before commencing mechanical tests on these joints, control specimens of room-
ternperature superelastic NiTi alloy with no joints were first tensile-tested to determine
the stress-strain characteristics of the base-NiTi, and how the braze process and aging
treatments affect the base-alloy’s mechanical properties. Figure 4.4 shows the stress-
strain curve of an as-received superelastic NiTi alloy subjected to tensile
loading/unloading cycles (with progressively larger strains from 6% to 12%) before being
strained to failure. The material suffered an unrecoverable strain of more than 2% after
the first strain cycle of 6%, with each subsequent cycle only producing about 0.5%
permanent strain. The last cycle yielded a recoverable strain of about 8%. Despite

superelastic recovery, no distinct plateau stresses were observed in the stress-strain plot.

143

Fracture occurred at about 1.10 GPa at a strain of 37%. Elastic modulus measured from
the second unloading slope revealed a value of about 17 GPa, which appears to be
significantly lower than values obtained from nanoindentation experiments. Given that
extensometers were not used in the tensile experiments, substantial errors in strain
measurements can be expected, which in turn, will affect all modulus measurements. In
order for the elastic modulus of NiTi alloys measured fi'om tensile experiments to agree
with the widely reported range between 50 GPa to 70 GPa, the strain scale can be re-
calibrated by dividing the current strain magnitudes by a factor of 3 to 4. However, the
main goal of this work is to compare the relative tensile strengths of the braze joints,
rather than to accurately determine the strains and elastic moduli, no corrections will be

applied to the rest of the stress-strain plots.

 

1200 I
,’ Tensile Strength:
,’ 1.10 GPa

1000 -

 

Stress (MPa)
§

400-

200 .

 

 

 

 

15 20 25 30 35
Strain (96)

Figure 4.4 Stress vs. strain plot of as-received room-temperature superelastic base-
NiTi control specimen.

144

Next, an as-received superelastic control specimen was subjected to annealing at
350°C for 1.5 hours before undergoing similar tensile testing. The mechanical behavior
is shown in Figure 4.5. Again, initial loading up to 6% strain yielded an unrecoverable
2% strain upon unloading, and subsequent loading/unloading cycles each produces less
than 1% permanent strain. Mean elastic modulus derived from the unloading curves of
these subsequent cycles was 24 GPa. Maximum superelastic recovery of about 9% was

observed in the final cycle. Tensile stress and strain at fracture were approximately 1.35

 

GPa and 28%, respectively. A relatively more defined upper- and lower-plateau stresses
close to 300 MPa and 100 MPa, respectively, could now be seen in the initial cycle in
Figure 4.5. However, these plateau stresses decreases gradually as the strains associated

with each cycle increases.

1600 .
I Tensile Strength:
1400 - l 1.35 GPa

 

1200 a

3

Stress (MPa)

 

  

 

 

 

 

200 - I
’/ Modulus: 24 GPa
0 " I I I I
0 5 10 15 20 25 30
Strain (96)

Figure 4.5 Stress vs. strain plot of as-received room-temperature superelastic base-
NiTi control specimen that had been subjected to aging treatment at 350°C for 1.5 hours
(then water-quenched) prior to testing.

145

Monolithic NiTi sections that had undergone the typical braze regime were tested
under tensile loading/unloading conditions to evaluate the effects of brazing on base-NiTi
mechanical properties. Figure 4.6 shows permanent strains of 1% associated with each
progressive loading/unloading cycle. Recoverable strains of about 7.8% can be achieved
in the second cycle. No distinct plateau stresses were observed and the specimen

fractured at a stress and strain of about 800 MPa and 26%, respectively.

 

§

Tensile Strength: 800 MPa

§§

a:
8

 

Stress (MPa)

ANOD-bO‘I
088888

 

 

 

l l I I T

10 15 20 25 30
Strain (96)

Figure 4.6 Stress vs. strain plot of as-received room-temperature superelastic base-

NiTi control specimen that had been subjected to the typical braze process conditions

prior to testing.

0
01

Figure 4.7 shows the stress-strain behavior of a control specimen subjected to the
braze process and subsequently aged at 350°C for 1.5 hours prior to testing. Initial

loading/unloading cycle produces an unrecoverable strain of more than 1%. A

146

recoverable strain of about 7% could be achieved if total strain did not exceed 8%. The
stress-strain plot now shows flatter, well-defined upper— and lower-plateau stresses at
about 350 MPa and 150 MPa, respectively, in the first cycle. They were observed to
decrease gradually as the strain associated with each loading/unloading cycle increases.

The control specimen failed at about 900 MPa at a strain of 23%.

 

Tensile Strength: 900 MPa

Stress (MPa)

 

200 -

 

 

 

 

O ‘I I I I I I
0 2 4 6 8 10 12 14 16 18 20 22 24

Strain (96)
Figure 4.7 Stress vs. strain plot of as-received room-temperature superelastic base-

NiTi control specimen that had been subjected to the typical braze process followed by
aging at 350°C for 1.5 hours (then water-quenched) prior to testing.

From Figure 4.4 to Figure 4.7, it can be deduced that the braze process did result
in mechanical degradation in terms of strength and ductility, but typical superelastic

response can be restored in the base-NiTi after post-braze aging treatments. Note that

this alloy is designed for superelasticity at 37°C, while all the above tensile tests were run

147

 

at 20°C. It is believed that superelastic response would have been superior if tests were

performed at 37°C.

4.4 Tensile Tests on Eutectic Braze Butt-Joints

The reactive eutectic braze process was next employed to prepare butt-joint
specimens. Most of the specimens made for this series of the study were the room-
temperature superelastic NiTi used for the base-line tests described in Section 4.3. The
initial joint clearance was approximately 400 pm (unless otherwise specified). Sufficient
Nb foils were used to fill in the gap. Figure 4.8 shows the representative stress-strain
behavior of a typical as-brazed superelastic butt-joint. The butt-joint specimen failed
consistently within the eutectic joint, which will give a good measure of the joint
mechanical properties. Based on the loading curve, the superelastic butt-joint specimen
possessed an elastic modulus of 22 GPa. This value is fairly consistent with the modulus
obtained from the control specimen in Figure 4.4. The superelastic butt-joint failed at a

stress and strain of 680 MPa and 21%, respectively.

Comparison between stress-strain characteristics of as-brazed superelastic butt-
joint (Figure 4.8) and that of the monolithic control (Figure 4.6) shows high degree of
similarity, with the exception that the former only have 85% of the base-NiTi’s strength
and ductility. This indicates that the eutectic joint possesses mechanical properties that
are comparable, but still inferior to that of the base-NiTi alloy. Given that pure Nb is

weaker than pure NiTi, with a room-temperature tensile strength of about 170 MPa

148

 

[Gupta and Suri 1994], it is expected that the strength-limiting phase in the joint is the

eutectic-Nb.

800

 

Tensile Strength: 680 MPa

(3 8 OJ ‘1
O o o
r r r r r

Tensile Stress (MPa)

N

O

C
l

8

 

 

 

 

O

2 4 6 8 10 12 14 16 18 20 22
Tensile Strain (96)

Figure 4.8 Stress vs. strain plot of as-brazed room-temperature superelastic butt-joint

specimen. (Initial joint clearance: Approximately 400 pm.)

0

However, it is also possible that the embrittling nature of TizNi-type precipitates

that are randomly distributed in the joint may also form crack initiation sites that will
result in premature joint failure during tensile loading. Figure 4.9 shows the micrograph
of a pre-polished section of a lap-joint that had undergone shear-loading but not to the

point of failure. The joint was deliberately fabricated under conditions that favor

precipitation of TizNi—type grains. Distinct cracks were evident in the facetted

precipitates even before joint failure occurred. Composition mapping at site ‘B’ and ‘C’

by XRF (see Table 4.1) confirmed that these facetted grains are of TizNi-type.

149

Fracture surface of TizNi-free butt-joint specimens typically revealed two distinct

fracture surface morphologies, which are shown in Figure 4.10a and b. In Figure 4.10a,
the fracture mode is primarily ductile cleavage, whereas in Figure 4.10b, highly-dimpled
surface and micro-void coalescence indicates a fracture mode with extensive ductile
tearing. The former was determined by XRF to be attributed to failure predominantly
within the pro-eutectic NiTi phase, while the latter was identified with the eutectic micro-

constituents.

  

" I. ""‘i .-_._ Nb 1.0.
Figure 4.9 Photograph of a room-temperature shape-memory single-lap joint that had
experienced onset of failure under shear loading. SEM micrograph of cracks traversing
facetted Ti-rich precipitates in the joint region. Elemental XRF maps of the same area
are also included.

 

150

 

l \l /\'r/.

  
   

 

(Continued). '1 I

 

Figure 4.9

Table 4.1 X—ray fluorescence analysis on selected Spots in Figure 4.9.
Spot Nb (at%) Ni (at%) Ti (at%)

 

 

A 3 51 46
B 13 36 51
C 12 36 52

 

    

5. ’ 5,, ‘. ' Q. . g t 3 E,
Figure 4.10 Fracture surfaces of the (a) pro-eutectic NiTi; and (b) eutectic micro-
constituents in a typical superelastic butt-joint specimen that failed under tension.
Elemental XRF maps of (a) are also shown.

151

        

Figure 4.10 (Continued).

Butt-joint specimens were also tested afier thermal annealing. The aging regimes
employed are as follow: (1) 350°C for 1.5 hours; (2) 500°C for 20 minutes; and (3)
750°C for 10 minutes, water-quenched, followed by 350°C for 1.5 hours. All specimens
are annealed in air and water-quenched. Figure 4.11 shows the representative tensile
behaviors of as-brazed and the aged butt-joints. As mentioned previously, the typical
plateau stress, found in room-temperature superelastic NiTi, were not observed in as-
brazed butt-joint specimens. In contrast, stress-plateau was observed in each of the
annealed butt-joints, with lower aging temperature yielding a better-defined plateau. All
aged specimens appear to fracture at a strain of about 13%. The highest plateau and
failure stresses of about 380 and 770 MPa, respectively, were attained by specimens aged
at 350°C. Higher annealing temperatures of 500°C and 750°C resulted in degradation of

both the plateau stress and ultimate strength.

152

 

800
3500/1 .5hrs As-Brazed

680 MPa)

700 _ (770 MPa) \ (
500C/20min
(630 MPa) ...-

'(

°8’

8
o

 

 

Tensile Stress (MPa)
.5
8

 

 

 

300 ~ g
200 - /
~ 7500/10min/35OC/1.5hrs in.
100: / (420 MPa)
0 l I I I I I I I 7 I I I

 

0 2 4 6 8 1O 12 14 16 18 20 22
Tensile Strain (96)

Figure 4.11 Stress vs. strain plots of as-brazed and aged superelastic butt-joint
specimens. (Aging regime: 350°C for 1.5 hours; 500°C for 20 minutes; 750°C for 10
minutes, followed by 350°C for 1.5 hours. All anneals performed in air and water-
quenched. Initial joint clearance: Approximately 400 um.)

The degradation of mechanical pr0perties in the 7 50°C solution treatment-then-
350°C aging specimen was unexpected because aging after a solution treatment should
yield mechanical results comparable to a specimen annealed solely at 350°C. There are

two possible explanations: (1) the duration for solution treatment may not be sufficiently

long to solutionize the Ni4Ti3 phase. Hence, subsequent aging actually resulted in

growth of pre-existing Ni4Ti3 phase; (2) significant oxidation might have occurred in the

joint, causing embrittlement.

153

Since superior plateau and failure stresses seem to be associated with lower aging
temperatures, butt-joint specimens were aged at 300°C for 8 hours and then tested in
tension. On the contrary, the initial cycle stress-strain characteristics of the 300°C-aged
joint, given in Figure 4.12, reveal upper- and lower-plateau stresses of less than 300 MPa
and 50 MPa, respectively. After the first loading/unloading cycle, the specimen incurred
about 1.5% unrecoverable strain. Subsequent 8 cycles yielded a consistent strain
recovery of about 6.5%. The upper-plateau stress decreased gradually as cycle increases,

while lower-plateau stress remained rather constant throughout the 8 cycles. On the 10th

 

cycle, the specimen was tensile-loaded to failure at a stress of 660 MPa at a strain of

10.6%.

 

700
600 r
500 _
400 r
300 ..

Stress (MPa)

200 -:

 

100 -

 

 

 

0 I l I I I l I I I I I

012 34 56 78 9101112
Strain(%)

Figure 4.12 Stress vs. strain plots of an aged superelastic butt-joint undergoing 10
tensile loading/unloading cycles. (Aging regime: 300°C in air for 8 hours followed by a
water—quench. Initial joint clearance: Approximately 400 um.)

154

The mechanical properties of room-temperature superelastic butt-joints with

narrower initial joint clearance of approximately 300 um were also investigated. The

stress-strain plots of as-brazed and 350°C-aged butt-joints are shown in Figure 4.13. For

comparison, results from the wider (~ 400 um) joint clearance specimens are also plotted

in Figure 4.13. The narrow as-brazed joint fails slightly above 500 MPa, while the aged

narrow joint fractures at about 680 MPa. On the other hand, the wider counterparts have

higher fracture strengths. The plateau stress of the narrow butt-joint was observed to be

lower (less than 300 MPa) than the wider joint (about 380 MPa). No significant

difference in fracture strain was observed.

 

 

 

900 '
3500 (wide) AS'B'azed (me)
800 - (770 MPa) \ (680 MPa)
35(6)? (rm-30W) . As-Brazed (narrow)
700 _ ( 5 a) : (515 MPa)
E 600 ‘ ' b
a 'i
Q ~ :'
E 500 l A :
g3..- i s
200 d :
100 -
0 W I I I I I I r I I

 

 

 

 

Tensile Strain (%)

Figure 4.13 Stress vs. strain plots of as-brazed and aged superelastic butt-joints with
initial joint clearances of approximately 300 um (narrow) and 400 um (wide). (Aging
regime: 350°C for 1.5 hours in air, followed by a water-quench.)

4.5 Tensile Tests on Tungsten Inert Gas-Welded Butt-Joints

As discussed previously, the braze process involved heating entire test specimens
to braze temperature in order to initiate contact melting between NiTi and Nb at the joint
area, which inevitably degrades the base-NiTi’s mechnical properties. Hence, an attempt
was made to explore joining techniques that only involve localized heating at the joint
region. The Tungsten-Inert-Gas (TIG) welding method was employed and adapted for
this purpose, where 0.51 mm-diameter Nb wires were manually-fed into the joint by the
welder during welding for eutectic liquid formation. The disadvantage of this approach is
that there is no direct control over welding temperature, while welding speed is solely
dependent on experience and dexterity of the welder. Furthermore, the argon gas shield
of the TIG welding process was not sufficient to suppress oxidation of NiTi, especially in

the joint regions.

The general microstructure and a higher-magnification of an as-welded room-
temperature superelastic butt-joint are shown in Figure 4.14a and b, respectively. Typical
pro-eutectic NiTi dendrites and eutectic phase were observed in the welded joint.
However, a second dendritic phase also coexisted randomly with the two common
microstructures. The fact that this phase was dendritic indicated that it formed before
general eutectic freezing occurred. Stereology performed on these micrographs provided

an estimated 5% volume fraction of this anomalous phase in the joint.

156

 

 

 

 

 

Ni Ka

 

X Y

Figure 4.14 SEM micrographs of (a) general as-welded joint microstructure, and (b)
high-magnification of the indicated area of (a). (XRF line profiles showing composition
variations across X-Y are also provided.)

 

Table 4.2 X-ray fluorescence analysis on selected spots in Figure 4.14.
Spot Nb (at%) Ni (at%) Ti (at%)

 

A 4 52 44
B 7 39 54
C 5 38 57

 

Line profiles from XRF analysis, indicated by ‘X-Y’ in Figure 4.23b, revealed
higher Ti content, but lower Nb concentration in this unknown phase as compared with

the eutectic micro-constituents, while XRF spot analysis (Table 4.2) shows that this phase

157

also has higher Ti concentration but comparable Nb content with the pro-eutectic NiTi
phase. However, these dendritic precipitates do not share similar chemical compositions
with those Ti-rich phases discussed in Chapter 3. It is, hence, suspected that the argon
gas shielding from the TIG welding process was unable to prevent extensive oxidation

from occurring in the joint. The significantly high oxygen contamination might have

resulted in formation of a complex quaternary oxide that differs from the (Ti,Nb)2Ni or

(Ti,Nb)4Ni20 stoichiometry.

J!

Room-temperature superelastic butt-joints were prepared using the TIG welding
technique. Initial joint clearance was approximately 400 pm, which is similar to most of
the brazed butt-joints. Niobium filler wires of 0.51 mm-diarneter were used instead of
Nb foils. Some welded joints were subjected to thermal aging at 350°C for 1.5 hours and
500°C for 20 minutes in air, followed by water-quench before tensile testing. Figure 4.15
shows that the as-welded butt-joint had failure strength of only 590 MPa and had limited
ductility (< 9% strain at failure). This may be attributed to presence of residual stresses
in the welded joint due to large thermal gradients associated with localized heating and
the presence of oxides in the joint. Aging welded joints at 350°C resulted in distinct
improvement in tensile strength (~ 750 MPa) but not in ductility. In addition, a plateau
stress of about 350 MPa could be achieved. However, joint strength and ductility
deteriorated to only 310 MPa and 6%, respectively, when welded joints were annealed at
500°C. Its plateau stress (less than 250 MPa) was also lower than the 350°C-aged joint.
The degradation of joint mechanical properties at aging temperatures above 350°C in

Figure 4.15 is consistent with those observed for brazed joints in Figure 4.1 1.

158

800

 

3500/1.5hrs

700 - As-Brazed (740 MP8)
(585 MPa) \

600 — 5000/20min \
(320 MPa)

500 ~

400 ~

300*

Tensile Stress (MPa)

200-

 

100-

 

 

 

 

 

o " I I I I I I I r I I
O 1 2 3 4 5 6 7 8 9 10 1 1
Tensile Strain (%)

Figure 4.15 Stress vs. strain plots of as-welded and aged superelastic butt-joints.
(Aging regimes: 350°C for 1.5 hours; 500°C for 20 minutes. All anneals performed in air
and cooling by water-quench. Initial joint clearance: Approximately 400 um.)
Fractography of the welded joint, given in Figure 4.16, reveals rrricro-void
coalescence in most areas, which suggests that the failure mode is predominantly ductile
tearing. However, ‘caterpillar track’ features (indicated by ‘A’ in Figure 4.16) can also
be seen on the fracture surface and some of them have brittle-mode cracks in close
proximity (indicated by ‘B’ in Figure 4.16). Based on the morphology, this failure mode
is probably associated with the Ti-rich complex oxides that were previously observed
(Figure 4.14). Therefore, it is highly probable that this dendritic phase initiates failure by
either inter- or trans-granular fracture. The former gives rise to the ‘caterpillar track’

features, while the latter manifests as cracks. In either case, it can be deduced that this

anomalous phase possesses brittle characteristics.

159

  
    

:5 V 5’ 7 . .
Figure 4.16 SEM micrographs of the fracture surface of a welded room-temperature
superelastic butt-joint that failed under tension. (‘A’ indicates ‘caterpillar track’ features;
‘B’ indicates brittle-mode cracks associated with the presence of ‘A’.)

 

4.6 Eutectic Braze Joints of Dissimilar Materials
In this section, a preliminary attempt is made to extend the reactive eutectic braze

process to join NiTi to dissimilar materials. In particular, the mechanical properties of
NiTi-Ta and NiTi-A1203 hetero-joints will be briefly discussed. There are two reasons
for the choice of materials for investigation. Firstly, Ta and A1203 are both bio-

compatible and commonly used as part of medical implants that also contain NiTi alloys.

For example, high-atomic number elements, such as Ta, coated onto NiTi stents, can
improve radiopacity [Schetky et al. 2004], while A1203 coating on NiTi bone implants

can inhibit release of Ni into human bodies [Ozeki et al. 2003]. Secondly, they represent

two classes of materials of vastly different mechanical properties, so it will be interesting

160

to evaluate the mechanical responses when these two types of materials are joined to NiTi

 

alloys. Furthermore, A1203 is an insulator that could be integrated with NiTi electrically-

excited actuators.

4. 6. 1 Ni T i-T antalum Joints
Tantalum coupons with nominal thickness of 1.588 mm was brazed to room-
temperature superelastic NiTi of similar thickness using 127 um—thick Nb foil as the

braze filler metal. The two base-metals were placed ‘edge-on’ against each other with

. I“?

the braze foil between them. Braze temperature and duration were 1185°C and 5 minutes,
respectively. Butt-joints of NiTi-Ta were prepared and then tested under tensile loading
to evaluate the joint mechanical properties. Figure 4.17 shows the stress-strain behavior
of an as-brazed NiTi-Ta butt-joint. The ultimate tensile strength of the joint is
approximately 210 MPa, which coincides with the reported tensile strength of high-purity
Ta (~ 210 MPa) [Sisco et a1. 1963]. Necking was observed in the Ta section of the
specimen around 20% strain (see inset of Figure 4.17). The strain at failure was close to
30%. Results and observations in Figure 4.17 clearly imply that the eutectic joint is

mechanically superior to the base-Ta metal.

161

 

250

 

 

 

 

 

200 -
i?
5
V150 - \1
3
I15
0
g 100 —
S
l-
50 _
0 I I I I | I
0 5 1O 15 20 25 30 35

Tensile Strain (%)

Figure 4.17 Stress vs. strain plot of as-brazed superelastic NiTi-Ta butt-joint. (Inset
shows necking occurred in the Ta section of the specimen.)

4.6.2 NiTi-Alumina Joints

For this study, 99%-dcnsity A1203 plates and room-temperature superelastic NiTi
strips were employed. The former has dimensions of 15 x 10 x 3.175 mm, while the
latter has dimensions of 30 x 4.7 x 0.2 mm. Instead of the conventional butt-joint or
single-lap joint configurations, the NiTi-A1203 joint was fabricated according to Figure
4.18. The overlap distance was 5 mm and the gage length on each NiTi strip was 15 mm.
Niobium foil with a thickness of 51 um was used for brazing, while braze temperature
and duration were set at 1185°C and 3 minutes, respectively. The specimen was cooled
naturally in the vacuum furnace before removing them from the chamber (see Figure 2.1

for the temperature-time profile of the braze process).

162

Reserved for AI203

 

 

 

 

 

 

 

 

anping / , A I
4.7 mm II E i NiTi H | NiTi ll 10 mm
1 ' l/
15 mm '5 mm ,3 Thickness of Al203: 3.175 mm

 

15 mm : Thickness of NiTi: 0.2 mm

Figure 4.18 Schematic diagram showing the setup of NiTi-A1203 lap shear specimen.

A metallographically-prepared cross-section of the NiTi-A1203 joint (Figure 4.19)

I‘m—u;

reveals meniscus forming at the edge of the joint with minimal voids forming in the braze
area. Figure 4.1% is a duplicate of Figure 4.19a that had undergone contrast

enhancement to reveal the eutectic micro-constituents. These observations indicate good

wetting and flow of the eutectic liquid on A1203. The joint consists mainly of eutectic

rrrixture and pro-eutectic NiTi dendrites, where the latter can be seen to solidify against

the base-NiTi. The Ale3-liquid interface does not appear to recede like the NiTi

counterpart and only the eutectic micro-constituents was observed to solidify against the

base-A1203, which indicates that A1203 does not participate strongly in braze liquid
formation. Some discrete facetted precipitates, believed to be either (Ti,Nb)2Ni or

(Ti,Nb)4Ni20, were also present in the joint. Cracks in the base-A1203 can be seen

extending from the meniscus (indicated in Figure 4.19), which might have formed as a

result of thermal shock due to low toughness (2 -— 3 MPa.m1/2), low thermal conductivity

163

(8 — 10 W/m.K) and a moderately high coefficient of thermal expansion (~ 8 x 10.6 K4)

[Munro, 1997] of A1203.

      

Figure 4.19 SEM micrograph showing (a) the typical cross-section of a NiTi-A1203
braze joint. (b) is the contrast-enhanced duplicate of (a).

The stress-strain plots of the NiTi-A1203 shear specimen are shown in Figure

4.20. It should be highlighted that tensile stress and strain in Figure 4.20 were calculated

based on the assumption that elongation was entirely concentrated in the NiTi sections.
This is reasonable given that the thickness of A1203 is significantly larger than the NiTi
strips. The specimen failed at a tensile stress of slightly less than 160 MPa, while
fracture strain was less than 1%. Fracture was observed to occur in the base-A1203 but

close to the joint. The poor mechanical properties and fracture site suggest probable
weakening near the joint interface due to thermal shock during cooling, which is

consistent with the observations of cracks in Figure 4.19.

164

180

 

160 MPa
160 d

140 -

 

120 -

8

Stress (MPa)
on
o

«b O)

O O

l l
...,

N
O
l

 

 

 

O

 

I I I I

0.002 0.004 0.006 0.008 0.01
Strain (%)

0

Figure 4.20 Tensile stress vs. strain plots of as-brazed superelastic NiTi-A1203 shear
specimen.

Nanoindentation was performed at the NiTi-A1203 joint area (see Figure 4.21).

The indentation holding depth for this study was set at 500 nm with a strain rate of 0.02
per second. Results in Figure 4.21 shows that the base-NiTi alloy and the eutectic micro-

constituents possess indentation hardnesses of 4.5 GPa and 3.0 GPa, respectively, which

are consistent with results in section 4.2. Hardness of the base-A1203 away from the

joint area was measured to be 27.8 GPa, which also agrees well with literature-reported

values [Krell et al. 2001]. However, A1203 in close proximity to the joint appears

significantly weaker with a hardness of only 18.7 GPa. Therefore, localized mechanical

properties obtained fiom this nanoindentation study further support the claim that poor

165

shear and tensile strengths of NiTi-A1203 joint is indeed attributed to weakening of the

A1203 near the joint.

Hardness (GPa)

 

 

 

Base- Alumina Eutectic Base~ » .- P .
Alumina (NearJoint) NITi ‘ :. . 7. . . . .
Figure 4.21 Bar-chart (left) illustrating the hardness of various phases in a NiTi-A1203
joint obtained from nanoindentations performed at the area shown in the optical
micrograph (right). Array of indents across the joint can be seen (except the indents in

the base-A1203).

4.7 Summary

The mechanical properties of eutectic joints were investigated at both micro- and
macro-scales. Results show that the braze process will cause degradation in the base-
NiTi’s strength and ductility, but superelastic response can be restored in the base-NiTi
through post-braze thermal aging. It was found that the optimum aging temperature is

around 350°C, where highest tensile strength and plateau stresses were observed.

166

However, the eutectic joints are still mechanically-inferior to the base-NiTi alloy, which
can be attributed to the strength-limiting eutectic-Nb phase or the presence of embrittling

TizNi-type precipitates serving as crack initiation sites for premature joint failure.

The TIG welding approach was successfully adapted to form eutectic liquid in
joint area. Despite the presence of a possible complex quaternary oxide phase,
mechanical strengths of welded joints are equivalent to the brazed counterparts, except

for a reduction in ductility.

The reactive eutectic braze process has also been successfully extended to joining

NiTi to Ta. Joint strength was observed to be well in excess of the ultimate tensile

strength of pure Ta. For joining NiTi to A1203, it was found that the eutectic liquid does

wet and flow on A1203. Weakening of the A1203 near the joint interface was observed,

which consequently results in poor shear and tensile strengths.

167

CHAPTER 5: MICRO-ALLOYING OF N IOBIUM-EUTECTIC BRAZE
F OILS AND AN ALTERNATIVE EUTECTIC BRAZE

SYSTEM

5.1 Introduction

The study on mechanical properties of braze joints in Chapter 4 shows that the

1

joints possess strengths that are comparable but still inferior to that of the base-NiTi alloy.

- Him.

Two phases in the joints were believed to limit mechanical strength. They are (1) the

TizNi-type precipitates within the eutectic phase that can act as crack initiation sites; and

(2) the eutectic-Nb phase, which is mechanically-weaker than the eutectic-NiTi phase.
Therefore, work reported in this chapter was undertaken to explore micro-alloying
strategies in an effort to improve mechanical properties. Towards the end of this chapter,
the NiTi-V eutectic system, which is qualitatively similar to the NiTi-Nb system, will

also be explored for its potential as an alternative braze material.

5.2 Micro-Alloying of the Niobium-Eutectic Micro-constituents with Zirconium
and Tungsten
Since eutectic-Nb is believed to be the strength-limiting phase in the eutectic
micro-constituents, it is logical to devise a strengthening approach to improve its
mechanical properties. In particular, solution-strengthening was selected for this study,
where Zr and tungsten were separately micro-alloyed into eutectic braze joints. For
micro-alloying with Zr, Nb-er (at.%) foils (from ATI-Wah Chang) were employed as

the braze filler metal, while tungsten micro-alloying was carried out via thin-film

168

sputtering of tungsten on pure Nb braze foils in order to obtain an approximate Nb-2W

(at.%) composition. Details on the sputtering procedures are highlighted in Chapter 2.

Zirconium and tungsten were selected as alloying candidates for the following
reasons: (1) they are both soluble in Nb; (2) the atomic radius of Zr and tungsten are
about 10% larger and smaller, respectively, than that of Nb, which lead to a solution-

strengthening effect; (3) they have been reported to produce solution-strengthening in Nb

a“... a

alloys [Wilcox et al. 1967, J ax 1971, Davidson et al. 1992]; and (4) they are both bio-

compatible metals, which make the strengthening approach compatible with bio-medical
applications. In addition, Zr has been reported to have the ability to suppress brittle

(Ti,Nb)2Ni phase formation in Ni-Ti-Nb alloys [J ia et al. 2000].

5. 2. I Effects on Microstructures

Reactive eutectic brazing of near-equiatomic NiTi (nominal thickness of
3.175mm) were performed using the Zr- and W-alloyed Nb foils (nominal thickness of
Slum), separately, using braze temperature and duration of 1185°C and 120 seconds,
respectively. Scanning electron microscopy revealed only pro-eutectic NiTi and eutectic
phases present in the Zr-alloyed joints (see Figure 5.1). Similar to the pure Nb-based
eutectic system, the Zr-alloyed eutectic possesses primarily lamallar and rod-like

morphologies. The average eutectic spacing was approximately 0.6 pm, which is also

similar to those observed in the pure Nb counterpart. No TizNi-type precipitates were

observed in any of the Zr-alloyed eutectic joints studied, strongly suggesting that the

presence of Zr can suppress TizNi-type phase formation.

169

       

V f3. 9 I W -- . -.
Figure 5.1 SEM micrographs showing (a) global joint morphology; and (b) eutectic
microstructure of a Zr—alloyed eutectic joint.

 

In order to confirm this observation, Zr-alloyed braze specimens, with
deliberately pre-oxidized NiTi surfaces, were prepared with prolonged braze duration of

1200 seconds under ‘poor’ vacuum conditions (similar to those described in Chapter 3).
The aim was to create a favorable condition for the precipitation of Ti2Ni-type grains.

Microstructures of the Zr-alloyed joint were then compared against the pure Nb braze foil

counterpart of similar process conditions. As discussed in Chapter 3, the pre-oxidized
NiTi surface had resulted in significant amounts of TizNi-type precipitates and voids in
the pure Nb braze joint (Figure 5.2a). In contrast, no Ti2Ni-type compounds were

observed in the Zr-alloyed joint and voids were rarely encountered (Figure 5.2b).

Therefore, it is safe to deduce that addition of Zr in the eutectic phase does indeed

suppress TizNi-type phase and void formation.

170

    

Figure 5.2 Microstructures of braze joints fabricated using (a) pure Nb foil; and (b)
Nb-er foil. (Mating surfaces were deliberately pre-oxidized before brazing under ‘poor’
vacuum conditions.)

The general joint morphology and eutectic micro-constituents of W-alloyed joints
are shown in Figure 5.3a and b, respectively. Again, the joint consists mainly of pro-

eutectic NiTi and eutectic phases. In this case, the joints were prepared with chemically-
cleaned NiTi under high vacuum conditions, so proliferation of TizNi-type precipitates

and voids should not be expected under such brazing conditions. However, Figure 5.3a

shows that despite careful specimen preparation and good vacuum conditions, facetted
TizNi-type phase and voids were observed in the W-alloyed joint. Such phase formation
might be attributed to a reduction in oxygen solubility in the eutectic micro-constituents

when tungsten was present. Precipitation of TizNi—type phase would occur in an effort to

incorporate excess oxygen.

171

The eutectic micro-constituents also possess lamellar and rod-like morphologies
(Figure 5.3b), but average eutectic spacing was determined to be about 1.3 pm. This
average spacing is distinctively wider than those of Zr-alloyed and pure Nb-eutectic
(inset of Figure 5.3b), which are approximately 0.6 pm. Eutectic spacing is known to be
governed by growth velocities and degree of undercooling in the liquid ahead of
solidification fronts [Flemings 1974]. Given that the rate of cooling was consistent
across all types of braze joint (cooled in the same vacuum braze chamber), it is proposed
that the larger eutectic spacing observed in the W-alloyed braze joint may be due to a

decrease in undercooling that is somehow associated with the addition of tungsten.

     

Figure 5.3 SEM micrographs showing (a) global joint morphology; and eutectic
micro-constituents of a W-alloyed eutectic joint. (Inset of (b) shows the eutectic micro-
constituents of a pure Nb-eutectic joint under similar magnification as (b).)
5. 2.2 Eflects on Mechanical Properties

Nanoindentation experiments were carried out on Zr- and W-alloyed

eutectic joints prepared in section 5.2.1 in the as-brazed condition. At least 40 indents

were performed on each specimen in order to obtain meaningful statistics. The

172

[I

representative load vs. displacement plots for pure Nb-eutectic, Zr-alloyed eutectic and

W-alloyed eutectic are shown in Figure 5.4.

 

 

 

0.25
Nb-ZW
0.2 4 I
uNb-‘IZI'
£0 15 - ..' Nb
5 -
U .
u , .
O ‘ o
'1 0.1 - ..' ,
0.05 — / _.‘ ,°
0 ‘1 I I I I I I :9. I O I I

 

 

 

0 51015 20 25 30 35 40 45 50 55 60
Displacement (nm)

Figure 5.4 Load vs. displacement plots of pure Nb-eutectic, Zr-alloyed eutectic and
W-alloyed eutectic. (Indentation holding depth: 50 nm. Strain rate: 0.001 per second.)
Figure 5.5a and b illustrate the histograms of calculated indentation hardnesses
and elastic moduli of Zr- and W-alloyed eutectic joints, respectively. For comparison,
results fiom pure Nb-eutectic joints are also plotted with existing data. The histograms in
Figure 5.5a show that alloying eutectic joints with Zr yielded an increase in average
hardness from 2.0 GPa to 2.6 GPa, while W-alloyed eutectic joints possess hardness of
2.9 GPa, which is almost 50% increase over the pure Nb counterpart. Only subtle
increases in elastic moduli, from 87 GPa to about 99 GPa, were observed in both types of

alloyed joints.

173

 

 

 
  
 

 

 

 

  

 

 

 

 

 

 

14- 12-
. [:1 Nb (2.0 GPa) Nb
12, (a) Nb-1 Zr - Nb-1Zr (2.6 GPa) 10: (b) 1:12;; GPa)
10‘ - Nb-zw (2.9 GPa) ‘ - NMZr
j Nb 3. (95 GPa)
es. s .
o ' a) 6-
%5- g .
I: ‘ E 4-
4.
2: 2-
0 o.

 

 

 

 

 

 

0.51015 2.0 2.5 3.0 3.5 4.0 4.5 5.0 20 40 50 80 100120140160180 200
Hardness (GPa) Modulus (GPa)

Figure 5.5 Histograms of (a) hardness; and (b) elastic modulus of pure Nb, Zr- and
W-alloyed eutectic braze joints. Continuous lines are Gaussian distribution curves fitted
over the histograms of the three eutectic braze joints. (All tests performed on as-brazed
joints. Data derived from indentation holding depth of 50nm.)

 

Alloyed joint specimens were then subjected to Vickers micro-hardness testing
using indentation load of 300 grams to assess the effect of rrricro-alloying on overall
eutectic properties. Approximately 10 indents were made on each type of joints. In order
to minimize sampling errors, indents were performed on areas solely of eutectic phase.
Optical and electron nricrographs showing typical Vickers indents on ideal test areas (i.e.
fully eutectic) are provided in Figure 5.6a and b, respectively. However, given that these
indents had depth of approximately 10 um, there is still the possibility that the indenter
might have probed pro-eutectic NiTi phase underneath the test area, causing erratic

results. Indentations were also performed on the near-equiatomic base-NiTi alloy.

Vickers indentation results are summarized in Figure 5.7. At a load of 300 grams,
the Vickers hardness number (VHN) of base-NiTi was determined to be 250. The pure
Nb-eutectic has a VHN of about 205, while Zr- and W-alloyed eutectic possess VHNs of
215 and 226, respectively. The micro-hardness analysis is consistent with the

nanoindentation results.

174

 
  
   
   

Figure 5.6 (a) Optical micrograph; and (b) SEM micrograph showing Vickers indents
within the eutectic micro-constituents.

 

@3000
N N N N N
N O) -l> 01 0')
IO 0 O 0 Cl)

NM
0.3
00

  

 

\fickers Hardness (

d
CDLO
0?

 

_BaseN-iTi Nb Nb—1Zr szw

Eutectic Micro-constituents

Figure 5.7 Bar chart showing the Vickers hardness of pure Nb, Zr- and W—alloyed
eutectic braze joints. (All tests performed on as-brazed joints at 300 grams.)

Hence, it can be deduced that alloying Zr and tungsten will yield improvements in
the mechanical properties of eutectic joints, and based on indentation studies, tungsten
may be a more effective solution-strengthening candidate than Zr. However, the

indentation experiments only probed mechanical characteristics of the eutectic micro-

constituents, where effects of those TizNi-type precipitates on overall joint mechanical

175

[1

properties did not come into play. Therefore, tensile tests on butt-joints, fabricated from
the alloyed Nb braze foils, were also conducted in order to determine how improvements
in the nano- and micro-mechanical properties of these eutectic joints were manifested in

mechanical tests on bulk sections.

 

Zirconium-alloyed eutectic butt-joints were brazed at 1185°C for 6 minutes using
room-temperature superelastic NiTi (nominal thickness of 1.588 mm). Initial joint

clearance was approximately 400 pm. Their tensile properties were studied in as-brazed

 

and thermally-aged conditions, where the latter range from 300°C to 500°C for 1 to 4
hours. All aging regimes were performed in air, followed by water-quench. Figure 5.8
shows that as-brazed Zr-alloyed joints exhibit tensile strength close to 840 MPa at a
failure strain of 30%. The stress-strain plot did not exhibit any stress-plateau, which is
consistent with observations of as-brazed butt-joints in Chapter 4. The fracture strength
and strain is about 100 MPa and 10% higher than the pure Nb counterpart, respectively.
However, the strain at failure in most aged joints remained at about 13%, which is similar

to the pure Nb-eutectic joints.

Room-temperature superelastic behavior was observed in the base-NiTi of all

 

thermally-aged Zr-alloyed joints. Consistent with results in Chapter 4, aging
temperatures from 300°C to 350°C tend to produce higher and flatter upper-plateau
stresses, while stress-plateau become lower and less pronounced at higher aging
temperatures. However, fracture strengths of Zr-alloyed joints did not seem to be

affected by any of the aging regimes and remained close to 800 MPa even when armealed

176

at 500°C. In contrast, pure Nb-eutectic joints experienced a marked decrease in tensile

strength when they were subjected to annealing at 500°C prior to tensile testing.

500C/1 hr 3500/1.5hrs 3000/4hrs
(785 MP3) (755 MP3) (800 MPa)

 

  
    

 

 

 

 

900
4000/4hrs /

800 a (775 MP3)\ 1/

700 - // As-Brazed
.. / / (840 MP3)
«I at"
m 600 '
E
8 500 -
3
'2 400 -
E
'3 300 -

200 -

100 —

o I I I I I I I I F I I I I I

 

 

 

 

0 2 4 6 8101214161820222426283032
Tensile Strain (‘16)

Figure 5.8 Stress vs. strain plots of Zr-alloyed as-brazed and aged superelastic butt-
joint specimens. (Aging regime: 300°C for 4 hours; 350°C for 1.5 hours; 400°C for 4
hours; 500°C for 1 hour. All anneals performed in air and cooling by water—quench.
Initial joint clearance: Approximately 400 um.)

Fracture surface of Zr-alloyed eutectic joints that failed under tensile loading were
examined. The rough global fracture morphology in Figure 5.9a indicates ductile-mode
failure. Higher-magnification of the fracture surface (Figure 5.9b) reveals extensive
dimpling from micro-void coalescence, which is consistent with the ductile-tearing

behavior observed in the pure Nb-eutectic joints discussed in Chapter 4. No brittle-mode

fracture was detected in any of the Zr-alloyed eutectic joints. This was expected since the

joint microstructures did not reveal any presence of TizNi-type grains.

177

 

 

   

       

  
  

     

 
      

‘I\. t . :-‘ c

 

' .. _ . sex/5i ’ ,a .~

Figure 59 SEM micrographs howig (Z) gloal fracture mohology; ‘ and )
higlrly-dimpled fi'acture surface of a Zr-alloyed eutectic butt-joint that failed under tensile
loading.

Tungsten-alloyed eutectic butt-joints, prepared under similar braze conditions as
those described for the Zr counterpart, were also tested under tensile loading. Results are
shown in Figure 5.10. The stress-strain plot of as-brazed butt-joint revealed the tensile
strength to be 680 MPa and a failure strain of about 15%. The aged specimen failed at
tensile stress and strain close to 800 MPa and 13%, respectively. Hence, W-alloyed
joints share rather similar mechanical characteristics with pure Nb-eutectic butt-joints
with no distinct improvement in fracture strength. This is surprising since
nanoindentations and micro-Vickers hardness test results unanimously suggested
pronounced improvement in hardness of eutectic micro-constituents from micro-alloying

with tungsten.

178

 

 

 

 

 

 

 

900
3500/1.5hrs
800 - (800 MPa)
As-Brazed
700 4 (680 MPa)
A M
g 600 -
E
g 500 —
‘3 400 1
,g 300 - i
200 - L
100 . " ~
0 I l I I T I I
0 2 4 6 8 10 12 ' 14 16

Tensile Strain (%)

Figure 5.10 Stress vs. strain plots of W-alloyed as-brazed and aged superelastic butt-
joint specimens. (Aging regime: 350°C for 1.5 hom‘s. Anneal performed in air and
cooling by water-quench. Initial joint clearance: Approximately 400 um.)

Fractography of the failed W-alloyed eutectic butt-joints revealed a highly-
undulating surface (Figure 5.11a), which is a common observation among all eutectic
joints. Detailed study of the fracture morphology shows discontinuous regions with flat
and smooth topographies among the predominantly dirnpled surface. Trans-granular

cracks were also observed within these flat regions (indicated by ‘A’ in Figure 5 .1 1b),

which are believed to be associated with initiation of fracture across the TizNi-type

precipitates. Therefore, it can be deduced that despite pronounced strengthening of the

eutectic micro-constituents, presence of TizNi-type precipitates in W-alloyed joints could

have acted as crack initiation sites, consequently causing premature joint failure to occur.

179

    

    

I" .‘ fl.
Ln ‘

Figure 511 MrrnogrEcrap s hown(a) obal fractureIImohoCI, and (b) brittle
cracks (indicated by ‘A’) among dimpled fracture surface of a W-alloyed eutectic butt-
joint that failed under tensile loading.
5.3 Vanadium-Based Eutectic: An Alternative Braze System

It now seems apparent that solution-strengthening Nb-based eutectic braze joints
via micro-alloying is a promising approach to improve their mechanical properties.
Concurrently, there is also motivation to explore alternative braze filler metals that will
react and produce braze liquid in an analogous fashion with respect to Nb, but yield
superior joint strength and ductility. Since vanadium and Nb belong to the same group in
the periodic table, and literature [Prima 1992] suggests that many phase features in the
900°C isothermal section of the Ni-Ti-V ternary system are in common with those of the
Ni-Ti-Nb system, vanadium was selected as an alternative braze metal to be investigated.
Furthermore, vanadium possesses a room-temperature tensile strength of 350 MPa
[Maksimovich et al. 1982], which is higher than that of Nb (170 MPa) [Gupta and Suri
1994]. In the following sub-section, the microstructural and mechanical properties of

braze joints, prepared using pure vanadium filler metal, will be discussed.

180

5. 3. 1 Joint Microstructural and Mechanical Properties

Braze joint specimens were prepared using room-ternerature shape-memory NiTi
(nominal thickness of 3.175mm) as the base-alloy. Brazing procedures were similar to
those employed for Nb-eutectic joints, except that vanadium foils (nominal thickness of
51 um) were used instead. It was found that contact melting between NiTi and vanadium
occurs between 1160°C to 1180°C. The typical joint microstructures fabricated fiom a
braze temperature of 1185°C and duration of 120 seconds are shown in Figure 5.12.
Complete exhaustion of the vanadium foil was observed, which resulted in a final joint
width of about 240 um. With a braze foil thickness of 51 um, this implies that

approximately 190 pm of NiTi was consumed for braze liquid formation.

The braze region consists mainly of dendritic phase and eutectic micro-
constituents (Figure 5.12a). Elemental XRF mapping (Figure 5.13) and spot analysis
(Table 5.1) indicated that the dendritic phase is pro-eutectic NiTi with approximately 1

at.% V. This possibly suggests that solubility of vanadium in NiTi is lower than that of

Nb. Only trace amount of TizNi-type precipitates were observed in the V-eutectic joints,

which were believed to be analogous with the Ti-rich phase observed in Nb-based
eutectic joints. The NiTi-V eutectic micro-constituents was observed to possess a
‘network’-type morphology (Figure 5.12b), with the brighter phase being eutectic-V and
the gray phase being eutectic-NiTi. Average eutectic spacing within the grain boundaries
was determined to be 0.4 um. However, the eutectic phase tends to transit towards rod-

like near some of the boundaries with average spacing increasing beyond 1 pm.

181

 

 

 

" 5
,(V r
$19,; MAL“ ‘

{IN
‘

mg \
'\‘,\‘3

\

   

Figure 5.12 SEM micrographs showing (a) global joint microstructures; and (b)
eutectic micro- constituents of a V-eutectic braze joint.

 

Figure 5.13 Elemental XRF mapping of a V-eutectic joint.

182

Table 5.1 X-ray fluorescence analysis on selected spots in Figure 5.13.
Spot V (at%) Ni (at%) Ti (at%)
A 1 48 51
B 1 48 51

Nanoindentation was performed on as-brazed and thermally aged joint specimens,
where the latter was annealed at 350°C in air for 1.5 hours, followed by water-quench.
At least 20 indents were made on each specimen. The representative load vs.
displacement plots associated with indentations on as-brazed V- and Nb-eutectic are
given in Figure 5.14. It is clear that higher loads are required for the V-eutectic in order

to achieve the same indentation holding depth as the Nb-eutectic.

 

0.25
V-Eutectlc
0.2 e .-
' Nb-
_' Eutectic
E 0.15 r .
E ' °
1: ' °
u 0
O o
-’ 0.1 1 .' ,
0.05 4 .' .°
0 1 F I I I r I :1 I O I I

 

 

 

0 5 10 15 20 25 30 35 40 45 50 55 60
Displacement (nm)

Figure 5.14 Load vs. displacement plots of pure Nb- and V-eutectic micro-constituents.
(Indentation holding depth: 50 nm. Strain rate: 0.001 per second.)

183

 

Figure 5.15a and b illustrates the hardness and modulus of as-brazed joint in the
form of histograms, respectively. Data from the pure Nb counterpart are also included
for comparison. Calculations from the nanoindentation result showed that as-brazed V-
eutectic micro-constituents possess an average hardness of 3.6 GPa, which is
significantly higher than the mean hardness of Nb-eutectic micro—constituents. The
elastic modulus of V-eutectic was determined to be 78 GPa, being slightly lower than the
Nb counterpart. For the thermally-aged V-eutectic joint, no change in the hardness was
observed (Figure 5.16a). In contrast, the Nb-eutectic joint experienced a significant
improvement in hardness fi'om the same aging regime. The calculated elastic modulus of

both V- and Nb-eutectic joints appeared to decrease after annealing at 350°C.

 

 

 
 
  
 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

12- - - 8-

. (a) — N'T"Nb 1:]NiTI-Nb I (b) I::INm-Nb
10- / (2.0. GPa) 7: (57 GPa)

. - NiTi-V 6- NiTi-V - NiTi-V

s- (3.6 GPa) 5; (78 GPa)
5‘ . 6‘ .
gm — NiTi-V § 41 \ Nrri-Nb
E41 g3‘
LL ‘ u- .

. 2.

2. .

. n 1:

0 . 0 . -

1 4o 60 80 100 120 140

 

 

 

 

2Hardness (GPa)4 5 Modulus (GPa)

Figure 5.15 Histograms of (a) hardness; and (b) elastic modulus of as-brazed pure Nb-
and V-eutectic braze joints. Continuous lines are Gaussian distribution curves fitted over
the histograms of the two eutectic braze joints. (Data derived from indentation holding

depth of 50nm.)

184

 

 

 

 

 

12- (:1 NiTi-Nb (4.0 GPa) 14‘ [:1 NITI-Nb (82 GPa)

 

 

 

 

 

 

 

   

 

 

 

1011 (a) -NiTi-V(3.7 GPa) 12; (b) -Nrrrv (64 GPa)
s- 1°j
r ' 68
g 5‘ g NiTi-Nb
CT 1 0'6
9 41 2
LL J u- 4.
21 21
0 ‘ 0- ...
2 3 4 5 6 20 30 4o 50 60 70 80 90100110120
Hardness (GPa) Modulus (GPa)

Figure 5.16 Histograms of (a) hardness; and (b) elastic modulus of thermally-aged
pure Nb- and V-eutectic braze joints. Continuous lines are Gaussian distribution curves
fitted over the histograms of the two eutectic braze joints. (Aging regime: 350°C in air
for 1.5 hours, followed by water-quench. Data derived from indentation holding depth of

50nm.)

 

Next, butt-joint specimens were prepared for tensile testing. Room-temperature
superelastic NiTi was employed as base-alloy. Joint clearance was kept at about 400 pm,
which is consistent with most of the brazed butt-joints. Pure vanadium foils were used as
the braze filler metal. Brazing was performed at 1185°C for 6 minutes. Eutectic joints
were also subjected to aging at 350°C in air for 1.5 hours, followed by a water-quench.
Figure 5 .17 shows that the as-brazed V-eutectic joint failed at 780 MPa with fracture .
strain close to 20%, while aging the V-eutectic butt-joints at 350°C resulted in a tensile

strength of 920 MPa and a fracture strain close to 18%.

Both as-brazed and aged joints exhibited failure strengths fliat are comparable to
their respective control specimens (Figure 4.6 and 4.7, respectively). Hence, the tensile
test results strongly suggest that the V-based eutectic system can serve as a braze
alternative with the potential to have superior joint mechanical properties to the Nb

counterpart. However, vanadium is a toxic element that is not biocompatible.

185

Fracture occurred in the joint region of the as-brazed V-eutectic joint and the
facture surface was examined by SEM. Micro graphs revealed undulating general fracture
morphology (Figure 5.18a), which was a common observation among the Nb-based
eutectic joints and indicates primarily ductile failure. No brittle cracks were detected in
the as-brazed joint. However, hi gher-magnification of the fracture surface showed

interesting ‘web’-like features, which also suggests ductile cleavage (Figure 5.18b).

 

1000
350C/1 .5hrs

800 _ (780 MPa)

W
700 -

600 ~

400 1

Tensile Stress (MPa)
01
8

300 ~

200 -

100 —

 

 

 

 

o T I I I I I I T I I
0 2 4 6 8 10 12 14 16 18 20 22
Tensile Strain (%)

Figure 5.17 Stress vs. strain plots of as-brazed and aged room-temperature superelastic
butt-joints fabricated using V-based eutectic. (Aging regime: 350°C for 1.5 hours.
Anneal performed in air and cooling by water-quench. Initial joint clearance:
Approximately 400 um.)

 

Similar ‘web’-like features were observed in both low- and high-magnification
micrographs of the fracture surface originating from the 350°C-aged butt-joint (Figure

5.19a and b). This confirms that such features are not artifacts but truly pertain to failure

186

in the V-based eutectic system. Based on the dimensions and their resemblance with the
morphology of the V-based eutectic, these features are likely associated with failure in

the eutectic micro-constituents.

       

      
   

  

. oral . ~
Figure 5.18 SEM micrographsshowrng (a) globalfracture morphology; and (b) ‘web’-
like ductile tearing features associated with the eutectic micro-constituents of an as-
brazed V-based eutectic butt-joint that failed under tensile loading.

\

   

r . , “2;" f .j‘ .7 1 ‘

~‘- ' ' rift" . .1 - ". "If“ e I
Figure 5.19 SEM micrographs showing (a) global fracture morphology; and (b) ‘web’-
like ductile tearing features associated with the eutectic micro-constituents of a 350°C-
aged V-based eutectic butt-joint that failed under tensile loading.

187

5.3.2 Ternary Phase Equilibria in the Ni-Ti-VSystem

The Ni-Ti-V ternary system has received little attention and there is limited
literature on phase equilibria in this system. Prima and coworkers [Prima 1992] made a
preliminary attempt to study partial regions of the 900°C- and 1000°C- isothermal
sections. Besides the known binary phases, no ternary intermediate compounds were
reported. There is also no information on the liquidus projections or solidus boundaries
in the phase diagram. Based on Prima’s findings, the 920°C-isothermal section of the Ni-

Ti-V system was constructed and is shown in Figure 5.20. Note that 920°C was selected
for illustration because NizV phase melts congruently at about 915°C and phase fields
will appear unnecessarily complicated in the Ni-rich region if the 900°C-isothermal
section was drawn instead. Furthermore, the Ni-rich region of the phase diagram serves

no purpose in discussions pertaining to contact melting, which occurs close to the NiTi-V

isopleth.

Phase field features in the NiTi-o’-V-Ti section of Figure 5 .20 appear to be
analogous with the NiTi-Ni7Nb6-Nb-Ti region in Figure 3.31. Hence, itis reasonable to
hypothesize the presence of a similar quasibinary eutectic point at compositions close to
the NiTi-V isopleth. Indeed, experimental evidence had indicated that NiTi-V eutectic
formation in braze joints, which confirms the validity of the hypothesis and suggests

similarities in the driving force for contact melting and dissolution kinetics with the NiTi-

Nb eutectic system.

188

 

 

   
  
 

I 1 - Phase Fields
El 2 - Phase Fields

N
3 - Phase Fields

Figure 5.20 Estimated 920°C-isothermal section of the Ni-Ti-V ternary system [Prima
1992].

Stereology of the NiTi-V eutectic mixture revealed an estimated 57% volume
fraction of eutectic-V (Figure 5.21), which, upon density considerations, places the
quasibinary eutectic point approximately at 20Ni-20Ti-60V (at.%). However, the
thicknesses of NiTi and vanadium consumed for braze liquid formation did not seem to
agree with results obtained from stereology. Based on measurements of the receded
thicknesses of base-NiTi alloy derived from Figure 5.123 and other similar micrographs,

the quasibinary eutectic composition should approximately be 40Ni-40Ti-20V (at%).

189

 

'0
I

013

  
 
  

\
I

pl.
nu 9‘

2'
and

m .a
‘ A.“ '
t‘O‘s,'I
\1'
fr‘
‘
t’a‘f

I;
3%; e .
’15?

3;;

I

‘ N
\ r

s 1
.535

‘ b
fir:

.' Iv.
"fig;
'33 2
’fi

 

IM . \ ’5". 1' \

£38; 9;" ’ :1 ‘e‘ \t:
Wheat-1.2 is" #21:, at!
fifteen?!” "5“ ~73“;

... .. arikatastes - x . .
Actual SE Image Flooded

Flooded Area: Eutectic-NiTi I

Unflooded Area: Eutectic-V D

Percentage of Unflooded Area: approx. 57%
Figure 5.21 Stereology on NiTi-V eutectic micro-constituents (performed using
WSXM software from Nanotec Electronica).

It is believed that measurement errors are larger in stereology than in the
receding-thickness measurement method because the former approach relies on
measuring the 2D projections of 3D features on a plane, which is known for its
inaccuracy when complex and highly-anisotropic features are involved. An example is
illustrated in Figure 5.22. Volume fraction estimations on the 3D network are highly-
dependent on the exposed surface employed for stereology. Figure 5.22a shows that an
under-estimation of volume fi'action of the phase will occur if stereology was performed
on a surface that exposes only the out-of-plane rods of the network. On the other hand,

Figure 5.22b indicates that volume fraction of the phase will be over-estimated if

190

stereology was performed on a surface that exposes only in-plane rods. Given that the
eutectic nricrostructural features in Figure 5.12 resemble more of those illustrated in
Figure 5.22b, it is possibly a case of over-estimating the vanadium content of the eutectic

micro-constituents.

Exposed surface
for stereology

 

 

Under-estimation of
volume fraction

 

 

 

 

3-Dimensional network
of a phase

 

Over-estimation of
volume fraction

Figure 5.22 Schematic diagram showing a 3-dimensional network of a phase. (a)
Under-estimation of volume fraction when stereology is performed on out-of-plane rods
of the network. (b) Over-estimation of volume fraction when stereology is performed on
in-plane rods of the network.

The receding-thickness measurements on joints that had experienced complete
dissolution of vanadium will tend to under-estimate the vanadium content of the eutectic

liquid. The reason is that part of the dissolved base-alloy was spent in enriching the

liquid with Ni and Ti so as to drive the liquid composition against the NiTi liquidus

191

surface after the complete exhaustion of vanadium. However, melting kinetics

simulations in Chapter 3 had shown that the thickness of base-NiTi alloy required for this

 

purpose is only of the order of about 10 um. Hence, the error incurred in receding-
thickness measurement should not be significant, and the quasibinary eutectic

composition predicted using this method will be employed for further discussion.

Experimental observations have already suggested that NiTi-V contact melting 1'“
occurs between 1160°C and 1180°C. Therefore, it can be speculated that the NiTi-V
quasibinary eutectic temperature lies approximately at 1170°C, which is similar to the
case of NiTi-Nb. Based on Alkarnade’s principles and the knowledge of the quasibinary

eutectic point, two possible liquidus projections are proposed. Figure 5.23a shows a Ni-

Ti-V system with two class-I (E 1 and E 2) and two class-II (P 1 and P2) reactions, while
Figure 5.23b shows a ternary system with one class-I (E 1) and three class-II (P 1, P2 and

P3) reactions. Despite the differences in reaction schemes, the liquidus lines dernarcating

NiTi and V(Ni,Ti) phases possess similar temperature maxima of 1170°C at about 40Ni-
40Ti-20V (at.%) in both cases. In addition, both scenarios involve two four-phase class-

II reaction planes on either sides of the NiTi-V isopleth (Figure 5.24) that are similar to

 

those in the Ni-Ti-Nb system (Figure 3.32). The NiTi + V(Ni,Ti) + Liquid phase fields in
both reaction planes will converge towards the NiTi-V isopleth and degenerate into a
single NiTi + V(Ni,Ti) tie-line onto the plane of the said isopleth at approximately

1170°C, producing a perfect quasibinary eutectic isopleth (Figure 5 .25).

192

 
  
 
 
 
  
  
 
  
   
  

(a) .
I 1 - Phase Fields
El 2 - Phase Fields

Ni3Ti
3 - Phase Fields

0 Quasibinary
Eutectic Point

Ti

(b) "i
I 1 - Phase Fields

El 2 - Phase Fields

Ni3TI
El 3 - Phase Fields

0 Quasibinary
Eutectic Point

Figure 5.23 Two possible liquidus projections for the Ni-Ti-V system: (a) two class-I
(E 1 and E 2), two class-II (P 1 and P2) reactions; (b) one class-I (E 1), three class-II (P 1, P2
and P3) reactions. (Quasibinary eutectic points are also indicated.)

193

 

Figure 5.24 A four-phase class-II reaction plane (P1 and P2) on either sides of the
NiTi-V isopleth in the ternary system. Both possess common NiTi + V(Ni,Ti) + Liquid
phase fields. (The reaction plane temperatures are speculative.)

 

Centigrade 2500

Figure 5.25 A perfect NiTi-V quasibinary eutectic isopleth in the ternary system.

194

5.4 Vanadium-Alloyed Niobium Braze System

It was demonstrated in section 5.3 that a pure V-based quasibinary eutectic, which
yields promising joint mechanical characteristics, can be employed as an alternative braze
system. Besides having analogous mechanisms for eutectic liquid formation, the
coordinates of the quasibinary eutectic points in both ternary phase diagrams are also
believed to be close. Given that vanadium and Nb are mutually soluble, an attempt was
made to explore solution-strengthening of Nb-based eutectic joints by alloying with
vanadium. The microstructural and mechanical properties of Nb-based braze joints, with

different degrees of V-alloying, will be discussed in this section.

5. 4. 1 Joint Microstructural and Mechanical Properties

Room-temperature superelastic butt-j oints, with initial joint clearance of about
400 um, were fabricated using Nb and vanadium foils at the same time. Braze
temperature and duration were 1185°C and 6 minutes, respectively. Prior to brazing, the
filler metal foils were weighed and calculated to yield an approximate overall
composition of 54Nb-46V (at.%). Braze joints were metallographically-prepared and

studied by SEM.

Joints fabricated using 54Nb-46V braze foils possess predominantly eutectic
micro-constituents (Figure 5.26a) and pro-eutectic NiTi phase. However, the eutectic did
not possess attributes that are associated with either the NiTi-Nb or NiTi-V systems.
Instead, it is a mixture of rod-like and labyrinth morphologies. The eutectic spacing was

observed to be much wider than the two previously-discussed eutectic systems. The

195

darker phase in Figure 5.26 is believed to be pro-eutectic NiTi or eutectic-NiTi, while the
lighter phase is unidentified. For ease of discussion, the latter will be termed ‘x’ phase.
Discontinuous colonies of NiTi-Nb eutectic micro-constituents (region ‘B’ in Figure
5.26b) also coexisted with the NiTi-x eutectic micro-constituents (region ‘A’ in Figure
5.26b), but were not prolific. The microstructural transition from NiTi—Nb eutectic to x
phase in Figure 5.26b strongly implies that the latter might have formed subsequently
from a peritectoid reaction between prior-solidified eutectic-NiTi and eutectic-Nb. That

is, NiTi + Nb ——> x phase during cooling.

       

 

   

. A, a V "I ’. ‘ ‘ 4:) I \ 1‘.“- 1 . I.” 1
Figure 5.26 SEM micrographs showing (a) NiTi-x eutectic micro-constituents; and (b)
coexistence of NiTi-x eutectic (indicated by ‘A’) and NiTi-Nb eutectic (indicated by ‘B’)
micro-constituents in joints fabricated using 54Nb-46V braze foils.

The experiment was repeated using an overall braze foil composition of 77Nb-
23V (at.%) and its joint microstructures are provided in Figure 5.27a and b. More
colonies of NiTi-Nb eutectic micro-constituents were observed in the braze joint (Figure

5.27a), while majority of the x phase (indicated by ‘A’ in Figure 5.27a) no longer form

the eutectic micro-constituents previously observed in Figure 5.26a. Instead, the x phase

196

 

manifested as elongated grains that are usually coupled with the pro-eutectic NiTi phase

(Figure 5.27b).

          

x... .‘

Pie .2 1 SEM micrographs showing (a) el ated
‘A’) among dense colonies of NiTi-Nb eutectic; and (b) higher-magnification of the
elongated grains of x phase in joints fabricated using 77Nb-23V braze foils.

     

Elemental mapping of the microstructures in Figure 5.27b are provided in Figure
5.28. Based on the fluorescence intensities, the x phase possesses lower Ni and Ti
concentrations than the pro-eutectic NiTi phase, but the former has higher Nb and
vanadium concentration than the latter. When compared against the eutectic-Nb phase,
the x phase appears to have lower Nb concentration. The element maps for vanadium and

Ti did not provide clear intensity distributions between 9; phase and eutectic-Nb phase.

197

 

     

11 1rd ‘ , ' Ni/x'u.
Figure 5.28 Elemental XRF mapping of a joint fabricated using 77Nb-23V braze foil.
Next, nanoindentations were performed on as-brazed 54Nb—46V joints.
Indentation holding depth was kept at 100 nm with a strain rate of 0.001 per second. At
least 50 indents were made on the specimen. For hardness and elastic modulus
calculations, Poisson’s ratio of 0.35 was assumed for all the phases. Results were
illustrated in the form of histograms and shown in Figure 5.29a and b. A bi-modal
distribution was observed in the calculated hardness of the 54Nb-46V joint (Figure 5.29a),

suggesting that the nanoindenter was sampling two different phases. The first mode

198

 

appeared at about 4.5 GPa, while the second mode was estimated to be 9.5 GPa. Based
on previous nanoindentation results, it is reasonable to assume that the first mode is
associated with either pro-eutectic NiTi or eutectic-NiTi. Therefore, the second mode

must be attributed to the x phase.

In contrast, the histogram of calculated elastic modulus did not appear to be bi-
modal, but rather manifested as one broad distribution (Figure 5.2%). This is likely
attributed to closeness of elastic moduli of both phases causing an overlap of distributions.
From Figure 5.2%, it seems that the first and second mode occurred around 90 GPa and
150 GPa, respectively. The first mode agrees reasonably well with the elastic moduli of
NiTi-Nb eutectic mixture and base-NiTi alloy. Hence, it is provisionally concluded that

the x phase possesses an approximate modulus of 150 GPa.

        

 

  

      

 

18- 14-
16‘ 12-
14‘ 10
12- ‘
gm- g 8-
8 3‘ 8 6-
2 6- 2
LL u. 4-
44
2.
l 2 8 10 12 14 16 40 60 80 100 120 140 160 180 200

6
Hardness (GPa) Modulus (GPa)
Figure 5.29 Histograms of (a) hardness; and (b) elastic modulus of as-brazed 54Nb-
46V joints.
Tensile tests performed on aged butt-joints (3 50°C in air for 1.5 hours, followed
by a water-quench) fabricated using 54Nb-46V and 77Nb—23V foils revealed poor

fi'acture strength and ductility (Figure 5.30). The former failed at 200 MPa with a strain

199

of less than 3%, while the latter fractured at 390 MPa with a strain slightly beyond 3%.
The decrease in tensile strength and ductility as volume fiaction of x phase increases
strongly suggests that x phase is hard and brittle. This deduction is consistent with the

nanoindentation results.

 

400
77Nb—23V

350 ~ (335 Mpa)

54Nb-46V
(200 Mpa)

NNOO
888

 

1504

Tensile Stress (MPa)

 

 

 

os§

 

I I I I I I

0.5 1 1.5 2 2.5 3 3.5
Tensile Strain (%)

Figure 5.30 Stress vs. strain plots of aged room-temperature superelastic butt-joints

fabricated using 54Nb-46V and 77Nb-23V braze foils. (Aging regime: 350°C for 1.5

hours. Anneal performed in air and cooling by water—quench. Initial joint clearance:
Approximately 400 um.)

O

Fractography performed on the aged 54Nb-46V butt-joint revealed more brittle-
like general fiacture morphology as compared to the typical eutectic joints. Flat and
smooth topographies can be observed frequently in Figure 5.31 a. A hi gher-magnification
micrograph of the fracture surface shows discrete and sparse regions of ductile tearing

embedded in a matrix of cleavage fracture (Figure 5.3 lb). Brittle cracks were also

200

observed close to the ductile-brittle interfaces. Given the similarity in the features
between Figure 5.3 lb and 5.26a, it can be deduced that the ductile-mode fractures in
Figure 5.31b were attributed to failure in the pro-eutectic NiTi or eutectic-NiTi phase,

while the brittle cleavages were associated with the x phase.

      

- I r.
I I‘ " ._ A . ‘ ‘
...\. ‘ - V , .. '
. .«4 .- I . ... J A ~

Figure 5.31 SEM microaphs showing (a) glbbal fracture morphology; an (b) mixed
ductile/brittle fracture features in a 350°C-aged room-temperature superelastic butt-joint,
fabricated using 54Nb-46V foils, which failed under tensile loading. (Ductile features are
indicated by ‘A’, while brittle features are indicated by ‘B’.)

The ultimate tensile strengths of Nb braze joints were plotted as a function of
vanadium composition. Figure 5.32 shows that the ultimate tensile strength of Nb braze
joint decreases steadily from 770 MPa to 200 MPa when vanadium composition increases
from 0 to 46 at.%. However, pure vanadium braze joint possesses an ultimate tensile
strength of 920 MPa, which is superior to the pure Nb counterpart. The experimental
data were fitted with a parabolic trend (solid line in Figure 5.32), which predicts that the

weakest joint of 220 MPa will be produced at a vanadium composition of 47 at.% V.

This prediction is reasonable because joints fabricated from 54Nb-46V foils possesses an

201

almost 100% volume fraction of NiTi-x eutectic mixture, signifying that the brittle 1
phase has a predominantly continuous microstructure in the joints. The consequence of a

continuous brittle phase is a joint with lowest ductility and strength.

 

1000 I Experimental .’ ‘
Fitted Trend ,’
a ‘ _ _ _ _ . l
800 ' ,\ Fitted. Trend '1
I \\ (solution- . ’1 u
I strengthening at I ~
500. ‘1‘ low cone.) 1'

400 -

 

200 -

Ultimate Tensile Strength (MPa)

 

 

0 ' 20 ' 4'01 6'0 ' 8'0 ' 100
Braze Foil Composition (at%V - Bal. Nb)

Figure 5.32 Ultimate tensile strength of Nb braze joints as a function of vanadium
composition (experimental and fitted data shown).

Based on the solid trend line in Figure 5.32, increasing vanadium content beyond
47 at.% should improve the ultimate tensile strength. Such increase in ultimate tensile
strength can be explained by the emergence of NiTi-V eutectic and their progressive
substitution for the NiTi-x eutectic as vanadium composition increases. This predicted
trend, however, did not consider the possibility of solution-strengthening at very low
concentrations of vanadium or Nb. If Nb braze joints can be solution-strengthened by
low amounts of vanadium, vice versa, the dotted line in Figure 5.32 will be a more
accurate representation than the solid trend as it contains a maximum close to each end of

the composition scale.

202

5. 4.2 Quaternary Phase Equilibria in the Ni-Ti-Nb- VSystem

Given the complete solubility of vanadium in Nb, presence of x phase in the V-
alloyed Nb braze system was unexpected. However, after analyzing quaternary phase
equilibria of Ni-Ti-Nb-V, it became apparent that x phase is actually a quaternary
intermetallic compound. In this section, an attempt will be made to discuss phase
equilibria in the quaternary system and the origin of x phase. The understanding of the
Ni-Ti-Nb-V system will serve as an important basis for future attempts to alloy Nb-based

braze joints with other alloying elements of similar properties with vanadium.

With the addition of vanadium, Gibb’s phase rule dictates that an extra degree-of-
fieedom will be introduced in the phase equilibria. Hence, up to five phases can be in
mutual equilibrium. Phase fields in the quaternary system will be represented by the
following (corresponding degrees-of-freedom indicated in parenthesis):

One-phase — Entire phase volume (four degrees-of—frwdom);

Two-phase — Series of tie-lines (three degrees-of-freedom);

Three-phase — Series of tie-triangles (two degrees-of-freedom);

Four-phase — Single tie-tetrahedrons (one degree-of-fieedom);

Five-phase — Single-point (no degree-of-freedom).

A tetrahedron representing the 920°C-isotherm of the Ni-Ti-Nb-V quaternary
phase diagram is constructed and shown in Figure 5.33, with only essential phases

illustrated. From Figure 5.33, it can be seen that it is impossible to construct four-phase

tie-tetrahedrons among the NiTi, BCC-(Nb,V,Ti,Ni), TizNi, X5 and 0’ phases without

203

introducing a quaternary intermediate phase in the NiTi-Nb-Ti-V space of the phase

diagram (Figure 5.34).

Ti
Figure 5.33 920°C-isotherm of the Ni-Ti-Nb-V quaternary phase diagram showing
only the NiTi, BCC-(Nb,V,Ti,Ni), TizNi, X5 and 0’ phases and their tie-triangles.

 

This intermediate compound is, in fact, the x phase observed in previous section.
The position of x phase in the quaternary space is consistent with the elemental mapping
results in Figure 5.28. Figure 5.34 shows how x phase establishes equilibrium and forms
four tie-tetrahedrons with its surrounding phases. Given the complexity of illustration,

two phase diagrams were constructed to show how )5 phase establishes equilibrium with

NiTi, BCC-(Nb,V,Ti,Ni), TizNi and o’ in the Ni-Ti-V system (Figure 5.34a), and NiTi,

BCC-(Nb,V,Ti,Ni), TizNi and X5 in the Ni-Ti-Nb system (Figure 5.34b).

204

 

 

 

Ni
ti

 

Ti Nb
Figure 5.34 Possible 920°C-isotherm of the Ni-Ti-Nb-V quaternary phase diagram
showing how the x phase could establish equilibrium with (a) NiTi, BCC-(Nb,V,Ti,Ni),

TizNi and 0’ phases in the Ni-Ti-V system; or (b) NiTi, BCC-(Nb,V,Ti,Ni), TizNi and
X5 in the Ni-Ti-Nb system.

205

The two tie-tetrahedrons illustrated in Figure 5.34b each possess a tie-triangle of
NiTi + BCC-(Nb,V,Ti,Ni) + x as one of its faces. Since there are two degrees-of-freedom
for three-phase equilibrium in a quaternary system, the NiTi + BCC-(Nb,V,Ti,Ni) + x
phase field will be made up by the volume shown in Figure 5.35b, while the NiTi + BCC-
(Nb,V,Ti,Ni) phase field will be confined in a space between the NiTi + BCC-
(Nb,V,Ti,Ni) + 3; phase field and the Ni-Ti-Nb ‘face’ of the quaternary phase diagram

(Figure 5.35a). The space configuration of the NiTi + BCC-(Nb,V,Ti,Ni) phase field is

1 2
(b) NiTI + c + 800 - (Nb,V,TI,Ni)
3
4 1 2

(c) _ NiTi+BCC-(Nb,V,TI,Ni)

illustrated in Figure 5.35c.

 

       

Ti -
Figure 5.35 (a) Possible 920°C-isotherm of the Ni-Ti-Nb-V quaternary phase diagram
showing common NiTi + BCC-(Nb,V,Ti,Ni) + x tie-triangles as one of their faces of the
tie-tetrahedrons illustrated in the bottom of the phase diagram in Figure 5.31. Space
configurations of the (b) NiTi + BCC-(Nb,V,Ti,Ni) + x - phase field; and (c) NiTi +
BCC-(Nb,V,Ti,Ni) - phase field.

The confinement of the NiTi + BCC-(Nb,V,Ti,Ni) phase field to low vanadium
compositions implies limited V-solubility. Since braze liquid compositions of V-alloyed

Nb-based joints lies in the vertical NiTi-Nb-V section of the quaternary diagram and

206

should be in the vicinities of both quasibinary eutectic compositions, it is easy to see that
excessive vanadium will result in liquid composition crossing into the NiTi + BCC-
(Nb,V,Ti,Ni) + 75 phase field, consequently producing joint microstructures observed in
Figure 5.26 and 5.27. Stereology performed on the 54Nb-46V (Figure 5.26) and 77Nb-
23V (Figure 5.27) braze joints revealed approximate x phase volume occupancies of 60%
and 20%, respectively. By linearly extrapolating x phase volume fiaction to zero, the
composition of vanadium is calculated to be about 12 at.%. Therefore, the degree of V-
alloying must be lowered below 12 at.% in order to produce predominantly Nb-based
eutectic rrricro-constituents and, at the same time, possibly solution-strengthen the alloyed

braze joints.

5.5 Summary
It is evident fi'om indentation experiments that solution-strengtlrenin g of
mechanical properties of Nb-based eutectic phase can be achieved through micro-

alloying with Zr and tungsten, with the latter being a more effective strengthener.
However, TizNi-type phase formation can be suppressed by addition of Zr but not with
tungsten. Consequently, strengthening effect only manifested in tensile tests of Zr-
alloyed joints. On the other hand, crack initiation by embrittling TizNi-type precipitates

were believed to have occurred in W-alloyed joints, which caused the potentially-stronger

eutectic joint to fail prematurely.

A vanadium braze foil was explored as an alternative braze material. Contact

melting between NiTi and vanadium was observed, which also resulted in a

207

predominantly eutectic joint microstructure. Experimental evidence and phase
equilibrium analysis showed that a NiTi-V quasibinary eutectic point exists at
composition and temperature of approximately 40Ni-40Ti-20V (at.%) and 1170°C,
respectively. Nanoindentation and tensile test indicated that mechanical properties of
these joints are superior to the Nb-based counterpart. Given the toxic nature of vanadium,
it will not be compatible with bio-medical applications, but may find use in aerospace,

transportation or energy fields.

Alloying of Nb-based joints with substantial amounts of vanadium (23 at.% and
46 at.%) yielded joints with an unidentified quaternary intermetallic (termed ‘x’ phase)
and NiTi-Nb eutectic micro-constituents. The former was observed to increase in volume
fraction with increasing vanadium content. Nanoindentation studies and fiactography
showed that the x phase is hard and brittle, which seriously degraded tensile strength and
ductility. Analysis of the Ni-Ti-Nb-V quaternary phase equilibria indicated that 1 phase
is a quaternary intermediate compound and its formation in V-alloyed Nb-based joints
was attributed to excess vanadium. Hence, despite mutual solubility of Nb and vanadium,
the degree of V-alloying should be reduced further in order to prevent x phase formation,

while effecting solution-strengthening.

208

CHAPTER 6: CONCLUSIONS AND FUTURE WORK

6.1 Conclusions

A novel reactive eutectic brazing technique has been successfully devised to form
metallurgical bonds between NiTi alloys. The joining method relies on contact melting
between pure Nb filler metal and the base-NiTi to form the braze liquid. Brazing was
carried out under moderate high vacuum at temperatures above 1150°C. Contact melting
between NiTi and Nb is governed by the NiTi-Nb isopleth in the ternary phase diagram,

which closely resembles a perfect quasibinary eutectic isopleth.

When 51 rim-thick Nb foils were employed as the braze filler metal, the brazing
process can be divided into 3 stages: (1) initial melting, (2) liquid propagation, and (3)
termination. The driving force for initial melting is due to the presence of a
thermodynamically-forbidden local concentration at the initial NiTi-Nb mating interface.
The braze liquid formed fi'om initial melting can dissolve any surface oxide present.
Subsequent melting or liquid propagation is driven by the system’s attempt to
simultaneously maintain different NiTi and Nb equilibrium compositions at the NiTi-
Liquid and Nb-Liquid interfaces, respectively. The rate of melting is not controlled by
heat transfer kinetics, but by diffusion of Ni, Ti and Nb across the liquid. A diffusion-
based model was established to predict the contact melting kinetics. Simulation suggests
that rate of melting decreases rapidly as braze time and liquid width increases. The

diffusion-based model also predicted the diffirsivities of Ni (and Ti) and Nb in the liquid

209

 

to be 2.10 x 10.9 mZ/s and 1.55 x 10'9 m2/s, respectively. Experimental results revealed

that the NiTi and Nb melts at a constant molar ratio of about 1.4NiTi: le (at.%).

Upon complete melting of Nb, the base-NiTi alloy continued to melt.
Termination of contact melting occurred only when the liquid attained the NiTi liquidus
composition. Hence, a 51 [rm-thick Nb foil will ultimately produce a maximum liquid
width that is approximately 200 um. Due to the slow diffusivity of Nb in NiTi,
isothermal solidification was insignificant within an hour of braze time when such filler
foil thickness was used. On the other hand, isothermal solidification was observed afier a

braze duration of 15 minutes when 6 [rm-thick Nb films were employed as the filler.

The presence of a quasibinary eutectic point at 38Ni-36Ti-26Nb (at.%) in the
ternary system caused the braze liquid to solidify predominantly into eutectic NiTi-Nb
micro-constituents and pro-eutectic NiTi phase upon cooling. The pro-eutectic NiTi
phase assumed a planar solidification fi'ont if fi'eezing occurred prior to Nb exhaustion,
but promptly evolved into a dendritic morphology if the Nb were melted completely prior
to solidification. The transition from a planar to a dendritic morphology is attributed to
instability at the solidification front due to constitutional supercooling. Due to fast
cooling rate associated with the termination of brazing, sub-grain boundaries appeared
within the pro-eutectic NiTi dendrites. Some of the sub-grain boundaries may be

decorated by fine Nb-rich precipitates.

210

 

The majority of the eutectic-Nb phase in the joint possessed lamellar morphology,
although there were also regions that were rod-like. Trace amounts of facetted
precipitates were observed within the eutectic phase in joints that were brazed for more

than 120 seconds. Selected-area diffraction from TEM revealed that these precipitates

have crystal structure that is related to the TizNi phase. Further experimental work
suggests that they are (Ti,Nb)4Ni20 sub-oxides, which formed due to stabilization of

(Ti,Nb)2Ni phase by oxygen contamination.

No microstructural changes were observed in SEM when brazed joints were

thermally aged below 800°C. But at 800°C, (Ti,Nb)4Ni20 sub-oxides precipitates began

coalescing to form larger grains. At 920°C, these sub-oxides appeared to be solutionized
in the eutectic matrix at 920°C, while the eutectic-Nb phase evolved from rod-like into
spheroidal morphology. Spheroidization of the eutectic-Nb phase can most likely be
attributed to the minimization of interfacial energy between eutectic-NiTi and eutectic-

Nb phases.

Nanoindentation studies revealed that as-brazed eutectic joints possess an average
hardness of 2.0 GPa. When thermally aged at 350°C for 1.5 hours, joint hardness can be
increased to 4.0 GPa, which is comparable to the base-NiTi alloy hardness. Elastic
moduli of as-brazed and aged joints measured by the Oliver and Pharr method yield
values in the range of 80 to 95 GPa, which are also comparable to the moduli of base-

NiTi alloys of corresponding conditions. Nanoindentation experiments also revealed the

211

mechanical properties of (Ti,Nb)4NizO precipitates. Average hardness and elastic

modulus of these sub-oxides were estimated to be 12.8 GPa and 175 GPa, respectively.

Results from tensile tests performed on butt-joints show that the braze process
caused degradation in the base-NiTi’s strength and ductility, but superelastic response
could be restored in the base-NiTi alloy through post-braze thermal aging. It was found
that the optimum aging temperature is around 350°C for 1.5 hours, where the highest
tensile strength and plateau stresses of 770 MPa and 380 MPa were observed,
respectively. However, the eutectic joints are still mechanically weaker than the base-
NiTi alloy, which has a ultimate tensile strength of about 900 MPa after being subjected
to the braze and optimum aging temperatures. The lower eutectic braze strength may be

attributed to either the strength-limiting eutectic-Nb phase or the presence of embrittling

TizNi-type precipitates serving as crack initiation sites for premature joint failure.

Fractography revealed extensive ductile tearing and micro-void coalescence, strongly

suggesting that the joint is primarily ductile.

Tungsten-inert gas welding was successfully adapted for the formation of eutectic
liquid at the joint area. The microstructure at the joint revealed pro-eutectic NiTi
dendrites, eutectic micro-constituents and another anomalous dendritic phase that was
believed to be a complex quaternary oxide. An ultimate tensile strength of 750 MPa was
achieved using this approach, which is comparable to the brazed counterparts. However,
TIG welded joints suffered poorer ductility, probably due to the presence of the complex

quaternary oxide.

212

 

A preliminary attempt was also made to join NiTi alloys to Ta and alumina using
the reactive eutectic braze process. Necking was observed in a NiTi-Ta butt joint and the
stress-strain plot revealed an ultimate tensile strength of about 210 MPa, which strongly
indicates that the joint possesses strength that exceeds that of the Ta base-metal. Wetting
and flow of the braze liquid was observed on alumina substrates. However, thermal
shock apparently caused weakening of the alumina near the joint interface. As a result,

the NiTi-alumina lap-joint specimen suffered poor shear and tensile strengths.

Solution-strengthening of Nb-eutectic braze joints through micro-alloying with Zr
and tungsten was studied. The addition of l at.% Zr to the Nb braze resulted in a
microstructure free fi'om the Ti-rich sub-oxide precipitates and porosity even when
brazing was performed on oxidized samples under poor vacuum conditions. On the other
hand, the addition of tungsten (~ 2 at.%) appeared to encourage precipitation of the Ti-
rich sub-oxide and also increased the eutectic spacing. The latter is likely caused by the
slow-diffusing nature of tungsten. Micro- and nanoindentation studies showed that
hardness of the eutectic micro-constituents can be improved by micro-alloying with both
Zr and tungsten. Distinct improvements in tensile strength were, however, only observed
in Zr-alloyed eutectic joints, where an ultimate tensile strength of 850 MPa was achieved.

Tungsten-alloyed joints did not produce an increase in strength, probably because the

adverse effects of TizNi-type precipitates outweigh the improvement in strength of the

eutectic-Nb phase.

213

 

An alternative braze material system was also explored. Pure vanadium was
employed as the braze filler metal for the reactive eutectic braze process. Contact
melting between NiTi and vanadium was observed to occur around 1170°C. The contact
melting mechanism is believed to be similar to that of the NiTi-Nb system. Because of
the likely presence of a perfect quasibinary eutectic isopleth in the Ni-Ti-V system, V-
eutectic braze joints have a predominantly eutectic microstructure, which has a network-
like morphology. Pro-eutectic NiTi dendrites were also observed to solidify against the
unmelted base-NiTi alloy. The NiTi-V quasibinary eutectic point was estimated to be at
40Ni-40Ti-20V (at.%) at a temperature of 1170°C. Nanoindentation results revealed that
as-brazed V-eutectic joints have higher hardness than the Nb counterparts, while no
significant difference in hardness was observed in the 350°C-aged joints. However,
tensile tests indicated that 350°C-aged V-eutectic butt-joints possess higher ultimate
tensile strength as compared to as-brazed Specimens. An ultimate tensile strength of 920
MPa was achieved in the aged V-eutectic joint, which is the highest reported in this work.
Similar to the Nb-eutectic joints, fractography performed on the V-eutectic joint also

revealed primarily ductile-tearing.

Alloying of Nb braze joints with vanadium was also explored in this work. Two
alloying compositions were studied. They were 54Nb-46V and 77Nb-23V (at.%). The
former produced predominantly NiTi-x eutectic micro-constituents in the joint, while the
latter resulted in elongated grains of x phase in a mainly NiTi-Nb eutectic microstructure.
Nanoindentation revealed that the x phase have hardness and elastic modulus of 9.5 GPa

and 150 GPa, respectively. Tensile tests on these V-alloyed Nb braze joints indicated

214

poor tensile strength and ductility for both alloying compositions. The 54Nb-46V joint
has a tensile strength of 200 MPa, while the 77Nb-23V joint failed at 390 MPa.
Fractography performed on the 54Nb-46V and 77Nb-23V joints revealed predominantly
smooth and flat cleavage surfaces. Hence, nanoindentation, tensile tests and fractography

all suggest that the x phase is hard and brittle.

Phase equilibrium analysis of the Ni-Ti-Nb-V system indicated that x phase is a

quaternary intermediate compound that mediates equilibrium among the NiTi, BCC-

(Nb,V,Ti,Ni), TizNi, X5 and 6’ phases. The x phase was believed to have formed via a

peritectoid reaction between the NiTi and BCC-(Nb,V,Ti,Ni) phases druing cooling due

to high V-composition in the braze joints.

6.2 Future Work

Given the fact that Nb is mutually soluble with many elements, solid-solution
strengthening appears to be the promising choice to improve the mechanical properties of
brazed joints. Indeed, nanoindentation and tensile results have demonstrated that micro-

alloying Nb filler metal with Zr and tungsten have the potential to produce stronger braze

joints. Furthermore, it was observed that (Ti,Nb)4NiZO formation can be suppressed by
the addition of Zr. Positive results from both alloying candidates suggest that further
improvements in the mechanical properties of eutectic braze joints may be achievable

through simultaneous alloying of Nb with Zr and tungsten. Since vanadium can also be

solution-hardened using Nb and tungsten [Maksimovich et al. 1982], strengthening of

215

 

vanadium-eutectic joints may also be achievable through micro-alloying using these

elements. Tantalum and Zr are also potential candidates.

The ability to re—solutionize the TizNi-type precipitates in the Ni-Ti-Nb system
opens up the possibility of strengthening Nb-eutectic joints by precipitation-hardenin g.

The precipitation of TizNi-type phase after solution-treatrnent can be controlled by

thermal aging. To maximize the strengthening effect, coherent TizNi-type precipitates

should be finely-dispersed in the eutectic matrix as GP-zones. With a thorough
knowledge of the quaternary phase equilibria and the ability to control x phase formation

in the Nb braze joints, it is believed that precipitation-hardening of Nb-eutectic joints can
also be achieved with the x phase using a similar approach for the TizNi-type precipitates.

Therefore, it is critical to understand the physical metallurgy of x phase in the joint.

Despite the poor mechanical properties of the NiTi-alumina joint, reasonably
good wettability and flow of the braze liquid on alumina show that the braze process is
still a promising approach for joining NiTi alloys to alumina and possibly other ceramic
materials. Thermal shock, responsible for weakening of the alumina near the mating
interface during cooling, may be minimized by employing a lower cooling rate. Also,
recent preliminary work has also demonstrated wettability and flow of the braze liquid on
hydroxyapatite (HAP), although there is no mechanical data from a joint between NiTi
and HAP. Hence, it will be of great interest to investigate the tensile and shear properties

of the NiTi-HAP joint, and melting reaction among NiTi, Nb and HAP.

216

 

APPENDIX A
Numerical calculations on heat transfer kinetics, involved in the ‘incubation’

period of initial melting, were written in FORTRAN 77. An explicit finite-difference

scheme using the 2nd-order Runge-Kutta method was applied to Eq. (3.1) — (3.3).

The FORTRAN code, input parameters and summarized results are listed below.

PROGRAM incubation time

 

Program created by Kebin Low
CHEMS Department, MSU
March 10th, 2008

This program performs numerical iterations to calculate
the incubation time for initial melting.

 

*i§*§*§*§*§

Declarations

implicit none

integer‘4 t,k,numstep

real‘8 conducgheatcap,cte_niti,tempext,tempint(3 0000 1)
real‘8 density,correctdensity,deltime,tottime

real"'8 delq(300001),nitithiclmess,tempdifl(300001)
real*8 nbthickness,melttemp

* Print on-screen menu
write("',"')"”
writc(*’*)n n
mite(",*)"This program performs numerical iterations to"
mite("',*)"calculate the incubation time for initial melting."
write("',*)" "
write(*,*)""

 

 

* Prompt user inputs
mite(*,"')"1nput desired time frame (sec):"
read(*,*) tottime
write(",*)"lnput desired total number of steps (0-300000):"
read(*,"') numstep
write(*,¢)nn
write(*,*)"Please Wait. . . "
WTIIC(T,*)""

Dividing the total time by the number of steps
deltime=tottime/float(numstep)

* Setting all time-variables to zero initially
do 10 t=l,numstep

217

tempdiffIt)=0
delq(t)=0

10 continue

Ill
1|:

Continue with user prompts

write(*,*)"lnput NiTi Plate Thickness (m):"

read(",*) nitithickness

write(*,")"Input Thickness of Nb Foil (m):"

read("‘,*) nbthickness

write(*,"')"lnput Thermal Conductivity of NiTi (W /m.K):"
read(*,"') conduct

write("',"‘)"Input Specific Heat Capacity of Nb (J/g.K):"
read(*,*) heatcap

write(*,"')"lnput External Temperature (Degrees Celsius):"
read("',"') tempext

write("',"')"lnput Initial NiTi-Nb Int. Temp. (Degrees Celsius):"
read("',*) tempint(l)

writc(fi,¢)rm

write("',*)"Calculation in Progress . . . "

writecp’iprm
“w(#,*)"***####*****###******#***#**#*****#**********#***#****"
write(*,*)"'raw.dat', 'output.dat' and 'results.dat"'
write("',"')"will be generated at the end of calculations."
WTIIC(‘,.)""

write("',*)"See 'results.dat' for details."
“w(fi,l)"*ti*¥*¥**#$¢¥*****t*¢*********#*********tt********#*#*"

Opening output files

open(unit= l ,file='raw.dat’,status='unknown')
open(unit=2,file='outputdat',status=‘unlmown')
open(uniF3 ,file='results.dat',status='unknown')

Writing inputs into results.dat

write(3,")"lnput Parameters"
wfitc(3,e)vaWWW"
write(3,*)"Time Frame:",tottime," seconds"

write(3,"')"Total Number of Steps:",numstep

write(3,")"Nb Foil Thiclmess:",nbthickness," m"
write(3,*)"Thermal Conductivity of NiTi:",conduct," W/m.K"
write(3,*)"Specific Heat Capacity of Nb:",heatcap," J/g.K"
write(3,"')"External Temperature:",tempext," C"
write(3,*)"lnitial NiTi-Nb Temperatrrre:",tempint(l)," C"
write(3,¢)mi

Density of NiTi (g/m3)
density=8.57

Define melting temperature
melttemp=l 170.0

Calculating thermal expansion at braze temperature
cte_niti= 1+(7. 1E-6 *tempext)

Corrected density
correctdensity=density/(cte_niti"3)

218

 

* Calculating the temperature difference
tempdiff(l)=tempext-tempint(l)

e
"‘ Print in-calculation message

write(t,*)mi

write(*,*)"Please wait... Calculation in progress..."
*

* Numerical iterations
do 20 t=l ,numstep

* Check if NiTi thickness limit has been exceeded
if(tempint(t).lt.melttemp) then

 

 

Q

* Calculating change in heat
delq(t)=(conduct*tempdifl‘(t)*deltime)/nitithickness

*

"‘ Calculating new interface temperature
tempint(t+1)=((delq(t)/heatcap)/(correctdensity*nbthickness* 100))
C+tempint(t)
tempdiff(t+ 1 )=tempext-tempint(t+ 1)

t

"' Write results to raw.dat
write( 1 ,")t"'deltime,delq(t),tempint(t),tempdiff(t)

1.
endif

*
if(tempint(t).gtmelttemp) then

i

* Write data points for plotting into outputdat

"' do 21 k=l ,t,200

* write(2,*)k‘deltime,delq(k),meltthickness(k),vniti(k)

* 21 continue

it

"' Print important information with regards to Nb exhaustion
write(3,*)"lnforrnation Associated With Incubation Time"
WTIIC(3,*)" 11
write(3,‘)"Time:",(t-l)‘deltime," seconds"
write(3,*)"NiTi-Nb Interface Temperature",tempint(t-l)," C"

It

"‘ Print End-of-Program message
write(3,*)""

write(3,*)"Files Generated"
wfite(3’#)"*****titit.titttttiit****#*#****#¢*#*#***************¢"
write(3,*)'"raw.dat', 'output.dat', and 'results.da "'

write( 3 , l")" have been generated."

write(3 ,"')" "

write(3,"‘)"'raw.dat' contains all time-evolution data."

write(3 , *)"Data is in the following sequence: "

write(3 ,"')"Time, Change in Heat, Interface Temp. and Temp. Diff."
write(3 ,*)""

write(3 , *)'"results. dat' contains most-recent input parameters"
write(3,*)"and calculated values."

write(3,*)""

write(3 ,‘)"END OF PROGRAM"

write(3 *)"****###**t$************#******#**********************#"
9
write(3,*)""

219

1|-

Close output files
close(unit=1)
close(unit=2)
close(unit=3)

*

stop
in:

endif

it

20 continue
*

end

Table A-1 Input parameters for calculating incubation time for initial melting.

 

 

 

Input Parameters Values
Time Frame (secs) 1E-03
Total Number of Steps 50000
Nb Foil Thickness (m) 2.55E-05

Thermal Conductivity of NiTi (W lm-K) 41
Specific Heat Capacity of Nb (J/g—K) 0.265

 

 

 

External Temperature (°C) 1185
Initial NiTi-Nb Temperature (°C) 300
Table A-2 Results from the incubation time calculation.
Results Values
Target NiTi-Nb Interface Temperature (°C) 1169.9

Time Taken to Attain Target Temperature (secs) 1.76E-06

220

APPENDIX B
Numerical calculations on melting kinetics before complete melting of the solid

Nb, on the assumption that it is rate-limited by diffusion in the liquid, were written in

FORTRAN 77. An explicit finite-difference scheme using the 2nd-order Runge-Kutta

method was applied to Eq. (3.13) -— (3.17). The FORTRAN code, input parameters and

summarized results are listed below.

PROGRAM eutectic

 

 

Program created by Kebin Low
CHEMS Department, MSU
July 27th, 2007

This program performs numerical iterations to simulate
the NiTi-Nb eutectic half-width evolution as a function of time
before exhaustion of the Nb interlayer. '

The program assumes:

(1) Density of solid and liquid phases are the same;

(2) The NiTi/L and Nb/L interfaces attempt to maintain
their respective liquidus compositions;

(3) A linear concentration gradient hence exists across
the 2 solids;

(4) Picks 1st law holds in the eutectic liquid;

(5) Dissolution process is diffusion-controlled;

(6) Heat transfer does not play a rate-limiting role; .

(7) Solid-state diffusion is negligible within the simulation
time-frame.

 

*ifIIIIIG'I'II'N’CI'Iitl‘O'I'I‘N'GO‘IQO

Declarations

implicit none

integer*4 t,k,numstep

real’8 jnb(300001),jniti(300001),xnb(30000l),xniti(300001)
real*8 w(300001),vnb(300001),vniti(300001),cte_nb,cte_niti,temp
real‘8 nbconc,latticenb,latticeniti,deltime,tottime

real*8 nbthickness,c l ,c2,c3 ,e4,rho_nb,rho_niti,rho_nb1 ,rho_nitil
real*8 difiusion_nb,diffusion_niti

* Print on-screen menu
write(*,*)mo
Wfitfi(*,‘)" n
write("',‘)"This program performs numerical iterations to simulate
write(‘,*)"the NiTi-Nb eutectic half-width evolution"
write(*,*)"as a firnction of time before exhaustion of the"
write("',*)"Nb interlayer."

 

221

 

“$0.31)" 11
write(*,"')" 11

Prompt user inputs

write(*,"‘)"1nput desired time frame (sec):"

read(*,*) tottime

write(*,*)"1nput desired total number of steps (0-300000):"
read(*,*) numstep

Wl'it8(*,*)" 11

write("‘,"')"Please Wait. . . "

WFIIC(T,T)""

Dividing the total time by the number of steps
deltime=tottime/float(numstep)

Setting all time-variables to zero initially
do 10 t=l,numstep
inb(0=0
jniti(t)=0
vnb(t)=0
vniti(t)=0
xnb(t)=0
xniti(t)=0
W(t)=0

10 continue

*
t

*

Continue with user prompts

write( *,"')"Input Initial Nb Interlayer Half-Thickness (m) : "
read("',"') nbthickness

write("‘,*)"lnput Initial Liquid Half-Width (m): "

read(*."') W(1 )

write("', *)"Input lattice Parameter of Nb (angstrom): "
read(*,*) latticenb

write(*,*)" Input Lattice Parameter of NiTi (angstrom): "
read(*,"‘) latticeniti

write(*, *)" Input Braze Temperature (Degrees Celsius): "
read(*,*) temp

write(‘,*)" Input NiTi Solidus Composition (atomic % Nb): "
read?!) cl

write( *, *)"Input Nb Solidus Composition (atomic % Nb):"
read(*,*) 04

write(*,"‘)" Input Self-Diffusion Coefficient of Nb (m"2/s) : "
read("',"') diffusion_nb

write("',*)"Input Self-Diffusion Coefficient of NiTi (m"2/s) : "
read(*,"') diffusion_niti
WTitC(*,*)" or

write(", *)"Calculation in Progress . . . "

write(¢,¢)mi
write(Il,*)" tit**************#***titittittiiitiflttt##1##*¢***##*****"
write(* , *)"'raw.dat', 'output.dat', and 'results.dat"'
write(*,*)"will be generated at the end of calculations."
write(ip, arm
write( *, *)" See 'results. dat' for details."
write(t,*)"*ttlhliflllt*Ilnt**********#*******#***Iliakttilttllnklhttitttlhhtt"

Calculating the liq. comp. of Nb and NiTi at selected temp.

222

c2=( 13 l 0-temp)*(float(26)/float( 140))
c3=(temp-(2468-(float( l 298)‘ 100/float(74))))/
C(float( l 29 8)/float(74))

Opening output files

open(uniF l ,file='raw.dat',status='unknown')
open(unit=2,file='output.dat',status=‘unknown')
open(unit=3,file='results.dat',status='unknown')

It

* Writing inputs into results.dat
write(3,*)" Input Parameters"
mite(3,*)"WWWWW"
write(3,*)"Time Frame:",tottime," seconds"
write(3,*)"Total Number of Steps:",numstep
write(3,*)"lnitial Nb Interlayer Half-Thickness:",nbthickness," m"
write(3,‘)"Initial Liquid Width:",w(1)," m"
write(3,*)"Lattice Parameter of Nb:",latticenb," Angstroms"
write(3,*)"Lattice Parameter of NiTi:",latticeniti," Angstroms"
write(3,‘)"Braze Temperature:",temp," Degrees Celsius"
write(3,‘)"NiTi Solidus Composition:",cl
write(3,*)"Nb Solidus Composition:",c4
mite(3,"')"Nb Diffusion:",difiusion_nb," m"2/s"
write(3,*)"NiTi Diffusion:",diffusion_niti," m"2/s"
write(3,¢)vm

* Calculating thermal expansion at braze temperature
cte_nb=l+(7.3E—6"'temp)
cte__niti=1+(6.6E-6"'temp)

"' Calculating the number density of Nb and NiTi
rho_nb=2/(latticenb"‘lE-lO’cte__nb)“3
rho_niti= l /(latticeniti"' lE-10*cte_niti)"' *3

rho_nb l =((c4/ 100.0)*rho_nb)+(((100.0-04)/100.0)*rho_niti)
rho_niti1=((c1/100.0)*rho_nb)+(((100.0-c l)/100.0)*rho_niti)

1"

Calculating the Nb concentration difference in #Nb/m"3
nbconc=(c3/(((c3/2.0)*((latticenb*lE-lO‘cte_nb)"'*3))
C+((100.0-c3)"‘((latticeniti*lE-10*cte_niti)*"'3))))
C-(c2/(((62/2.0)*((latticenb"'lE-10*cte_nb)**3))
C+((100.0-c2)*((latticeniti"1E-lO‘cte_niti)**3))))
a:
* Print in-calculation message
write(*,¢)1m
write(*,*)"Please wait... Calculation in progress..."

* Numerical iterations
do 20 t=l,numstep

* Check if Nb interlayer thickness has been exceeded
if(xnb(t).lt.nbthickness) then

* Calculating the Nb and NiTi flux in the liquid

jnb(t)=diffusion_nb*nbconc/w(t)
jniti(t)=diffusion_niti*nbconc/w(t)

223

G

‘I’

t

21

as
It!

Calculating the Nb/L interface velocity
vnb(t)=jnb(t)/rho_nb1
vniti(t)=jniti(t)/rho_nitil

Calculating the position of Nb/L and NiTi/L interfaces
xnb(t+l)=xnb(t)+(vnb(t)*deltime)
xniti(t+l)=xniti(t)+(vniti(t)*deltime)

Calculating the width of the liquid
w(t+1)=xnb(t+l)+xniti(t+ 1 )

Write results to raw.dat
write(l,*)t*deltime,jnb(t),jniti(t),vnb(t),vniti(t),xnb(t),
aniti(t),w(t)

endif
if(xnb(t).gtnbthiclmess) then

Write data points for plotting into outputdat

do 21 k=l,t,200

write(2,*)k*deltime,vnb(k),vniti(k)
write(2,‘)k*deltime,w(k),xnb(k),xniti(k),vnb(k),vniti(k)
continue

Print important information with regards to Nb exhaustion

write(3,*)"Ca1culated Parameters"
wfite(3,*)"WWW~M~W~m~"
write(3,*)"Density of Pure Nb:",rho_nb," #/m3"

write(3,*)"Density of Pure NiTi:",rho_niti," #/m3"

write(3,‘)"Density of Pro-eutectic Nb:",rho_nbl," #/m3"

write(3,*)"Density of Pro-eutectic NiTi:",rho_nitil ," #lm3"

write(3,*)"NiTi Liquidus Composition:",c2

write(3,*)"Nb Liquidus Composition:",c3

write(3,"')"Nb-Concentration Diff. Across Liquid:",nbconc," #/m3"

write(3,"')" II

write(3,‘)"lnformation Associated With Nb Exhaustion"
write(3,‘)"mfi-~vv~m-WW"
write(3,*)"Time:",t*deltime," seconds"

write(3,*)"Thickness of Nb Exhausted:",xnb(t)," m"

write(3,*)"Thickness of NiTi Exhausted:",xniti(t)," m"

write(3,*)"Eutectic Liquid Half-Width:",w(t)," m"

Print End-of-Program message

write(3 ,*)n It

write(3,*)"Files Generated"
write(3,*)"*ilittttt*i****##*##*¥***¥*¢ititttti*****#************#"
write(3,"')"'raw.dat', 'output.dat', and 'results.da '"

write(3 ,*)" have been generated."

write(3 , *)n n

write(3 ,*)"'raw.dat' contains all time-evolution data."
write(3 ,*)"Data is in the following sequence"

write(3 ,*)"Time, Nchlux, NiTi-Flux, Nb/L Velocity,"
write(3 ,*)"NiTi/L Velocity, Nb/L Displacement,"

write( 3 , *)"NiTi/L Displacement, Eutectic Liquid Width"
write(3, *)""

224

write(3,‘)'"output.dat' contains plot-fiiendly data points."
write(3,*)"Data is in the following sequence:"

write(3,"')"Time (200-step interval), Eutectic Liquid Half-Widtlr,"
write(3,*)"Nb/L Displacement, NiTi/L Displacement,"
write(3,*)"Nb Interface Velocity, NiTi Interface Velocity"
Wfite(3,*)""

write(3,*)"'results.dat' contains most-recent input parameters"
write(3,*)"and calculated values."

mite(3,"‘)""

write(3,*)"END OF PROGRAM"

write(3 e)"eeseeesetetteessssseettsaltar-estates:are-Hsuwstntutuwtsw
’
write(3,"‘)""

* Close output files
close(unit= 1 )
close(unit=2)
close(unit=3)

1.1

 

stop
*

endif

41

20 continue
*

end

Table B-1 Input parameters for simulating the melting kinetics before the complete
melting of Nb (assume rate-limited by liquid phase diffusion).

 

Input Parameters Values
Time Frame (secs) 60

Total Number of Steps 3E05
Initial Nb Interlayer Half-Thickness (m) 2.55E-05
Initial Liquid Width (m) 1E-08

Lattice Parameter of Nb (Angstroms) 3.307
Lattice Parameter of NiTi (Angstroms) 3.015

Braze Temperature (°C) 1185
NiTi Solidus Composition (at.% Nb) 7

Nb Solidus Composition (at.% Nb) 91

Nb Diffusion Coefficient (mz/s) 1.55E-09
NiTi Diffusion Coefficient (mZ/s) 4.10E-09

 

Table B-2 Results fi'om the melting kinetics calculation (before the complete melting
of Nb, assume rate-limited by liquid phase diffusion).
Results Associated with Complete Melting of Solid Nb Values

 

Thickness of Melted Nb (m) 2.55E-05
Thickness of Melted NiTi (m) 9.55E-05
Liquid Half-Width (m) 1.21E-04
Elapsed Time (secs) 33.51

 

225 I

APPENDIX C
Numerical calculations on melting kinetics after complete melting of the solid Nb,

on the assumption that it is rate-limited by diffusion in the liquid, were written in
FORTRAN 77. An explicit finite-difference scheme using the 2nd-order Runge-Kutta

method was applied to Eq. (3. 14) — (3.19). The FORTRAN code, input parameters and

summarized results are listed below.

PROGRAM eutectic_late_stage

 

Program created by Kebin Low
CHEMS Department, MSU
July 27th, 2007

This program performs numerical iterations to simulate
the NiTi-Nb eutectic half- width evolution as a function of time
after exhaustion of the Nb interlayer.

The program assumes:

(1) Density of solid and liquid phases are the same;

(2) The NiTi/L and Nb/L interfaces attempt to maintain
their respective liquidus compositions;

(3) A linear concentration gradient hence exists across
the 2 solids;

(4) Fick's lst law holds in the eutectic liquid;

(5) Dissolution process is difiusion-controlled;

(6) Heat transfer does not play a rate-limiting role;

(7) Solid-state diffusion is negligible within the simulation
time-frame.

 

Iii-{It'l‘ii'III-itl'l'i‘l-‘I‘I‘IQII'I’II'D

Declarations
implicit none
integer‘4 t,numstep
real’8 jnb(300001),jniti(300001),xnb(300001),xniti(30000l)
real’8 w(3 0000 1 ),vnb(3 0000 l ),vniti(300001 ),cte_nb,cte_niti,temp
real*8 latticenb,latticeniti,deltime,tottime
real*8 nbconc(300001),c1,c2,c3,c4,rho_nb,rho_niti
real*8 diffusion_nb,diffusion_niti,rho_nbl rho_nitil

* Print on-screen menu
WTIIC(*,*)""
write(*,*)" n
write(*,*)"This program performs numerical iterations to simulate
write(*,"')"the NiTi-Nb eutectic half-width evolution"
write("',"')"as a function of time after the exhaustion of the"

 

226

 

write(*,"‘)"Nb interlayer."
write(j, g)" n
WI'HC(*,*)""

 

Prompt user inputs

write("',*)"1nput desired time frame (sec):"

read("‘,"') tottime

write(*,*)"Input desired total number of steps (0-300000):"
read(*,"') numstep

Dividing the total time by the number of steps
deltime=tottime/float(numstep)

Setting all time-variables to zero initially
do 10 t=l,numstep
nbconc(t)=0
jnb(t)=0
jniti(t)=0
vniti(t)=0
xniti(t)=0
W(t)=0

10 continue

*
t

It

II

Continue with user prompts

mite(‘,"')"lnput Liquid Half-Width Upon Nb Exhaustion (m):"
read("',"') Wfl)

write(*,*)"lnput Lattice Parameter of Nb (angstrom):"
read(*,*) latticenb

write(‘,*)"lnput Lattic Parameter of NiTi (angstrom):"
read("',") latticeniti

write(*,*)"1nput Braze Temperature (Degrees Celsius):"
read(*.*) temp

write(‘,*)"lnput NiTi Solidus Composition (atomic % Nb):"
read(*,"') cl

write(*,"')"lnput Nb Solidus Composition (atomic % Nb):"
read(*,*) 04

write(‘,*)"1nput Self-Diffusion of NiTi (mA2/s):"

read(*,"') diffusion_niti

Calculating the liq. comp. of Nb and NiTi at selected temp.
c2=( l 3 lO-temp)*(float(26)/float( 140))
c3=(temp-(2468-(float(1298)‘100/float(74))))/

C(float( l 29 8)/ float(74))

Opening output files
open(unit= l ,file='output_late.dat',status=‘unknown')
open(unit=2,file='results_late.dat',status='unknown')

Writing inputs into results_late.dat

write(2,*)"lnput Parameters"
wfite(2,*)"WM-w--—-~MW~WW-~"
write(2,*)"Time Frame:",tottime," seconds"

mite(2,"')"Total Number of Steps:",numstep

write(2,"‘)"Liquid Half-Width Upon Nb Exhaustion:",w(l)," m"

write(2,’)"Lattice Parameter of Nb:",latticenb," Angstroms"

write(2,*)"Lattice Parameter of NiTi:",latticeniti," Angstroms"

227

 

‘I'

a.

write(2,*)"Braze Temperature:",temp," Degrees Celsius"
write(2,*)"NiTi Solidus Composition:",cl," atomic % Nb"
write(2,*)"NiTi Liquidus Composition: ",62," atomic % Nb"
write(2,"‘)"Nb Liquidus Composition:",c3," atomic % Nb"
write(2,*)"Nb Solidus Composition:",c4," atomic % Nb"
write(2,*)"NiTi Difiusion:",difiusion_niti," m"2/s"

Calculating thermal expansion at braze temperature
cte_nb=l+(7.3E-6*temp)
cte_niti=l+(6.6E-6*temp)

Calculating the number density of Nb and NiTi
rho_nb=2/(latticenb* l E- 1 0*cte_nb)* *3
rho_niti=l/(latticeniti* lE-10*cte_niti)"3

rho_nb l =((c4/ 100.0)*rho_nb)+(((l 00.0-c4)/100.0)*rho_niti)
rho_nitil=((c1/100.0)*rho__nb)+(((100.0-c1)/100.0)*rho_niti)

Calculating the Nb concentration difference in #Nb/m"3
nbconc(l)=(c3/(((c3/2.0)*((latticenb*lE-lO‘cte_nb)**3))
C+((100.0-c3)*((latticeniti*1E-10‘cte_niti)**3))))
C-(c2/(((c2/2.0)*((latticenb*lE-10*cte_nb)**3))
C+((100.0-c2)*((1atticeniti*lE-lO‘cte_niti)**3))))

Print in-calculation message
WIIIC(*,T)""
write(*,*)"Please wait... Calculation in progress..."

Numerical iterations
do 20 t=l,numstep

Calculating the NiTi flux in the liquid
jniti(t)=difl’usion_niti‘nbconc(t)/w(t)

Calculating the NiTi/L interface velocity
vniti(t)=jniti(t)/rho_nitil

Calculating the width of the liquid

w(t+ l )=w(t)+(vniti(t) *deltime)

Calculating the new concentration difference in #Nb/mA3
nbconc(t+ l)=nbconc(t)c((jniti(t)*deltime)/w(t+1))

20 continue

*
#

Write data points for plotting into output_late.dat
do 30 t=l,numstep,200
write( 1 ,*)t*deltime,w(t)

30 continue

*
*

Print calculated values to results_late.dat

write(2,*)""

write(2,*)"Calculated Parameters"

write(2,*)" ..
write(2,‘)"Density of Pure Nb:",rho_nb," #/m"3"

write(2,*)"Density of Pure NiTi:",rho_niti," #/m"3"

 

228

write(2,*)"Density of Pro-eutectic Nb:",rho_nbl ," #/m"3"
write(2,*)"Density of Pro-eutectic NiTi:",rho_nitil ," #/m"3"
write(2,*)"Self-Diffusion of NiTi:",diffusion_niti," m"2/s"

Print End-of-Program message

write(¢’g)n n
write(*,*)"*#********#****Iliiltllttillihlflllttfilhhhhhk**********#******#"
write(*,*)"'output_late.dat' and 'results_late.dat'"
write("',"')"have been generated."

WI'ItC(*,*)" 11

write( ‘,*)"Data in 'output_late.dat' is in the following"
write(*,*)"sequence:"

write(*,*)"Time, Eutectic Half-Width"

wfiw(*")" n

write(",*)"'results_late.dat' contains most-recent input"
write(*,"')"parameters and calculated values."
write(*,*)" 11

write(*,*)"END OF PROGRAM"

write(t,*)"tititifittiiifiii*##*t*******#**it*ttttt*********t***t¥*"
write(‘,*)""
* Close output files
close(unit=1)
close(unit=2)

stop

end

Table C-l Input parameters for simulating the melting kinetics after the complete
melting of Nb.

 

 

 

 

 

Input Parameters Values
Time Frame (secs) 80
Total Number of Steps 3E05
Liquid Half-Width Upon Nb Exhaustion (m) 1.20951E-04
Lattice Parameter of Nb (Angstroms) 3.307
Lattice Parameter of NiTi (Angstroms) 3.015
Braze Temperature (°C) 1185
NiTi Solidus Composition (at.% Nb) 7
Nb Solidus Composition (at.% Nb) 91
NiTi Diffusion Coefficient (mZ/s) 4-101'3-09
Table 02 Results from the melting kinetics calculation (after the complete melting
of Nb).
Results Associated with Post-Exhaustion of Solid Nb Values
Liquid Half-Width at Exhaustion of Solid Nb (m) 1.21E-04
Approximate Thickness of Melted NiTi (m) 513-06
Final Liquid Half-Width (m) 1.26E-04

Approximate Time Taken to Attain Final Liquid Half-Width (sees) 15

229

Liquid Half-Width as a Function of Time
After Complete Melting of Solid Nb

 

1 .27E-04

 

 

1 .26E-04

 

 

1 .25E-04

 

1 .24E-04

 

1.23E-04

 

1.22E-04 ~

Liquid Half-Width (m)

 

1.21E-04 -

 

 

 

 

 

 

 

 

 

 

 

1 .20E-04
30 40 50 60 70 80 90 1 00 1 10 120

Braze Time (secs)

Figure C-l Predicted liquid half-width as a function of braze time.

230

APPENDIX D

Numerical calculations on melting kinetics before complete melting of solid Nb,

on the assrnnption that it is rate-limited by heat transfer, were written in FORTRAN 77.

An explicit finite-difference scheme using the 2nd-order Runge-Kutta method was

applied to Eq. (3.20) — (3.21). The FORTRAN code, input parameters and summarized

results are listed below.

“‘2;

PROGRAM heat transfer

 

Program created by Kebin Low
CHEMS Department, MSU
March 10th, 2008

This program performs numerical iterations to calculate
the NiTi-Liquid interface position and velocity

as a function of time if heat transfer, instead of diffusion,
is the rate-limiting factor for contact melting.

 

****§*§§*§*§*

Declarations

implicit none

integer‘4 t,k,numstep

real*8 meltthickness(300001),xniti(30000l),vniti(300001)
real*8 conduct,latent,cte_niti,tempext,tempint,tempdiff
real’8 density,correctdensity,deltime,tottime

real*8 delq(300001),nitithiclmess(300001)

real*8 nbthickness,thicknesslimit

* Print onvscreen menu
write(ip,*)im
write(*,*)" "
write(“',"')"This program performs numerical iterations to"
write(*,*)"calculate the NiTi-Liquid interface position and"
write(*,"')"velocity as a function of time if heat transfer,"
write("',*)"instead of diffusion, is the rate-limiting factor"
write(*,*)"for contact melting."
write(ip,ip)rr n
wfitc(t,¢)im

 

 

* Prompt user inputs
write(‘,*)"lnput desired time frame (sec):"
read("',*) tottime
write("',*)"Input desired total number of steps (0-300000):"
read(*,*) numstep
write(ip,*)rm

231

write(*,*)"Please Wait. . . "
WT1tC(*,*)" n

Dividing the total time by the number of steps
deltime=tottime/float(numstep)

Setting all time-variables to zero initially
do 10 t=l,numstep

nitithickness(t)=0

meltthickness(t)=0

vniti(t)=0

xniti(t)=0

delq(t)=0

10 continue

as
*

Continue with user prompts

write(*, *)"Input Initial NiTi Plate Thickness (m) : "

read(*, *) nitithickness( l )

write( * , *)" Input Limiting Thickness for NiTi Melting (m): "

read("', *) thicknesslimit

write(* , *)"Input Thermal Conductivity of NiTi (W /m. K) : "

read(*, *) conduct

write(*, *)"Input Latent Heat of Fusion of NiTi (J / g) : "

read(*,*) latent

write( *, *)"Input External Temperature (Degrees Celsius): "

read(*, *) tempext

write(* , *)" Input NiTi-L Interface Temperature (Degrees Celsius): "
read(*,*) tempint

WTItC(* ,*)11 II

write(*, *)"Calculation in Progress . . . "

WTItC(*,*)" n
write(II,*)"tilt**#**¢**#t*It*tt*tttttttttifllttttttlktttlhk***#****#***"
write("',"‘)"'raw.dat' and 'results.dat'"

write(*, *)"will be generated at the end of calculations."
Wf1t6(*,*)""

write(*,*)"See 'results.dat' for details."
writc(*’*)"#**************#*******¢******************************"

Opening output files

open(unit= l ,file='raw.dat',status='unknown')
open(unit=2,file='output.dat',status='unknown')
open(unit=3,file='results.dat',status='unknown')

Writing inputs into results.dat

write(3,*)"Input Parameters"
wfite(3,*)"M~WWWW--~"
write(3,*)"Time Frame:",tottime," seconds"

write(3,*)"Total Number of Steps:",numstep

write(3,*)"1nitial NiTi Thickness:",nitithickness(l)," m"
write(3,*)"Thickness Limit for NiTi Melting:",thicknesslimit," 1n"
write(3,*)"Thermal Conductivity of NiTi:",conduct," W/m.K"
write(3,*)"Latent Heat of Fusion of NiTi:",latent," J/g"
write(3,*)"External Temperature:",tempext," Degrees Celsius"
write(3,*)"NiTi-L Temperature:",tempint," Degrees Celsius"
write(3,*)""

232

 

t

1|-

21

It
3

Density of NiTi (g/m3)
density=6.5

Calculating thermal expansion at braze temperature
cte_niti=l+(6.6E-6*tempext)

Corrected density
correctdensity=density/(cte_niti**3)

Calculating the temperature difference
tempdiff=tempext—tempint

Print in-calculation message
write(*,*)mr
write(*,"')"Please wait... Calculation in progress..."

Numerical iterations
do 20 t=l ,numstep

Check if NiTi thickness limit has been exceeded
if(meltthickness(t).lt.thicknesslimit) then

Calculating change in heat
delq(t)=(conduct*tempdifl‘deltimeynitithiclmessa)

Calculating thickness of NiTi melted
xniti(t+lF(delq(t)/latent)/(correctdensity"'100)
nitithickness(t+ 1)=nitithiclmess(t)-xniti(t+ 1)
meltthickrress(t+ l )=meltthickness(t)+xniti(t+ 1)

Calculating the velocity of NiTi-L interface
vniti(t-l-l)=xniti(t+1)/deltime

Write results to raw.dat
write(l ,*)t*deltime,delq(t),meltthickness(t),vniti(t)

endif
if(meltthickness(t).gt.thicknesslimit) then

Write data points for plotting into outputdat

do 21 k=1 ,t,200

write(2,*)k*deltime,delq(k),meltthickness(k),vniti(k)
continue

Print important information with regards to Nb exhaustion

write(3,*)"Information Associated With Nb Exhaustion"
wfite(3,t)"WWWWW"
write(3,*)"Time:",(t-1)*deltime," seconds"

write(3,*)"Thickness of NiTi Melted:",meltthiclmess(t-l)," m"

write(3,’)"Final Interface Velocity:",vniti(t—l)," m/s"

Print End-of-Program message
write(3,"‘)""

write(3,"')"Files Generated"
write(3,*)"*******************II**************************#*******"

233

 

Oil-{*5-

II

t

write(3,*)'"raw.dat’, 'output.dat', and 'results.dat"'
write(3,*)" have been generated."
write(3,*)""
mite(3,*)"'raw.dat' contains all time-evolution data."
write(3,*)"Data is in the following sequence:"
write(3,*)"Time, Change in Heat, NiTi Melt Thickness and Velocity"
write(3,*)mr
write(3,*)'"output.dat' contains plot-fiiendly data poin ."
write(3,*)"Data is in the following sequence:"
write(3,*)"Time (200-step interval), Change in Heat,"
write(3,*)"NiTi Melt Thickness and Velocity"
W11t8(3,.)""
write(3,*)"'results.dat' contains most-recent input parameters"
write(3,*)"and calculated values."
write(3,*)mr
write(3,*)"END OF PROGRAM"

write(3,*)"titit***#******#***t$**#**t*****************¢******#**"
write(3,*)mr

Close output files

close(unit=1)

close(unit=2)

close(unit=3)

stop

endif

20 continue

It

end

Table D-l Input parameters for simulating the melting kinetics before the complete
melting of Nb (assume rate-limited by heat transfer).

 

 

Input Parameters Values
Time Frame (secs) 0.1

Total Number of Steps 50000
Initial NiTi Thickness (m) 3.175E-03

Thickness Limit for NiTi Melting (m) 9.6E—05
Thermal Conductivity of NiTi (W/m-K) 41

Latent Heat of Fusion of NiTi (J/ g) 357
External Temperature (°C) 1 185
NiTi-Liquid Temperature (°C) 1170

 

Table D-2 Results from the melting kinetics calculation (after the complete melting
of Nb, assume rate-limited by heat transfer).

 

 

Results Values
Thickness of NiTi Melted (m) 9.54E-06
Elapsed Time (secs) 1.12E-O4

 

234

 

 

BILIOGRAPHY

Akari T., Hirose A., Uchihara M., Kohno W., Honda K., Kondh M., Materials, 38(428)
(1989), p478.

Alemany M. M. G., Rey C., Gallego L. J ., Journal of Chemical Physics, 109(13) (1998),
p5175.

Alkorta J ., Martinez-Esnaola J. M., Gil Sevillano J ., Acta Materialia, 54 (2006), p3445.
Bastin G. F., Rieck G. D., Metallurgical Transactions, 5 (1974), p1827.

Bird R. K., Hoffman E. K., “Evaluation of the Transient Liquid Phase (TLP) Bonding
Process for T i 3A l-Based Honeycomb Core Sandwich Structure” in NASA Report,
(NASA/TP-l998-208421), (1998).

Broughton W. R., Hinopoulos G., “Evaluation of the Single-Lap Joint Using Finite
Element Analysis” in National Physical Laboratory Report, CMMT(A) 206 Report No.
15 (1999), p1.

Callister W. D., “Phase Transformations in Metals” in Materials Science and Engineering
— An Introduction, (John Wiley and Sons, New York, 2003), p298.

Campbell C. E., Boettinger W. J., Metallurgical and Materials Transactions A, 31A
(2000), p2835.

Chau E. T. F ., Friend C. M., Allen D. M., Hora J ., Webster J. R., Materials Science and
Engineering A, 438-440 (2006), p589.

Chuprina V. G., Shalya I. M., Powder Metallurgy and Metal Ceramics, 41(1-2) (2002),
p85.

Darken L. S., Gurry R. W., “Liquids” in Physical Chemistry of Metals, (McGraw-Hill,
New York, 1953), pl 10.

Davidson M. J ., Biberger M., Mukherjee A. K., Scripta Metallurgica et Materialia, 27(12)

(1992), pl 829.

Duerig T. W., Melton K. N., Proft J. L., “Wide Hysteresis Shape Memory Alloys” in
Engineering Aspects of Shape Memory Alloys, edited by T. W. Duerig, K. N. Melton, D.
Stockel, C. M. Wayrnan, (Butterworth-Heinemann, Essex, UK, 1990), p130.

Einziger R. E., Mundy J. N., Hoff H. A., Physical Review B, 17(2) (1978), p440.

235

 

b;

 

Eyring H., Journal of Chemical Physics, 4 (1936), p283.

F irstov G. S., Van Humbeeck J ., Koval Y. N., Materials Science and Engineering A, 378
(2004), p2.

Flegler S. L., Heckrnan J. W., Klomparens K. L., Scanning and Transmission Electron
Microscopy: An Introduction, (Oxford University Press, New York, 1993), p192.

Flernings M. C., “Plane Front Solidification of Polyphase Alloys” in Solidification
Processing, (McGraw-Hill, New York, 1974), p93.

Gale W. F., Orel S. V., Journal of Materials Science, 31 (1996), p345.
Gale W. P., Guan Y., Journal of Materials Science, 32 (1997), p357.
Getazheev K. A., Savintsev P. A., Soviet Physics Journal, 13(5) (1970), p620.

Grummon D. S., Zhang J ., Pence T. J ., Materials Science and Engineering A, 273-275
(1999), p722.

Grummon D. S., Shaw J. A., Foltz J. W., Materials Science and Engineering A, 438-440
(2006), p1113.

Gupta C. K., Suri A. K., “Properties and Applications of Niobium” in Extractive
Metallurgy of Niobium, (CRC Press, Florida, 1994), p1.

Gupta K. P., Phase Diagrams of Ternary Nickel Alloys, Part2 Section D2, (Indian
Institute of Metals, Calcutta, India, 1991), p163.

Hall T. A., U. S. Patent No. 5242759, 1993.

Hillert M., “Projected and Mixed Phase Diagrams” in Phase Equilibria, Phase Diagrams
and Phase Transformations: Their Thermodynamic Basis, (Cambridge University Press,
UK, 1998), p226.

Hirose A., Akari T., Metals (in Japanese), 59(8) (1989), p61.

Honma T., “T iNi-Based Shape Memory Alloys” in Shape Memory Alloys, edited by
Funakubo H., (Gordon and Breach Science Publishers, New York, 1987), p61.

Hsieh S. F., Wu S. K., Journal of Alloys and Compounds, 266 (1998), p276.

Hsu Y. T., Wang Y. R., Wu S. K., Chen C., Metallurgical and Materials Transactions A,
32(3) (2001), p569.

236

 

Hume-Rothery W., Raynor G. V., “The Atomic Radii and Some Physical Properties of
the Elements” in The Structure of Metals and Alloys — 3rd Edition, (The Institute of
Metals, London, 1954), p60.

Iida T., Guthrie R. I. L., “Surface Tension” in The Physical Properties of Liquid Metals,
(Claredon Press, Oxford, 1988), p134.

Iida T., Guthrie R. I. L., “Difiirsion” in The Physical Properties of Liquid Metals,
(Claredon Press, Oxford, 1988), p199.

Iida T., Guthrie R. I. L., “Viscosity” in The Physical Properties of Liquid Metals,
(Claredon Press, Oxford, 1988), p147.

J ax V. P., Zeitschrififiir Metallkunde (in German), 62(1-14) (1971), p279.
Jia D., Liu W., Han M., Wang D., Dong Z., Zeitschrzfifiir Metallkunde, 91 (2000), p258.

Johnson A. D., Martynov V. V., Minners R. S., Journal de Physique IV, C8 (5) (1995),
p783.

Kapgan M., Melton K. N., “Shape Memory Alloy Tube and Pipe Couplings” in
Engineering Aspects of Shape Memory Alloys, edited by T. W. Duerig, K. N. Melton, D.
Stockel, C. M. Wayman, (Butterworth-Heinernann, Essex, UK, 1990), p137.

Krell A., Shiidlich S., Materials Science and Engineering A, 307 (2001), p172.

Lehnert T., Crevoiserat S., Gotthardt R., Journal of Materials Science, 37 (8) (2002),
p1523.

Li M. G., Sun D. Q., Qiu X. M., Sun D. X., Yin S. Q., Materials Science and
Engineering A, 424 (2006), p17.

Lieberman D. S., Wechsler M. S., Read T. A., Journal of Applied Physics, 26 (4) (1955),
p473.

Lindquist P. G., Wayman C. M., “Shape Memory and Transformation Behavior of
Martensitic T i—Pd-Ni and T i-Pt-Ni Alloys” in Engineering Aspects of Shape Memory
Alloys, edited by T. W. Duerig, K. N. Melton, D. Stockel, C. M. Wayman, (Butterworth-
Heinemarm, Essex, UK, 1990), p58.

London B., Fino J ., Pelton A., Fuller C., Mahoney M., “Friction Stir Processing of

Nitinol " in “Proceedings of The Minerals, Metals and Materials Society Conference —
San Francisco, California, USA ”, (2005), p67.

237

 

MacDonald W. D., Eagar T. W., “Transient Liquid Phase Bonding Processes " in The
Metal Science of Joining, edited by Cieslak M. J ., Perepezko J. H., Kang S., Glicksman
M. E., (The Minerals, Metals and Materials Society, 1992), p93.

MacDonald W. D., Eagar T. W., “Low Temperature Transient Liquid Phase Bonding of
T i-6Al-4 V” in “Proceedings of the 3rd International Conference on Trends in Welding
Research — Gatlinburg, Tennessee, USA”, (1992), p1083.

MacDonald W. D., Eagar T. W., Metallurgical and Materials Transactions A, 29A
(1998), p315.

Maksirnovich G. G., Lyutyi E. M., Shirokov V. V., Druzhinina I. P., Per’kova G. N.,
Soviet Materials Science, 18(2) (1982), p150.

Massalski T. B., Binary Alloy Phase Diagrams 2"“ Edition (American Society for Metals,
Metals Park, Ohio, 1990).

Matsumoto S., Tokunaga T., Ohtani H., Hasebe M., Materials Transactions (Special
issue on “Materials Science of Bulk Metallic Glasses ”), 46(12) (2005), p2920.

Mei J ., Davenport J. W., Physical Review B, 42(15) (1990), p9682.

Melton K. N., “Ni-T i Based Shape Memory Alloys” in Engineering Aspects of Shape
Memory Alloys, edited by T. W. Duerig, K. N. Melton, D. Stockel, C. M. Wayman,
(Butterworth-Heinemann, Essex, UK, 1990), p21.

Miyazaki S., Ohmi Y., Otsuka K., Suzuki Y., Journal de Physique, 43 (1982), pC4-255.
Miyazaki S., Imai T., Igo Y., Otsuka K., Metallurgical Transactions, 17A (1986), p115.

Mueller M. H., Knott H. W., Transactions of the Metallurgical Society of AIME, 227
(1963), p674.

Munro R. G., Journal of American Ceramic Society, 80(8) (1997), p1919.
Murai M., U. S. Patent No. 4952044, 1990.

Myneni G. R., Agnew S. R., American Institute of Physics Conference Proceedings, 671
(2003), p227.

Nakata Y., Tadaki T., Shimizu K., Materials Transactions JIM, 32 (1991), p580.
Nakata Y., Tadaki T., Shimizu K., Materials Transactions JIM, 32 (1991), pl 120.

Nevitt M. V., Transactions of the Metallurgical Society of AIME, 218 (1960), p327.

238

Niemann J. T., Garrett R. A., Welding Journal, 53(8) (1974), p35 ls.
Nishida M., Wayman C. M., Honma T., Metallurgical Transactions, 17A (1986), p1505.
Oliver W. C., Pharr G. M., Journal of Materials Research, 7(6) (1992), p1564.

Otsuka K., Wayman C. M., “Introduction ” in Shape Memory Materials, edited by
Otsuka K., Wayman C. M., (Cambridge University Press, Cambridge, UK, 1998), p1.

Otsuka K., Ren X., Progress in Materials Science, 50 (2005), p511.
Owczarski W. A., Welding Journal, 42(2) (1962), p78s.

Ozeki K., Yuhta T., Aoki H., Fukui Y., Bio-Medical Materials and Engineering, 13(3)
(2003), p271.

Piao M., Miyazaki S., Otsuka K., Materials Transactions JIM, 33(4) (1992), p346.

Piao M., Miyazaki S., Otsuka K., Nishida N., Materials Transactions JIM, 33(4) (1992),
p337.

Piao M., Otsuka K., Miyazaki S., Horikawa H., Materials Transactions JIM, 34(10)
(1993), p919.

Plietsch R., Ehrlich K., Acta Materialia, 45(6) (1997), p2417.

Porter D. A., Easterling K. E., “Solidification” in Phase Transformations in Metals and
Alloys — 2"“ Edition (Reprinted), (Nelson Thornes, Cheltenham, UK, 2001), p185.

Prima s. B., Metally, 5 (1992), p120.

Prima S. B., Tret’yachenko L. A., Petyukh V. M., Powder Metallurgy and Metal
Ceramics, 34 (1995), p155.

Profi J. L., Duerig T. W., “The Mechanical Aspects of Constrained Recovery” in
Engineering Aspects of Shape Memory Alloys, edited by T. W. Duerig, K. N. Melton, D.
Stockel, C. M. Wayman, (Butterworth-Heinemann, Essex, UK, 1990), p115.

Rhines F. N., “Ternary F our-phase Equilibrium, Class 1’ ’ in Phase Diagram in
Metallurgy — Their Development and Application, (McGraw-Hill, New York, 1956),
p159.

Rhines F. N., “Ternary F our-phase Equilibrium, Class 11’ ’ in Phase Diagram in
Metallurgy — Their Development and Application, (McGraw-Hill, New York, 1956),
p176.

239

Rhines F. N., “Ternary F our-phase Equilibrium, Class III” in Phase Diagram in
Metallurgy — Their Development and Application, (McGraw-Hill, New York, 1956),
p186.

Rhines F. N., “Congruent Transformation in Ternary Systems” in Phase Diagram in
Metallurgy — Their Development and Application, (McGraw-Hill, New York, 1956),
p193.

Roy A. K., Chhabra R. P., Metallurgical Transactions A, 19A (1988), p273.

Saburi T., Tatsumi T., Nenno S., Journal de Physique, 43 (1982), pC4-261.

Saburi T., Yoshida M., Nenno S., Scripta Metallurgica, 18 (1984), p363.

Savintsev P. A., Akhkubekov A. A., Getazheev K. A., Rogov V. 1., Savvin V. S., Soviet
Physics Journal, 14(4) (1971), p467.

Schetky L. M., Wu M. H., “Issues in the Further Development of Nitinol Properties and
Processing for Medical Device Applications” in “Proceedings of the Materials and
Processes from Medical Devices Conference, Anaheim, California”, (2004), p271.

Schlossmacher P., Haas T., Schuessler A., Journal iof Physics Colloquium, C5
(Supplementary) (1995), p251.

Schwartz M. M., “Vacuum Brazing” in Modern Metal Joining Techniques, (John Wiley
and Sons, New York, 1969), p278.

Seki M., Yarnamoto H., Nojiri M., Uenishi K., Kobayashi K. F., Journal of Japan
Institute of Metals, 64(8) (2000), p632.

Shaw J. A., Grummon D. S., Foltz J. W., Smart Materials and Structures, 16 (2007), p
S170.

Shiue R. H., Wu S. K., Materials Transactions, 46(9) (2005), p2057.

Sinclair C. W., Purdy G. R., Morral J. E., Metallurgical and Materials Transactions A,
31A (2000), p1187.

Sisco F. T., Epremian E., “Physical and Mechanical Properties” in Columbium and
Tantalum, (John Wiley and Sons, New York, 1963), p198

Sneddon I. N., International Journal of Engineering Science, 3 (1965), p47.

Su C. Y., Chou C. P., Wu B. C., Lih W. C., Materials Science and Technology, 15 (1999),
p316.

240

 

Surbled P., Clerc C., Le Pioufle B., Ataka M., Fujita H., Thin Solid Films, 401 (1-2)
(2001), p52.

Tuah-Poku 1., Dollar M., Massalski T. B., Metallurgical Transactions A, 19A (1988),
p675.

Vadovic C. J ., Colver C. P., Physical Review B, 1(12) (1970), p4850.

Van der Eijk C., Fostervoll H., Sallom Z. K., Akselsen O. M., “Plasma Welding of MT i
to Ni T i, Stainless Steel and Hastelloy C2 76” in “Proceedings for ASM Materials
Solutions 2003 Conference, Pittsburgh, Pennsylvania”, (2003).

Wang G., Zhou Z., Zhang C., Acta Metallurgica Sinica (in Chinese), 41(7) (2005), p703.
Wechsler M. S., Lieberman D. S., Read T. A., AIME Transactions, 197 (1953), p1503.
Wilcox B. A., Allen B. C., Journal of the Less-Common Metals, 13 (1967), p186.

Winzek B., Sterzl T., Quandt E., “Proceedings of the International Conference on Solid-
State Sensors and Actuators, Transducers ’01 — E urosensors XV, Munchen, Germany ”,
(Springer-Verlag, Berlin, Heidelberg, 2001), p706.

Wollants P., De Bonte M., Roos J. R., Z Metallkunde, 70 (1979), p113.

Wu M. H., “Fabrication of Nitinol Materials and Components” in “Proceedings for
International Conference on Shape Memory and Superelastic Technologies, Kunming,
China”, (2001), p285.

Yan P., Wallach E. R., Intermetallics, 1 (1993), p83.

Yang G., Hao S., Journal of Alloys and Compounds, 297 (2000), p226.

Yang T. Y., Shue R. K., Wu S. K., Intermetallics, 12 (2004), p1285.

Yung K. C., Zhu H. H., Yue T. M., Smart Materials and Structures, 14 (2005), p337;
Zadno G. R., Duerig T. W., “Linear Superelasticity in Cold- Worked Ni-Ti” in
Engineering Aspects of Shape Memory Alloys, edited by T. W. Duerig, K. N. Melton, D.
Stockel, C. M. Wayman, (Butterworth-Heinemann, Essex, UK, 1990), p21.

Zhang C. S., Zhao L. C., Duerig T. W., Wayman C. M., Scripta Metallurgica, 24 (9)
(1990), p1807.

Zhao L. C., Duerig T. W., Justi S., Melton K. N., Profi J. L., Yu W., Wayman C. M.,
Scripta Metallurgica, 24 (2) (1990), p221.

241

 

   

"'llfillllljllfillflljillglfllfll