 

,L. w.

 

w

qg-4M...—“.

‘_.a—~~« {WOO-.‘th

 

 

 

VERSITY LIBRARIES

 

‘92.;

 

 

 

 

 

 

 

 

 

Ill8

 

 

 

 

 

Illlllllllllllllllllll

31293 01683

This is to certify that the

dissertation entitled

Processing, Microstructure and Thermomechanical

Behavior of Superelastic Nickel-Titanium Thin Films

Sputter-Deposited at Elevated Temperatures

presented by

Li Hou

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Materials Science

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution

 

LIBRARY
Michigan State
University

 

 

 

 

0-12771

PROCESSING, MICROSTRUCTURE AND THERMOMECHANICAL
BEHAVIOR OF SUPERELASTIC NICKEL-TITANIUM THIN FILMS
SPUTTER-DEPOSITED AT ELEVATED TEMPERATURES

By

Li Hou

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Materials Science and Mechanics

1998

ABSTRACT

PROCESSING, MICROSTRUCTURE AND THERMOMECHANICAL BEHAVIOR
OF SUPERELASTIC NICKEL-TITANIUM THIN FILMS SPU'ITER-DEPOSITED
AT ELEVATED TEMPERATURES

By

Li Hou

Near-equiatomic N iTi is capable of fully recoverable strains associated with the
formation and reversion of a stress-induced martensite phase from a B2 (CsCl) parent
phase. Recently, sputter-deposited N iTi thin ﬁlms have drawn attention for potential
application to microactuators in microelectromechanical systems (MEMS). As—sputtered
NiTi ﬁlms are amorphous when deposition temperature is below 0.4TIn and an additional
thermal treatment is required to achieve the crystalline BZ phase. The objective of this
study was to investigate the microstructure and thermomechanical behavior of crystalline
NiTi ﬁlms where were directly produced by sputter deposition at elevated substrate
temperatures.

A series of NiTi ﬁlms were sputter deposited at various substrate temperatures
between 573 K and 773 K using a NiSITi49 alloy cathode. In order to study the effects of
grain boundaries and precipitates on deformation behavior, some as-sputtered crystalline
ﬁlms were given a thermal treatment to form diverse microstructures which were
systematically examined by transmission electron microscopy. Electron microdiffraction
and XRD methods were employed to identify precipitate structures. Phase transformation
characteristics were obtained by electrical resistivity and DSC measurements. Stress-strain

experiments were conducted in uniaxial tension on free—standing NiTi ﬁlms. Finally,

fracture surfaces and the microstructure of deformed ﬁlms were examined by TEM and
SEM for the purpose of verifying structural stability.

A fully crystallized N iTi ﬁlm could be produced at a deposition temperature of 623 K
which is about 150 K lower than the minimum temperature to form B2 phase by post-
deposition annealing in a reasonable time. Crystalline ﬁlms displayed microstructures
consisting of very ﬁne columnar BZ grains with 0.1-0.3 um in-plane diameter and
dispersed Ni4Ti3 particles. Films deposited at 703 K and 723 K exhibited well-deﬁned
transformational superelasticity. The stability of microstructure against permanent
deformation was found to result from the combination of a ﬁne-grained structure with
moderate-sized transgranular precipitates. As a practical demonstration of applicability to
MEMS systems, a prototype electrically-excitable NiTi/polyimide actuating element was

fabricated by hot-substrate deposition and a pattern-etch technique.

To my Wife and Daughter with Appreciation

iv

ACKNOWLEDGMENTS

I am deeply grateful to Dr. David S. Grummon, my major professor, for his advice
and continuing support. Dr. Thomas J. Pence is greatly acknowledged for sharing
invaluable ideals in our N iTi thin ﬁlm group meetings. Sincere appreciation is expressed to
my other guidance committee members: Dr. Kali Mukherjee and Dr. Subhendra D. Mahanti
for their important comments and valuable suggestions.

I would like to thank Dr. Andre Y. Lee for his generosity in providing liberal access
to the Minimatm microtensile apparatus. 1 also like to thank Dr. John Mansﬁeld in
University of Michigan for his instruction of manipulating Jeol 2000 FX TEM in his
laboratory.

Mr. Syed Shahzaman is acknowledged for his kindly help in Photolithography
Laboratory at Dept. of Electrical Engineering, Michigan State University. I am grateful to
my colleagues in the NiTi Thin Film Group, Dr. Xiaochuang Wu for his assistance in
using MarkTM ﬁnite element analysis software, and Mr. Jinping Zhang for his help in
characterizing the X-ray spectra of precipitates.

Finally, I would like to thank my wife, Jiefei, for her support and patience during the

preparation of this work.

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES
INTRODUCTION

CHAPTER 1. BACKGROUND

1.1. Thermoelastic Martensitic Transformation
1 .1 . 1 . General Characteristics
1.1.2. Crystallography of Martensitic Transformation
1.1.3. Frictional Work and Stored Elastic Energy
1.1.4. Shape Memory and Superelasticity
1.1.5. The Clausius-Clapeyron Equation for Uniaxial Stress

1.2. Transformations in NiTi Alloys
1.2.1. Martensitic and R-Phase Transformations
1.2.2. Diffusional Transformations
1.2.3. The Effects on Phase Transformations

1.3. formation of Metallic Crystals
1.3.1. Role of Grain Boundaries in Plastic Deformation
1.3.2. Precipitation Hardening

1.4. Defamation of NiTi Alloys
1.4.1. Recoverable Tensile Strains
1.4.2. Slip Deformation
1.4.3. Factors Inﬂuencing SME and Superelasticity

1.5. Formation and Structures of Thin Films
1.5.1 . Sputter Deposition
1.5.2. Inﬂuence of Temperature on Phase Formation
1.5.3. Structural Zone Models
1.5.4. The Secondary Grain Growth in Thin Films
1.6. Studies on NiTi Films and the Purpose of Present Work
CHAPTER 2. EXPERIMENTAL METHODS
2.1. Sputter Deposition Facility

2.2. Fabrication of NiTi/Polyimide Prototype Microactuator
2.2.1. Crystallizing NiTi by Post-Deposition Annealing
2.2.2. Metallization of Kapton by Hot-Substrate Deposition
2.2.3. Photolithography Process

vi

viii

ix

2.3. Preparation of Free-Standing N iTi Films
2.3.1. Sputter Deposition Procedure
2.3.2. Thermal Treatment of As-Sputtered NiTi Films

2.4. Characterization of NiTi Films
2.4.1. Energy Dispersive X-Ray Microanalysis (EDX)
2.4.2. Film Thickness Measurement
2.4.3. Electrical Resistimetry
2.4.4. Differential Scanning Calorimetry (DSC)
2.4.5. X-Ray Diffraction (XRD)
2.4.6. Transmission Electron Microscopy (TEM)
2.4.7. Tensile Tests
2.4.8. Scanning Electron Microscopy (SEM)

CHAPTER 3. RESULTS AND DISCUSSION

3.1. Microstructure of As-Sputtered NiTi Films
3.1.1. TEM Observation
3.1.2. Thin Film Microstructure: Discussion
3.1.3. Formation of Texture
.1.4. Growth of Precipitates

3
3.2. Microstructure of Annealed NiTi Films
3.2.1. Morphology
3.2.2. Structure of Precipitates in Annealed Films

3.3. Phase Transformation Characteristics
3.3.1. Phase Transformations in As-Sputtered Films
3.3.2. Phase Transformations in Annealed Films

3.4. Deformation Behavior of NiTi Films
3.4.1. Ultimate Elongation of N iTi Films
3.4.2. Isothermal Stress-Strain Curves of NiTi Films
3.4.3. Classiﬁcation of Stress-Strain Curves
3.4.4. Thermomechanical Behavior of NiTi Films
3.4.5. Transformational Superelastisity

3 .5. Fractography of NiTi Films
CONCLUSIONS
APPENDICES

A.1. Electrically-Excitable Shape-Memory Actuator
A.2. Thermodynamic Analysis by Jacobian Notation

BIBLIOGRAPHY

vii

Table 1-1

Table 1-2

Table 2-1

Table 2-2

Table 3-1

Table 3-2

Table 3-3

Table 3-4

Table 3-5

LIST OF TABLES

Austenite-to-martensite lattice correspondences of NiTi [after
Matsumoto et al. 1978].

The 24 habit plane variants and related correspondence variant-
combinations. The habit plane indices were calculated from the
phenomenological crystallographic theory, and correspond to
the experimental data for solution-treated specimens [after
Miyazaki et al., 1978].

Sputter-deposition conditions and resultant ﬁlm compositions.
The notation ‘H---’ is used in the present work where the ‘H’
signiﬁes ‘hot deposition’ and three digits signify the substrate
temperature of deposition in Kelvin. The substrate for all ﬁlms
were quartz except for ﬁlm H698 which was deposited on
Kapton ﬁlm.

Heat-treatment conditions of annealed NiTi ﬁlms. For annealed
NiTi ﬁlms, the notation ‘A—’ is used where the ‘A’ represents
‘annealing’ and a digit designates the detailed heat-treatment path
as indicated in the table.

Summary of microstructures of as—deposited NiTi ﬁlms. In the
table, D32, DP“ and PS, respectively, designate the in-plane grain
diameter of the B2, the diameter of precipitates in longitudinal
direction, and precipitate spacing.

Summary of NiTi matrix and precipitate morphologies of
annealed NiTi ﬁlms. In the table, 032, DP“ and PS, respectively,
designate the in-plane grain diameter of the BZ, the diameter of
precipitates in longitudinal direction, and precipitate spacing.
Phase transformation temperatures of as-deposited NiTi ﬁlms.

Transformation characteristics of the ﬁlm H703 obtained from
electrical resistivity and DSC measurements.

Phase transformation temperatures of annealed NiTi ﬁlms.

viii

112

113

114

115

116

117
118
119

120

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

LIST OF FIGURES

Schematic drawings of a two-variant matensitic transformation
[Chang, 1993]. (a) The lattices of the parent phase and the
martensite. (b) The Bain distortion, which converts the parent
phase into the martensite, does not yield an undistorted plane
associated with the habit plane. (c) A lattice invariant shear,
through a reversible manner of twinning, produces an undistorted
plane. ((1) The rigid body rotation, which brings the transformed
lattice back, so that the interface is in contact with the parent.
lattice

A schematic representation showing the inﬂuence of frictional
work and stored elastic energy on therrnoelastic martensitic
transformations [modiﬁed from Salzbrenner and Cohen, 1979].

Schematic drawings summaries the inﬂuence of frictional force
and stored elastic energy on thermoelastic martensite
transformations.

Upper ﬁgure: schematic drawing of shape-memory effect: (a-c)
the deformation is accomplished by detwinning or reorientation
of variants in martensite lattice. M is twinned martensite and M'
is deformed martensite. Bottom ﬁgure: (a-c) the deformation

is accomplished by transformational twining in the austenite.

The steps illustrating the transition from B2 to B 19' lattices
[modiﬁed from Hehemann and Sandrock, 1971]. (a) B2 lattices;
(b—c) an orthorhombic distortion; (c-d) a monoclinic shear; (e)
two shufﬂe models.

An illustration representing geometry relationships between
planes and directions in twinning [Reed-Hill, 1964].

The self-accommodation morphology of the monoclinic
martensite, and a model describing crystallographic relationships
between the martensite variants in the triangular morphology
[Miyazaki et al. 1989].

The schematic diagram showing physical property changes
associated with R-phase and martesitic transformation: (a) lattice
distortion, (b) electrical resistivity change, and (c) latent heat
exchange.

Self-accommodating morphology and corresponding schematic

variants-combination of rhombohedral phase proposed by
Miyazaki and Wayman [1988].

ix

121

122

123

124

125

126

127

128

129

Figure 1-10

Figure 1-11

Figure 1-12

Figure 1-13

Figure 1-14

Figure 1-15

Figure 1-16

Figure 1-17

Figure 1—18

Figure 1-19

(a) Ni-Ti equilibrium phase diagram proposed by Massalski
[1987]; (b) Ni-Ti phase diagram in the vicinity of stoichiometric
composition proposed by Wasilewski et al. [1971].

Isothermal time-temperature-transformation (TIT) diagram for
Ni52Ti48 alloy [Nishida, Wayman and Honma, 1986].

The effect of Ni-concentration on the electrical resistivity vs.
temperature curves for NiTi alloys quenched from 1073 K
[Nishida and Honma, 1984]. The M, temperature decreases
rapidly with the increase of nickel content.

Electrical resistivity vs. temperature curves [Wu, Lin and Chou,
1989], of NislTi49 alloy aged at 673 K after a 1073 K, 2 hour
solution-treatment for various times, showing the effects of aging
on M, temperatures.

The effect of specimen thickness on the normalized ﬂow stress
in polycrystalline (a) A], (b) Cu, (c) Cu-13Al (at.%), and (d) Fe
[Miyazaki, Shibata and Fujita, 1979].

Three types of crystallographic relationships between the lattice
of the precipitate and the lattice of the matrix [Hirsch et al., 1977].
(a) coherent with negative misﬁt; (b) semi-coherent, and (c)
incoherent.

Interactions between dislocations and precipitates. (a) In lightly
aged alloys, the closely spaced stress ﬁelds make the dislocation
unable to bend between the particles. The dislocation has to
move out of its slip plane by climbing or cross-slipping. (b) In
over-aged alloys, the particles are much wider apart. The
dislocation can bend around each precipitate and leaving
dislocation loops [Orowan, 1948]. (c) A schematic representation
showing sheared N i3Al particles by passing dislocations in a Ni-
19% Cr-6% Al alloy after 2 % deformation [based on Gleiter and
Hornbogen, 1965].

A diagram schematically illustrating an idealized deformation
route from a B2 single crystal to a B 19' single crystal where the
transformation twinning and detwinning of martensite are well
separated. (a-b) The martensitic transformation proceeds through
the movement of parent/martensite interface. (b—c) The
deformation is realized by the coalescence of favorable twin
variants with respect to applied stress.

The orientation dependence of recoverable tensile strain for NiTi
single crystals in B2-to-B19' transformation: (a) The values
predicted from the lattice deformation matrix [Saburi and Nenno,
1981]. (b) The measured values from solution-treated single
crystals [Miyazaki et al. 1984]. Numbers in parentheses indicate
the measured results and contour lines represent predicted values.

(a) The orientation dependence of recoverable strain for B2-to-R

130

131

132

133

134

135

136

137

138

Figure 1-20

Figure 1-21

Figure 1-22

Figure 1-23

Figure 1-24

Figure 1-25

Figure 1-26

Figure 2-1

Figure 2-2

Figure 2-3

Figure 2-4

transition at a temperature of (TR - 35) K. Solid circles indicate
the experimental data, whereas contour lines indicate calculated
values; (b) The temperature dependence of the recoverable strain
for various orientations. The recoverable strain in all directions
increases gradually as temperature decreases from TR [Miyazaki,
Kimura and Otsuka, 1988].

(a) The effect of tension cycling on superelasicity characteristics
in Ni495Ti505 alloy [Miyazaki et al 1986]; (b) The effect of grain
size on superelasticity in Ni50,5Ti49_5 alloy [Saburi, Yoshida and
Nenno, 1984].

Schematic representation of (a) planar diode sputter deposition
[modiﬁed from Thornton, 1982], and (b) planar magnetron
[modiﬁed from Ahmed, 1987].

A schematic illustration of the effect of substrate temperature on
phase formation for a hypothetical polymorphic material
[Thomton, 1977].

Structural zone models for physical vapor deposition processes.
(a) The model proposed by Movchan and Demchishin [1969]
for electron beam evaporations; (b) The model proposed by
Thornton [1974] for sputtered metal coatings.

A computer modeling of ﬁlm growth using 2D hard spheres by
Muller [1985]. (a-c) show the inﬂuence of temperature on
packing density of ﬁlms. (d) displays that the temperature effects
packing density while the deposition rate changes the transition
temperature of density-to-unity density.

Schematic microstructures and grain distributions for thin ﬁlms
undergoing secondary grain growth [Thompson,1990]. (a)
Microstructure before the secondary grain growth occurs. (b)
Intermediate stage in which the grain sizes are bimodelly
distributed. (c) After the secondary grain growth microstructure
shows crystallographic (ﬁber) texture.

Schematic drawing of two devices [Johnson, 1992] using the
shape-memory effect of NiTi thin ﬁlms: (a) a microvalve, and (b)
a microactuator.

Photograph of magnetron sputter machine developed for this
study.

Schematic diagram of magnetron sputter system developed for
this study.

The weight loss characteristics of Kapton in air and helium at a
heating rate of 3 K/min [Product Bulletins, DuPont, 1993].

Differential scanning calorimetry (DSC) curves of sputter-
deposited binary and ternary NiTi ﬁlms at a heating rate of 20 K/

xi

139

140

141

142

143

144

145

146

147

148

149

Figure 2-5

Figure 2-6

Figure 2-7

Figure 2-8

Figure 2-9

Figure 2-10

Figure 2-11
Figure 2-12

Figure 2- 1 3

Figure 2-14

Figure 2-15

Figure 2-16

Figure 2-17

Figure 2-18

min showing the crystallization temperature.

Experimental set-up for in situ observation of electrical resistivity
change in crystallization annealing of NiTi/Kapton laminate.

The diagram showing the electrical resistivity of NiTi ﬁlm as a
function of temperature during crystallization annealing treatment
at a heating rate of 6.7 K/min.

Residual gas pressure vs. temperature curves of Kapton ﬁlm at
a heat-up rate of 5 K/min.

Schematic drawings of substrate heating devices which have been
tested in this study. (a) A conventional substrate holder with a
heater situated under the substrate. (b) A heating device consists
of four 625 watt quartz lamps clustered around the sputter source
to irradiate the substrate.

Schematic illustration showing the exploded view of two-sided
substrate heating device.

A patterning-etching process for fabrication of NiTi/polyimide
actuator elements.

Pattern for making eight NiTi/polyimide actuator elements.

A plot of substrate temperature and base pressure vs. time for a
typical deposition run (T = 670 K).

A quartz-tube vacuum furnace used for heat-treatment of N iTi
ﬁlms.

A drawing illustrating the preparation of TEM disks. Three
sampling locations along the depth of the NiTi thin ﬁlms could be
obtained by a masking technique.

(a) The reciprocal lattice of the B2 structure in [110] zone. (b)
The geometry for calculating the positions of diffracted arcs upon
tilting. Where (p is the angle between tilting axis and the arcs in
the image plane, 0 is the tilting angle of the specimen, dis the
distance between the zero-order reciprocal rings and the higher-
order reciprocal rings and R is the radius of arcs in the image
plane.

Schematic illustration of tensile test apparatus for evaluating thin
NiTi ﬁlms at various temperatures.

Finite element grid for analysis of rectangular thin ﬁlm tensile
specimens.

A contour map obtained from the ﬁnite element analysis showing
the maximal principal stress of thin ﬁlm tensile specimen. It can

be seen that the maximal principal stress reaches the maximum at
edges near the ends of the gauge section.

xii

150

151

152

153

154

155

156
157

158

159

160

161

162

163

164

Figure 2-19

Figure 3-1

Figure 3-2

Figure 3-3

Figure 3-4

Figure 3-5

Figure 3-6

Figure 3-7

Figure 3-8

A plot of the number of samples fractured versus the normalized
x-position where the fracture occurred. Each number is collected
from the interval of (x, x + 0.05).

TEM bright-ﬁeld (BF) image in (a) showing a completely
featureless structure, and the associated selected area diffraction
pattern (SADP) in (b) showing a broad, diffuse halo pattern
which are normally observed in the amorphous materials. The
micrographs were taken from the midplane of the ﬁlm H573 at a
tilt of 0°.

TEM BF image in (a) and the associated SADP in (b), taken from
the interface of the ﬁlm H598 at a tilt of 0°, showing the
amorphous structure.

TEM BF image and the associated SADP taken from the midplane
of the ﬁlm H598 at a tilt of 0°. The SADP reveals a single-crystal
diffraction pattern in the [110],” zone together with a
superimposed diffuse halo pattern scattered from the amorphous
background. This initial crystallite has an in-plane size greater
than 1 pm.

TEM BF images and the associated SADP’s taken from the
surface of the ﬁlm H598. The BF image (a) and (d), respectively,
were recorded at a tilt of 0 and 45°. Elongated grain images in ((1)
suggests a columnar grain structure. The SADP (b) and (c) were
recorded at a tilt of 0 and 30°, respectively, showing a strong
(110)ﬁber-texture. The non-even intensities of the diffraction
rings in (b) reﬂect the restricted in-plane grain orientations.

TEM BF image and the associated SADP’s taken from the
interface of the ﬁlm H623. The SADP (b) and (c) were recorded
at a tilt of 0 and 30°, respectively, showing a strong (110) ﬁber-
texture.

TEM BF images and the associated SADP’s taken from the
midplane of the ﬁlm H623. The BF image (a) and ((1),
respectively, were recorded at a tilt of 0 and 45°. Elongated grain
images in ((1) implies a columnar grain structure. The SADP (b)
and (c) were recorded at a tilt of 0 and 30°, respectively, showing
a strong (110) ﬁber- texture.

TEM BF image and the associated SADP’s taken from the surface
of the ﬁlm H623. The SADP (b) and (c) were recorded at a tilt of
0 and 30°, respectively, showing a strong (110) ﬁber-texture.

TEM BF image and the associated SADP’s taken from the
interface of the ﬁlm H673. The lenticular precipitates in (a) can
be distinguished (indicated by arrow). The SADP (b) and (c)
were recorded at a tilt of 0 and 30°, respectively, showing a
strong (110) ﬁber-texture.

xiii

165

166

167

168

169

171

172

174

175

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

TEM BF images and the associated SADP’s taken from the
midplane of the ﬁlm H673. The BF image (a) and (d),
respectively, were recorded at a tilt of O and 45°. A precipitate
particle displays a perfect ‘butterﬂy’ shape of the image contrast
as indicated by the arrow. Again, elongated grain images in ((1)
implies a columnar grain structure. The SADP (b) and (c) were
recorded at a tilt of 0 and 30°, respectively, showing a strong
(110) ﬁber-texture.

TEM BF image taken from the surface of the ﬁlm H673.

(a) The high magniﬁcation image of precipitate with ‘butterﬂy’
shaped contrast. The strain ﬁelds around Ni4Ti3 particles come
from the lattice mismatch between particles and B2 matrix. (b)
The habit plane of precipitate is {111}32 and the maximum
mismatch of -2.9 % is along the normal to the precipitate disk.
This minus value indicates a tensile strain perpendicular to the
precipitate disk [Kainuma and Matsumoto, 1988].

TEM BF image and the associated SADP’s taken from the
interface of the ﬁlm H723. Extremely ﬁne grains, 25 nm in
diameter, were nucleated on very thin initial amorphous layer.
The SADP (b) and (c) were recorded at a tilt of 0 and 30°,
respectively. The unchanged SADP's upon tilting imply the
absence of the grain texture.

TEM BF image and the associated SADP’s taken from the
midplane of the ﬁlm H723. The SADP (b) and (c) were recorded
at a tilt of 0 and 30°, respectively. The unchanged SADP's upon
tilting display the random grain orientations.

TEM BF image taken from the surface of the ﬁlm H723. The
precipitates show smaller size than that in the midplane. Three
ranges of precipitates can be seen in the low part of the
micrograph.

The micrograph and associated SADP taken at the midplane of
the ﬁlm H703 revealing crystalline B2 grains and transgranular
precipitates (indicated by arrows).

TEM BF image and the associated SADP’s taken from the
interface of the ﬁlm H773. The SADP (b) and (c) were recorded
at a tilt of 0 and 30°, respectively. The micrographs display
extremely ﬁne grains and the random grain orientations.

TEM BF image and the associated SADP’s taken from the
midplane of the ﬁlm H773. The SADP (b) and (c) were recorded
at a tilt of 0 and 30°, respectively. The micrographs show coarse
precipitates and the random grain orientations.

TEM BF image taken from the surface of the ﬁlm H773 showing
the coarse grains and precipitates.

xiv

176
178

179

180

181

182

183

184

185

186

Figure 3-19

Figure 3-20

Figure 3-21

Figure 3-22

Figure 3-23

Figure 3-24

Figure 3-25

Figure 3-26

Figure 3-27

Figure 3-28

Time-lapse sequence illustrating the development of the Johnson-
Mehl microstructure from the untransformed body [Mahin, 1980]. 187

A schematic diagram illustrating the formation of columnar

grained structures. By means of superposing two-dimension

grain growth upon ﬁlm growth (increasing thickness), a

columnar structure can be formed as shown. The shape of the
columnar grains is controlled by both the in-plane grain growth

rate and the deposition rate. 188

A schematic diagram showing the effect of in-plane grain size on

texture formation. (a) At higher substrate temperatures the surface

and interface energies of grains are not signiﬁcant due to smaller
fraction of the interface and surface areas of the grains and the

crystals show random orientations. (b) Films formed at lower
temperatures present a strong (110) ﬁber-texture, i.e., the (110)

plane of grains are restricted to parallel to substrate plane, but with

no preferred in-plane orientation. 189

SADP sampled at the midplane of the ﬁlm H673. The indices of

Ni4Ti3 phase were underlined for purpose of distinguishing them

from the B2 reﬂections. The ring pattern in the ﬁgure also shows

the (110) grain orientation, since the (310),” reﬂection having the

radius of 64.3 mm absents. 190

Two SADP's, each including approximately one grain,_ sampled
at two areas of the midplane of the ﬁlm H773. (a) [133 ]32 //
[1 1 11mm; zone, and (b) [ 132]32// [1 101mm: zone. 191

Diagrams showing the size changes of B2 grains and precipitates
as a function of deposition temperature in three layers. 192

A structural model for Ni-rich NiTi ﬁlms sputter-deposited at a
substrate temperature of 573-773 K. 193

Schematic presentations predicting the effects of Ni and Ti
diffusions on Ni-rich ﬁlm microstructure. In Region I, the surface
diffusions of Ni and Ti are not active simultaneously. Films
deposited at this region are generally amorphous in structure. In
sub-region 12, hyperstoichiometric precipitates may form via Ni
surface diffusion in amorphous phase. In Region H, ﬁlms are
mainly of columnar B2 grains. In sub-region 112, large
precipitates may be produced through Ni bulk diffusion and B2/
ppt dual-phase may form. In Region III, bulk diffusions for both
Ni and Ti start. Films with microstructures similar to that of fully
annealed and aged ﬁlms may result. 194

TEM bright-ﬁeld image of ﬁlm A1 which was produced by aging
the ﬁlm H623 at 858 K for 1 hour and furnace cooling. 195

(a) TEM bright-ﬁeld image and SADP’s of ﬁlm A2. (b) The

inserted picture taken at 45° specimen tilt revealing columnar grain
structure. (c), (d) and (e) are SADP’s at three tilt angles 0’, 30°

XV

Figure 3-29

Figure 3-30

Figure 3-31

Figure 3-32

Figure 3-33

Figure 3-34

Figure 3-35

Figure 3-36

Figure 3-37

Figure 3-38

and 45°, respectively, displaying (110) texture. The axis of tilt is
indicated in (d) and (e).

TEM bright-ﬁeld image and SADP of ﬁlm A3 which was
produced by aging the ﬁlm H723 at 858 K for 1 hour and furnace
cooling. The SADP in (b) was taken at 30° tilt showing random
grain orientation.

TEM bright-ﬁeld image of ﬁlm A4 which was produced by aging
the ﬁlm H723 at 813 K for 1 hour and furnace cooling.

TEM image of the ﬁlm A5 obtained by annealing ﬁlm H673 at
988 K for 1 hour and furnace cooling. The size of grains have
increased during annealing. Obviously, the bulk diffusion for
NiTi grain growth have occurred.

TEM image shows the microstructure of ﬁlm A6 produced by
annealing the ﬁlm H673 at 1123 K for 1 hour and furnace
cooling. The ﬁlm is consist of large grains and lenticular
precipitates. The precipitates align along well-deﬁned
crystallograghic planes of the matrix. Besides, very ﬁne
precipitate particles with strain ﬁeld conuast also can be noticed
in the area between the large precipitates.

TEM image of ﬁlm A7 obtained by annealing the ﬁlm H673 at
1073 K for 2 hours and air cooling. The precipitation tends to
occur more rapidly at grain boundary areas since the grain
boundary is a favorable site for heterogeneous nucleation of
precipitation.

Cross-section SEM micrograph of A7 displaying coarse-grained
microstructure. Deep etched surface morphology reveals the
crystallographic facets of the grains.

TEM image of ﬁlm A8 deposited at 1073 K, 2 hours and 673 K,
1 hour two-stage annealing.

BF image and two single-crystal SADP's taken from the large
precipitate plate in ﬁlm A3. (a) An isolated precipitate plate in
the edge of the TEM foil was examined in favor of excluding
diffraction spots from the matrix. (b) The [0-10] zone of Ni3Ti2
phase. (c) The [110] zone of Ni3Ti2 phase.

XRD curve of the ﬁlm A5 recorded at the room temperature
(298 K). Besides (110) B2 peak, four peaks match the
reﬂections of Ni4Ti3 phase at 29 angle between 35° and 60°
within 2 % error.

BF images and single-crystal SADP’s taken from the ﬁlm A6.
The index of large lenticular precipitates could be completed by
assuming that they possess the Ni4Ti3 structure. (b) and (d) are

SADP's of zone [012]!” // [001]N,41~,3 and zone [230]!” //
[221]”,4m, respectively, taken from indicated areas in (a) and (c).

xvi

196

198

199

200

201

202

203

204

205

206

Figure 3-39

Figure 3—40

Figure 3-41

Figure 3-42

Figure 3-43

Figure 3-44

Figure 3—45

Figure 3-46
Figure 3-47

Figure 3-48

The underlined indices represent the diffractions of precipitates.

XRD curve of the ﬁlm A7 recorded at 353 K showing (110) B2
peak and two Ni4Ti3 phase peaks at 20 angle between 35° and 60°.
The reﬂections of Ni4Ti3 phase are very weak due to its ﬁne size.

The XRD curves of the ﬁlm A8, taken from the same specimen at
353 K and the room temperature respectively. The curve (a)
identiﬁes the (110) B2 peak and ﬁve Ni4Ti3 peaks in 20 angle
between 35° and 60°. In the curve (b), the same set of the Ni4Ti3
reﬂections appears, however, the (110) B2 peak splits into two

NiTi R-phase peaks: (011) and (GI 1)with a separation about
0.4°.

Schematic drawings of in-plane views of various microstructure
formed in the post-deposition annealing.

Schematic diagrams illustrating the classiﬁcation of electrical
resistivity vs. temperature curves for NiTi alloys. (a) Two-stage
curve involving Martensite H R-phase H Austenite phase
transformations. (b, c) The M H R and R H A curves
decomposed from the curve (a). Curve (d) represents a two-stage
curve with an incomplete austenite —> R-phase transition. (e) The
curve involving R-phase H Austenite transformation which is the
same as the curve (c). (f) The curve involving Martensite H
Austenite transformation.

Electrical resistivity vs. temperature curves of as-sputtered NiTi
ﬁlms deposited at various substrate temperatures.

Bright-ﬁeld images of midplane of the ﬁlms H723 and H673.
The micrographs were recorded at 101 K, which is well below
their M, temperatures. (a) The image of ﬁlm H723. (b) The
image of ﬁlm H673.

DSC scans (a) and electrical resistivity vs. temperature
measurements (b) on ﬁlm H703 having composition of NisogTi493.
In (a) the second heating scan from 223 to 373 K is indicated by
double arrows.

Electrical resistivity vs. temperature curves of annealed NiTi ﬁlms.

Ultimate elongation curves tested at 423 K for ﬁlms H673, H723
and A3. According to Clausius-Clapeyron relations shown in
Figure 3-61, the critical stresses to induce martensite at 423 K
would be 1250, 1035 and 992 MPa for H673, H723 and A3
respectively. Clearly, the yield points in the ﬁgure were caused
by the signiﬁcant permanent deformation in the parent phase.

Ultimate elongation curves tested at 323 K for ﬁlms H673, H723
and A3. The curves show well-deﬁned three stages: ﬁrst the
elastic deformation of parent phase, then the plateau of parent-to-
martensite transformation and ﬁnally the elastic deformation of

xvii

207

208

209

210

211

212

213

214
215

217

Figure 3-49

Figure 3-50

Figure 3-51

Figure 3-52

Figure 3-53
Figure 3-54

Figure 3-55
Figure 3-56
Figure 3-57

Figure 3-58
Figure 3-59

Figure 3-60

Figure 3-61
Figure 3-62

Figure 3-63

martensite.

Isothermal stress-strain curves of ﬁlm H623. The yielding due to
the detwinning of martensite variants and stress-induced
martensitic transformation did not occur at the stress level above
300 MPa. The thin ﬁlm behaved as an ordinary metallic material.

Isothermal stress-strain curves of the ﬁlm H673.

TEM bright-ﬁeld images from deformed ﬁlm H673. Compared
with its original structure, it is seen that the microstructure
essentially remained unchanged.

A series of isothermal stress-strain curves generated from a single
tensile specimen of the ﬁlm H703 at various test temperatures.
Arrows indicate the critical stresses for the ﬁrst and second
yieldings.

Isothermal stress-strain curves of the ﬁlm H723.

TEM bright-ﬁeld images from deformed ﬁlm H723. Just like its
undeforrned structure, well-deﬁned precipitate particles and grains
without any dislocation can be seen.

Isothermal stress-strain curves of the ﬁlm A2.
Isothermal stress-strain curves of the ﬁlm A3.

(a) and (b) TEM bright-ﬁeld images taken from the deformed
ﬁlm A3. The absence of precipitates inside grains led the
formation of dislocations.

Isothermal stress-strain curves of the ﬁlm A5.

TEM bright-ﬁeld images taken from the deformed ﬁlm A5.

(a) Slip bands which formed by a number of passage of
dislocations spread throughout entire grains. The slip bands in
different grains tend to align in a common direction which is the
direction of maximum shear stress. (b) High density dislocations
can be seen.

Isothermal stress-strain curves of the ﬁlm A7. The loading
curves dose not show any transformation yielding due to the
extremely ﬁne Ni4Ti3 particles.

Isothermal stress-strain curves of the ﬁlm A8.

TEM bright-ﬁeld image from the deformed ﬁlm A7. The
microstructure was unchanged in stress-strain cycling.

TEM bright-ﬁeld image taken from the deformed ﬁlm A8. The
microstructure was unchanged in stress-strain cycling.

xviii

218

219
220

221

222
223

224
225
226

227
228

229

230

231

232

233

Figure 3-64

Figure 3-65

Figure 3-66

Figure 3-67
Figure 3-68
Figure 3-69
Figure 3-70

Figure 3-71

Figure 3-72

Figure 3-73

Figure 3-74

Figure 3-75

Figure 3-76

Figure A-l

A schematic a-e-T diagram showing six types of stress-strain
behavior depending on the deformation temperature relative to the
transformation temperatures in polycrystalline NiTi thin ﬁlms.

Critical stresses to induce the martensite vs. adjusted deformation
temperature curves for NiTi ﬁlms.

Tensile yield stress and reversion stress as a function of test
temperature for ﬁlm H703.

Stress-temperature diagram for ﬁlm H673.
Stress-temperature diagram for ﬁlm H723.
Stress-temperature diagram for ﬁlm A3.

(a) Output of mechanical energy vs. adjusted deformation
temperature for NiTi ﬁlms. (b) Energy efﬁciency vs. adjusted
deformation temperature curves.

SEM image showing the fractography of ﬁlm H623. The fracture
surface consisting of many microdimples exhibits characteristics
of mode I (the opening mode) ductile failure. The fracture is a
transgranular one since no feature of columnar grain structure can
be seen in the fracture surface.

SEM image showing the fracture surface of the ﬁlm H673. The
columnar structure is clearly seen and its grain size (abut 0.25 pm)
is consistent with that observed by TEM.

SEM images showing the fractography of ﬁlm H723. (a) is the
fracture surface viewed from surface side, and (b) is the fracture
surface viewed from substrate side. The external neck area in
surface side of the ﬁlm reveals a serial of valleys as indicated by
arrows, but that in substrate side shows even and ﬂat
morphology.

SEM image showing the fracture surface of ﬁlm A3. The fracture
is mainly transgrainular type but it is also inﬂuenced by equiaxial

grain structure which was formed in 858 K, 1 hour annealing (i.e.,

recrystallization) process.

The fracture surface of ﬁlm A5 shows large cup-like feature which
is due to a greater grain size. Evidently, the combination of large
grains and coarse precipitates posses poor structure stability.

The SEM image showing the fracture surface of the ﬁlm A7. The
fractography shows a ﬂat and straight fracture surface with many
small dimples. No neck is found in the edge of fracture surface.
The free surface of the ﬁlm is very smooth.

X-ray diffraction spectra of NiTi ﬁlm on Kapton substrate
sputter deposited at 698 K showing that the NiTi was fully

xix

234

235

236
237
238
239

240

241

242

243

244

245

246

Figure A-2

Figure A-3

crystallized during the deposition. 248

Photograph demonstrating electrically-excitable NiTi/polyimide

actuator. The prototype microactuator displayed an electrical

impedance of 30 ohm. (a) The actuator was deformed into an
approximately 100 mm radius hairpin bend at the LN2

temperature. (b) With the device still immersed in cold nitrogen

gas, the application of approximately 3 volts dc induced a rapid

and complete recovery of the bending strain. 249

A series of photographs showing a bending test on a free-

standing NiTi strip (ﬁlm H703). (a) The ﬁlm could be sharply

bent in the LN2 temperature. (b-c) When the LN2 ﬂask was

removed, the ﬁlm restored its original ﬂat shape. 250

XX

INTRODUCTION

There are certain interrnetallic compounds, such as NiTi, CuAlN i, CuZnAl and
Aqu, which undergo ﬁrst-order shear-dominated hysteretic displacive transformation
from a high-symmetry parent phase (an austenite) to a lower-symmetry martensite phase on
loading or cooling. Several of these possess a unique and useful strain-recovery ability, a
further subset of alloys, based on ordered equiatomic nickel-titanium, possesses good
strength and toughness as polycrystalline engineering materials and have found a number
of important industrial uses.

We designate the start and ﬁnish temperature of martensitic transformation on cooling
as M, and Mf, and the start and ﬁnish temperature of reverse transformation on heating as
A, and Af respectively.‘ When such an alloy is plastically deformed at below its Mf
temperature, deformation up to 5-8 %, far beyond normal metallic elastic limits, can be
produced at relatively low stress. Upon heating through its A, temperature, the specimen
may spontaneously recover its original pre-deformed shape, exerting great force if
constrained. This behavior has been termed the shape-memory eﬁect (SME).
Thermoelastic martensitic transformations can also be induced by applying stress to the
austenite phase. Within a certain temperature interval, rather large nonlinear elastic strains
(5-8 %) can be completely recovered upon unloading at a constant temperature. Such
materials are said to show transfonnationl superelasticity2 which may be viewed as an

isothermal shape-memory effect.

 

I As discussed in the Figure 3-42b of the Section 3.3.1, for NiTi system involving R-phase transformation,
the characteristic temperatures of A, and Ar actually represent the start and ﬁnish temperatures of M —> R
hase transition.
The terms superelasticity and pseudoelasticity have often been used interchangeably in the literature.
Generally, any non-linearity in a stress-strain curve upon unloading can be referred to as pseudoelasticity.

Among the many shape-memory alloys which have been developed as stress-strain-
temperature functional materials, nearly equiatomic N iTi alloys are the most commercially
important due to their superior mechanical properties, e.g. high ductility (having fracture
strains as great as 60 %) [Buehler and Wang, 1968], long fatigue life (up to 103 cycles at
A8 = 0.02) [Miyazaki, Sugaya and Otuska, 1989], good corrosion resistance (slightly more
noble than 316 stainless steel) [Melton, 1990] and biocompatibility [Haasters et al., 1990],
although there are some disadvantages in terms of high cost and difﬁculty in machining.
Applications of ingot-metallurgy NiTi alloys utilizing SME can be found in the areas of
pipe couplings, fasteners, actuators and elecuical interconnection. Applications utilizing
superelasticity include eyeglass frames and many medical devices, such as orthodontic
arches, stents, kink-resistant guidewires and bendable surgical tools.

Recently, N iTi thin ﬁlms with a thickness of several micrometers have attracted much
attention for use as actuating elements in microelectromechanical systems (MEMS). N iTi
shape-memory alloys have a potential advantage over electrostatic, piezoelectric or
bimetallic materials because of the large energy output associated with the reverse-
martensitic transformation [Johnson, 1991; Wolf and Heuer, 1995]. These
transformational displacements, if constrained, generate recovery stresses exceeding 650
MPa [Zhang, 1997]. In this case, one gram force could theoretically be produced by a
NiTi strip having a cross-section of 3 x 5 umz. Moreover, in thin ﬁlm form, due to a large
surface-to-volume ratio, heat transfer is fast and the intrinsic disadvantage of slow response
in bulk NiTi alloy actuators can in principle be greatly improved [J ardine, 1992].

In conventional NiTi shape-memory alloy manufacturing processes, pure nickel and
titanium are melted together under a vacuum or a protective atmosphere to produce NiTi
ingots. Then the ingots are forged, rolled or drawn to form N iTi alloy sheets and wires.

The minimum dimension of NiTi alloys produced is several tens of micrometers. In

 

In shape—memory alloys a enormous amount of deformation can be completely recovered during the reverse
transformation. This type of pseudoelasticity is often speciﬁed as superelasticity [Duerig and Zadno, 1990].

contrast, sputter deposition, a vacuum coating process, is able to form N iTi thin ﬁlms with
a thickness less than a few micrometers. Furthermore, NiTi thin ﬁlms are suitable for
batch microfabrication by a standard semiconductor integrated circuit (IC) processing
applied to silicon wafers.

A prerequisite of martensitic phase transformations underlying shape-memory effect
and superelasticity is an ordered crystalline structure. NiTi ﬁlms sputter—deposited at an
ambient temperature (< 623 K) typically amorphous. Although such as-sputtered
amorphous ﬁlms may be utilized as corrosion-resistant and wear-resistant coatings, a post-
deposition anneal at a temperature exceeding 773 K is necessary to achieve crystallization in
order to subsequently realize the shape-memory effect associated with martensitic
transformation. Recently, much effort has been expended to clarify the dependence of
microstructures, phase transformation characteristics and mechanical properties on post-
deposition annealing processes. On the other hand, crystalline NiTi ﬁlms may also be
directly produced by the sputter deposition at an elevated substrate temperature. From both
fundamental and practical viewpoints, it is interesting to study microstructure and related
thermomechanical properties obtained with heated substrate deposition. However, so far
no systematic work have been reported in the open literature.

The ﬁrst goal of this work is to establish a basic understanding of the evolution of
microstructure of NiTi ﬁlms as a function of deposition temperature. It is expected that the
structure be different from that of post-annealed ﬁlms because of the different kinetic
conditions involved in the thin ﬁlm process. The second objective is to qualitatively
determine the effects of microstructure (grain size and precipitates) on the phase
transformation, thermomechanical behavior and structural stability.

In this study, a series of NiTi ﬁlms were sputter deposited at various substrate
temperatures and characterized by TEM in terms of grain structure, preferred grain

orientation and precipitate morphology. Some as-sputtered ﬁlms were further given a post-

deposition annealing process in order to generate diversiﬁed set of microstructures for

exploring structure effects. The phase transformation temperatures of N iTi ﬁlms were
determined by four-point resistivity measurements. The thermomechanical response of the
ﬁlms was evaluated by isothermal tensile tests in a temperature-strain window where the
ﬁlms exhibited the shape-memory effect and/or superelasticity. Finally, NiTi ﬁlms that had
experienced cyclic deformation were reexamined by TEM and SEM for the purpose of
predicting the structural stability against permanent deformation. In addition, some subtle
experimental issues, such as depositing crystalline NiTi thin ﬁlms onto a polymeric
(Kapton3) substrate and the measurement of stress-strain curves of N iTi thin ﬁlms are
detailed.

The present work has demonstrated that a fully crystallized NiTi ﬁlm could be
produced at a deposition temperature of 623 K. This is about 150 K lower than the
minimum crystallization temperature for a post-deposition annealing processes. As-
sputtered ﬁlms displayed very ﬁne columnar-grained structures having in-plane grain sizes
of 0.1-0.3 um. Such ﬁne grained structure is not readily attainable by the prevailing N iTi
thin ﬁlm processes. Films sputter-deposited at 723 K displayed perfect superelasticity in
that 2.6 % transformational strain, at a stress level of 700 MPa, could be completely
recovered upon unloading. These results suggest that hot-substrate deposition is a viable
alternative NiTi thin ﬁlm process route, which has advantages in terms of eliminating the
post-deposition annealing, depressing the maximum process temperature and producing
favorable ultraﬁne grained structures.

The main body of the present thesis is organized into three chapters. Chapter 1 forms
the background necessary for discussing the experimental results obtained. The
experimental arrangements, procedures and analytical methods employed in this study is

covered in Chapter 2. Chapter 3 presents the experimental results and discussion.

 

3 Kapton is a commercial name of a series of pruducts produced by DuPont.

CHAPTER 1

BACKGROUND

This study mainly deals with the therrnoelastic NiTi thin ﬁlms. The background
necessary for understanding the experimental data of NiTi thin ﬁlms presented in this study
should include following three categories:

(1) Thermoelastic martensitic transformation: This has been well established for bulk
alloys and is the basis of shape-memory effect and transformational superelasticity of NiTi
thin ﬁlms.

(2) Mechanical response of metal crystals: In NiTi thin ﬁlms three kinds of
deformations occur, namely, the elastic deformation attributed to elasticity displacing
individual atoms from their equilibrium positions, permanent deformation caused by
dislocation-mediated slip between crystal lattice planes, and recoverable deformation due to
stress-induced displacive transformation from the parent phase to the martensite.

(3) The formation and structure of thin ﬁlms: Tire kinetic conditions goveming thin
ﬁlm formation during vapor deposition processes are quite different from those for melt-
solidiﬁcation process.

In this chapter, general aspects of therrnoelastic transformation are ﬁrst described
from thermodynamic and crystallographic viewpoints in Section 1.1. Then the special
features of displacive transformations (martensitic and R-phase transformations), as well as
diffusional transformations in near-equiatomic NiTi alloys, are brieﬂy reviewed in Section
1.2. Section 1.3 and 1.4, respectively, cover the characteristics of permanent and

recoverable deformations occurring in metallic crystals and NiTi alloys. Section 1.5

provides a review of thin ﬁlm formation process and the correlation between process
conditions and resultant ﬁlm structure. Finally, in Section 1.6 the current state of
knowledge regarding NiTi thin ﬁlms is summarized and the objectives of this work are

proposed. The organization of this chapter can be illustrated as follows:

Sec. 1.1. Martensitic
transformation

Sec. 1.3. Deformation of Sec. 1.6. Studies on NiTi films
metallic crystals

 

 

 

 

Thermoelastic NiTi thin ﬁlms

Sec. 1.2. Transformations in NiTi Sec. 1.5. Formation and structure
of thin ﬁlms

 

Sec. 1.4. Deformation of NiTi

1.1. Thermoelastic Martensitic Transformation
1.1.1. General Characteristics

There are two main types of phase transformation in the solid state: diffusional and
displacive. Diﬁ‘usional transformations are those in which a new phase can only be formed
by moving atoms over a long range. The new phase often has chemical composition
different from that of the parent phase. Since long distance atomic migration is involved,
the transformation is not only dependent upon the temperature but also dependent upon the
length of time. On the other hand, in displacive transformations, the atoms undergo a
cooperative atomic rearrangement over distances smaller than interatomic distances.
Because no long-range atomic movement is necessary, the progress is only dependent upon
temperature and the motion of the interface between the two phases is only limited by the

speed of sound in the lattice.

Martensitic transformations are ﬁrst-order,4 displacive phase transformations. The
martensite phase is formed upon cooling from a high temperature parent phase called the
austenite. Martensitic transformations are of two types: therrnoelastic and non-
therrnoelastic. For thermoelastic transformations the interface is quite mobile and the
renucleation of the parent phase during reverse transformation is not necessary. The
martensite plates form and grow continuously upon cooling, and shrink and revert to the
initial parent phase orientation upon heating by the exact reverse path. Just as stated by
Tong and Wayman [1974], the ﬁrst martensite plate appearing at M, is generally thought to
be the last one to disappear at Af, and the martensite plate appearing last at Mf is the ﬁrst to
disappear at A,. In contrast, for non-thermoelastic transformations, the interface between
the parent phase and martensite is relatively immobile. It does not undergo inverse
movement during heating, but instead the parent phase must be renucleated within the
martensitic phase. As a result, the parent plate does not maintain the original parent phase
crystal orientation [Kessler and Pitsch, 1967].

From energetic considerations, the thermoelastic transformation satisﬁes a condition
of local energy balance between chemical and non-chemical forces. Chemical forces are
attributed to the difference in Gibbs free energy between the parent phase and martensite,
and act as a driving force promoting the phase with lower energy at each temperature.
Non-chemical forces arise from three main contributions: interface energy, stored elastic
strain energy, and frictional work [Ortin and Planes, 1988 and 1991]. The interface energy
is caused by the lattice distortion in both phases in the interface area. The stored elastic
strain energy is associated with the accommodation of transformational shape and volume
changes. This energy is accumulated inside the lattices of the transforming specimen

during the forward (cooling) transformation, and released during the reverse

 

4 According to Ehrenfest’s classiﬁcation [Swalin, 1961], a ﬁrst-order transformation is one for which the
Gibbs free energy G as a function of a given state variable (V, P, T) is continuous, but the ﬁrst derivative
of the G with respect to the state variable is discontinuous. Other examples of ﬁrst-order transitions are
vaporization and fusion.

transformation. The frictional work is the energy dissipated in the specimen as heat and
acoustic emission [Bararn and Rosen, 1981] due to interfacial motion either during the
growth or during the shrinkage of martensite plates. Upon cooling, when a given
temperature is reached, the interfaces move a certain distance into the parent phase, such
that the chemical energy released is equal to the sum of interface energy and elastic energy
created in the lattices, plus the energy that is dissipated by frictional work. Further
progress of the forward transformation requires further decrease in the temperature.
Based on thermodynamic equilibrium formalisms, any dissipative processes such as
frictional work should be neglected, otherwise there will be some deviation from the
equilibrium. Compared to the elastic strain energy, the interface energy is very small in a
therrnoelastic transformation, due to a coherent nature of martensite/parent interface
[Wayman, 1983]. If the frictional work and interface energy are ignored, the state of a
system can be described merely by a thermal and an elastic term. This is the origin of the
terminology: thennoelastic. As will be discussed later, the temperature and external stress

are interchangeable state variables which can induce the martensite transformation.

1.1.2. Crystallography of Martensitic Transformation

The crystallography of martensitic transformations is now well understood in terms
of the phenomenological theory of Lieberman, Wechsler and Read [1955]. The basic
assumption of the theory is that the martensite/parent phase interface (habit plane) should be
macroscopically undistorted in order to minimize strain energy during phase
transformation. According to this theory, the transformation consists of three operational
processes:

(1) A Bain distortion which forms the martensite lattice parameters and symmetry
from the parent lattice.

(2) A lattice invariant shear, which maintains the lattice structure, and in combination

with the Bain distortion, produces an undistorted interface.

(3) A rigid body rotation which brings the transformed lattice back in contact with the
parent lattice.

Figure 1-1 schematically represents a simpliﬁed two-variant martensitic
transformation in terms of these three operations [Chang, 1993]. Figure 1-1a shows the
lattices of the parent phase and the martensite. In Figure 1-1b, a Bain distortion, which
converts the parent phase into the martensite, but does not yield an undistorted plane
associated with the habit plane. A lattice invariant shear, through a reversible mechanism
such as twinning, produces an undistorted plane as shown in Figure 1-1c. Finally, a rigid
body rotation is shown in Figure l-ld. There is an invariant-plane strain associated with
martensitic transformations as indicated in Figure 1—1d by the vector m.

In general, thermally induced martensite consists of a twin-related, multiple-variant
combination called self-accommodating morphology (an example is the triangular
morphology shown in Figure 1-7, Section 1.2.1 for NiTi alloy). The combination of
variants breaks down the invariant-plane strain into many smaller portions and arranges
them in such a way that the total stored elastic strain energy created by lattice invariant shear
is a minimum.

The shape reversibility in the reverse transformations is ensured by the combination
of three characteristics inherent in shape-memory alloys:

(1) Thermoelastic nature: All phase transformations which are able to show the
shape-memory effect are thermoelastic ones [Delaey, Krishnan and Tas, 1974]. As

mentioned earlier, in this case no renucleation in the parent phase occurs during the reverse

transformation because of the mobile interface. The ﬁrst martensite plate appearing at M, is
the last one to disappear at Ar, and the martensite plate appearing last at Mr is the ﬁrst to
disappear at A, [Tong and Wayman, 1974]. The martensite plates shrink and revert to the
parent phase upon heating along the same path as they have formed. In this circumstance

the original orientation of parent plates is retained in the reverse transformation.

10

(2) Large difference in lattice symmetry: The parent lattices of shape-memory alloys
have higher symmetry (cubic-based B2 or D03) compared to that of martensites
(monoclinic, B19 etc.). While a large number of crystallographically equivalent martensite
variants can form from the parent phase (a maximum of 24 martensite variants can form
from a cubic NiTi parent phase), variants of the parent phase formed in the reverse
transformation are highly restricted.

(3) Ordered structure: All shape—memory alloys have an ordered parent phase except
for the ones which exhibit fcc H fct transformations [Gokin, 1984]. Coupled with the
restriction of lattice symmetry, only one orientation of austenite can form in martensite
lattices to avoid creation of ‘wrong’ ordered structures which raise the free energy of the
lattice. In case of the fcc H fct transformations, instead of the ordering, its reversibility
stems from the fact that the lattice correspondence is unique in the reverse transformation
due to very simple lattice change and lower symmetry of the fct phase [Otsuka and

Shimizu, 1977].

1.1 .3. Frictional Work and Stored Elastic Energy

In thermoelastic martensitic transformations, there is a temperature hysteresis
associated with the transformations, and there is a temperature range over which the
martensite and the parent phase co-exist. The hysteresis and the co-existence of the
martensite and the parent phase can be explained by the effects of the frictional work and
the stored elastic energy respectively [Van Humbeeck et al., 1990].

Salzbrenner and Cohen [1979] investigated the inﬂuences of the frictional resistance
and the elastic strain energy on thermoelastic martensitic transformations in Cu-14Al-2.5Ni
(w/o) alloy. In their work, a single-interface transformation from the parent phase [31 with
a D03 structure to martensite 71' with a 2H structure was achieved by cooling a single
crystal (20 x 6 x 6 m3) at a rate of 5—10 K/min, in a temperature gradient of 5 K/cm along

the length of the specimen. During a given gradient—cooling/single-interface experiment,

11

the temperature of the interface consecutively passing four spot-welded thermocouples and
was found to remain the same. Based on their observations, the fraction of martensite vs.
temperature curves can be represented in Figure 1-2a, where T8 and Tr, respectively, are
designated as martensite growth and martensite reversion temperatures, and T o is the
temperature at which the parent and the martensitic phase have the same chemical free
energy. In this case, elastic energy was not generated since the transformational shape
changed merely against aunospheric pressure.5 Furthermore, there was no change in
interface energy because the area of parent/martensite interface that extended across the
specimen section remained the same during interface motion. Evidently, the hysteresis was
caused by frictional resistance to interfacial motion and To lies between T3 and T,. If we

assume that the friction force is the same in both directions, To can be taken as

 

To: 3 ' (1.1)

The same single crystal could also be made to undergo multiple-interface
transformations into many martensitic variants by cooling it uniformly instead of in a
temperature gradient. As shown in Figure l-2b, the stored elastic energy depressed the
martensitic transformation and enhanced the reverse transformation. The martensitic
transformation was completed at a lower temperature (Mf< M,), and the reverse
transformation ‘prematurely’ starts at lower temperature (A, < Af). The thermodynamic
equilibrium temperature can be approached by

_ M s + A f

T0———2——— (1.2)

which was proposed by Tong and Wayman [1974; Wayman and Tong, 1977]. Since at

M, and Ar, where the ﬁrst martensite plate appears in forward transformation and the ﬁrst

 

5 The martensite produced is internally twinned. However, the contribution of the strain energy in the twin-
related martensite to the system is negligible [Salzbrenner and Cohen, 1979].

12

formed martensite disappears in the reverse transformation respectively, there is no elastic
strain energy involved.

Transformation observations were also made on polycrystalline specimens with
different grain sizes (D = 0.5, 4.0 mm). Figure 1-2c and Figure 1-2d show a downward
trend in the transformation temperatures with decreasing grain size. As expected, this is
due to increasing elastic constraint of grain boundaries [Salzbrenner and Cohen, 1979].
Similar phenomena were observed in surface—constrained Nitinol6 alloy samples by
Hedayat, Rechtien and Mukherjee [1992]. In their work the adhering carbon coating and
TiC carbide layer formed after annealing introduced a constraint force against martensitic
transformation. The in situ TEM examination suggested that during the martensitic
transformation the last martensite to form was close to the constrained surface.

Based on the preceding discussion, we can summarize the inﬂuence of the frictional
force and the stored elastic energy on a thermoelastic martensite transformation. Figure 1-
3a shows an ideal transformation in which neither frictional work nor stored elastic energy
is involved. Figure 1—3b and Figure 1-3c are curves where only the frictional force or
elastic strain energy exists. Finally, Figure 1-3d represent general cases where both the
frictional force and elastic strain energy exist. The curves in Figure 1-3a, 3b. 3c and 3d,
respectively, correspond to the simplest, unfolding 2, unfolding 1 and dual unfolding
models for phase transition proposed by Pence and Ivshin [1994]. It is interesting to note
that the relative values of M, and A, for a system are only determined by the magnitudes of
frictional force and stored elastic energy. The former tends to make A, >Ms while the latter

tends to make M, > A,.

 

5 A commercial name Nitinol was given to a series of near-equiatomic Ni-Ti compounds. Nitinol stands
for nickel titanium, Naval Ordinance Laboratory [Robinson, 1987].

13

1.1.4. Shape Memory and Superelasticity

Typical uniaxial stress-strain response of the shape-memory process is schematically
shown in the upper drawing of Figure 1-4.7 When the specimen is cooled below Mr,
thermally-induced martensite forms in a self-accommodating conﬁguration containing twin-
related variants (only two variants are shown for the purpose of simplicity, in real materials
up to 24 variants are formed) and there is no macroscopic shape change (point 0). Upon
loading, the specimen ﬁrst deforms elastically (point o-a). Then deformation proceeds
through the movement of the intervariant boundaries, in a process often called detwinning;
i.e., reorientation of a martensitic structure (point a-c). After removing the load, the
deformed shape remains except for a small elastic recovery (point c-d). However, if the
remaining variants revert to the parent phase of the original orientation upon heating (point
d—o), due to the speciﬁc correspondence between the parent lattice and each martensite
variant, the remaining plastic strain is recovered.

Transformational superelasticity is the other unique deformation manifestation of
shape-memory alloys as shown in the bottom drawing of Figure 1-4. At a temperature T>
Af, the specimen is in the austenite phase (point 0). The loading initially produces elastic
distortions of the austenite (point o-a). At point a, stress-induced martensite (SIM) starts to
form and it spreads in the specimen under increasing load (point a-c). Unlike thermally-
induced martensite in which the variants are equally distributed in a self-accommodation
morphology, stress-induced martensite consists mainly of the most favored variants with
respect to the applied stress. Unloading the specimen ﬁrst results in elastic unloading of
martensite phase (point c—d). The martensite becomes unstable as the stress drops and the
specimen starts to transform back to austenite through reverse movement of
martensite/austenite interface (point d-e). Eventually, the specimen returns to zero strain

due to the elastic recovery of the parent phase (point e-o).

 

7 The inserted microstructure drawings only roughly describe the austenite-to-martensite structure change.
See Figure 1-1 or Figure 1- 17 for more precise representation.

14

1.1.5. The Clausius-Clapeyron Equation for Uniaxial Stress

Analogous to classical Clausius-Clapeyron equation describing the effect of a
hydrostatic pressure on transformation temperature, a relation describing the effect of a
uniaxial load on the temperature of a stress-induced thermoelastic martensitic
transformation has been developed by Wollants et al. [1979, 1993]. Next, we brieﬂy
review the transformation thermodynamics involving the external loading as a state
variable.

Assuming that the tensile specimen which undergoes phase transformation contains
one mole of material, the specimen has length L and is subject to a uniaxial load F. Other
symbols have their usual meanings in the thermodynamics. Since no chemical composition
change occurs during martensite transformation the system may be treated a single-
component system.

From the ﬁrst law of thermodynamics

du=5Q—6W (1.3)

where 5Q and SW represent inﬁnitesimal amount of heat and work respectively. They are
not state variables since they are dependent on the particular path chosen. However, the
difference between them is a state variable. If only reversible work PdV and FdL are

considered, then

dU=5Q-PdV+FdL (1.4)

We apply the second law of thermodynamics,

_.§Q_
as- T (1.5)

which produces the equation combining the ﬁrst and second laws,

15

dU = TdS — PdV+ FdL (1.6)

On the other hand, the generalized Gibbs free energy function,8 which involves F L-

work, is

Ggen=U+PV—TS—FL (1.7)

In order to shift the independent variables from the set (S, V, L) in Equation 1.6 to the set
(T, P, F), we apply a Legendre transformation, i.e., differentiate Equation 1.7 and using

Equation 1.6, we obtain,

nge" = — SdT+ VdP — LdF (1.8)

In the state (T, P, F), if the parent phase and the martensite reach the equilibrium condition,

Ggen, P = Ggen. M (19)

After an inﬁnitesimal change in temperature and loading, in the state (T + dT, P, F + (IF),

the system again attains the equilibrium condition,

Ggen, P + ngen, P = Ggen. M + ngen. M (1 .10)

From Equation 1.9 and 1.10, we have,

ngen- P = dGsen- M (1.11)

Substituting Equation 1.8 for the parent phase and the martensite respectively, at an

isobaric environment (i.e., at atmospheric pressure) (IF = O, we ﬁnd that,

—SPdT—LPdF=—SMdT—LMdF (1.12)

 

8 The generalized variables which are involving FL-work are superscripted with ‘gen’ in the present study.

16

After rearranging the last equation, the Clausius-Clapeyron equation for a uniaxial loading

can be obtained:

where ASP-W = S“ — SP and ALP—W = LM — LP. Substituting F/A = 0' and ALP-’M/L =
5"” into Equation 1.13 in favor of engineering applications, noting that AL 2 V, the

following equation is obtained,

do
217=_——_V£P_)M (1.14)

The term of ASP-9M needs to be replaced by a measurable value. From the deﬁnition

of the generalized enthalpy function
ngn=U+PV—FL (1.15)

and Equation 1.7 for isothermal martensitic transformations, the following equation can be

obtained

AGgen, P—>M = Angn. P—tM _ TASPHM (1 .16)

At the equilibrium condition AGgen- P-W = GS“. M — GS“. P = 0, the Equation 1.16
becomes,
Angn,P—)M

ASP-’1": (1.17)
T

 

Substituting Equation 1.17 into Equation 1.14, we obtain the Clausius-Clapeyron equation

having state variables of Tand 0',

17

do. Angn,P—)M
d_T _ _ TVePTM

 

(1.18)

The inﬂuence of stress con transformation temperature T is described by the preceding

equation in thermodynamical equilibrium sense. In Chapter 3, this equation is employed to

predict transformation strain using measured experimental data.

1.2. Transformations in N iTi Alloys
1.2.1. Martensitic and R-Phase Transformations

The crystal structures of the parent and the martensitic phase in NiTi alloys have been
studied by many investigators using X-ray and electron diffraction methods. It has been
clear that the parent phase of near—equiatomic NiTi alloys has a CsCl (B2) structure with a
lattice constant a0 = 0.3015 nm [Philip and Beck, 1957]. For the crystal structure of
martensite in NiTi, different models have been proposed. In 1971, Otsuka, Sawamura and
Shinrizu (088) [1971] and Hehemann and Sandrock (HS) [1971] arrived at virtually the
same lattice constants for the martensite. They all agreed that the martensite phase has a
distorted monoclinic Aqu (B 19) structure (denoted as B 19'). The monoclinic crystal of
N i49,75Ti50.25 with a = 0.2889 nm, b = 0.4120 nm, c = 0.4622 nm, and ﬂ = 96.80° reported
by 088 [1971] has been widely accepted as the standard unit cell. However, controversy
lies in the mode of atomic shufﬂe occurring during the phase transformation [I-IS, 1971;
088, 1971].

The conversion from the parent lattice to the martensite lattice can be illustrated in
Figure 1-5 .9 The ‘steps’ in the ﬁgure do not mean actual sequence of the lattice transition.
A Bain distortion, which consists of an orthorhombic distortion (Figure 1-5b to c) and a
monoclinic shear (Figure l-Sc to d), converts original tetragonal unit cell in B2 lattice

(Figure 1-5b) into a monoclinic cell (Figure 1-5d). Due to the different sizes between Ni

 

9 In HS’s work the monoclinic angle is 7, while in OSS’s work the monoclinic angle is ,3. In the present
study the second description is used for the comparison of two shuffle models.

18

and Ti atoms, an atomic shufﬂe is necessary for atoms to reach their ﬁnal stable positions
in the monoclinic lattice. The shufﬂe proposed by HS is (010)M[00I]M type with
magnitude of 313-[00 I]M, whereas that proposed by 088 is (001)M[010]M type (Figure 1-
5e).

There are total 12 lattice correspondences (also called correspondence variants of the
martensite) between the parent cell and the martensite as summarized in Table 1-1
[Matsumoto et al., 1987]. Figure 1-5 shows the variant 6. In the Table 1-1, the numbered
correspondences with and without prime symbols (e. g. 6 and 6') represent correspondence
variants having an opposite monoclinic shear along cM axis in the (bc)M plane.

Based on the unit cell parameters of the parent and the martensitic phase and the
phenomenological crystallographic theory, twinning modes in the martensite of NiTi have
been predicted by calculation by several groups. The calculated results were then compared
with experimental data. The deformation twinning has been described by three types: Type
1, Type II and compound twin as graphically illustrated in Figure 1-6 [Reed-Hill, 1964].
Type I twinning has a rational K1 plane and 112 direction; Type H twinning has the rational
K2 plane and 111 direction, and the compound twinning has all rational K1, K2, 1]] and n2.
With the aid of limited TEM and electron diffraction results, Knowles and Smith (1981)
ﬁrst proposed (011) Type II twinning. By using the X-ray diffraction data obtained from
NiTi single crystals Matsumoto et al. [1987] conﬁrmed Knowles and Smith’s prediction.
A { 11 I} Type I twinning mold was also frequently observed in TEM and it was thought to
be caused by ‘thin foil effect’lo [Matsumoto et al., 1987; Chang, 1993].

Using the (011) Type II twinning as a lattice invariant shear, the 24 habit-plane
variants (or plate variants) shown in Table 1-2 can be calculated [Miyazaki et al., 1978].
The habit plane variant 1(+) is the combination of lattice correspondences 1-2, and so on.

The ﬁrst and second numbers of the correspondence variant-combinations designate the

 

‘0 The thin foil mentioned here is the area near perforated area of TEM disks. The thickness in such areas is
usually under 0.5 11m. However, the thickness of thin ﬁlms for NiTi research is typically 2-10 pm.

l9

variant of the ‘major’ and ‘nrinor’ regions respectively. The calculated habit planes in
Table 1-2 ﬁt well with experimentally obtained data for solution-treated NiTi specimens.
Miyazaki, Otsuka and Wayman [1989] observed a typical triangular self—accommodating
morphology consisting of three variants in age-treated NiTi specimens as shown in Figure
1-7. They found that the magnitudes of the habit plane indices of 0.78, 0.39 and 0.48 in
age-treated specimens (Figure 1-7) correspond to that of 0.89, 0.22 and 0.40 in solution-
treated specimens (Table 1-2). Three junction planes between 1(-), 1‘(+) and 2'(-)
domains in Figure 1-7 were determined by trace analysis to be one (011) Type I twinning
plane and two (011) Type H twinning planes. They calculated the average shape-strain
matrix for the triangular morphology and found that the shear components of the matrix
approached zero. Thus it was justiﬁed that the triangular morphology satisﬁed the self-
accommodating conditions. Miyazaki, Otsuka and Wayman [1989] also investigated the
change in morphology during loading. Upon stressing the triangular morphology was
changed into the most favorable variant with respect to the principal shears through the
motion of the interfaces between the variants.

Besides the parent phase and the martensite, an intermediate rhombohedral phase (R-
phase) may appear upon cooling, prior to the martensitic transformation in NiTi systems.
Hwang et al. [1983] studied the R-phase transformation in TisoNi47Fe3 alloy and proposed
a transformation sequence upon cooling as follows: parent phase (B2) —> incommensurate
phase11 (I) -r commensurate phase (R-phase) —> martensite (M). Figure 1-8b [Hwang et
al., 1983] schematically shows an electrical resistivity vs. temperature curve in which the
R-phase transformation and the martensitic transformation are well separated. Figure l-8a
[Ling and Kaplow, 1981] and Figure 1-8c show the rhombohedral angle, 0:, vs.

temperature curve and differential scanning calorimetry (DSC) curves.

 

1‘ According to the study conducted by Hwang et al. [1983], on cooling from room temperature the initial
l/3(110) and 1/3(111) superlattice reﬂections deviated slightly from the exact 1/3 positions relative to the
parent B2 structure. They described such phase with non-integral fraction of a reciprocal lattice vector of the
parent phase as incommensurate phase.

20

Upon cooling, the incommensurate phase starts to form at TR' where the resistivity
abruptly increases as shown in Figure 1-8b and 1/3-(110)-type superlattice reﬂections
appear in select area diffraction pattern (SADP). In fact, the superlattice reﬂections are not
located in the exact 1/3 positions. There is no heat release associated with the B2-to-
incommensurate transformation due to its second-order nature. The onset temperature of
incommensurate-to—commensurate (R-phase) transformation is termed TR, which is a few
degrees below TR' and is close to the inﬂection point. At TR, the cubic lattice begins to
distort into the rhombohedral lattice (Figure 1-8a) and the superlattice reﬂections lock into
precise 1/3 positions and a ﬁnite latent heat starts to evolve (Figure 1-8c). The
rhombohedral angle continuously decreases upon further cooling until the martensite starts
to form at the M, temperature where the resistivity sharply drops.

Upon heating, the resistivity slowly increases up to the inﬂection point A,. At Af the
heating curve intersects the cooling curve and then shows a small hysteresis for the R-
phase H B2 transformation. As detailed in Section 3.3.1, the A, and A, in this case do not
represent the onset and the completion temperature of M-to—B2 transformation, but rather
that of M-to-R transformation. The DSC curve peak, with a shoulder indicating a
convolution of the M —> R and R —> B2 phase transformation enthalpies, is seen in Figure
1-8c. In contrast with the martensitic transformation in which the thermal hysteresis
exceeds 10 K the hysteresis of R-phase transformation is only 1.5 K.

Miyazaki and Wayman [1988] investigated the shape-memory mechanism associated
with the R—phase transition using age-treated Ni50_5Ti49.5 alloy. They observed four distinct
vaiants of the R-phase, designated as A, B, C and D in correspondence with the elongated
axes ['1'1 1],, [i T1132, [111]B2 and [i1 i132, respectively. The surface traces of
specimen indicated {011 }32 and {001 }32 plane-type compound twins. Figure 1-9 shows
self-accommodating morphology and a corresponding four-variant combination of

rhombohedral phase proposed by Miyazaki and Wayman [1988].

21

1.2.2. Diffusional Transformations

At high temperatures, sufﬁcient atomic mobility leads to diffusional transformation
and precipitate particles which are formed in NiTi alloys with composition deviating from
equiatomic stoichiometry. Several N i-Ti phase diagrams have been plotted which display
an eutectoid decomposition of NiTi phase into NiTi2 + Ni3Ti at lower temperatures (903
K), as seen in Figure 1-10a [Massalski, 1987; Taylor and Floyd, 1952]. A steep solvus
boundary on the Ti-rich side indicates a very limited solubility for titanium in the
compound. However, Wasilewski et al. [1971] reported that the NiTi phase may remain
down to the room temperature with a very narrow composition range as shown in Figure 1-
10b. In Ni-rich side, besides Ni3Ti phase, metastable phases Ni3Ti2 and Ningi42 (i.e.,
NI4TI3 phase”) also exist. These metastable phases were later conﬁrmed by Van Loo,
Bastin and Leenen [1978], Nishida, Wayman and Honma [1986] and Kim, Moine and
Stevenson [1986]. It is generally agreed that the nricrostructures of near—equiatomic NiTi
alloys are primarily B2 NiTi with small amounts of other phases distributed in the matrix.
The formation of these phases tends to move the composition of supersaturated matrix
toward stoichiometry and also create strain ﬁelds associated with volume changes and
interface coherency effects, and thus profoundly inﬂuences phase transformation
temperatures and deformation properties.

Nishida, Wayman and Honma [1986] studied the precipitation behavior of NisoTiso
and Ni52Ti43 bulk alloys, and constructed an isothermal time-temperature-transformation
(TIT) diagram for Ni52Ti4g alloy as illustrated in Figure 1-11. In furnace-cooled NisoTiso
alloy, no secondary phase was found except for some NizTi4Ox particles.13 However, in

the NiszTIzrg alloy, the precipitation took place and its sequence could be written as:

 

‘2 The precipitation Ni,Ti, has been denoted as Ningi“ [Nishida and Honma, 1984] or NiMTi” [Nishida,
Wayman and Honma, 1984] according to the chemical composition analysis of energy dispersive X-ray
(EDX). Later, the notation Ni4Ti3 prevailed in the literature based on its crystal structure [Sabrui, Nenno
and Fukuda, 1986].

13 Nevitt (1960) concluded that oxygen is soluble interstitially in NiTi2 up to 15 at.%. NizTi,O, phase is
isomorphous with NiTiz. It has an fee structure with lattice constant increasing slightly from 1.132 nm
(NiTiz) to 1.134 nm (N izTi40). Chang (1993) showed that most of grain boundary precipitates in sputtered

 

22

T < 953 K: NiTi (supersaturated) —> NiTi (matrix) + Ni4Ti3
—-> NiTi (matrix) + Ni3Ti2 —-> NiTi (matrix) + Ni3Ti (1.19)

953 < T < 1023 K: NiTi (supersaturated) —) NiTi (matrix) + Ni3Ti2
—> NiTi (matrix) + Ni3Ti (1.20)

T > 1023 K: NiTi (supersaturated) —> NiTi (matrix) + NIgTI (1.21)

Interrnetallic precipitates play an important role in near-equiatomic NiTi alloys when
N i-content is greater than about 50.3 %.14 It is well known that ﬁne precipitates can
severely depress martensitic transformation temperatures. On the other hand, the
precipitation process, i.e., aging after the solution treatment, may be utilized to improve the
deformation properties of NiTi alloys. Metastable Ni4Ti3 is the most important one because
it readily forms at an aging temperature as low as 623 K, and has a strong effect on
martensitic transformation and deformation behaviors. The Ni4T13 phase has a lenticular
morphology and a rhombohedral cell structure with a0 = 0.661 nm and a: 113.65°. The
orientation relationship between the B2 matrix and the Ni4Ti3 phase has been determined to
be (100mm, // (142),2 and [001 1,4,, // [012],2 and the precipitation habit plane is
{ l 11 }132 [Nishida, Wayman, Kainuma and Honma, 1986]. In the early stages of
precipitation, the Ni4Ti3 phase is coherent with the matrix and generates internal stress

ﬁelds due to the mismatch of d-spacing between the Ni4Ti3 phase and BZ matrix.

 

Ni,,_oTi,,_,Cu,_6 ﬁlm are NizTi4O, due to oxygen diffusion along the grain boundaries during post-deposition
annealing.

14 For bulk NiTi alloys, the chemical composition is generally determined by the weight percentage of
nickel and titanium elements to be melted together for ingot. The chemical analysis is also frequently
employed (e.g., in the works of Nishida, Wayman and Honma, 1986; Miyazaki, Igo and Otsuka, 1986).
For NiTi ﬁlms, energy dispersive X-ray (EDX) is mostly used for composition measurement, although
ﬂame atomic absorption was also used (Busch, et al., 1990). Determining accurate compositions is
difﬁcult especially in ﬁlms. For this reason, M, temperature may be used as a reference of composition
because it is very sensitive to Ni-contant in Ni-rich side. However, only the M, of solution-treated NiTi
may give a reasonable estimation since the M, is also strongly inﬂuenced by structure defects (Section
1.2.3).

23

The NI3T12 phase usually appears at higher aging temperatures (> 1000 K) after a
short period of annealing (1 hour). It has a rectangular plate-like shape. The unit cell was
deduced to be monoclinic with a = 0.414 nm, b = 0.828 nm, c = 1.352 nm, and 7: 893°.
The orientation relationship between the Ni3Ti2 phase and the B2 matrix is as follows:
(100)Ni3'1‘12 // {loolazr IOOIINisriz // (001)132 01’ (001)Ni31i2 // {1001132, I2101Ni3Ti2 // (001)132
[Nishida and Wayman, 1986].

1.2.3. Effects on Phase Transformations

Martensitic transformation in NiTi alloys can be either thermally-induced (decreasing
the temperature) or stress induced (applying an external load). In this section, the effects
on thermally-induced martensitic transformation are discussed, and the effects on stress-
induced martensitic transformation will be addressed in the Section 1.4.3. Several factors
inﬂuencing the temperature M, are listed as follows:

(1) Composition: Since solution-treated NiTi alloys have no special internal
structures such as precipitates and dislocations, in such systems the M, can sometimes
directly reﬂect the effect of chemical composition on the phase transformation. It has been
veriﬁed that the M, strongly depends upon composition on Ni-rich side in quenched
specimens after solution treatment. It decreases with the increase of Ni-content at a rate
over 100 Klat.%, primarily due to the fact that the extra nickel atoms in BZ lattice change
relative chemical energies of the parent phase and the martensite. Figure 1-12 illustrates
electrical resistivity vs. temperature curves for specimens having various nickel
concentrations quenched into ice water from 1073 K [Nishida and Honma, 1984]. In the
case of Ti-51.8 at.% Ni specimen, the M, was depressed beyond the range of
measurement.

(2) Precipitates: The internal strain ﬁelds associated with Ni4Ti3 precipitates created
during the aging process have a strong tendency to depress the M, temperature, since the

distorted parent lattice imposes a constraint force on the growth of martensite variants (The

 

24

strain ﬁelds associated with Ni4Ti3 will be further discussed in Section 3.1.1). Wu, Lin
and Chou [1989] studied variation of M, with aging time in NiSITi49 alloy as shown in
Figure 1-13. Curve (a), which is similar to the curve ((1) in Figure 1-12, is a typical
resistivity feature of solution-treated specimens where only the martensitic transformation
takes place. The M, was low due to a high Ni-content in NiTi solution. In curve (b), after
one hour of aging the R-phase transition appears and the M, is dramatically depressed (over
50 K) below the measurement range although the formation of Ni4Ti3 may reduce some N i-
content in the matrix. In curves (b-g) the M, increases with increasing aging time. This is
caused by the combination of stress ﬁeld relaxation due to the growth of precipitates and
the depletion of Ni concentration in the matrix.

(3) Dislocations: Similar to the effect of strain ﬁelds created by the precipitation, the
strain ﬁelds associated with dislocations which are introduced by either cold-work [Lin et
al., 1991], or thermal cycling, impede the development of martensite and thus depress the
M, temperature. Miyazaki, Igo and Otsuka [1986] observed that the M, decreased about 25
K after 100 thermal cycles in all solution-treated specimens irrespective of Ni-content.
TEM images revealed that the density of dislocations in the specimens increased with
increasing number of thermal cycles. Jean and Duh [1995] conﬁrmed above result in both
solution-treated Ni50,5Ti49,5 and Ni47T150Cu3 alloy.

(4) Grain size: Motohashi et al. [1991] developed a process involving sequential
stages of intermediate annealing and cold rolling followed by recrystallization. They found
that the M, temperature of Ni50,7Ti49_3 alloy increased from 200 K to 241 K when grain size
decreased from 21.7 pm to 4.0 um. Gil, Manero and Planell [1995] reported sirrrilar
results in Ni-42Ti wt.% (or Ni53Ti47) alloy. The M, monotonically rose from 260 K to 270
K when the perimeter of grains, P, decreased 14 um.15 The increment in M, was attributed

to grain boundaries that acted as nucleation sites for martensite.16

 

15 Gill et al. [1995] did not mention the initial grain perimeter P0 in their paper.
16 However, in Cu-l4Al-2.5Ni (w/o) alloy as discussed in Section 1.1.3, M, decreases as grain size
decreases due to increasing elastic constraint of grain boundaries [Salzbrenner and Cohen, 1979].

25

(5) Impurities: Metallurgical process contamination with impurities decreases M,
considerably. Oxygen is usually introduced from sponge titanium used as a raw material.
Almost all of the oxygen appears as precipitates of Ti4N 120. The oxide phase depletes the
Ti content in the matrix, and relatively increases the Ni concentration, thus decreases the M,
[Shugo, Hanada and Honma, 1985]. Other impurities, such as nitrogen from the
atmosphere and carbon from graphite crucibles also depress the M, but to a lesser degree,
because of the preferential consumption of titanium in formation of TiC and Ti4Ni2N
precipitates [Gokin, 1984].

(6) Ternary elements: As mentioned above, any structural defects (precipitates,
dislocations) and impurities tend to lower the M,. The highest M, that has been achieved in
Ni-Ti binary alloys by carefully controlled processing is around 353 K [Melton, 1990].
However, the addition of certain amount (> 5 %) of speciﬁcally selected ternary elements
such as hafnium which substitutes for Ti [Johnson, Martynov and Minners, 1995], and
palladium [Quandt et al., 1995; Miyazaki et al., 1995] which substitutes for Ni, can raise
the M, temperature to well above 373 K.

The R-phase and martensitic transformations are two competing transformations on
cooling from the parent phase. Whether or not the R-phase appears prior to the martensitic
transformation depends on the relative magnitudes of M, and TR. Any factors which can
depress M, and/or can raise TR may encourage R-phase transformation. Factors that lower
the M, with respect to the TR are:

(1) Introducing ﬁnely dispersed Ni4T13 precipitates by aging Ni-rich alloys after
solution treatment [Wu, Lin and Chou 1989].

(2) Introducing rearranged dislocations by annealing specimens below the
recrystallization temperature (about 773 K) after cold work [Miyazaki et al., 1982].

(3) Adding (< 5 at.%) a third element such as iron, aluminum or copper to the binary

N i-Ti alloy [Edmonds and Hwang, 1986].

26

1.3. Deformation of Metallic Crystals
1.3.1. Role of Grain Boundaries in Plastic Deformation

Effects of grain boundaries on yielding behavior have been widely studied in
bicrystals since they are the simplest form of polycrystals. One conﬁguration of bicrystal
has two grains A and B bound by a boundary being parallel to the tensile axis (equistrain
bicrystal). For each point six components of strain must be speciﬁed for arbitrary
deformation of a grain: three tensile (8,, e, and 6,) and three shear my, 73,, and 71,) ['I‘aylor,
1938]. But only ﬁve of these are independent because the dilational strains are related
through the constant-volume condition (8, + ey + e, = 0). To avoid the development of
voids between, or overlaps of the grains, the following plastic compatibility conditions
must be satisﬁed at the grain boundary region in order to provide continuity of strains

across it [Livingston and Chalmers, 1957; Hook and Hirth, 1967]:

cf = sf (1.22)
8? = 8? (1.23)
7?, = 7,1,9, (1.24)

Two types of bicrystals are interesting: symmetric and non-symmetric ones. In the
former, two crystals (A and B) are symmetric with respect to the boundary, while in the
latter the orientations of the two crystals are independent to each other. For symmetric
bicrystals, the operation of the slip systems in the two crystals is compatible. The grain
boundary does not have any effect on deformation, i.e., the bicrystal behaves like a single
crystal, as has been reported in zinc bicrystals by Gilman [1953]. However, for non-
symmetric bicrystals the deformation cannot be accomplished by the operation of one
system per grain. The activation of multiple-slip systems have been observed along the

grain boundary of Fe—3Si (at.%) bicrystal [Hirth, 1972].

27

The polycrystal is made up of identical crystals differing only in their orientations.
Optical microscopy examination of previously polished surface of deformed polycrystals
showed that for small grains the deformation is essentially by a multiple slip even at small
strains of 1-5 % [Honeycombe, 1968]. For very large grains, the center region of a grain
may be deform on a single slip system, but this will soon change with increasing
deformation. The displacements must be continuous across grain boundaries and as a
result, the grains deform in a cooperative manner. In general, the ﬂow stresses of
polycrystals are higher than those of single crystals of the same materials, since the
deformation in a grain aggregate is essentially through multiple slip from the start of the
deformation. Similar to the free-surfaces which serve as dislocation sources for slip in
single crystals, the grain boundaries are major sources of dislocations for plastic
deformation in polycrystals [Murr, 1975].

A well-known description of the grain-size dependence of yield stress is the Hall-

Petch relationship [Hal], 1951; Petch, 1953] which has the form
0', = 0'0 + kD‘" (1.25)

where 0'y is the yield stress, 00 is a frictional stress required to move dislocations, and D is
the average grain diameter. The index n, normally 0.5 in bcc metals, but is not so well
deﬁned in other systems [Honeycombe, 1968]. According to Petch’s explanation [Petch,
1953], plastic deformation commences ﬁrst in a single grain. When an applied stress
reaches yielding point, a Frank-Read dislocation source becomes unpinned and sends out
dislocation loops. The dislocation loops run up to the boundary, and form piled-up arrays
of dislocations. The pile—up produces a stress concentration in the adjacent grains. If this
stress concentration is sufﬁciently large, dislocations in a neighboring grain will become
activated and emit dislocation loops. In this way the yield spreads, grain by grain, through

the specimen. As indicated in Equation 1.25, the yield point is inversely proportional to the

28

square root of grain size. This is due to the fact that a polycrystal with smaller grains
allows less number of dislocations to pile-up, and imposes less stress concentration.

The Hall-Petch relationship has been applied to many systems with varying degrees
of success. Care must be taken when using this equation, such as:

(1) The specimens should be true polycrystals. The numbers of dislocation pile-up
tend to be greatly reduced in the grains residing in free surface, since the dislocations are
drawn toward the surface and leave the lattice by image-force. Kocks [1976] calculated
that wire specimen which has 100 grains per cross section has about 1/3 of its grains at the
free surface. This is a signiﬁcant value and could have an important effect on the
mechanical properties. A criterion to assume polycrystallinity suggested by him would be
to have a fraction of surface grains less than 10 %. In this case the diameter of cylindrical
specimen with equiaxed grains has to be at least 30-40 times the grain diameter.

(2) Specimens having different grain size should have the same level of purity and
texture, since impurities may segregate at boundaries and change the ﬂow stress.

In conventional metallurgy, the dimensions of a specimen are generally much greater
than the diameter of the grains. Consequently, when such a specimen is stretched, most
grains are constrained by surrounding grains except for a very small fraction of the grains
that reside near the free surface of the gauge section. However, when the dimensions of
specimens become comparable to the diameter of the grains, such as in the case of thin
ﬁlms, a considerable portion of constrained grain boundaries are replaced by free surfaces.
In these circumstances, slip tends to occur more easily due to reduced constraint force.
This point has been conﬁrmed by Miyazaki, Shibata and Fujita [1979]. They investigated
the ﬂow stress of polycrystalline Al, Cu, Cu-13 A1 (at.%) and Fe as a function of the grain
size and specimen thickness, and found that the yield stress decreased as the specimen
thickness decreased when specimen thickness to grain size ratio, t/d, is below 5-15 (see
Figure 1-14).

29

1.3.2. Precipitation Hardening

Precipitation from a supersaturated solid solution is a very versatile and common
strengthening technique. The basic requirement of precipitation system is that the solubility
of solid solutions should decrease as temperature decreases, in other words, the phase
diagram must show a declining solvus line. Fine scale precipitates are generally produced
by following three steps:

(1) Solutionizing any soluble particles by heating the alloy to the monophase region
and maintaining it for a sufﬁciently long time.

(2) Quenching the alloy so that the formation of coarse precipitates is avoided, and a
supersaturated solid solution is formed.

(3) Aging the alloy at moderate temperature to precipitate ﬁnely dispersed particles.

The crystallographic relationships between the lattice of the precipitate and that of the
matrix can be coherent, semi-coherent or incoheremt, as schematically shown in Figure 1-
15 [Hirsch et al., 1977]. A coherent precipitate has a one-to-one correspondence between
the precipitate and matrix lattices. The exact spacing of lattices may or may not be the
same, and the latter case introduces some elastic strains in both phases (Figure 1-15a). A
semi-coherent precipitate has typically one set of coherent interfaces and one set of semi-
coherent interfaces (Figure 1-15b). The misﬁts on semi-coherent interface are taken up by
an array of boundary dislocations. An incoherent precipitate has a completely different
crystal structure from the matrix and there is no speciﬁc orientation relationship between
two lattices (Figure 1-15c).

Precipitate hardening is governed by interactions between moving dislocations and
precipitate particles. For coherent and semi-coherent precipitates, when precipitate particles
lie across slip planes along which dislocations pass, dislocations must behave in one of two
ways: (a) take a path around the particles, and (b) cut through the particles. Accordingly,
there are two prevailing mechanisms of hardening: internal strain hardening and chemical

hardening as described below.

30

(1) Internal strain hardening: Internal stress ﬁelds arise from a slight misﬁt with
respect to the matrix. In lightly aged alloys, the average distance between particles, i.e.,
the ‘wavelength’ of the internal stress ﬁeld, is small. When a dislocation moves on a slip
plane containing precipitate particles, the closely spaced stress ﬁelds make the dislocation
unable to bend between the particles since a dislocation line always tends to reduce elastic
energy by shortening its length. The dislocation has to move out of its slip plane by
climbing or cross-slipping (Figure 1-16a). This is the peak-aged condition which provides
the highest yield stress. In over-aged alloys, the particles are much wider apart. It is easier
for dislocations to bow around each precipitate, leaves dislocation loops behind (Figure 1-
16b) [Orowan, 1948]. Orowan [1948] showed that the initial ﬂow stress, to, varies

inversely with the interparticle spacing, A, as shown below

2T
1' =1 +— 1.26
0 s DA ( )

where 1', is the critical resolved shear stress of the matrix, b is the magnitude of Burgers
vector and Tis the line tension of the dislocation.

(2) Chemical hardening: Figure l-16c schematically shows sheared Ni3Al particles
formed by passing dislocations in a Ni-l9Cr-6Al at.% alloy after 2 % deformation [Gleiter
and Hombogen]. For this process, sufﬁcient energy must be supplied to break favorable
bonds within the particles and increase the area of newly formed antiphase boundary.
Whether dislocations cut through the precipitate particles or take a path around them
depends on the applied stress and the nature of the precipitate. Meyers and Chawla [1984]
pointed out that the precipitates tend to be cut when the speciﬁc interface energy of the

particle and matrix is low and the particle size is very ﬁne.

31

1.4. Deformation of NiTi Alloys
1.4.1. Recoverable Tensile Strains

Recoverable tensile strains in single crystals of NiTi depends on the crystal
orientation with respect to the tensile axis, transformation shape strain, and the detwinning
strain. In a NiTi polycrystal, the deformation of each grain is constrained by neighboring
grains. The strain ﬁeld may not be homogenous due to the different orientations of the
grains. Its recoverable strain is the spatial average of those inhomogenous strains that may
be recovered in each grain by reverse transformation. In this section, we calculate the
tensile strains solely caused by, respectively, the B2-to-twin-related B 19' martensitic
transformation, and the detwinning of B 19' variants in NiTi single crystals. The sum of
these two strains in the most favorable direction is shown to be consistent with the
maximum recoverable tensile strains reported in literature. The recoverable strain of NiTi
polycrystals is also discussed.

Figure 1-17 schematically illustrates an idealized deformation route from a single
crystal of B2 NiTi to a single crystal of the B 19' phase where the transformation twinning
and the detwinning in twin-related martensite are well separated. From Figure 1-17a to 1-
17b, the martensitic transformation proceeds through the movement of parent/martensite
interface. The orientation of the interface is controlled by the requirement of an undistorted
interface. The strain associated with this process is referred to as transformation shape
strain. From Figure 1-l7b to 1-17c, the deformation is realized by the coalescence of
favorably oriented twin variants with respect to applied stress. Macroscopically,
martensitic transformations are accompanied by a shearing deformation along the habit
plane (Figure 1-17a). But it is not a pure shear since there is a component perpendicular to
the habit plane. The term of invariant plane strain is used to describe shape changes where
planes and straight lines are preserved during the transformation.

In Figure 1-17a, d, is the unit vector in the direction of the shape change, p1 is the

unit vector normal to the invariant plane (habit plane) and m, is the magnitude of

32

deformation. The deformation mldl (corresponding to m in Figure 1-1, Section 1.1.2)

may be decomposed into two components:
mld1= mIPdIP +m1"p1" (1.27)

Where mIPd ,P and mlnpl" represent the shear and dilatational components respectively.
m1" is simply the volume change of transformation since a shear component does not
attribute a volume change. From the lattice parameters of B2 and B 19' and X-ray
diffraction data, Matsumoto et al. [1987] found consistency between predictions of the
phenomenological theory and experimental results for { 5771 1 1}(01 I) Type-II twinning,
and calculated the crystallographic data for NI493TI502 alloy as follows:

Habit plane (p1): (-0.8889 0.4044 0.2152)”
Direction of shape strain ((11): [0.4345 0.4874 0.7574]32
Magnitude of shape strain (m1): 0.1308

From above data”, the angle between the normal of the habit plane and the direction of
shape strain is found to be 91 .49°, the magnitudes of shear strain m1P = 0.1307, and
normal strain m1" = 0.0034.‘8

In order to calculate the tensile strain associated with the transformation, Otsuka et al.
[1976; Otsuka and Shinrizu, 1986] modiﬁed Schnrid and Boas’s formula for slip and
mechanical twinning [Schmid and Boas, 1950] by adding a dilatational contribution written

as follows:

 

2
8 = J1+2mfsinxocoslo +(m1p sinxo) —1+m1"sinxo (1.28)

 

17 The correspondence variant 2' in Table 1-1 was chosen for calculating the data [Matsumoto et al., 1987].
The habit plane variant in this case is 2'(+) (see Table 1-2).

18 Since pure shear strain does not have any contribution to the volume change of austenite-to-martensite
phase transformation, the normal strain (= 0.0034) here is corresponding to the volume change (= 0.0041)
of the Bain distortion represented in Figure 1-5, Section 1.2.1. The slight difference between these two
values was probably caused by experimental errors.

33

where )10 is the angle between the tensile axis and the shear direction, 10 is the angle

between the tensile axis and the habit plane. We can simplify this expression by conditions
(mIPsinxo)2 <( I, and ml’lsinxo z 0

The Equation 1.28 becomes:

 

e=J1+2mfsinx0cosA,-1 (1.29)

Using the power series expression

Vl+x=l+§x~ﬁxz+ﬁx3mn (1.30)

for the right hand side of Equation 1.29 and truncating the higher power terms, since

x = 2m1Psin10cosiio S 2m1P(0.5) = mll’ = 0.13

is fairly small. We obtain

8 z mll’sinxocoslo (1.31)

Therefore, the transformation strain is approximately proportional to the product of the
shear strain and the Schmid factor. When the Schmid factor is 0.5, i.e., in the most
favorable orientation, the tensile strain solely caused by transformation shape strain is 8 =
0.5m1P = 0.0654.

The detwinning strain can be easily calculated from the crystallographic data as
shown in Figure 1-l7b. Since two twin regions have different volume fractions, x = 0.27
and (l-x) = 0.73, two opposite detwinning directions would lead to different detwinning
strains. Using {0.7721 1 1}(01 I) Type-II twinning as a lattice invariant shear, the

twinning elements for variant 2’ are [Matsumoto et al., 1987]:

34

Kt=<67211uu nt=tmiiu

K2=(011)M. n2=[1_.5_7—1 111..
gives the twinning strain s = 0.280. The detwinning strains (= average twinning strains of
two twin regions) are S, = [0, 0.145, -0.145] in the favored direction and s2 = [0, -0.053,
0.053] in the other direction. The magnitudes of s, and S, are s1 = 0.204 and s2 = 0.075
respectively. If the Schmid factor is 0.5, the magnitude of the tensile strains in the favored
direction and in the other direction, respectively, would be 0.1022 and 0.0375.

The total recoverable strain, which is the strain associated with the transition of

detwinned B 19' single crystal-to-B2 single crystal, is the sum of the transformation shape

strain and detwinning strain (Figure 1-l7c). The recoverable shear strain would be
y=mld1+s1 (1.32)

The magnitude of y is calculated to be 0.221 for the favored twinning direction and 0.163
for the opposite direction. Accordingly, by assuming the Schmid factor to be 0.5, we
obtain the maximum recoverable single-crystal tensile strains of 0.1105 for the favored
direction and 0.0815 for the Opposite direction.19

The recoverable tensile strains of B2-to-B19' transformation for NiTi single crystals
in different directions have been calculated from the lattice deformation matrix and
measured experimentally. Saburi and Nenno [1981] predicted orientation dependence of
recoverable tensile strain using the lattice deformation matrix as shown in Figure 1-18a.
Saburi, Yoshida and Nenno [1984] and Miyazaki et al. [1984] measured the dependence of
recoverable tensile strain on crystallographic orientation in Ni50_5Ti49.5 single crystals. The
strains of solution-treated specimens for various orientations were in good agreement with

the values calculated from the lattice deformation matrix. The maximum recoverable tensile

 

‘9 In single crystal NiTi, the recoverable strain depends on crystallographic orientation. The maximum
recoverable strain mentioned here is referred as the recoverable main in the most favored direction.

 

35

strain is greater than 0.105 as shown in Figure 1-18a and 1-18b which is consistent with
the result of preceding calculation (y = 0.1105).

From the previous discussion, it is apparent that: (a) The maximum tensile recovery
capacity of single-crystal NiTi is 0.1105. (b) For the NiTi system we cannot predict the
recoverable strain only from the transformation shape strain, since the detwinning strain
also needs to be taken into consideration.20

In analogy to the deformation of polycrystals by slip (Section 1.3.1), the
transformation twinning and the rearranging of variants in a polycrystalline shape-memory
alloy also take complex forms in grain boundary regions in order to maintain grain
boundary compatibility. Bhattacharya and Kohn [1996] proposed a general theoretical
prediction of recoverable strains for polycrystalline shape-memory alloys by utilizing the
Taylor estimate. They found that the recoverable strains of a polycrystal depends not only
on the texture of the polycrystal and the transformation strain of the underlying martensitic
transformation, but critically on the change of symmetry during the underlying phase
transformation. In cubic-to-tetragonal transformations occurring in MA] and FeNiC
alloys, the number of martensite variants, k, is 3. In this case the range of deformations
that each grain can undergo by rearranging variants is too small to allow cooperative
deformation between grains except for special textures. Hence, even though single crystals
of Ni-37Al (at.%) can recover tensile strains up to 13 % [Enanri et al., 1981], polycrystals
can only recover 0.2 % strain in compression [Kim and Wayman, 1992]. In contrast, in
the cubic-to-monoclinic transformation occurring in N iTi alloys, the large change in
symmetry (k = 12) provides a great range of cooperative deformations. Thus signiﬁcant

recoverable strains can be expected in NiTi polycrystals. Bhattacharya and Kohn [1996]

 

2° Otsuka et al. [1976] calculated the orientation dependence of B, —) B', transformation strain in Cu-Al—Ni
alloys by using the Equation 1.27. In their work, the measured elongation always slightly exceeded the
calculated ones. This might due to the fact that the contribution of the detwinning strain was not included
into the consideration.

36

estimated that NiTi alloys have at least 2 % recoverable strain in any polycrystal and at least
2.4 % along the length in ribbons with the Eucken-Hirsch texture.21

The recoverable tensile strain observed experimentally in NiTi polycrystals is in a
range of 5.5-8 % [Ling and Kaplow, 1981; Miyazaki, Otsuka and Suzuki, 1981; Saburi,
Tatsumi and Nenno, 1982]. The discrepancy is probably caused by the different levels of
material texture and/or nricrostructure defects (such as precipitates and dislocations).

NiTi alloys also exhibit the shape-memory effect and superelasticity associated with
R-phase transformation, although the recoverable strain is limited to 1 % due to small
change of rhombohedral angle (see Figure 1-8a). The mechanical behaviors associated
with the R-phase transformation have been investigated in single-crystals [Miyazaki,
Kimura and Otsuka, 1988] and polycrystals [Miyazaki and Otsuka, 1984, 1986]. In a
single-crystal, of course, the recoverable strain associated with the R-phase transition is
dependent on crystallographic orientation. The maximum and minimum values were found
in [111],” and [001],” dirctions respectively as shown in Figure 1-19a [Miyazaki, Kimura
and Otsuka, 1988]. The recoverable strain is also strongly dependent on the temperature,
based on the fact that the rhombohedral angle, a, is a function of temperature at T S. TR
(Figure 1-8a). The temperature dependence of the recoverable strain for various
orientations was calculated using the B2 —> R lattice distortion matrix as shown in Figure 1-
19b [Miyazaki, Kimura and Otsuka, 1988]. The recoverable strain in all directions
increases gradually as temperature decreases from TR. For NisopTi4gg alloy, the
temperature rate for B2 —9 R transformation is greater than that for R —-) M transformation

(15.5 vs. 5.3 MPa/K, Stachowiak and McCormick [1988]).

 

21 Eucken-Hirsch texture is a sheet texture in which all grains align the [001] direction perpendicular to the
plane of the ribbon and either the [100] or the [110] direction along the length of the ribbon [Eucken and
Hirsch, 1990].

37

1.4.2. Slip Deformation

Unlike most polycrystalline interrnetallic compounds which often have limited
ductility at room temperature due to their few and difﬁcult slip systems, B2 N iTi alloys
have signiﬁcant ductility up to 50 % [Moberly et al., 1990]. The permanent deformation
mechanism of B2 NiTi system is outlined as follows.

B2 is an ordered bcc compound with CsCl structure. In bcc metals, slip directions
are in general the closest packed direction, i.e., (111). Perfect dislocations may be
dissociated in order to reduce the energy. There are two possible ways for (111)

dislocations to dissociate in the B2 lattice [Li and Szpunar, 1992]. One way is

(111)—) (110)+(001) (1.33a)
followed by
(110)—->(100)+(010) (1.33b)

Each of these dislocation components is perfect and the motion of a dislocation does not

disorder the structure. The other way is

(111)->-§-(111)+%(111> (1.34)

These two partial dislocations with the same Burgers vectors are elastically repulsive, but in

the ordered structure, separation of partial dislocations will produce a ribbon of antiphase

boundary (APB) between them with an associated APB energy. The width of this ribbon

r0 can be obtained from the balance of the elastic energy of the repulsion between the partial

dislocations and the APB energy.

Li and Szpunar [1992] found that at room temperature the equilibrium separation r0 =

0.045 nm for B2 NiTi, which is smaller than the lattice parameter (a, = 0.301 nm). Such a

small separation implies that the APB energy is so high that the dissociation of two
repulsive %(111) is impossible. The high APB energy is attributed to a perfect order at

38

room temperature since the ordering temperature, T,, of N iTi alloy is 973 K [Wang,
Buehler and Pichart, 1965]. Therefore, at room temperature the (111) dislocations tend to
dissociate in the way represented by Equation 1.33. The (100) slip vector in NiTi has been
experimentally observed by Mori and Fujita [1984].

At high temperature the APB energy decreases greatly due to the loss of lattice
ordering energy, and the second dissociation described by Equation 1.34 may operate.
Koskimaki, Marcinkowki and Sastri [1969] observed that the %(l 11) dislocations were
introduced as interfacial dislocations by a second phase which precipitated at about 900
K .22 Li and Szpunar [1992] concluded that although the slip system in ordered NiTi alloy
is (100){011}, but when the temperature rises to near the order-to-disorder transition
temperature, the slip system shifts to %(111){110}.

Moberly et al. [1990] investigated the plastic deformation of N i47Ti50Fe3 alloy at room
temperature. (010) type dislocations started to appear after 1 % elongation, and
{114} (221) mechanical twinning, which was ﬁrst proposed in NiTi alloy by Goo et al.
[1985], could be identiﬁed in a 10 % cold worked bar. Both dislocation density and twin
density increased with an increase in cold working up to 40 %.

Generally, ﬁve independent slip systems (188) are needed to maintain boundary
strain compatibility requirements [Mises, 1928]. (010) dislocations provide for only three
188 in B2 materials and thus two more 188 must come on other sources [Moberly et al.,
1990]. From TEM observation, they suggested that the fourth 188 was provided by
mechanical twinning which occurred on only one set of parallel planes, and the newly
oriented twins could in turn undergo dislocation slip to provide additional slip systems.

The ductility of martensite NiTi is also greater than 50 %, as reported by Miyazaki,
Otsuka and Suzuki [1981] and Miyazaki, Kohiyama and Otsuka [1991]. However, no
detailed slip system for B 19' NiTi has been proposed. It is found that the critical stress for

 

22 The second phase was not identiﬁed. However, it might be N i,Ti3 phase which precipitated at a similar
temperature [Koskimaki, Marcinkowki and Sastri, 1969].

39

slip in the parent phase is generally greater than that in the martensite [Miyazaki et al.,
1983; Miyazaki, Kohiyama and Otsuka, 1991].

1.4.3. Factors Inﬂuencing SME and Superelasticity

If slips in the martensite and B2 lattice occur during stressing, N iTi will not fully
return to its original shape on heating or/and unloading. To realize a perfect shape-memory
effect and superelasticity, any permanent deformation must generally be avoided.23
Superelasticity occurs at T> Af, according to Clausius-Clapeyron equation the stress level
for superelasticity would be greater than that for shape-memory effect which occurs at T <
M,. Consequently, it is more difﬁcult to achieve the superelasticity which requires a higher
yield point (for slip) of both the martensite and B2 phase.

As will be seen in Section 3.4, in order to enhance the performance of NiTi alloys, a
modiﬁcation of microstructure should have a selected inﬂuence on the ﬂow stresses of the
BZ and martensite, and the critical stress to induce the martensite. In other words, it should
increase the former to a greater degree than the latter. Several rrricrostructural factors
affecting NiTi deformation behavior are summarized below:

(1) Solid solutions: B2 and B 19' lattices are ordered lattices [Gokin, 1984]. In
solution-treated Ni-rich alloys, extra nickel atoms substitute the sites normally occupied by
titanium atoms and introduce elastic distortion around them due to the difference in size
between nickel and titanium atoms. The strain ﬁelds of the distortion restrict the motion of
dislocations thus raise the ﬂow stress. Miyazaki, Kohiyama and Otsuka [1991] observed
that both ﬂow stresses of the parent and martensite phase of solution-treated specimens
(Ni-concentration between 50.0 and 52.0 at.%) monotonically increased as the nickel
content increased at the approximately same rate of 400 MPa/Ni at.%, and the ﬂow stress

of parent phase is about 100 MPa greater than that of martensite. However, they did not

 

23 In the case of superelasticity, unrecoverable strain can also be caused by retained martensite after
unloading [Miyazaki et al., 1984].

40

report the effect of composition on the critical stress to induce martensite. Saburi, Tatsumi
and Nenno [1982] recognized that as nickel content increased the alloys tended to show
superelasticity. Perfect superelasticity was observed even in quenched N 151.3TI43_7 alloy
after solution treatment in which it could be assumed that no precipitates and few
dislocations existed.

(2) Precipitates: Fine precipitates produced in the aging process can suppress
permanent deformation during stress-induced martensitic transformation by raising the ﬂow
stress for slip and thus enhance superelasticity. Miyazaki et al. [1982] reported that after an
aging treatment in polycrystal NiTi alloys at 673 K, N i-rich specimens (N i50,6Ti49.4 and
Ni51_6Ti4g4) showed superelasticity; while stoichiometric specimens (Ni4ggTi502 and
N i50.,Ti49.9) did not because of the absence of the precipitates. In NiTi single crystals
superelasticity could similarly be achieved only after aging treatments [Miyazaki et al.,
1983].

On the other hand, ﬁne precipitates reduce the recoverable strains. In the work
conducted by Saburi, Tatsurrri and Nenno [1982], the recoverable tensile strain in
polycrystalline N15].3TI43.7 alloy is 5.3 % in solution treated state, but only 3.7 % in 773 K,
1 hour aged sample. Similar results were found in single crystals. Miyazaki et al. [1984]
investigated the dependence of recoverable strain on crystallographic orientation in
N150_5Ti49,5 single crystals. The recoverable strains of solution-treated specimens for
various orientations were in agreement with the values calculated from the lattice
deformation matrix. But the recoverable strains of aged specimens were only about half of
the calculated values. They speculated that the strains were reduced by untransformed
regions or the regions with un-detwinned multi-variants of martensite around the ﬁnely
dispersed precipitates.

(3) Dislocations: A great quantity of dislocations can be introduced by cold work or
cyclic deformation in NiTi systems. Filip and Mazanec [1994] studied inﬂuence of work

hardening on deformation. A cold-rolled N i495Ti505 specimen with 20 % reduction

41

exhibited superelasticity, but the critical stress to induce martensite, as well as the slope of
plateaus associated with stress-induced martensitic transformation, and the reorientation of
martensite variants, was higher compared with usual superelasticity curves. Zadno and
Duerig [1990] stated that cold-worked NiTi alloys exhibit nearly hysteresis-free linear
superelasticity with recovered strains as high as 4 %. The plateau is so steep that it almost
merges into the elastic slope. Miyazaki et al. [1986] investigated the effect of tension
cycling on superelasticity characteristics in NiTi alloys. The stress-strain curves of
fumace-cooled Ni495T1505 alloy show that the plateau becomes steeper and steeper as the
number of cycles increases and the change is most rapid in the initial cycling (Figure 1-
20a). The steep plateau was believed to be caused by a high density of dislocations which
limit B2/B 19' interface mobility and resist the reorientation of martensite variants.

Miyazaki et al. [1982] proposed a thermomechanical treatment method, i.e.,
annealing at temperatures below the recrystallization temperature immediately after cold
work, for realizing superelasticity in stoichiometric NiTi specimens. TEM image of
nricrostructures prepared this way showed that the high density of dislocations remained in
a specimen annealed at 673 K was similar to that in an as-rolled specimen. However,
sharp spots in the diffraction patterns implied that the dislocations had been rearranged
from the cold-worked state to release strain [Miyazaki, Igo, and Otsuka, 1986]. The
detailed mechanism has not been thoroughly investigated, nevertheless, the annealing
process improved strain recovery. The maximum recoverable strain of polycrystalline
Ni4ggTi503 alloy in their study was about 5.2 %, which is comparable to solution-treated
specimens reported by Saburi, Tatsumi and Nenno [1982].

(4) Grain size: An experiment conducted by Saburi, Yoshida, and Nenno [1984]
clearly demonstrated that the reduction of grain size was effective in improving
superelasticity, as shown in Figure 1-20b. Three types of single crystal specimens
(dimensions: 1 x 0.4 x 20 m3), a polycrystal with grain size of 1 mm, and a polycrystal

with grain size of 50 um, were made from the same ingot of the N150_5Ti49,5 alloy. The

42

single crystal specimen did not behave superelastically at any test temperature. As the grain
size decreased superelasticity became pronounced. For a specimen with a grain size of 50
um, there were 200 grains per cross section in the tensile sample and the criterion to
assume polycrystallinity was nominally satisﬁed. Apparently grain boundary constraint
played an important role in resisting of slip as discussed in Section 1.3.1.

In short, for NiTi alloy both recoverable deformation which underlies the
transformation shape strain and detwinning strain, and ordinary permanent deformation
need to be considered. In order to achieve perfect shape-memory effect and superelasticity
the NiTi process should result in low critical stress to induce martensite and high ﬂow
stress for permanent deformation. The recoverable strain in NiTi single crystal in the most
favorable direction is 22.10 % in shearing or 11.05 % in extension, while the maximum
recoverable tensile strain in NiTi polycrystals is 8 %. Solid solutions, precipitates,
dislocations and grain boundaries have effect of enhancing shape-memory effect and

superelasticity. However, they also tend to reduce the magnitude of recoverable strain.

1.5. Formation and Structures of Thin Films
1.5.1. Sputter Deposition

To meet an increasing demand for various functional coatings, e.g. optical, electrical,
mechanical and chemical coatings, a large number of deposition techniques have been
developed in the last 40 years. Methods of preparing thin ﬁlms can be divided essentially
into two groups, namely, cherrrical methods and physical methods. The chemical methods
basically consist of chemical vapor deposition (CVD), and electroplating. CVD is a
chemical process which takes place in vapor phase near or on the substrate so that a
reaction product is condensed onto the substrate. Electroplating is an electrochemical
process in which two electrodes are immersed in an electrolyte of an ionic salt and positive
ions are deposited onto the cathode (substrate). Physical methods are mainly the physical

vapor deposition (PVD) processes of evaporation, sputtering and ion plating. In the

43

evaporation process, atoms to be deposited are thermally transported from a source material
onto the substrate by various heating methods, such as direct resistance, electron beam (E-
beam) or an arc discharge. In the sputtering process, gas ions produced in a glow
discharge bombard the source material, called the target or cathode, dislodging atoms
which pass into the vapor phase and deposit on the substrate. Ion plating is a hybrid
process that combines the beneﬁts of evaporation and sputtering. In ion plating, the
coating source can be evaporation or sputtering, the surface to be coated is subjected to a
ﬂux of high energy ions and neutrals before and during the deposition.

There are some advantages which make sputter deposition one of the most versatile
techniques in thin ﬁlm technology. First, virtually any material, including metals, ceramics
and organics are candidates for a coating. The coating material is transported into the vapor
phase by a mechanical rather than thermal or chemical process. Second, sputtering is a
relatively high energy process. The energies of sputtered atoms are in the range of 1040
eV, whereas the energies of evaporated atoms are typically in the range of 0.2-0.3 eV. The
momentum of sputtered atoms enhances the mobility of adatoms, and often provides
superior ﬁlm quality and good adhesion between the coating and substrate materials.
Finally, the sputtering process has capacity to transfer alloys and compounds into the vapor
phase while preserving their composition. This feature is of importance in the fabrication
of alloy or compound ﬁlms where the composition needs to be controlled.

The planar diode shown in Figure 1-21a is the simplest sputtering conﬁguration. The
cathode serves as the source of coating material. It is usually water cooled to prevent
outgassing and the occurrence of diffusion in the case of alloy cathodes.24 The substrates
facing the cathode are mounted on the anode.” By applying a dc (direct current) potential

of the order of 1-5 kV across the electrodes, a glow discharge is initiated at an argon

 

24 Cathode (target) cooling is a necessary requirement to produce an altered layer having the reduced
concentrations of high yield species at the surface of an alloy cathode [Bunshah et al., 1982]. At the steady
state, this altered layer will provide a vapor ﬂux having the chemical composition of originating solid.

25 In some cases, the substrates are independantly biased (bias sputtering) or allowed to ‘ﬂoat’.

44

pressure of 1-10 Pa. Most of the electrical potential between the anode and cathode is
consumed in a ‘cathode dark space’ near the cathode, due to the relatively low mobility of
ions compared to the electrons. Positive Art ions are accelerated by this strong electric
ﬁeld and strike the cathode, and ballistically eject atoms from the surface of the cathode.
The sputtered atoms condense as a thin ﬁlm on the substrate. During ion bombardment, a
small number of electrons are also emitted from the cathode, and these electrons are
accelerated in the cathode dark space and enter the plasma volume. They collide with argon
atoms and generate the volume ionization required to sustain the glow discharge.

Recently, magnetron sputter sources have been developed. Magnetrons can be
deﬁned as diode devices in which magnets are situated behind the cathode and the
combination of magnetic and electrical ﬁelds force electrons into helical trajectories near the
cathode surface (Figure 1-21b). In this way, the electron’s path length is greatly increased,
thus raising the ionization probability. Most magnetron sources operate at a pressure in a
range of 0.1-10 Pa and a cathode potential of 300-700 volts. Compared with conventional
diode sputtering, magnetron sputtering is capable of high deposition rates and lower
operating pressures. This allows longer free path of atoms and reduces collisions therefore

gives higher adatom energy.

1.5.2. Inﬂuence of Temperature on Phase Formation

Most solid metallic materials are produced by melting/solidiﬁcation technology which
approximate equilibrium condition because of the normally slow cooling rate from high
temperature. The resultant phases can generally be predicted from equilibrium phase
diagrams. In physical vapor deposition processes, at high substrate temperature,
equilibrium phases similar to their melt-produced counterparts form due to sufﬁcient
diffusion processes. At low substrate temperatures the surface mobility of adsorbed
particles is sufﬁciently limited that the disordered state may be frozen before the adatoms

are able to reach energetically preferred sites, and non-equilibrium metastable phases,

45

which do not exist in equilibrium phase diagram, may form. In this subsection we
consider the inﬂuence of substrate temperature on phase formation.

Figure 1-22 is schematic illustration of the effect of substrate temperature on phase
formation for a hypothetical polymorphic material26 [Thomton, 1977]. Substrate
temperature is indicated by the horizontal axis and a series of equilibrium phases or, Band 7
are shown along the vertical axis. The stable phase at a given temperature will be the one
with minimum Gibbs free energy, G, given by the relation G = H - TS.27 At high
temperature, the -TS term dominates and the phase with larger entropy S will be more
stable. Thus, high temperature phases are characterized by large entropies, i.e., more open
and disordered structures and in the limit become an amorphous liquid.

At low substrate temperatures, a highly disordered, amorphous-like phase forms
because of low mobility of atoms. However, whether or not an amorphous ﬁlm forms also
depends on the nature of atomic bonding. It is found that in contrast to elements forming
covalent bonds such as in compounds, pure metals form crystallites even at liquid helium
temperature [Buckel, 1959; Mader, Nowick and Widmer, 1967]. The cause of the
difference is that in covalently-bonded materials the formation of crystalline structure
requires a relatively large displacement of atoms due to their small coordination number.
On the other hand, the formation of close-packed structures for pure metals needs only
small displacements which may occur even at very low temperatures.

When substrate temperature is elevated to an intermediate level (0.3 < Tﬂ‘m < 0.5,
where Tm is the melting point in Kelvin) [Thomton, 1977], diffusion processes become

signiﬁcant. Diffusivity generally obeys the Arrhenius law:

D: Do exp(—%). (1.35)

 

26 Polymorphic materials, such as iron and tin, are capable of existing in more than one solid phase.
27 The equation G = H - TS only provides a thermodynamic tendency at given temperature. Whether a
stable phase forms also depends on the kinetic conditions, i.e., the diffusion process.

46

Since surfaces have higher energy than the bulk, the extra energy reduces the amount of
activation energy which must come from other sources before the atom will diffuse. The

activation energies for surface and bulk diffusions can be ordered as

qurf. < Qbulk. (136)

From Equation 1.35 and 1.36, evidently, surface diffusion is favored when substrate
temperature is not high enough. The adatom mobility in growing ﬁlm surfaces is generally
adequate to form equilibrium phases even at relatively low substrate temperatures where
equilibrium phases are impossible to form in the bulk via bulk diffusion.

When deposition is done at high substrate temperatures (0.5 < T/I‘m < 1) [Thomton,
1977], diffusion through the volume tends to overpower the surface diffusion, a structure
characteristic of conventional melting/solidiﬁcation processing results.

From preceding discussion it is clear that the structure of thin ﬁlms is governed by
both the thermodynamics which determines the driving force, and the kinetics which
deﬁnes the available rate controlling mechanisms at a given temperature. Low substrate
temperatures limit kinetic processes and result in amorphous phase in alloys and
compounds. Moderate substrate temperatures only promote the surface diffusion and
produce equilibrium phases having unique columnar-grained structure (as will be discussed
in Section 3.1.2). High substrate temperatures greatly remove the kinetics limitation
through bulk diffusion process. The structures formed usually display equiaxed grain

morphology similar to that formed in the melting/solidiﬁcation process.

1.5.3. Structural Zone Models
Structural zone models are empirical diagrams representing the morphology of

coatings with respect to deposition conditions (substrate temperature and working-gas

47

pressure). The resultant ﬁlm structures are primarily controlled by shadowing effects,28
surface diffusion, and bulk diffusion. For many pure metals the activation energy for
moving adatoms is proportional to the melting point [Brophy, Rose and Wulff, 1964].
Therefore, various atomic processes can be expected to occur over different temperature
regions of T/l‘m in zone models.29

Movchan and Demchishin [1969] proposed a structure zone model based on their
experimental observations for electron beam evaporation as shown in Figure l-23a. The
model predicts three regions which depend on the substrate temperature. In Zone 1 (T/I'm
< 0.3), the ﬁlm consists of tapered crystals with domed tops which are separated by voided
boundaries. The crystals have a high dislocation density and poorly deﬁned structure.
Zone 2 (0.3 < Tﬂ‘m < 0.5) consists of columnar grains separated by distinct and dense
boundaries. The ﬁlms have properties similar to cast metals and in this region the surface
diffusion dominates. Zone 3 (0.5 < Tﬂ‘m < 1) consists of equiaxed grains with a bright
surface. The ﬁlms are formed through bulk diffusion and have properties similar to a fully
annealed metal.

Thornton [1977] extended this zone classiﬁcation to sputter deposition by adding an
axis to account for working gas pressure as shown in Figure 1-23b. A transition zone
(Zone T) between Zone 1 and Zone 2, consisting of a dense array of poorly deﬁned ﬁbrous
grains without voided boundaries, was also proposed. A higher working gas pressure
tends to shift the zone structures toward lower T/Tm since the higher argon pressure
increases the chance of collision between sputtered particles and argon atoms and reduces
the energy of condensing particles. Thus, at a given substrate temperature denser ﬁlms will

form at low argon pressure conditions. Thornton [1977] further pr0posed a generalized

 

23 Shadowing effect is a geometric interaction between the rough growing surface and the angular directions
of the coating ﬂux.

29 In structural zone models the substrate temperature is usually normalized as a fraction of the absolute
melting point of the metals.

48

schematic representation showing the superposition of effects of coating-ﬂux shadowing,
surface diffusion and bulk diffusion.

A computer model of ﬁlm growth using two-dimensional (2D) hard spheres by
Muller [1985] demonstrated that a transition from low packing density ﬁlms (0.7) to unity
density ﬁlms occurred as a function of temperature and deposition rate (Figure 1-24). It
was assumed in his model that depositing spheres intersect the growing surface at random
positions and stick in place except for relaxation into nearby ‘pockets’. This would
produce a structure having ‘pipes’ inclined along the angle of incidence of spheres. If a
thermal activation process was added to simulate the temperature dependence, a dense

microstructure can be formed at the equivalent temperature of 0.3T",

1.5.4. Secondary Grain Growth in Thin Films

Grain growth in polycrystalline metals is accomplished through grain boundary
nrigration which results from the diffusion of atoms across the grain boundary region.30
The driving force for normal grain growth lies in the interface energy of the grain
boundaries. As the grains grow in size and their numbers decrease, the grain boundary
area diminishes and the total interface energy is lowered.

Recently, Thompson [1990] developed a concept of secondary grain growth.
Secondary grain growth is a grain growth mode in which only a subpopulation of grains
grow at the expense of a static matrix of grains. The driving force for the secondary grain
growth is the surface energy anisotropy of the grains. Generally, closely packed planes
have a relatively low surface energy since the bonding forces of an atom within those
planes are utilized so fully by neighbors in its own plane that the absence of similar atoms

outside that plane introduces only relatively little local distortion. The most densely packed

 

3° The grain growth in polycrystals should be referred as the coalescence which describes the grain growth
in non-crystalline matrix (will be discussed in Section 3.1.2). However, the term of grain growth is more
prevail in the literature.

49

planes in bcc are (110), in fee are (111), and in hcp with ideal c/a ratios they are (0002).
These are the planes which have the lowest surface energy.

In thin ﬁlms, if the major diameter of grains is comparable to ﬁlm thickness, a
considerable part of the grain boundary is replaced by free surfaces or the interfaces
between the ﬁlm and substrate as seen in Figure 1-25a. The rate of grain growth in this
case is a function of not only grain size relative to the mean grain size as the situation of
normal grain growth, but also the surface energy of grain relative to the average surface
energy of grains. The grains with orientations that lead to low surface energies have an
energetic advantage over others during growth (Figure 1-25b). Finally, a uniform
crystallographic orientation, i.e., texture may develop (Figure 1-25c). Secondary grain
growth in thin ﬁlms has been experimentally observed in several systems, e.g., Si
[Thompson and Smith, 1984], Ge [Palmer, Thompson and Smith, 1987] and Au [Wong,
Smith and Thompson, 1986].

1.6. Studies on N iTi Films and the Purpose of Present Work

Since the near-equiatomic NiTi alloy, or Nitinol was ﬁrst deve10ped by Buehler and
Wang at the US. Naval Ordnance Laboratory [1968], this unique material has been
extensively studied by many researchers. By the late 1980’s, the phase transformation
characteristics, crystallographic aspects and thermoelastic deformation behaviors of bulk
NiTi alloys were well established.

Studies of N iTi thin ﬁlms relevant to the shape-memory effect commenced in the
1980’s and much effort has been expended to clarify the dependence of nricrostructures and
phase transformation temperatures on thin ﬁlm process conditions. Shifts in composition
with respect to the sputter target, and the structural and transformation characteristics of
sputtered Ti(Ni1-,Cu,) ﬁlms were investigated [Chang, Simpson and Grummon, 1989;
Chang et al., 1990; Chang and Grummon, 1997]. In their work, single layer (SL) films
and periodic multilayer (PML) ﬁlms, respectively, were fabricated by sputtering from

50

TisoNi45Cu5 ternary targets and by co-sputtering from alternating ternary and pure titanium
targets. Atomic composition analysis showed that titanium loss due to preferential
resputtering was compensated for by periodic multilayering with pure titanium.
Precipitates in annealed SL ﬁlms produced marked effect on its martensitic transformation
behavior, whereas the PML ﬁlnrs showed evidence of appropriate phase transformations
during DSC, resistivity, and cold-stage TEM experiments. The crystal structure of ternary
precipitate (N i+Cu)2Ti found in Tia-,5 (Ni+Cu)52,5 ﬁlms was identiﬁed by Chang and
Grummon [1991] using convergent beam electron diffraction (CBED).

Busch et al. [1990] deposited NiTi ﬁlms using a dc magnetron sputtering. The
temperature of unheated substrates during deposition was approximately 423 K due to
glow discharge heating. As-sputtered ﬁlms were amorphous in structure and did not
exhibit any shape memory characteristics. The heating curve of differential scanning
calorimetry (DSC) of as-sputtered ﬁlm exhibited one distinct exothermic peak at 753 K
which corresponds to the crystallization of the ﬁlm from amorphous structure. Below this
temperature, even to 723 K, the ﬁlm still remained amorphous. Films annealed at 823 K
for 30 minutes showed appropriate B2 diffraction peaks in X-ray diffraction spectra. A
free-standing annealed ﬁlm with a thickness of 10 um demonstrated a shape-memory
recovery in a conventional bending test.

Moberly et al. [1992] conducted an in situ observation of the crystallization of as-
sputtered amorphous N iTi ﬁlms in high voltage electron microscope. A electron
accelerating voltage of 1500 kV was used in order to observe crystallization in ﬁlm regions
as thick as 1.5 um. The crystallization process occurred over a range of temperatures from
773 K to 873 K. The nucleation of crystallites in thicker regions of TEM specimens,
which occurred prior to nucleation in thinner regions, implied that nucleation did not
proceed preferentially at surface, but rather homogeneously in the bulk of ﬁlms.

Ishida, Takei and Miyazaki [1993] sputtered NiTi ﬁlms under various argon

pressures and cathode powers. They found that argon gas pressure had a critical effect on

51

the mechanical properties of the ﬁlms although the cathode power also affected it. The
ﬁlms formed at a high argon pressure ( 13.3 Pa) were brittle due to its porous structure,
while the ﬁlms formed at a low argon pressure (0.67 Pa) exhibited a good shape-memory
behavior comparable to that of bulk NiTi alloys after annealing. The high argon pressure
decreased the energy of sputtered atoms by collision and limited surface diffusion. This
result was consistent with the Thomton’s structural zone model [Thomton, 1977].

Hua, Su and Wuttig [1993] investigated the effects of isothermal annealing on
NiTi/SiOZISi heterostructure by cross-sectional TEM observations. The amorphous
structure remained in the ﬁlm annealed at 733 K for 1 hour. The crystallites formed at
annealing temperatures of 833 K and 933 K had average grain sizes of 50 nm and 200 nm
respectively. Two kinds of nucleation centers were suggested. One was homogeneous
nucleation in the bulk of the ﬁlm, the other was heteronucleation at the ﬁlm/substrate
interface. The grain morphology displayed a columnar structure.

Gyobu et al. [1996] studied the inﬂuence of chemical composition and heat-treatment
on phase transformations in sputtered NiTi ﬁlms. During crystallization heat treatment, no
precipitation occurs in near-equiatomic NiTi ﬁlms, whereas NiTi; particles precipitated in
the Ti-rich ﬁlms and Ni4Ti3 precipitated in the Ni-rich ﬁlms. The R —> B 19' or B2 —> B 19'
transformation temperatures of 773 K, 1 hour annealed ﬁlms reached the maximum at the
equiatomic composition, decreased with increasing Ni—content in the Ni-rich side, and kept
constant in the Ti-rich side.

There are only few reports in the literature relating to N iTi ﬁlms sputter-deposited at
elevated substrate temperatures to directly form crystalline NiTi ﬁlms. Ikuta et al. [1990]
deposited NiTi thin ﬁlms at various substrate temperatures and observed that crystalline
ﬁlms were formed only at temperatures exceeding 673 K. At lower substrate temperatures,
the ﬁlms were amorphous and no signiﬁcant crystalline peaks were found in X-ray
diffraction spectra. Gisser et al. [1992] employed X-ray diffraction method and observed
that crystalline ﬁlms with (110) texture were produced by sputter deposition of NiTi alloy

52

onto (100) silicon substrate at a temperature range of 623-733 K. Krulevitch et al. [1996]
sputter-deposited 1.3-um and 5.1-ttm thick Ni-Ti-Cu ﬁlms at 723 K. The ﬁlms have
excellent shape-memory properties, but the columnar rrricrostructure of the ﬁlms showed a
signiﬁcant increase in surface roughness with ﬁlm thickness. Recently, F. Chang, [1997]
showed that the stiffness of the substrate may affect the minimum temperature to produce
crystalline NiTi. Fully crystalline NiTi ﬁlms could be formed on Kapton ﬁlm whose elastic
modulus is only 1.6 GPa [Product Bulletin, DuPont, 1993] at substrate temperature as low

as 560 K.

In parallel with the fundamental investigation, several prototypes based on shape-
memory NiTi thin ﬁlms were proposed and developed. A micro-planar spring of NiTi ﬁlm
was fabricated by Walker, Gabriel and Mehregany [1990]. In their study, 2 um—thick NiTi
ﬁlm was deposited onto a silicon wafer which had been coated with a 3 rim-thick polyirrride
ﬁlm (Kapton). The as-sputtered ﬁlm was annealed at 623 K for 1 hour. A wet-etch was
used to give a zigzagged spring pattern of the ﬁlm. After releasing the ﬁlm pattern using a
plasma etch to remove the polyirrride spacer, the spring pattern displayed some curling due
to residual stress. The curled ﬁlm then returned to its original, unreleased conﬁguration
after applying a heating current. Some researchers argued that the ‘shape-memory effect’
observed in the ﬁlm may actually be caused by thermal expansion effects since the N iTi
ﬁlm should have remained amorphous at 623 K [Ikuta et al., 1990]. A lack of basic
structure analyses made this study uncertain. Nevertheless, that study demonstrated the
compatibility of sputtered N iTi ﬁlms with current silicon-based nricromachining

technology.

Kuribayashi, Yosaaki and Ogawa [1990] fabricated a NiTi thin ﬁlm actuator showing

a two-way shape-memory effect.“ They conducted a constrained-aging anneal which was

 

3‘ In two-way shape-memory effect, the shape change of specimens also occurs on cooling. The two-way
shape-memory can be generated by different methods such as deformation cycling [Schroeder and Wayman,
1977] and constraint aging [Nishida and Honma, 1984]. In the case of constraint aging, Ni,Ti3 precipitates

53

directly adopted from that for bulk material [Nishida and Honma, 1984]. Ni-rich ﬁlms
with 10 um thickness were sputtered onto NaCl plates which were washed off in distilled
water. The free-standing ﬁlms were annealed at 1073 K for 10 minutes, and then were
curled to 3.7 mm diameter and constrained in a glass pipe for a second annealing at 673 K
for 6 hours to produce desired precipitates. The aged ﬁlms were then cut into a horseshoe-
like shape for testing. The ﬁlm horseshoe demonstrated a reversible bending motion with a
maximum frequency of 5 Hz when passing a rectangular wave of electrical heating current

through its pads.

Johnson [1991] reported two prototype devices developed at the NiTi Alloy
Company using the shape—memory effect in NiTi thin ﬁlms. One device has a membrane
of NiTi stretched over an oriﬁce to form a valve as shown in Figure 1-26a. The membrane
was heated by a dc current. An operating prototype has been cycled more than two million
times in their laboratory and a cycling rate of twenty Hz could be attained. The other one is
a microactuator for a rrrirror to be used as a refreshable optical storage device shown
conceptually in Figure 1-26b. Some other types of silicon-based rrricrovalves were also

explored [Johnson et al., 1992; Ray et al., 1992].

A thermomechanical NiTi/Si composite thin ﬁlm switch was fabricated and tested by
Kim, Su and Wuttig [1995]. In their work 1 pm NiTi ﬁlm was deposited onto 90 um
silicon substrate and the composite was given a heat treatment at 873 K. The deﬂections of
a cantilever composite were observed by cycling it between low and high temperature. It is
believed that the volume change of about 0.5 % during B2-to-R or B2-to-B19'

transformation was fully accounted for such deﬂections.

The development of residual stresses in NiTi ﬁlms sputtered on (110) silicon wafer
during isochronal and isothermal annealing of both amorphous and crystalline ﬁlms has

been studied using a scanning laser substrate curvature method [Grummon, et al. 1995;

 

grow preferentially under constraint stress. On cooling, the growth of martensite variants will be biased by
the internal stress ﬁelds associated with the precipitates and gives a macroscopic shape strain.

54

Zhao, 1996; Zhang and Grummon 1996, Zhang, 1998]. The sputter-deposited thin ﬁlms
showed large compressive intrinsic stresses which tend to increase with increasing
deposition temperature. Stress-relaxation induced by austenite-to-martensite transformation
on cooling can be fully recovered on heating at rates as high as 50 MPa/K. This effect may
form the basis for reversible cyclic actuation of NiTi/Si composites applicable to silicon-

based rrricroelectromechanical system.

Krulevitch et al. [1996] quantitatively compared several microactuation schemes and
investigated the thermomechanical properties of Ni-Ti-Cu thin ﬁlms using the wafer-
curvature method. The ternary ﬁlms exhibited recoverable stresses up to 510 MPa and
transformation temperatures above 305 K. A fatigue test, which was measured by the
curvature of chips transferred between two water baths one at room temperature and
another at 353-373 K, revealed that the stresses up to 350 MPa could be sustained for
thousands of cycles. Two nricromachined devices, a rrricrogripper and rrricrovalve, were

also demonstrated.

Since NiTi ﬁlms are expected to be utilized as force-displacement functional
components, their thermomechanical responses are of interest. In order to evaluate
deformation properties of NiTi thin ﬁlms, one must overcome two difﬁculties. The ﬁrst
difﬁculty is to suppress contamination during the sputter deposition and post-deposition
crystallization/annealing processes, since even a trace of impurities incorporated into the
ﬁlms will make them brittle. The second is the preparation and handling of free-standing
thin ﬁlm specimens, which are 1000 times thinner than conventional tensile specimens.
Probably for these reasons, only very recently has deformation data of the NiTi thin ﬁlms
became available. Miyazaki et al. [1994], Ishida et al. [1995] and Nomura and Miyazaki
[1995] studied the effects of chemical composition, heat treatment and deformation cycling
on deformation properties associated with martensite and R-phase transformations using

iso—stress tests. Miyazaki et al. [1994] reported an achievement of superelasticity in an

55

annealed (at 973 K) and aged (at 773 K) Ni49,7T150.3 ﬁlm. Recoverable strain (including

elastic strain) about 4 % was attained at an applied stress greater than 600 MPa.

Grummon, Nam and Chang [1992; Nam, 1995] proposed a superelastic NiTi
surface coating approach to prevent fatigue crack initiation and improve the fatigue lifetime
of polycrystalline copper. NiTi thin ﬁlms were deposited onto the gauge section of
polycrystalline copper dog-bone specimens by ion beam sputtering in which nickel and
titanium atoms are sputtered off by argon ions generated from a Kaufman-type ion source.
After crystallization annealing, the specimens were subjected to fatigue cycling under a
constant strain amplitude at various temperatures with respect to the phase transformation
temperatures of the ﬁlms. The development of fatigue cracks was observed by optical
microscopy and SEM. All crystalline ﬁlm coated specimens showed improved fatigue
crack initiation lifes over bare copper specimens. The greatest increase in fatigue life, about
100 % improvement over bare specimens, was found in Ni49,6Ti50_4 ﬁlm coated specimens.
The ability of NiTi crystalline ﬁlm to suppress fatigue crack initiation was believed to be
attributed to both superelastic deformation and ductility of the NiTi ﬁlms.

Typically, the M, of NiTi binary alloys are near ambient or sub-ambient. In order to
extend the application of NiTi thin ﬁlms to higher ambient temperatures as well as achieve a
high cycle-rates, it is necessary to increase the transformation temperatures. Partial
substitution of nickel or titanium by a ternary element additions has been found to be an
effective way. Miyazaki et al. [1995] sputtered Ti-26.4Ni-21.8Pd (at.%) ternary thin
ﬁlms. The transformation temperatures of the ﬁlms were characterized by the DSC
measurement. For the ternary ﬁlms the martensitic and austenite transformation peak
temperatures were 385 K and 401 K, respectively. Johnson, Martynov and Minners
[1995] sputtered ternary alloy ﬁlms using a Ti-50Ni-10Hf (at.%) target having a M,
temperature of 393 K. It was found that the Ms of deposited ﬁlms could be adjusted in a

temperature range from 373 to 473 K by placing additional Ti or Hf pieces on the target’s

56

surface. The dependency of transformation temperatures upon the Ni/Pd ratio in sputtered
Ti(Nil-,Pd,) ﬁlms with x = 0, 0.2, 0.6 and 1.0 was investigated by Quandt et al. [1995].
The Af and Mf temperatures, respectively, were increased with the increment of the
palladium content from 305 K and 235 K in NiTi ﬁlm to 843 K and 771 K in TiPd ﬁlm.
The study showed an achievement of widely adjustable transformation temperatures, from
room temperature to over 773 K, which are the highest transformation temperatures ever
realized in shape-memory thin ﬁlms.

Most NiTi ﬁlms that have been discussed in the literature have been produced by two
steps. The ﬁrst is the sputter deposition of N iTi onto an unheated substrate (T < 573 K)
which results in an amorphous NiTi ﬁlm. The second step is crystallization annealing at
temperatures well above 773 K. Alternatively, crystalline NiTi ﬁlms can be directly
produced by sputter deposition at elevated substrate temperatures as suggested by the
works of Ikuta et al. [1990], Gisser et al. [1992], Hou and Grummon [1995], Hou, Fence
and Grummon [1995] and Krulevitch et al. [1996]. However, so far little detail has been
reported in open literature.

The ﬁrst objective of present work is to establish a basic understanding of the
connection between the parameters of hot—substrate deposition processes and resultant
rrricrostructure of NiTi ﬁlms. A series of deposition temperatures were chosen in a range
where the surface diffusion of NiTi at growing ﬁlm dominates and no signiﬁcant bulk
diffusion of NiTi occurs. It is expected that crystalline ﬁlms with a microstructure different
from that produced through bulk diffusion can be formed at relatively low temperatures.
From engineering viewpoint, hot-substrate deposition may have advantages in terms of
reducing maximum required process temperature, which is critical for deposition of
crystalline NiTi ﬁlms onto polymeric components, and may also produce favorable
rrricrostructures unattainable from prevailing two-step NiTi ﬁlm processes.

The second objective is to clarify the relationship between the nricrostructure of NiTi

thin ﬁlms and their thermomechanical behavior. Although there is a considerable volume

57

of literature on the subject of deformation of NiTi alloys, the data obtained from bulk
specimens cannot always be directly applied to thin ﬁlms. As mentioned earlier, for a
given grain size, slip in lattice tends to become easier when the thickness of specimens
decreases (Section 1.3.1). Dislocation slip, which degrades the shape-memory effect and
superelasticity may become a controlling factor in thin ﬁlm cases where dimensions are
very small. Ni-rich precipitates in NiTi have been found to exert a strong inﬂuence on
mechanical response. For this reason, Ni-rich thin ﬁlms sputtered from Ni51Ti49 alloy
cathodes are of interest. In order to separate the effects of grain size and precipitate on the
deformation, some as-sputtered ﬁlms were further given a post—deposition annealing at
different temperatures to produce nricrostructures with various combinations of grain size
and precipitate morphology.

The completion of above two objectives has identiﬁed a basic thin ﬁlm
process/structure/property relationship which provides useful information for achieving an
optimum thermomechanical performance of NiTi thin ﬁlms by controlling process
conditions. In parallel with the above study, the fabrication and test of a prototype of

electrically excitable NiTi/polymeric actuator has been conducted.

CHAPTER 2

EXPERIMENTAL METHODS

The experiments carried out in the present study consist mainly of two parts: the
preparation and the thermomechanical characterization of the NiTi ﬁlms. The sputter-
deposition equipment developed for this study is ﬁrst discussed (Section 2.1). Then
methods used for fabrication of N iTi/polyimide prototype nricroactuator and the preparation
of free-standing NiTi ﬁlms are presented in Section 2.2 and 2.3. In the second part,
several basic characterization methods employed in the present study are brieﬂy described

in Section 2.4.

2.1. Sputter Deposition Facility

The sputter machine developed for this study is shown in the photograph in Figure 2-
1 and schematically illustrated in Figure 2-2. A polished stainless-steel bell-jar having a
volume of about 100 liters was evacuated by a 10 inch diameter silicone-oil diffusion pump
(Varian HS-10) with a pumping speed of 40001iter/sec,32 backed by a rotary-oil-pump
(W elch 1397) through a copper-wool oil vapor trap. In order to establish a stable working
pressure around 5 x 10'3 torr in the chamber during coating stage, a 10 inch throttle valve

(gate valve) was placed between the chamber and the diffusion pump. Four 1500 watt

 

32 The pumping speed can be calculated using the following equation [Technical Tables, Vacuum Product,
Kurt J. LeskerCompany, 1993]:
S = ¥ln(i)

P

where S: pumping speed in liter/sec, V: chamber volume in liters, t: time to reduce pressure from P0 to P,
in seconds, Po: starting pressure in torr and P: ending pressure in torr after t second.

58

59

resistive heater tapes were mounted around the chamber wall for baking out the chamber.
Copper tubing with 3/8 inch diameter was soldered to an outer copper jacket for water-
cooling the chamber wall during coating operation. A K-type thermocouple gauge, a
capacitance manometer (MKS Baratron) and an ion gauge were connected to the chamber to
monitor the pressure in different ranges from the atmospheric pressure to 10'7 torr (1.3 x
10'5 Pa), and a Quadrupole mass spectrometer was connected to the chamber to indicate the
partial pressure of residual gases and serve as a leak detector.

A Torus-2C magnetron sputter source (Kurt J. Lesker Co.) was powered by an
Advanced Energy NDX- 1K dc power supply. Most deposition runs in this work were
operated at power control mode, i.e., during the entire deposition run the product of
cathode voltage and cathode current was kept constant. A stainless-steel substrate holder
was originally designed to be electrically insulated from other ﬁxtures in the chamber for a
purpose of applying a biasing voltage to the substrates during coating.33 Argon gas with a
purity of 99.9996 % was further puriﬁed by passing it though a titanium getter heated to
over 1023 K.34 The ﬂow rate of argon was controlled by a MKS mass ﬂow controller and
the outlet of the working gas was located in the vicinity of the sputter source. A stainless-
steel shutter which could be manipulated through a rotary feedthrough was placed between
the sputter source and substrates. The target-to-substrate distance was kept at 65 mm for

all runs.

 

33 Bias sputtering is the technique of maintaining a negative potential (generally in a range of 50-500 volts)
on the substrates during deposition. The positively charged ions (primarily Art) are attracted toward the
substrates and give the surface of growing ﬁlms a low energy bombardment. Bias sputtering has been
found to have an effect on suppressing the development of Zone 1 structure at low TIT“, [Bunshah, et al.,
1982]. In the preliminary work of this study, the effect of biasing on depressing the deposition temperature
to form crystalline N iTi ﬁlms has been explored. Nine deposition cycles were conducted with substrate
temperatures between 523 and 673 K and biases between -15 and -300 volts. It was found that the biasing
did not have much effect on promoting crystallization. The ﬁlms deposited at -15 volt bias showed slightly
better quality in teams of ductility which may due to the removal of loosely bonded contamination on the
growing surface of the ﬁlm [Thomton, 1982]. The ﬁlms formed at the higher bias (2 250 volts) displayed
severe fracture and separation from the substrates which suggests very high internal stress.

34 By estimating using the ideal-gas state function, at a typical deposition pressure of 5 nrillitorr, the partial
pressure of impurity gas from working gas is 2 x 10'8 torr (2.6 x 104 Pa). Therefore, it is beneﬁcial to
utilize the titanium getter in the present work.

60

2.2. Fabrication of N iTi/Polyimide Prototype Microactuator
2.2.1 . Crystallizing NiTi by Post-Deposition Annealing

The Kapton ﬁlm is synthesized by polymerizing an aromatic dianhydride and an
aromatic diarnine. It is claimed that there are no known organic solvents for the ﬁlm
[Product Bulletin, DuPont, 1993]. Kapton polyimide ﬁlm has an ability to maintain its
physical, electrical, and mechanical properties over a wide temperature range from 4 K to
673 K. Kapton has been used in a variety of electrical and electronic insulation
applications: wire and cable tapes, substrates for ﬂexible printed circuits and as a
transformer and capacitor insulation.

Since Kapton is stable at a temperature as high as 673 K, this property makes the
polyinride ﬁlm a useful substrate material for metallizing or coating with crystalline NiTi
alloy. In the present study, the feasibility of integrating crystalline NiTi ﬁlm on Kapton
ﬁlm was explored. First, an attempt was made to sputter NiTi onto Kapton substrate at the
ambient temperature”, followed by annealing the NiTi/Kapton laminate to crystallize NiTi
ﬁlm. In this case, the metallization of Kapton with crystalline NiTi thin ﬁlm requires that at
a given heating rate, the crystallization temperature of NiTi ﬁlm should be lower than the
decomposition temperature of the Kapton. The weight loss characteristics of Kapton in air
and helium at a heating rate of 3 K/min is shown in Figure 2-3 [Product Bulletin, DuPont,
1993]. At a temperature of about 773 K, the weight loss increases rapidly which indicates
the start of decomposition.

The crystallization behavior of NiTi ﬁlms sputter deposited using Ni51Ti49 and
N i45Ti50Cu5 targets were investigated by differential scanning calorimetry (DSC) at a
heating rate of 20 K/min as shown in Figure 24.36 The exothermic peak appearing in each
curve represents an amorphous-to—crystalline reaction. The temperatures for start of

crystallization of binary and ternary ﬁlms were 770 K and 759 K respectively. The data

 

35 It was found from this study that due to the bombardment of sputtered material ﬂux, the temperature of
unheated substrate could increase from room temperature to 353 K at the sputter power of 250 watts.
35 The heat ﬂow is deﬁned to be positive if the system release heat to the environment.

61

show that crystallization temperatures are lower than the Kapton decomposition temperature
of the polyinride, since high heating rate in the DSC experiment would have raised the
crystallization start temperatures, according to the study made by Chang [1993].37 From
this preliminary examination it seemed worthwhile to conduct a further trial.

Compared with its amorphous structure, crystalline N iTi possesses a lower electrical
resistivity as revealed by Ikuta et al. [1990]. In their work, resistivities of amorphous and
crystalline NiTi ﬁlms at the room temperature were stated to be 1.77 x 103 ohm cm and
0.87 x 10'3 ohm cm respectively. Consequently, a resistivity drop should be accompanied
with an amorphous-to—crystalline transition. An in situ observation of electrical resistivity
change was performed during annealing treatment. When a rapid resistivity drop appeared,
the heating power was reduced instantly in order to preserve the Kapton substrate from
being overheated. Figure 2-5 shows the set-up of the experiment. The NiTi/Kapton
laminate formed in cold-substrate deposition was cut into a dog-bone shape. In the vacuum
chamber the specimen was irradiated by two 625 watt quartz lamps powered by a
controllable power supply.

Figure 2-6 gives the electrical resistivity curve of NiTi ﬁlm and temperature curve in
annealing at a heating rate of 5.6 K/min. At room temperature, the resistivity of NiTi
amorphous ﬁlm is 1.6 x 10'3 ohm cm. In the temperature between 301 K and 767 K, the
coefﬁcient of resistivity is about -6.65 x 105 ohm cm/K. The negative temperature
coefﬁcient can be attributed to structure recovery on increasing temperature. At 767 K, a
large resistivity drop occurred due to crystallization and the temperature was maintained
until the resistivity drop slowed down. After crystallization the curve exhibited a positive
temperature coefﬁcient, which is normally found in metallic crystals. The crystalline
structure of NiTi ﬁlm was conﬁrmed by X-ray diffraction spectra. However, both NiTi

ﬁlm and Kapton displayed discoloration and brittleness after the crystallization anneal. The

 

37 In Chang’s work [Chang, 1993], crystallization temperature of single-layer NiTi ﬁlm measured by DSC
is increased with increasing heating rate from 749 K at heating rate of 2 K/min to 780 K at heating rate of

50 K/min.

62

NiTi ﬁlm then was contaminated since N iTi is very reactive at the elevated temperature.
Figure 2-7 shows the residual gas pressure vs. temperature curves of Kapton ﬁlm at

the heat-up rate of 5 K/min. The residual gas pressures start to increase at the temperature

around 700 K. It is possible that before the temperature reached the point at which the

weight loss increases rapidly the Kapton had been outgassing signiﬁcantly.

2.2.2. Metallization of Kapton by Hot-Substrate Deposition

For a given material, compared with the bulk diffusion surface diffusion is important
at the lower temperature as discussed in Section 1.5.2. A crystalline ﬁlm can be formed by
surface diffusion in a growing ﬁlm, while crystallization of an amorphous ﬁlm can only
proceed via bulk diffusion. Thus, it is expected that hot-substrate deposition can produce
crystalline N iTi thin ﬁlms at relatively low temperature. In the study reported by Ikuta et
al. [1990], crystalline N iTi was produced during deposition at substrate temperature of 673
K.

The condition of successfully depositing crystalline NiTi ﬁlm onto Kapton is that
during deposition the temperature of Kapton substrate should be maintained between the
minimum temperature to form a fully crystallized NiTi ﬁlm and the maximum temperature
at which Kapton is stable (= 703 K).38 A conventional substrate holder (Figure 2-8a) with
a heater situated behind the substrate did not work well with polyirrride substrates which
have poor thermal conductivity. For instance, if the side of substrate facing sputter source
was heated to 673 K, the other side of substrate, facing the heater, will have a far higher
temperature than the decomposition temperature of Kapton. Another heating device which
has been tested, consisted of four 625 watt quartz lamps clustered around the sputter source
to irradiate the substrate (Figure 2-8b). Thermocouples were located in front of the

substrate and in contact with it. However, the problems encountered were that the

 

33 The requirement of the stable of Kapton" is not only for preventing polyinride ﬁlm from thermal damage
(decomposition), but also for preventing NiTi ﬁlm from the contamination by outgassing of the Kapton”.

63

temperature of substrate was not uniform due to imperfect focusing of quartz lamps and in
some local areas the Kapton overheated.

Finally a heating device which has two separately controllable heating elements was
developed as shown in Figure 2-9. The front heater was placed between the sputter source
and substrate holder. The spacing between heating wires permitted sputtered atoms to pass
through the heater and to deposit onto the substrates. The heating wires were shielded by
alumina tubes to eliminate contamination of the ﬁlm and to prevent embrittlement of the
heating wires caused by in-diffusion of sputtered atoms. The size of both front and back
heaters was made to have an area about four times greater than that of the substrate, so that
in lateral direction a nearly uniform substrate temperature could be expected. The substrate
temperature was monitored by two shielded thermocouples39 which were placed in the
vicinity of each side of the substrates. By individually adjusting power to the two heaters
so that readings from the front and back thermocouples were identical, the true substrate
temperature could be read.

Crystalline NiTi ﬁlm with a thickness of 3.1 pm was successfully sputtered onto
Kapton-2OONH ﬁlm (having a nominal thickness of 7.6 pm), at 698 K. The detailed
deposition procedure will be described in Section 2.3.1. The as-sputtered NiTi ﬁlm was
conﬁrmed to have crystalline structure (Appendix A. 1). NiTi ﬁlms revealed a shiny
mirror-like surface and excellent adhesion. When the ﬁlm was scratched by a ﬁne metal
point such as ﬁne tip of tweezers, only a thin deformed groove was formed and no
delarninating or cracking occurred. The composite displayed a 3 to 5 mm radius of
curvature at room temperature which was convex on the metal side indicating that the NiTi
ﬁlm maintained a slight residual compressive stress arising from the mismatch in the
thermal coefﬁcient of linear expansion between NiTi (01.32 = 11.0 x 1045 /K, 01319 = 6.6 x

10‘5/K) and Kapton (ct = 20 x 10-5/K).

 

39 It was found that during the deposition unshielded thermocouple probe would pick up the wrong signal
generated by the plasma.

64

2.2.3. Photolithography Process

The technique for fabricating a NiTi/polyimide microactuator prototype is directly
adopted from conventional integrated circuit (IC) process. The experiment of patteming-
etching process was conducted at the Photolithography Laboratory in Department of
Electrical Engineering, Michigan State University. Several trials were necessary for
establishing proper working parameters. The process used in this study is illustrated in
Figure 2-10 and outlined as follows:

(1) An 18 x 18 mm2 square of NiTi metallized Kapton was cemented onto a 22 x 22
mm2 square slide cover glass with a collodion solution with the NiTi ﬁlm facing outward.
A moderate pressure was applied on rnicrolanrinate until the solution had solidiﬁed.

(2) A photoresist [Shipley Co., PR1813] was spun over onto the surface of NiTi
ﬁlm for 30 seconds at 3000 rpm in a spin coating machine. The thickness of photoresist
was approximately a few micrometers.

(3) The photoresist was soft-baked for 15 minutes at 358 K in a convection oven.

(4) The specimen was aligned with a pattern photomask and exposed under a 7.6
mw/cm2 ultraviolet source for 60 seconds by using a Karisuss-MJB3 apparatus. The
photomask shown in Figure 2-11 was made by photo fabrication.

(5) The specimen was developed for 35 seconds in a MF319 developer at room
temperature [Hoechst Celanese Corp., Chatham, NJ].

(6) Again the rrricrolaminate was post-baked at 393 K for 20 nrinutes in another
convection oven to cure the pattered photoresist.

(7) The NiTi thin ﬁlm was etched by a mixture of equal parts of a titanium enchant
and a nickel enchant [Transene Company, Inc., Rowley, MA] for 3 to 5 rrrinutes.
(8) Finally, the rrricrolanrinate was released from slide cover glass and the photoresist

was dissolved with the acetone.

65

2.3. Preparation of Free-Standing NiTi Films
2.3.1 . Sputter Deposition Procedure

Several 50 mm diameter, 3 mm thick disk-shaped cathodes (targets) were sliced from
a wrought NiTi bar having a nominal chemical composition of Ti-51.0 at.% Ni [Special
Metals Corporation, New Hartford, NY] by electrical discharge machining (EDM). The
surface of cathodes was ground using 400 grit abrasive grit papers. After cleaning, the
cathode plate was clamped onto the water cooled copper heat sink base of the sputter gun.
Virgin cathodes were given a preliminary sputter ‘wear in’ for six hours in order to
establish a favorable geometrical conﬁguration prior to deposition. Each cathode was only
used for two deposition runs, for a total of about 5 hours, to minimize variations caused by
cathode wear.

Free-standing NiTi ﬁlms were needed for TEM observation and tensile tests. The
N iTi ﬁlms were sputtered onto cleaved potassium chloride (KCl) single crystals, onto
micro-polished copper sheets, and onto glass microscope slides. Films could be separated
from KC] using water and nitric acid could be used to dissolve the copper sheets. Films on
the glass substrates could be manually peeled off. However, the ﬁlms were very brittle
when deposition temperatures were above 623 K. Eventually, high purity NiTi ﬁlms were
deposited on quartz plates. Then good quality free-standing NiTi ﬁlms could be
mechanically separated from the quartz substrates.

Quartz plates with a thickness of 1.5 mm were cut into 22 mm x 22 mm squares.
Before loading them onto the substrate holder the plates were cleaned with acetone and
methanol successively in an ultrasonic cleaner for ﬁve rrrinutes to remove contamination.
After setting up the specimens and pumping down the chamber, the chamber wall was
baked by heating tapes at approximately 353 K and substrates were baked by two-sided
substrate heaters (Figure 2-9) at 623 K. Baking for 18 hours was necessary to drive out
water vapor absorbed on the wall of belljar and other ﬁxtures inside the chamber. Then the

cooling water was introduced into the tubing surrounding the chamber and power for the

66

substrate heaters was gradually adjusted to the selected deposition temperature. Two hours
was allowed to elapse in order to establish steady state heat transfer conditions. During the
sputtering the temperature readings from both front and back thermocouples were
controlled to within :5 K of the desired value.

The base pressure attained in the vacuum system was less than 3 X 10'6 torr (4 x 104
Pa) at process temperatures between 573 K and 773 K. Argon working gas was bled into
the deposition chamber though the preheated titanium getter. The argon pressure was
chosen as a minimum to sustain a plasma discharge at each selected temperature. Prior to
the exposure of the substrates the water-cooled target was presputtered for 15 minutes at
full power lever to clean the target surface and to attain a steady-state sputtering condition.
The ﬂoating potential of the substrates during the sputter deposition was +6 to +8 volts.
After sputtering for 2 hours, which typically yielded NiTi ﬁlms with thickness between 4
and 7 pm, the substrates were cooled to room temperature in the vacuum chamber at an
initial rate of 0.5 K/sec. A plot of the substrate temperature and base pressure vs. time for
a typical deposition run is shown in Figure 2-12.

The experimental set-up and deposition procedure for Kapton substrates were
essentially the same as that for quartz substrates just described. The deposition parameters
and resultant ﬁlm compositions are tabulated in Table 2-1. For NiTi ﬁlms produced by the
hot-substrate deposition, the notation ‘H---’ is used in the present work where the ‘H’
signiﬁes ‘hot deposition’ and three digits signify the substrate temperature of deposition in

Kelvin.

2.3.2. Thermal Treatment of As-Sputtered NiTi Films

Some as-sputtered NiTi ﬁlms were given a post-deposition anneal at various
temperatures between 858 K and 1123 K for the purpose of producing rrricrostructures
with a variety of grain sizes and precipitate morphologies. A quartz-tube vacuum furnace

was used for annealing as schematically illustrated in Figure 2-13. The vacuum chamber

67

was pumped by an Alcatel 5400 turbomolecular pump having a pumping speed of 400
liters/sec. The external electrical resistance furnace could be raised or lowered around the
quartz tube manually. A titanium getter which contained pure titanium wires was set in the
center of the quartz tube to reduce oxidation of the NiTi ﬁlms. Free-standing NiTi ﬁlms
were loaded on a titanium wire rack.

The furnace temperature was maintained by a programmable Omega CN 8600
controller. The vacuum chamber the was ﬁrst linearly ramped to a temperature of 575 K in
8 hours and baked at that temperature for 4 hours, then linearly ramped to selected
annealing temperatures in 2 hours. After annealing for one or two hours the power for
furnace was shut off. Two cooling methods, radiative and furnace cooling, both were kept
in vacuum condition, were used in the present study. In radiative cooling, the furnace was
elevated quickly and the hot quartz tube was exposed to the ambient immediately after
terminating the annealing. The quench rate achieved was 2.0-1.4 K/sec in the temperature
range of 1044-799 K. In furnace cooling, it usually took two hours to cool the specimens
down to below 373 K. During the entire annealing process, the base pressure was
maintained under 10-5 torr (1.33 x 103 Pa).

For annealed N iTi ﬁlms, the notation ‘A-’ is used where the ‘A’ represents

‘annealing’ and a digit designates the detailed heat-treatrnent path as indicated in Table 2-2.

2.4. Characterization of N iTi Films
2.4.1. Energy Dispersive X-Ray Microanalysis (EDX)

The chemical composition of the NiTi ﬁlms was determined by the energy dispersive
X-ray rrricroanalysis (EDX) using ZAF“0 corrections in a Hitachi S-2500 scanning electron
microscope (SEM). The N iTi ﬁlms and the calibration materials were mounted onto a

graphite specimen stage using double-coated carbon conductive tape. The calibration

 

4° ZAF is a correction method to correct measured relative intensity ratio. In this method three effects are
corrected: (1) atomic number effect, Z; (2) absorption of x-rays within the specimen, A; and (3) ﬂuorescence
effects. F.

68

spectrum was acquired from a piece of pure nickel. A piece of standard N150.1Ti49_9 alloy
was used to determine the intensity ratios of Ni and Ti elements. Prior to taking spectrum
the LaBg ﬁlament of the SEM was stabilized for at least one hour. All spectra were
collected at 20 kV electron accelerating voltage, 15 mm working distance and 2000 x
magniﬁcation. The spectra were then analyzed using a ZAF-4 program supplied by Link
Analytical Incorporation. For the purpose of checking the accuracy of the analysis, two
target alloys having known compositions were also measured at the same time. The
compositions of two target alloys indicated in the manufacturer’s speciﬁcation were
NisoTiso (nominal) and N151_3Ti4g-,, whereas that determined by this study were N150,01Ti49.99

and Ni50,96Ti49_04 respectively. Scatter of about 0.5 at.% are considered to be normal in the

ZAF method.

2.4.2. Film Thickness Measurement

The thickness of ﬁlms was measured using a Dektak II proﬁlometer. The surface
proﬁle of thin ﬁlms on quartz substrates was determined by scanning a ﬁne stylus across a
step produced by masking a part of the substrate during deposition. The scanning distance

was 3 mm and scanning speed was 1 mm/min.

2.4.3. Electrical Resistimetry

The phase transformation temperatures of free-standing NiTi ﬁlms were determined
from electrical resistivity vs. temperature curves. The curves were measured by four-point
electrical resistimetry on the ﬁrst thermal cycle after cooling the ﬁlms to the room
temperature from high temperature of hot-substrate deposition or annealing. The
experimental set-up was similar to that presented by Nam [1995]. A 5 x 1.5 mm2 strip of
N iTi thin ﬁlm was connected to the ﬁne gauge copper wires which linked to the electrical
circuit by pure indium. The measurement cycled between 150 K and 380 K with a ramping

rate of 3-5 K/min, achieved by adjusting the ﬂow rate of liquid nitrogen to the test tube,

69

and the dc current of power supply to the heating wire mounted on the test tube. Signals
from the voltage drop across the N iTi ﬁlms and the output of the thermocouples, i.e.,

temperature were plotted by a HP 7046A x-y recorder.

2. 4.4. Differential Scanning Calorimetry (05 C)

A TA Instruments Model 2920 modulating differential scanning calorimeter was
employed to determine martensitic and R-phase u'ansformation temperatures and associated
enthalpies of the NiTi ﬁlm sputtered at 703 K. For the measurement, a 10.90 mg NiTi ﬁlm
sample was cut into small pieces and encapsulated into a 5 mm diameter aluminum pan.
Another empty pan was used as the blank reference. The pans were cycled between 223 K
and 373 K at the rate of 5 K/min. The transformation start and ﬁnish temperatures were

determined by the tangential extrapolation method.

2.4.5. X-Ray Diffraction (XRD)

The crystal structure of as-sputtered NiTi ﬁlms on Kapton substrates and free-
standing NiTi ﬁlms were veriﬁed by a Scintag XRD 2000 diffractometer using the Cu K,
radiation. The machine was operated at an accelerating voltage of 35 kV and a tube current
of 25 mA. The scanning range and the scanning speed were 20 = 35-60° and 0.3°/min
respectively. In some case the specimens were heated to 353 K during data collection
using a custom-built electrical resistive heater mounted at the back side of the sample stage.

The temperature of specimens was measured by a J-type thermocouple.

2. 4. 6. Transmission Electron Microscopy (TEM)

TEM disc specimens, 3 mm in diameter, were shaped by a pair of sharp scissors.
Then the discs were thinned with a Tenupol-3TM [Struers A/S. Denmark] twin-jet
electropolisher using 12 % nitric acid and 88 % methanol by volume at 223 K. Using a

masking technique, foils parallel to the plane of the deposit were obtained at three locations

70

along the depth of the ﬁlm: one at the ﬁlm/substrate interface, one at the nominal midplane
of the ﬁlm, and one at the free-surface of the ﬁlm, as shown in Figure 2-14. Conventional
rrricrostructure observation, precipitate analysis and grain texture determination were
conducted in a Hitachi H-800 TEM operated at 200 kV and equipped with a double-tilt
specimen stage. Cooling stage observation was performed in a Jeol 100 CX TEM operated
at 100 kV and equipped with a single-tilt cooling stage. Precipitate analysis of as-sputtered
NiTi ﬁlms was made in a Jeol 2000 FX TEM operated at 200 kV, using the smallest
available selected area diffraction (SAD) aperture (d = 30 pm).

In physical vapor deposition, most deposits exhibit a strong preferred grain
orientation at low-to-interrnediate deposition temperatures and tend to a more random
orientation with increasing deposition temperature [Bunshah, 1982]. Several techniques
are available to determine preferred grain orientation (i.e., texture) in thin ﬁlms. The pole
ﬁgure method in the X-ray diffraction can provide a detailed orientation distribution.
However, it collects diffracted beams from the specimen to a depth of several micrometers
and does not resolve the signals from the different layers of the thin ﬁlms. Electron
rrricrodiffraction in TEM may give ambiguous result unless either the textures are very
strong or diffraction patterns are taken from many individual grains. In contrast, the
method of analyzing diffraction patterns with tilting the specimen over a wide range in the
TEM can effectively determine the textures of individual layers if the specimen with various
sampling layers can be prepared. Tang and Thomas [1993] have successfully analyzed the
texture evolution versus thickness of Cr thin ﬁlms sputtered onto NiP—coated AlMg
substrates using the sample tilt method.

It is known that the reciprocal lattices of thin ﬁlms with randomly oriented grains can
be obtained by rotating the reciprocal lattice of single crystal around the origin, whereas,
the reciprocal lattices of thin ﬁlms with ﬁber-texture can be obtained by rotating reciprocal
lattice of single crystal around the texture axis. In the reciprocal space the former is

concentric spheres while the latter consists of layered concentric rings. The selected area

71

diffraction patterns (SADP’s) are the intersections of those reciprocal spheres or rings and
the Ewald sphere.41 For the ﬁlms with randomly oriented grains, there is no change of
SADP’s upon the specimen tilting. But for the textured ﬁlms, tilting moves zero-order
layer rings away except at the locations along the tilting axis and makes the rings in the
adjacent layers intersect the Ewald sphere.

Based on the experimental results made by Gisser et al. [1992], NiTi ﬁlms formed at
the elevated temperatures have (110) ﬁber texture. As a starting point, it is logical to
assume that the same texture will present in the present study. Figure 2-15a shows the
reciprocal lattices of single—crystal (B2) with zone axis [110] and Figure 2-15b shows the
geometric representation for calculating the positions of diffracted arcs.42 The following

formula gives the positions of diffraction arcs:

 

(a = sin“( ) (2.1)

Rsin 6

where (p is the angular distance of the arcs measured from the tilting axis in the image plane
(Ewald sphere), d is the distance between the zero-order reciprocal rings and higher-order
reciprocal rings, R is the radius of diffracted arcs in the image plane and His the tilting
angle of the specimen. For the (110) texture, at a tilting angle of 30°, angles between the
tilt axis and the diffraction arcs in image plane are q) = 900° for the {110} ring, (p = 353°
for the {211} ring and (p = 266° for the {310} ring etc. By comparing these calculated
results with the tilted diffraction patterns, whether a (110) ﬁber texture presented could be

determined.

 

41 The Ewald sphere can be considered as a plane because of the short wavelength of the electrons.
42 Instead of diffracted spots, the diffracted arcs were observed in practice. This is due to a imperfect ﬁber
texture in which the ideal concentric rings of the reciprocal lattice spread into the concenuic spherical bands.

72

2.4. 7. Tensile Tests

In the present work, the isothermal tensile tests were carried out in a MinimatTM
miniature tensile tester [Expanding Polymer Science] using a strain rate of 6.8 x 105 to 3.4
x 104 lsec at various temperatures.43 The machine was equipped with a IBM compatible
PC and an electronic controller having displacement or load controlled testing capacity.

The 20 N load cell and the displacement of the crosshead were independently calibrated by
using standard weights and an extensometer respectively. A pair of grips were designed
and machined from a 1000-series aluminum alloy plate as shown in Figure 2-16. The grips
were made to be much stiffer than the thin ﬁlm specimen in order to minimize the
elongation caused by the grips. The minimum cross-section of grips was 2 x 8 mm2 which
is about 1300 times greater than that of specimens. Two design features of the grips were
important to make the tensile tests successful as seen in inserted sketches in Figure 2-16:
(a) Both grips were rotatable around the joint pin which prevented the specimen being torn
and bent due to an imperfect alignment in specimen loading. (b) One surface of the grips
was machined with a radius about R = 50 mm. This surface provided a great compressive
force and eliminated the slippage of the specimen.

The thin ﬁlm specimens were cut using a pair of ﬁne blade scissors. The gauge
length and width of the specimens were approximately 6 mm and 1.5 mm. Test
temperatures were maintained within 12.5 K by a methanol ﬂow, pre-cooled to the desired
temperature by liquid nitrogen, for temperatures below ambient, and by an electrically
heated chamber at higher temperatures. The lowest temperature that could be controlled in
the methanol ﬂow was 198 K which is below the M, temperatures of most ﬁlms in this
work. For each temperature-series of stress-strain curves, a single specimen was used,

with the ﬁrst of the tests conducted at the lowest temperature in the series. In each case,

 

43 In bulk NiTi, it has been shown that the strain rate has a strong inﬂuence on stress-strain responses
[Mukherjee, Sircar and Dahotre, 1985; Shaw and Kyriakides, 1995]. This is because stress-induced phase
transformations also liberate heat and the heat transfer in bulk tensile specimens is slow. In the present
study, however, the specimens are very thin and the effect of straining rate can be neglected. The
temperature of the thin ﬁlm specimens can be considered equal to that of the surrounding.

73

after completion of a stress-strain loop, the specimen was heated to 373 K, and then cooled
to the temperature selected for the subsequent test with care taken to avoid overshoot. To
prevent premature fracture, maximum total strain for any individual test was limited to 4 %.

In conventional tensile tests, the test bar has a reduced gauge cross-section for the
purpose of elirrrinating the stress concentration at the grip areas and an extensometer is
attached at the center region of the specimen where a uniform deformation is assumed. In
the present work, however, rectangular thin ﬁlm strips were used as the tensile specimens
due to the difﬁculties of handling the thin ﬁlms. Besides, the tensile strain was evaluated
from the measured crosshead displacement, since the extensometer would impose a
considerable constraint force on the thin ﬁlm specimen. Next, the stress state of tensile
specimens is estimated by ﬁnite element analysis and the initial sites of fracture are
predicted.

The dimensions of the gauge section were approximately 6 mm x 1.5 mm x 6.5 pm.
The problem can be treated as the plane-stress state since the thickness of specimens was
about 1000 times smaller than other two dimensions and there were no loads applied in the
coordinate direction parallel to the thickness. Figure 2-17 shows the ﬁnite element grid of
the specimens. For boundary conditions it can be assumed that at the ends of gauge section
neither x-displacement nor y-displacement exists owing to clamping of grips. In the arid-
section of the test strip (A-A plane in Figure 2-17) the nodes have the same x-displacement.
For simplicity, elastic deformation is considered in the analysis. Entering mechanical
parameters E = 50 GPa, v = 0.3 for NiTi alloys and e = 0.25,44 a contour map of the
magnitude of the maximum principal stress can be plotted by Marc® ﬁnite element analysis
package as shown in Figure 2-18. The dashed frame in the background represents the
image of the specimen before stretching.

The fractography, which will be discussed in Section 3.5, reveals a well-deﬁned

ductile fracture characteristic referred to as dimples. The fracture was formed through void

 

44 A large tensile strain was chosen in favor of illustrating deformation in each point of the specimen.

74

nucleation, growth and coalescence under normal tensile stress [Dodd and Bai, 1987]. The
sites of the fracture can be predicted from the maximal principal stress contours. As
indicated in Figure 2-18, the fracture should start at edges near the ends of the gauge
section. Figure 2-19 is a plot of a number of samples fractured vs. the normalized x-
position where the fracture occurred. It is in agreement with the prediction in spite of some

scattering which may be caused by the imperfect by cut edges of the specimens.

2.4. 8. Scanning Electron Microscopy (SEM)

The secondary electron images of cross-section of the annealed ﬁlm and the fracture
surface of tensile samples were observed in the Hitachi S-2500 SEM. The SEM was
operated at 20-25 kV at a 15 mm working distance. To examine the cross-sectional
morphology of the annealed ﬁlm, free-standing ﬁlm was mounted in epoxy resin and
ﬁnally polished using a water suspension of 0.05 um alumina powders. The polished
surface was then chemically etched in a solution consisting of 4 parts I'INO3, 2 parts H20
and 1 part I-IF by volume for 15 seconds at the room temperature. To observe the
fractography of N iTi ﬁlms, the fractured ﬁlms were mounted onto an aluminum stage using
carbon conductive tape with the fractured surface upward. To get the best image contrast,

the specimen stage was tilted in a range between 20° and 45°.

CHAPTER 3

RESULTS AND DISCUSSION

In this chapter, experimental ﬁndings and related discussion are presented. First, the
microstructures of as-sputtered and annealed ﬁlms, respectively, are presented in Section
3.1 and 3.2. The phase transformation characteristics of the ﬁlms are described in Section
3.3. Deformation behavior of NiTi ﬁlms is reported in Section 3.4. Finally, the

fractography of NiTi ﬁlms is discussed in Section 3.5.

3.1. Microstructure of As-Sputtered N iTi Films
3.1.1. TEM Observation

On examining three regions of ﬁlm H57345 (the interface, nridplane and surface), it
was found that an amorphous structure formed through the entire thickness of the ﬁlm.
Figure 3-la and 3-1b, respectively, are a bright-ﬁeld (BF) image and a corresponding
selected area diffraction pattern (SADP) sampled at the nridplane of the ﬁlm. The
nricrographs show a completely featureless structure (Figure 3-1a) and a broad, diffuse
halo diffraction-pattem46 (Figure 3-1b), which is normally observed in amorphous
materials.

Figure 3-2a and 3-2b are BF image and the SADP taken at the interface of the ﬁlm
H598. The nricrographs display the same features as shown by the ﬁlm H573, indicating

that the interface of the ﬁlm H598 has an amorphous structure. Figure 3-3a and 3-3b were

 

45 As indicated in Section 2.3, the notation ‘H---’ is used in the present work where the ‘H’ signiﬁes ‘hot
deposition’ and three digits signiﬁe the substrate temperature of deposition in Kelvin.
46 The average interatomic distance of amorphous NiTi determined from this pattern is 0.212 nm.

75

76

taken at the rrridplane of the ﬁlm H598. The SADP reveals a single-crystal diffraction
pattern in the [110],” zone, together with a superimposed diffuse halo pattern scattered
from an amorphous undelayer. Since the SADP was taken from the whole scope of the
Figure 3-3a, the initial crystallite with an in-plane size of 750 nm is a monocrystal although
the image contrast is different within the crystallite. Several such crystallites have been
examined. They were all in the [110] zone but with random in-plane orientations. It is
clear that at the substrate temperature of 598 K the crystalline NiT i phase started to nucleate
on the amorphous precursor near the rrridplane and with their (110) planes aligned parallel
to the plane of the ﬁlm. At the surface of ﬁlm H598, a very ﬁne structure of crystal grains,
95 nm in average diameter, can be seen (Figure 3-4a). Strain contrast arising from the
precipitates can be discerned (the strain ﬁeld contrast will be discussed later in this section).
The elongated grain images in Figure 3-4d obtained after tilting the specimen through 45°
suggest a columnar grain structure. Figure 3-4b and 3-4c are the associated SADP’s taken
at a tilt of 0° and 30° respectively. Using the method of determining grain orientation
described in the Section 2.4.6, a strong (110) ﬁber-texture can be veriﬁed. The non-even
intensities of the diffraction rings in Figure 3-4b reﬂect the restricted in-plane grain
orientations.

Figure 3-5a, which is taken at the interface of the ﬁlm H623, exhibits a ﬁne-grained
structure having average in-plane grain size of 85 nm and very ﬁne precipitates 12 nm in
length, also showing evidence of strain-ﬁeld contrast. Careful examination around the
perforation area of the TEM foil revealed that the layer ﬁrst deposited, which had a
thickness of less than 200 nm, is amorphous.47 Figure 3-6a and 3-6d are BF images of the
rrridplane of the ﬁlm H623 recorded at the tilt of 0° and 45°, respectively, showing the
columnar structure. At the surface of the ﬁlm H623 (Figure 3-7a), a well-deﬁned
polycrystalline structure with ﬁne precipitates can be seen. The associated SADP’s of each

layer indicate that the (110) ﬁber-texture prevails throughout the bulk of the ﬁlm.

 

47 The thickness of amorphous was estimated from the curved edge in perforated areas of TEM foil.

77

Figure 3-8a, sampled at the interface of the ﬁlm H673, shows the initial crystallites
having the average grain size of 240 nm developed from the very thin amorphous layer.
Although having low contrast, precipitates averaging 55 nm in length can be distinguished.
Figure 3-9a is BF image taken from nridplane of the ﬁlm H673. The interior of the grain
exhibits the ﬁne precipitates 20 nm in length surrounded by strain ﬁelds. The precipitates
were aligned roughly in the same directions within each grain. As indicated by the arrow
A, a perfect ‘butterﬂy’ shape of the image contrast arising from lattice nrisﬁtting strain ﬁeld
can be seen around a precipitate particle.48 Again, the columnar structure in the rrridplane of
the ﬁlm was veriﬁed by tilting the specimen (Figure 3-9d). Figure 3-10 gives the BF
image of the surface plane. The tilted SADP’s (Figure 3-8c and 3-9c) again show the (110)
ﬁber-texture.

Figure 3-11a is a high magniﬁcation image of precipitate with ‘butterﬂy’ shaped
contrast having been shown in Figure 3-9a. The particles could be indexed as the Ni4Ti3
phase, as will be discussed in Section 3.1.4. The strain ﬁelds around Ni4Ti3 particles
comes from the lattice rrrismatch between particles and B2 matrix. The habit plane of
precipitate is {111}32 and the maximum mismatch, 6,, = (d(11])an3 - dunmzydunjgz =
-2.9 % [Li and Chen, 1996], is along the normal to the precipitate disk as shown in Figure
3-1 lb [Kainuma and Matsumoto, 1988]. This minus value indicates a tensile strain
perpendicular to the precipitate disk.

A further increase of the deposition temperature above 673 K led to a dramatic change
in the nricrostructure. At interface of the ﬁlm H723 (Figure 3-12a), extremely ﬁne grains,
25 nm in diameter, were nucleated on very thin initial amorphous layer estimated to be 20
nm in thickness. The unchanged SADP’s upon tilting implies the absence of texture
(Figure 3-12b, 3-12c). A BF image from the rrridplane (Figure 3-13a) shows grains with
clear and smooth boundaries and the lenticular-shaped precipitates with slightly wavy

 

43 Ashby and Brown [1963] have predicted the image proﬁles for particles with spherically symmetrical
strain ﬁelds. The proﬁles displayed the characteristic ‘butterﬂy’ shape of the image contrast.

78

boundaries and longitudinal size comparable to the matrix grain sizes. Neither strain
contrast nor misﬁt dislocations was found in the regions surrounding the precipitates. A
random grain orientation also appears in the rrridplane (Figure 3-13b, 3-13c). The surface
plane of ﬁlm H723 exhibits a rrricrostructure similar to that of the nridplane except for
somewhat larger grains and smaller precipitates (Figure 3-14). Three ranges of precipitates
can be seen at the lower part of the micrograph. (The growth of precipitates will be
discussed in Section 3.1.4.) Figure 3-15 shows micrograph and associated SADP taken at
the nridplane of the ﬁlm H703 revealing crystalline B2 grains and transgranular precipitates
which are similar to that of ﬁlm H723.

The microstructures and the grain orientations in the three layers of the ﬁlm H773
(Figure 3-16, 3-17 and 3-18) are essentially the same as those of the ﬁlm H723 except for
slightly larger grains and precipitates (270 nm vs. 250 nm for B2 grains, and 220 nm vs.
150 nm for precipitates). Finally, the NiTi mauix and precipitate morphologies of as-

deposited NiTi ﬁlms are summarized in Table 3-1.

3.1.2. Thin Film Microstructure: Discussion

The TEM observation presented in the preceding section revealed that a fully
amorphous NiTi ﬁlm was produced at the deposition temperature of 573 K (ﬁlm H573).
As discussed in Section 1.5.2, amorphous ﬁlms are generally formed under conditions
where the surface mobility of arriving species is very limited. In interrnetallic compounds
the requirement for coordinated adatom motion to form an ordered crystalline lattice
increases the tendency to form amorphous structure. As a result, crystalline NiTi ﬁlms can
only form at high enough temperatures. In the present work, the NiTi crystals started to
nucleate at a deposition temperature of 598 K (ﬁlm H598), and a fully crystalline NiTi ﬁlm
was produced at a deposition temperature of 623 K (ﬁlm H623).

NiTi crystallites were observed to nucleate on an initially amorphous layer during hot

substrate deposition (T> 598 K). Generally, the resultant grain size of the initial

79

polycrystals is controlled by three kinetic phenomena: crystal nucleation, growth and
coalescence [Thompson, 1994]. The grains nucleate and grow on the untransformed
material (the amorphous phase) until they ﬁll space by impinging on one another, followed
by grain boundary motion.

For nucleation and growth processes, if crystals are assumed to nucleate at a constant
rate, I, at random locations, and to grow isotropically at a constant rate G, the well-known
Johnson-Mehl structure results [Johnson and Mehl, 1939]. Mahin, Hanson and Morris
[1980] analyzed the Johnson-Mehl model through computer simulation and presented the
growth sequence leading to Johnson-Mehl structure as illustrated in Figure 3-19. It has
been shown that the ﬁnal average in-plane grain diameter D is given by [Gilbert, 1962]:

D = 1.4486?)3 (3.1)
It can be seen from this equation that a higher nucleation rate and/or lower grain growth rate
produce smaller ﬁnal grain size.

From Figure 3-3, 3-8 and 3-12, it is apparent that the in-plane grain size of the initial
crystalline phase rapidly decreases as the deposition temperature is increased. The
nucleation rate must therefore increase monotonically as the temperature increases, since
both the grain growth and the coalescence are thermally activated processes and would not
be expected to decrease as the temperature increases. Furthermore, the nucleation rate must
grow faster than grain growth and the coalescence. The higher nucleation rate of N iTi
crystals at the higher temperature is also consistent with the TEM observation that a thinner
initial NiTi amorphous layer is formed at higher deposition temperatures.

The specimen tilting method indicated a columnar grain structure in ﬁlms H598,
H623 and H673. Such columnar structures are commonly found in thin ﬁlms formed in
physical vapor deposition (PVD) at relatively low substrate temperatures. The grain

growth in polycrystals involves grain boundary migration which results from diffusive

80

jumps of atoms across the grain boundary. In normal grain growth, the curvature of the
grain boundary provides the driving force which minimizes total grain boundary area.
Based on experimental results discussed further in Section 3.2.1, the ﬁne columnar grain
structure of NiTi ﬁlms was found to be stable in a one-hour, 858 K post-deposition
annealing process. This fact suggests that little bulk diffusion for signiﬁcant grain
boundary migration would occur in the deposition time/temperature range of the present
work (T S 773 K). However, the activation energy for surface diffusion is lower than that
for bulk diffusion. Consequently, even when the temperature is too low for signiﬁcant
diffusion in bulk, the diffusion necessary for grain growth during deposition can occur
readily at a free surface.

Srolovitz [1986] coupled grain boundary migration and thin ﬁlm growth and
proposed a statistical model using a Monto Carlo approach for the evolution of grain
structure during deposition. The columnar grain structure formed during deposition can be
interpreted as follows: Within a certain temperature range, the structure below the free
surface is frozen-in because bulk diffusion is sluggish and two—dimensional grain growth
develops only at the free surface of growing ﬁlm. By superimposing two-dimensional
grain growth upon ﬁlm growth (increasing thickness), a columnar structure can be formed
as shown in Figure 3-20. The shape of the columnar grains is controlled by relative
magnitudes of in-plane grain growth rate and the deposition rate.

The columnar grain structure in ﬁlms H723 and H773 could not be veriﬁed by the
tilting method due to large in-plane grain size. However, as mentioned earlier, signiﬁcant
bulk diffusion did not occur in a one hour, 858 K annealing process. It is thus reasonable
to conclude that very little bulk diffusion took place in sputter processes even at substrate
temperatures of 723 K and 773 K. Thus, the columnar grain structure could also be
expected.

Comparing Figure 3-16, 3-17 and 3-18, which were taken from the interface,
midplane and surface of the ﬁlm H773 respectively, in-plane grains grew rapidly in ﬁrst

81

half of the growth period (from D32 = 30 nm to 032 = 270 nm), then the grain growth
slowed down in the second half (from 032 = 270 nm to D32 = 320 nm). This fact can be
explained as follows: The growth rate of the mean diameter of the grains is proportional to
the curvature of the grain boundaries. When grains grow larger, total areas of grain
boundaries and thus grain boundary energy becomes smaller and the driving force

decreases.

3.1.3. Formation of Texture

It has been shown that the ﬁrst crystalline grains to fornr nucleated from the
amorphous layer, and that (1 10) texture developed in the ﬁlms H598-H673, deposited in
the temperature range for which surface diffusion dominates. This means that the atoms in
the surface have the ability to arrange themselves into the lowest energy conﬁguration.
Consider the disk-shaped initial crystallites having the thickness within which the atoms are
able to move via the surface diffusion and having the major diameter of the in-plane grain
size. On one side the crystallites are in contact with the amorphous NiTi and on another
side they are in contact with the vacuum environment. Because the interface and surface49
energies of the crystals depend strongly on the crystallographic face, the different
orientation of the crystallites will lead to different interface and surface energies
[Thompson, 1990].

For the ﬁlms deposited at higher temperatures (T 2 723 K), the initial grains impinge
quickly to form extremely small in-plane grain size (DB, = 25 nm) due to the high
nucleation rate. In this case, the surface and interface energies for such crystallites are less
signiﬁcant because the interface and surface areas of the disk are a relatively small fraction

of the total boundary area. The driving force arising from the surface energy anisotropy is

 

49 The boundary between the gas phase and the solid or liquid phases is refereed as the surface. The interface
is the interphase boundary. In this sense the surface is also the interface but a special interface.

82

not high enough to align the initial grains to the preferred orientations. The ﬁlms with
random grain orientations are produced as schematically illustrated in Figure 3-21a.

For ﬁlms deposited at lower temperatures (T S 673 K), the initial grains have in-
plane grain size greater than 85 nm due to a decreased nucleation rate. The interface and
surface energies are important because these disk-shaped crystallites have considerable
areas in the interface and surface, and grains with orientations that minimize total surface
and interface energy are favored. Consequently, (110) texture develops in hot-deposited
NiTi ﬁlms. Moreover, the amorphous substrate and vacuum environment have no interface
and surface energy anisotropy for in-plane grain rotations, ﬁlms formed at lower
temperatures present a (110) ﬁber-texture, i.e., the (110) planes of grains are restricted to
being parallel to substrate plane but with no preferred in-plane orientation. This
circumstance can be schematically illustrated in Figure 3-21b.

The ﬁne grains (032 = 95 nm) displayed on the surface of the ﬁlm H598 (Figure 3-4)
nucleated from the initial crystalline grains having large in-plane size of D32 = 750 nm
(Figure 3-3). Locally, the initial large crystalline grains can be considered as single-crystal
substrates for the ﬁne grains nucleated on them. As mentioned by Thompson [1990], for
polycrystalline ﬁlms formed on single-crystal substrates there should always be a tendency
to form a three-dimensionally constrained relationship, i.e., epitaxial orientations between a
grain and the substrate in order to minimize the interface energy. In Figure 3-4b, the
selected areas diffraction rings having non-uniform intensity, reﬂect a constrained in-plane
grain orientation. The reasons that ﬁne grains re-nucleated from initially crystalline grains
are not clear, however, it may be that the surface diffusion at 598 K is be sufﬁcient for

grains continuously growing at large scales.

3.1.4. Growth of Precipitates
The morphologies of precipitate observed in the hot-deposited ﬁlms are of two types:

the ﬁne precipitates surrounded by strain ﬁelds as shown in the Figure 3-9a, and coarse

83

transgranular “wavy plates” as shown in Figure 3-17. In order to index the ﬁne
precipitates, the SADP taken from the midplane of the ﬁlm H673 (Figure 3-9b) was
magniﬁed as shown in Figure 3-22. The (110),” ring in the diffraction patterns was used
as an internal standard for calibration. Within the variation of 2 %, each diffraction ring in
the patterns could be indexed by assuming either the B2 structure of the NiTi with a, =
0.3015 nm [Philip and Beck, 1957], or the rhombohedral Ni4Ti3 structure with a0 = 0.6610
nm and a: 113.65° [Nishida, Wayman, Kainuma and Honma, 1986]. The indices of
Ni4Ti3 phase were underlined in the Figure 3-25 for a purpose of distinguishing them from
the B2 reﬂections. It is noted that the ring patterns in the Figure 3-22 conﬁrm the (110)
grain orientation since the (310)32 reﬂection having a radius of 64.3 mm is absent.50

For the coarse precipitates formed at the rrridplane of the ﬁlm H773 (Figure 3-17),
two SADP’s were recorded using the smallest SAD aperture in Jeol 2000 FX TEM, as
shown in Figure 3-23. The SADP’s were sampled from two areas of the foil each
approximately including one grain. The strong reﬂections in the SADP’s can be indexed
using the B2 structure of the NiTi, while the weak reﬂections (the indices are underlined)
can again be indexed using the rhombohedral Ni4Ti3 structure. The Figure 3-23a shows
[13'3'],,// [1 11],,,,, zone and the Figure 3-23b shows [1'3 '21,, // [1 101m, zone.
Some diffraction spots that do not belong to these zones were reﬂected from partially
sampled neighboring grains. The orientation relationships between Ni4Ti3 precipitates and

the B2 matrix can be converted to:

(T01)Nitri3// (351)82a I0101Ni4ri3 // [0151132

for the zone [133’],2 in Figure 3-23a, and

(OOT)N14Ti3// (42mm, I1001Ni4r13” [0151132

 

50 If the lattice planes {hkl} are reﬂected in the image plane (Eward sphere) having the zone axis (uvw), the
scalar product of the zone axis with each individual reﬂection {th} has to be zero, i.e., uh + vk + wl = 0
[Heimendahl, 1979].

84

for the zone [132],” in Figure 3-23b. The above orientation relationships were ﬁrst
proposed by Nishida and Wayman [1987]. The boundaries between precipitate and matrix
are very clear and no boundary dislocation or strain ﬁeld contrast can be seen. It can be
speculated that the precipitates have lost the coherency during ripening. However, the
observed orientation relationships between the Ni4T13 precipitates and B2 matrix, which
were established during the nucleation phase of the precipitation, are seen to persist.

Since as-sputtered ﬁlms have overall composition of 51.0-52.3 at.% Ni, the ﬁlms are
hyperstoichiometric in NiTi. During the hot-substrate deposition, Ni-rich particles will
precipitate in the ﬁlm to reduce the free energy of the matrix. On the other hand, the
formation of the second phase creates additional interface energy. There is thus always a
tendency to increase the size of the individual particle but decrease the number of particles.
The diffusion of Ni atoms makes it possible for Ni-rich precipitate growth to continue.

Several investigations have shown that the Ni4Ti3 phase readily precipitates through
bulk diffusion in solution-treated Ni-rich N iTi alloys at an annealing temperature as low as
673 K [Nishida and Wayman, 1984; Wu, Lin and Chou, 1989; Xie, Zhao and Lei,
1989]. The TEM rrricrographs of the ﬁlm H723 give evidence for growth of Ni4Ti3 phase
in the ﬁlms through the bulk diffusion. The sizes of the NI4TI3 phase is 150 nm in the
rrridplane (Figure 3-13) but 65 nm in the interface (Figure 3-13). The precipitates in the
nridplane thus apparently coarsened during deposition.

At a given temperature surface diffusion proceeds more rapidly than dose bulk
diffusion. It is expected that the precipitates nucleated and grew also via surface diffusion.
Ni4Ti3 precipitates having a considerable size are observed in the surface of the ﬁlms H723
(Figure 3-14) and H773 (Figure 3-18). The particles must have developed during the
deposition, since after shutting off power to the sputter source and substrate heaters the
ﬁlm was cooled down to below 573 K (where no signiﬁcant diffusion is assumed) in only

few nrinutes.

85

Similar to the grain growth of NiTi, the growth rate of precipitates increases as the
substrate temperature increases. The BF images of the rrridplane of ﬁlms H673 (Figure 3-
9) and H723 (Figure 3-13) show that the larger particles in smaller numbers were produced
at the higher substrate temperature where the rate of diffusion is greater.

It is interesting to note that the precipitates did not nucleate within extremely ﬁne
initial grains (D32 = 25 nm) even at the deposition temperature of 723 K, as shown in
Figure 3-12 (ﬁlm H723). There are a few possible ways to accommodate the extra Ni
atoms in the ﬁlm: forming supersaturated NiTi grains, increasing solubility in grain
boundaries and creating a N iTi-Ni4Ti3 dual phase. However, the detailed structure is not
clear yet at this point due to the difﬁculty of deconvoluting the precipitate reﬂections from
the B2 reﬂections.

Experimental observation allows the size changes of B2 grains and precipitates as a
function temperature in the three layers to be plotted as in Figure 3-24. There was a rapid
change in B2 grain size in the ﬁrst half of deposition course. Then B2 grain size only
increased slightly in the second half of deposition course. It seems that the deposition
temperature had strong effects on the nricrostructure of the initial crystalline layer. Figure
3-25 is a proposed structural model for Ni-rich NiTi ﬁlms sputter—deposited at the
temperature of 573-773 K based on TEM observation conducted in this study.

It is likely that the growth of ordered B2 grains involves the cooperative movement of
both nickel and titanium atoms across the grain boundary, whereas the growth of Ni4Ti3
phase in the Ni-rich matrix mainly relies mainly on the movement of nickel atoms. The
diffusivity of nickel in B2 NiTi is approximately an order of magnitude greater than that of
titanium [Bastin and Rieck, 1974]. In amorphous phase, titanium is also thought to be a
relatively slower diffuser [Chang, 1993]. For the Ni-rich NiTi system, a possible resultant
ﬁlm structure would be determined by the deposition temperature, and the active regions

for surface and bulk diffusion of Ni and Ti. Figure 3-26 schematically represents such

86

relationships.51 In Region I, the surface diffusion of Ni and Ti are not active
simultaneously. Films deposited at this region are generally amorphous in structure.
However, in sub-region 12, hyperstoichiometric precipitates may form via Ni surface
diffusion in amorphous phase. In Region 11, ﬁlms consist mainly of columnar B2 grains.
In sub-region 112, large precipitates may be produced through Ni bulk diffusion and B2/ppt
dual-phase nricrostructures similar to that of annealed ﬁlms Al-A4 which will discussed in
the next section may form. In Region III, bulk diffusion for both Ni and Ti dominate.

Films with rrricrostructures similar to that of fully annealed and aged ﬁlms may result.

3.2. Microstructure of Annealed NiTi Films

In this study, the microstructures obtained from hot-deposition at below 773 K
basically consist of ﬁne columnar B2 grains and ﬁne Ni4Ti3 second phase distribution.
According to Ni-Ti phase diagram (Figure 1-10), Ni will redissolve at T> 900 K. An
additional anneal to as-sputtered ﬁlms at higher temperature (> 773 K) can provide a variety
of additional combinations of grain and precipitate due to the activation of bulk diffusion.

In this section, the nricrostructures of annealed NiTi ﬁlms are analyzed.

3.2.1. Morphology

The ﬁrst set of heat treatment experiments was conducted at 858 K for 1 hour.
Figure 3-27, 3-28 and 3-29 are TEM bright-ﬁeld images of ﬁlms A152, A2 and A3 which
were produced by aging the ﬁlms H623, H673 and H723, respectively. They all show a
dual-phase nricrostructure consisting of ﬁne grains and coarse platelike precipitates. It
appears that during the annealing the ﬁne precipitates inside grains coalesced into large

independent plates (compare the original microstructures in Figure 3-6, 3-9 and 3-13).

 

5‘ The temperature boundaries for each regions in Figure 3-26 were estimated from the present work. Since
diffusion is controlled by both temperature and time, a given rnicrostucture may be formed either at lower
temperature after longer time or at higher temperature after shorter time.

52 As indicated in Section 2.3, for annealed NiTi ﬁlms, the notation ‘A-’ is used where the ‘A’ represents
‘annealing’ and a digit designates the detailed heat-treatrrrent path as indicated in Table 2-2.

87

However, the grain structure of the B2 NiTi phase remains nominally unchanged since
neither in-plane grain size nor columnar grain structure has increased (Figure 3-28b). The
ﬁlm texture, of course, should not have been altered in the heat treatment, as is veriﬁed by
Figure 3-28c, 3-28d, 3-28e and 3-29b.

To explore the nrinimum temperature for forming a dual-phase structure, the ﬁlm A4
was produced at an annealing temperature of 813 K. Film A4 also displays the same dual-
phase morphology (Figure 3-30). It can be concluded that at the annealing temperature of
813-858 K diffusive processes for B2 grain boundary motion have not become signiﬁcant,
whereas bulk diffusion for precipitate growth was quite rapid.

Figure 3-31 is an image of the ﬁlm A5 obtained by annealing ﬁlm H673 (Figure 3-9)
at 988 K for 1 hour. The size of B2 grains increased from 0.22 um to 0.64 um during the
annealing. Obviously, bulk diffusion for NiTi grain growth has occurred. Large lenticular
precipitates 180 nm in length started to form within grains. According to the NiTi phase
diagram proposed by Wasileski et al. [1971] (Figure 1-10b), at 988 K, nickel in excess of
50 at.% should have dissolved in the B2 matrix (the composition of A5 was Ti-51.0 at.%
Ni), and the formation of precipitates must have occurred on cooling from 988 K to room
temperature. For larger B2 grains (diameters > 0.5 um) like those in ﬁlm A5 and other Ni-
rich ﬁlms reported in literature [Nishida and Honma, 1984; Xie, Zhao and Lei, 1989],
there seems a tendency for the precipitates to form inside the grains than outside grains
(i.e., B2/ppts dual phase) in thermal treatment.

Figure 3-32 shows the microstructure of the ﬁlm A6 produced by annealing ﬁlm
H673 at 1123 K for 1 hour. The ﬁlm consists of large grains 2-5 um in diameter and large
lenticular precipitates about 1 mm in longitudinal direction.53 The precipitates align along

well-deﬁned crystallograghic planes of the matrix.54 In addition, very ﬁne precipitate

 

53 The sizes of grains and precipitates in the ﬁlms A6-A8 were measured from the rrricrographs having
lower magniﬁcation (10,000 x). For a purpose of direct comparison, the majority of BF images in this
work were prepared at the same magniﬁcation (165,000 x).

54 The formations of plate- or needle-shaped precipitate particles in such a manner that they are aligned
along speciﬁc crystallographic planes is called widmanstdtten structures [Reed-Hill, 1964].

88

particles with strain ﬁeld contrast can be noticed in the area between the large precipitates.
A band of matrix free of precipitation was left on each side of large precipitates due to the
depletion of nickel atoms in these areas.

According to the TIT diagram (Figure 1-11) proposed by Nishida, Wayman and
Honma [1986], any Ni4Ti3 particle should be dissolved at a temperature above 1023 K.
The binary NiTi phase diagram indicates that solutionization should occur at 903 K. Film
A7 was produced by annealing ﬁlm H673 at 1073 K for 2 hours to form single NiTi phase
and then subjected to radiative cooling. Figure 3-33 is the BF image of the ﬁlm A7
showing the microstructure consisting of coarse grains (1.5 um) and ﬁne precipitates. The
ﬁne particles probably grew in the course of the cooling because it took about 5 minutes to
cool the ﬁlm to below 623 K. Precipitation tends to occur more rapidly at grain boundaries
since the grain boundary is a favorable site for heterogeneous nucleation. Development of
precipitation in grain boundaries depleted the Ni atoms in the vicinity of the boundaries.
The coarse grain structure in the ﬁlm A7 is also conﬁrmed in a cross-section SEM
micrograph as shown in Figure 3-34. Deep etched surface morphology reveals the
crystallographic facets of the grains.

Film A8 was processed under the conditions similar for that of the ﬁlm A7 except for
an additional 673 K, 1 hour aging treatment. As seen in Figure 3-35, the grain size is
about the same as that of ﬁlm A7 because 673 K was too low for grain growth via bulk
diffusion, but the precipitate particles had ripened. This microstructure is similar to that of
Ni-rich bulk NiTi alloy formed at low temperature (673-723 K) aging after solution
treatment [Nishida et al., 1986; Xie, Zhao and Lei, 1989].

3.2.2. Structure of Precipitates in Annealed Films
The lattice structure of precipitates in annealed ﬁlms were characterized using either
electron diffraction in TEM or the XRD method. Films A1-A4 have the same dual-phase

structure. The precipitate plate in ﬁlm A3 was selected for indexing lattice structure

89

because its greater size. Figure 3-36b and 3-36c are two single-crystal SADP’s taken from
the region indicated in Figure 3-36a. So as to exclude diffraction spots from the BZ matrix,
the SADP’s were taken on an isolated precipitate plate situated in the edge of the TEM foil.
The precipitate can be indexed by assuming Ni3T12 phase having a monoclinic cell with
constants: a0 = 0.441 nm, b0 = 0.882 nm, co = 1.352 nm and 7: 893° [Nishida and
Wayman, 1987]. As a self-consistency check, the magnitudes of the tilt angle between the
[O 10] zone in Figure 3-36a and the [110] zone in Figure 3-36b, and the theoretical angle
between these two zone axes for N i3Ti2 are compared. The former was measured to be
29.5°, while the latter is calculated from crystallographic data to be 26.4°. About 10 %
deviation may be caused by improperly adjusted manipulator of double-tilting stage and/or
imperfect equalization of diffracted spot intensity at 0° tilt. Thus it is concluded that the
precipitates in ﬁlm A3 are the Ni3Ti2 phase.

Figure 3-37 shows the XRD curve of ﬁlm A5 recorded at room temperature (298 K).
Besides the (110) B2 Bragg peak, four peaks match the reﬂections of Ni4Ti3 phase in 20
angle between 35° and 60° within 2 % error. This indicates that the large lenticular
precipitates formed within the grains are Ni4Ti3 particles.

Indexing of large lenticular precipitates in ﬁlm A6 could be completed by assuming
that they possess the NI4TI3 structure. Figure 3-38b and 3-38d are SADP’s of zone
[0 I2]32 // [001]Nim3 and zone [230]32 // [221]Nim3, respectively, taken from areas
indicated in Figure 3-38a and 3-38c. The underlined indices represent the diffraction from
the precipitates.

Figure 3-39 shows the XRD curve of the ﬁlm A7. In order to reduce the complexity
of the curve, the NiTi film was scanned at 353 K, which is about 40 K higher than its A555

to eliminate the reﬂections from the R-phase of NiTi. The spectrum shows the (110) B2

 

55 The characteristic temperatures associated with the phase transformations will be represented in the
Section 3.3.

90

peak and two Ni4Ti3 phase peaks in 20 angle between 35° and 60°. The reﬂections of
Ni4Ti3 phase are very weak due the small particle size (20 nm in length).

The XRD curves of the ﬁlm A8 are shown in Figure 3-40. Curves (a) and (b) were
taken from the same specimen at 353 K and room temperature respectively. The curve (a)
identiﬁes the (110) B2 peak and ﬁve Ni4Ti3 peaks in 26 angle between 35° and 60°. In
curve (b), the same set of the Ni4Ti3 reﬂections appears, however, the (110) B2 peak splits
into two NiTi R-phase peaks: (011) and (011) with a separation about 0.4°. This is due to
the B2 —) R-phase transformation when the temperature is below TR. Ling and Kaplow
[1981] ﬁrst found this interesting character in Ni50,13Ti49_96 bulk alloy. Recently, Nam
[1996] reported similar XRD spectra in a Ni50_3Ti49,7 thin ﬁlm specimen.

The TTT diagram of Ni52Ti4g bulk alloy shown in Figure 1-11 [Nishida, Wayman
and Honma, 1986] veriﬁes that compared with Ni4Ti3, Ni3Ti2 phase forms after longer
aging process at higher temperature. The Ni3Ti2 phase is closer to the thermodynamic
equilibrium state, but there is a higher energy barrier for its formation. In Ni52T14g bulk
alloy, it would take over 100 hours for the completion of Ni4Ti3 —) Ni3Ti2 transition at 858
K. However, at the same temperature it only took one hour to transform Ni4Ti3 phase into
Ni3Ti2 in the ﬁlms Al-A4. The reasons for enhancing the formation of Ni3Ti2 precipitates
are not fully understood. Nevertheless, one pronounced difference is the grain
morphology. In Nishida, Wayman and Honma’s work [1986] the grain size were greater
than 100 um and the Ni3Ti2 phase grew within grains and along speciﬁed crystallographic
planes. However, in the present work the in-plane grain size of ﬁlms Al-A4 is only about
0.1-0.3 um and N i3Ti2 particles formed in dual-phase structure. It is likely that the
interface energy of the second phase and the matrix or parent grains played an important
role in the precipitate formation.

As a summary, the in-plane views of various rrricrostructure formed in the post-

deposition annealing are schematically represented in Figure 3-41.

91

3.3. Phase Transformation Characteristics
3.3.1. Phase Transformations in As-Sputtered Films

In NiTi-based systems, many measurable physical values have been utilized to study
the martensitic and R-phase transitions, such as electrical resistivity (p), enthalpy (H),
internal friction (Q1) [e. g. Hasiguchi and Iwasaki, 1968; Wu, Lin and Chou, 1990], and
magnetic susceptibility (x) [Loloee, 1995]. Among them measuring electrical resistivity is
most widely used because the simplicity of required equipment. For this reason the
resistimetry was selected, together with the DSC (for measuring transformation AH), as
techniques for determining transformation temperatures in the present work. Before
discussing the experimental results, it is useful to discuss some general features of
resistivity vs. temperature curve related to the phase transformation in NiTi system. By
examining a large number of resistivity vs. temperature curves in the literature, it is found
that:

(1) Compared with the parent phase and martensite, the R-phase has the highest
electrical resistivity. During the R-phase transformation, Ap/AT becomes negative.

(2) The slope of the curve in martensite region is much steeper than that in austenite
and R-phase regions.

(3) The hysteresis of R H B2 transitions is only about one tenth of that of M H R
and M H B2 transitions.

Figure 3—42 schematically illustrates the classiﬁcation of electrical resistivity vs.
temperature curves for NiTi. Curve (a) shows a two-stage transformation sequence
involving M H R H B2 phase transformations as discussed in the Section 1.2. 1. Curve
(a) can be decomposed into curves (b) and (c), which involve M H R and R H B2
transformations respectively. The curve (b) clearly shows that the characteristic
temperatures A, and Af, respectively, indeed represent the start and ﬁnish temperatures of
M -> R transformation. Curve ((1) is a two-stage curve but with an incomplete B2 —> R

transition. The martensitic transformation starts within the R-B2 mixture before the

92

completion of the R-phase transition. Curve (e) shows the R H B2 transformation which
is essentially the same as the curve (c). This kind of curves can be frequently seen in the
literature [e.g. Edmonds and Hwang, 1986; Stachowiak and McCormick, 1988; Wu, Lin
and Chou, 1989; Xie, Zhao and Lei, 1989]. Curve (f) reﬂects the M H B2
transformation. Similar curves have been reported for the Ni-Ti-Cu ternary system [e. g.
Edmonds and Hwang, 1986; Chang and Grummon, 1992].

Figure 3-43 shows the electrical resistivity vs. temperature curves for several as-
deposited ﬁlms. Film H57 3 exhibits a negative temperature coefﬁcient of resistivity in the
temperature range of measurement (140-380 K).56 This is a typical resistivity-temperature
behavior of amorphous N iTi as reported by Ikuta et al. [1990], Nam [1995], and as
discussed in previous crystallization annealing trial in this study (Figure 2-6, Section
2.2.1). The amorphous structure of the ﬁlm H573 was conﬁrmed by the TEM observation
(Section 3.1.1).

The resistivity curve of ﬁlm H623 shows that the M, temperature was severely
depressed to below the measurement range and only the R H B2 transition occurred. This
is the case of curve (e) in Figure 3-42. The curves of this sort have been reported for Ni-
rich (2 50.5 at.% Ni) bulk N iTi alloys aged at 673-773 K for 1-2 hours after solution
treatment [Xie, Zhao and Lei, 1989; Wu, Lin and Chou, 1989]. In the early stage of
precipitation, the nickel atoms have not been fully depleted and the matrix has high nickel
content which strongly suppresses M,. Also, small precipitates still remain coherent with
parent matrix. The combination of a higher Ni-content matrix and ﬁnely dispersed
precipitates with coherency strain ﬁelds in the ﬁlm is responsible for depressing the M,
temperature.

Films H673, H723 and H773 all display classical M H R H B2 two-stage

transformation sequence commonly observed in Ni-rich bulk alloys aged at 673-773 K for

 

56 The magnitude of the resistivity of ﬁlm H573 in Figure 3-43 has been compressed three times with
respect to other curves.

93

_>_ 5 hours. Compared with H723 and H773, the H673 has highly dispersed fine
precipitates with evidence of strain contrast. The N i-content of the matrix may still have
been hyperstoichiomeuic since the precipitates were small. Consequently, the M, of ﬁlm
H673 was more depressed than that of other two ﬁlms. The ﬁlms H723 and H773 have
similar microstructures, but the latter has a slightly higher Ni-content and thus displays
lower M, temperature.

A cold—stage TEM experiment was conducted to gain a better understanding of the
effect of precipitates on martensite transformation. Figure 3-44 shows bright-ﬁeld images
of the nridplane of H673 and H723 taken at 101 K which is well below their M,
temperatures. In the ﬁlm H723 (Figure 3-44a), martensite twins were well-deﬁned and
spread over the entire grain regardless of the array of precipitate plates. This implies that
the ripening of precipitates relaxes the lattice mismatch stress in particle/matrix interface. In
contrast, the martensite twins in the ﬁlm H673 (Figure 3-44b) were conﬁned in small areas
due to the stress ﬁeld imposed by ﬁnely dispersed precipitates. As a result, the
microstructure of H723 was more favorable for the formation of martensite thus led a
higher M, temperature. The phase transformation temperatures of as-deposited ﬁlms are
tabulated in the Table 3-3.

Both DSC and electrical resistimetry were performed to study the transformation of
film H703. In the DSC scan shown in Figure 3-45a, initial scanning from room
temperature to 373 K revealed an endothermic reaction at 314 K associated with the R-
phase transformation having an enthalpy, AHR-M, of 2.8 J/g. In the subsequent cooling
scan, two well separated exothermic reactions were found with enthalpies of -3 .0 J/g at 310
K and -4.1 J/g at 251 K, which are identiﬁed as AHA-’R and AHR"’M respectively.

Reasons for the low magnitude of the latter datum are unclear, but on a second heating scan
from 223 K to 373 K (indicated by double arrows), an endothermic peak at 310 K was

found and is associated with the reversion enthalpy, AHM-M, of 17.8 J/g. A small

94

shoulder appears on this peak which coincides with AHR-M on the heating scan, indicating
that the overall peak is a convolution of the M —) R and R —> A transformation enthalpies.
The value of AHM-tk can be calculated, AHM-W = Mill—V| - AH‘HA = 15.0 J/g. Electrical
resistivity curves are plotted in Figure 3-45b. The transformation temperatures determined
from the electrical resistimetry are in good agreement with that of DSC results. Table 3-4
summarizes the transformation temperatures and enthalpies determined from above

experiments.

3.3.2. Phase Transformations in Annealed Films

Films A1-A3 all displayed a similar NiTi-NigTiz dual-phase structure. As expected,
they possess the same general type of electrical resistivity behavior (curves (1, b and c,
Figure 3-46). This type of curve has been deﬁned in the Figure 3-42d as indicating that the
martensitic transformation starts prior to the completion of R-phase transformation.
Normally, both solution-treated specimens irrespective of the composition, and aged
equiatonric specimens reveal this characteristic [e.g. Miyazaki, Igo and Otsuka, 1986;
Miyazaki et al. 1982]. It is believed that the absence of precipitates in NiTi grains of the
ﬁlms Al-A3 promoted the martensitic transformation. The compositions of B2 grains in
these ﬁlms was very close to equiatonric since the M, temperatures (= 271-282 K) are very
close to those reported for solution-treated N1493T1502 (3 280 K) [Miyazaki et al., 1982].
The platelike Ni3Ti2 particles residing between matrix grains have no effect of depressing
the M, temperature. Film A1 has a greater Ni-content than that of ﬁlms A2 and A3 (52.3
at.% vs. 51.0-51.1 at.%), however, they all have almost the same M, temperatures. The
greater volume of precipitates in the ﬁlm A1 (Figure 3-27) substantiates that the extra Ni
atoms migrated into the Ni3Ti2 phase and maintained the composition of matrix at a

relatively constant value. The unique structure of ﬁlms Al-A3 provides a potential

95

advantage to NiTi ﬁlms such that the M, temperature is independent of the composition in a
certain range.

The resistivity curve of the ﬁlm A5 (curve d, Figure 3-46) is similar to the curves of
ﬁlm Al-A3 as just discussed. The feature that the martensitic transformation starts prior to
the completion of R-phase transformation is primarily due to the depletion of the extra Ni
content in the matrix during the formation of coarse Ni4Ti3 particles.

Film A6 displayed the two-stage transformation behavior (curve e, Figure 346),
although the composition of the matrix should have been near-equiatonric state due to the
formation of large lenticular precipitates. The ﬁne coherent particles with local strain ﬁelds
are thought to lower the M, temperature. The changes in resistivity for A —> R and R —9 M
transitions on cooling are very slow. This implies that the A -> M transition proceeded in a
broad temperature range. It is possible that two regions, the region of precipitate-free
bands near the large precipitates which may have higher transformation temperatures and
the region having ﬁne precipitates which may have lower transformation temperatures,
consequently transformed into the martensite on the cooling.

Film A7 showed ill-deﬁned transformation behavior (curve f, Figure 3-46). The M,
temperature could not be unambiguously determined from the curve. The strain ﬁeld
surrounding each particles apparently impeded the transformations on cooling.

Both the microstructure and electrical resistivity scan (curve g, Figure 3—46) of ﬁlm
A8 are similar to that of bulk Ni-rich materials aged at 773 K after solution treatment
[Saburi, Tatsurrri and Nenno, 1982]. The dispersed lenticular precipitates with a moderate
size enhances the R-phase transition. The rapid change in resistivity resulted from the
uniform microstructure which allowed the ﬁlm to transform in a narrow temperature

interval.

96

3.4. Deformation Behavior of N iTi Films

Therrnomechanical properties determine the usefulness of NiTi thin ﬁlms as a stress—
strain-temperature function material. In this section, stress-strain curves made at various
temperature are discussed. Then correlation of measured thermoelastic parameters are
examined by Clausius-Clapeyron equation. Finally, transformational superelasticity is

discussed in terms of stress-temperature diagrams.

3.4.1. Ultimate Elongation of Ni Ti Films

Figure 3—47 shows ultimate elongation curves of ﬁlms H673, H723 and A3 at a test
temperature of 423 K. Initially, the ﬁlms were in the B2 state since Af temperatures for
these three ﬁlms are 309, 313 and 307 K, respectively. Each ﬁlms display a yielding at a
stress over 800 MPa. It is interesting to know whether the yielding was caused by stress-
induced martensitic transformation or caused by slip in B2 lattice. According to Clausius-
Clapeyron relations (as will be discussed in Section 3.4.4), the critical stresses to induce
martensite at 423 K (#333,) would be 1250, 1035 and 992 MPa for H673, H723 and A3
respectively. Clearly, the yielding in Figure 3-47, which occurred below the
corresponding critical stresses to induce martensite for each ﬁlms, was caused by
signiﬁcant permanent deformation in the parent phase. In other words, ﬁlms H673, H723
and A3 at 423 K behave like conventional materials.

Figure 3-48 shows ultimate elongation curves of the same set of ﬁlms at a test
temperature of 323 K. All curves show three well-deﬁned stages: ﬁrst the elastic
deformation in the parent phase, then a plateau associated with the parent-to—martensite
transformation, and ﬁnally the elastic deformation of the martensite. The curves did not
show signiﬁcant yield in the martensite phase before fracture at stresses above 1000 MPa.
Figures 3-47 and 3-48 indicate that the martensite yield point is greater than that of the
parent phase. It also can be seen in the Figure 3-48 that the Young’s modulus of austenite

is 2-3 times greater than that of martensite. Similar result was reported in literature [e. g.

97

Shaw and Kyriakides, 1995]. The elastic modulus are functions of two things: the nature
of atoms and their lattice structure [Smith, 1956]. In this case, austenite and martensite
have the same atoms but different lattices and thus result in a great change in elastic

modulus.

3.4.2. Isothermal Stress-Strain Curves of NiTi Films

In this section, the isothermal stress-strain curves obtained in this work are reviewed.
The transformation temperatures of ﬁlms were ﬁrst determined by electrical resistivity
measurement. Figure 349 shows stress-strain responses of ﬁlm H623 at test temperature
increments of 30 K.57 Yielding due to the detwinning of martensite variants and stress-
induced martensitic transformation did not occur even at stresses above 300 MPa. The thin
ﬁlm simply behaved as an ordinary metallic material. In ﬁlm H623, the interspacing of
precipitates is only 7 nm. High density of ﬁne Ni4Ti3 particles which are rigid to
deformation apparently blocked the development of martensite variants.

Figure 3-50 exhibits the stress-strain curves of ﬁlms H673 at test temperature
increments of 20 K. The ﬁlm shows yielding plateaus related to the detwinning of
martensite (at T < M,) and stress-induced martensitic transformation (at T) A,). Curves f,
g and h display classical transformational superelasticity. Figure 3-51 shows the
rrricrostructure of ﬁlm H673 which had experienced eight load/unload cycles as shown in
Figure 3-50. Compared with its original structure (Figure 3-9), it is seen that the
nricrostructure essentially remained unchanged. No dislocations were introduced by
stressing that can be seen under high magniﬁcation. However, as indicated in Figure 3-50,
a strain about 0.2 % was left after the last cycle. The slip at stress concentrations (such as

the grip areas) of the tensile specimen, which may not have been detected in the TEM foil,

 

57 As seen in the Figure 3-49, the difference between the slopes of loading and unloading curves are evident
in (a) and (b) but not in (c) and (d). This phenomenon has been reported for gold polycrystalline ﬁlm by
Neugebauer [1960]. The lower slope on loading curves is thought to relate to the creep in grip region
and/or uneven fastening of thin ﬁlm specimen.

98

and unrecovered martensite after unloading may be responsible for this rather small
permanent strain.

Figure 3-52 presents a series of twelve stress-strain curves generated from a
specimen of ﬁlm H703 at deformation temperature increments of 15 K ranging from 198 K
(< M,) to 363 K (> A,).58 The curves are substantially similar to those reported for bulk
NiTi alloys [e.g., Miyazaki, Ohmi, Otsuka and Suzuki, 1982; Miyazaki and Otsuka, 1986;
Shaw and Kyriakides, 1995]. Figure 3-53 shows the stress-strain curves of ﬁlm H723.
Similar to ﬁlm H703, H723 also displays superior thermoelastic properties in terms of very
small residual strains in the superelastic region (T> Af) and less steep slope of yielding
plateau (curves e-h). The deformed microstructure of ﬁlm H723 is shown in Figure 3-54.
Just like its undeforrned structure (Figure 3-13), displaying well-deﬁned precipitate
particles and grains without any dislocations. Thus the ﬁlm H723 also possesses a very
stable structure against slip during stress-cycling.

Annealed ﬁlms A2 and A3 possessed NiTi/Ni3Tiz dual phase structure. Their stress-
strain responses are shown in Figure 3-55 and 3-56 respectively. Residual strains about
0.2-0.3 % can be seen after unloading. Figure 3-57a and 3-57b are the images of
deformed nricrostructure of ﬁlm A3. Lack of transgranular precipitates led to the formation
of dislocations and thus reduces structural stability against permanent deformation.

Figure 3-58 shows the stress-strain curves of the ﬁlm A5 which consists of coarse
N iTi grains and Ni4Ti3 particles. As NiTi grains grew coarser in the ﬁlm A5, slip bands
which formed by passage of a number of dislocations spread throughout entire grains as
seen in Figure 3-59a. This is consistent with its very low critical stress for slip (0', = 271
MPa). As seen in the micrograph, the slip bands in different grains tended to align in a

common direction in order to accommodate the maximum shear stress.

 

53 At the highest test temperature T: 363 K, the ﬁlm fractured at 4 % elongation and only the loading
curve was recorded.

99

Figure 3-60 and 3-61 show stress-strain curves of ﬁlms A7 and A8 respectively.
Both ﬁlms are composed of coarse NiTi grains with but different precipitate morphology.
The loading curves of ﬁlm A7 does not show any transformation yielding due to the high
density of ﬁne Ni4Ti3 particles with spacing of 12 nm. The ﬁlm A8 having precipitate
spacing of 60 nm shows a limited stress-induced martensitic transformation. The images
of the deformed ﬁlms (Figure 3-62 for A7 and Figure 3-63 for A8) indicate that, the
microstructures were virtually unchanged in stress-strain cycling.

It may be concluded that, both small grain size and ﬁne precipitates were found to
promote a perfect transformational superelasticity. On the other hand, when the
interspacing of precipitates was less than 12 nm (in ﬁlms H623 and A7), the precipitates
tended to impede stress-induced martensite transformation. The combination of ﬁne B2
grains (0.25 um) and precipitates with moderate interspacing (3040 nm) in ﬁlms H703 and

H723 provided superior thermoelastic properties.

3.4.3. Classification of Stress-Strain Curves

One striking feature of shape-memory alloys is that in a certain temperature range the
stress-strain behavior changes dramatically with temperature as seen in the preceding
section. From Figure 3-52 (ﬁlm H703) the stress-strain behavior of polycrystalline NiTi
thin ﬁlm associated with the martensitic transformation can be classiﬁed into six types
depending on the value of deformation temperature, T, with respect to transformation
temperatures M,, M,, A,, and A,.59

Type I (T < M,, curves a-c): In the initial, unstressed condition, the specimen is

entirely in the martensite phase with a self-accommodating morphology. Plastic strain due

 

59 According to a study conducted by Miyazaki and Otsuka [1984], in the regime TS T1,, the strain
associated with R-phase decreases with increasing temperature and reaches the minimum at T = T,. This is
consistent with the decrease in rhombohedral distortion with increasing temperature. In the regime T> T,,
the associated strain is constant irrespective of the test temperature. Overall, the strain associated with R-
phase is so small (8 < 0.5 % in polycrystals) that it is not easy to detect even in the specially-prepared
specimen. The strain associated with the martensite dominates the stress-strain behavior of NiTi system.
For this reason, the classﬁcation in this work is only focused on the martensitic transformation.

100

to detwinning and variant coalescence on loading is followed by elastic unloading of the
martensite and a large plastic su’ain. Serrated feature in the loading curve implies the burst
progress of variant rearrangement.

Type H (M, < T < M,, curves (1 and e): The martensite and R-phase coexist in this
temperature range. Curve (e) reveals variations in the tangent modulus during plastic ﬂow,
i.e., two-stage yielding. According to the results reported by Miyazaki and Otsuka [1986],
the ﬁrst yielding is associated with the coalescence of R-phase variants, while the second
yielding is caused by the growth of the stress-induced martensite as well as the continuation
of variant coalescence in the martensite.

Type III (M, < T < A,, curves f and g): The specimen is fully in the R-phase state
since only the R-phase is stable in a temperature between M, and TR (> A,). The very low
initial tangent modulus of the loading curves may be associated with R-phase variant
reorientation. Each of these curves shows a nearly ﬂat plateau connected with stress-
induced martensite transformation. Only elastic recovery of the martensite occurs upon
unloading since the temperature is still below the A,.

Type IV (A, < T < A,, curve h): The R-phase and austenite coexist in this
temperature range. The deformation is caused by stress-induced martensite formation as
well as R-phase variant reorientation. After elastic unloading of the martensite, the
deformation strain is only partially recovered by M —> A reverse transformation because the
temperature is above A,, but below A,.

Type V (A, < T < M,,“) curves i and J): Since stress-induced martensite is unstable in
the absence of external stress at temperatures above A,, the deformation strain which results
from stress-induced martensite tends to be fully recovered upon unloading. The residual
strain drops to less than 0.5 % in curve i. Further increasing test temperature yields nearly

perfect surperelasticity as shown in curve j.

 

‘50 The M,, temperature is designated as the temperature at which the stress to induce martensite equals the
critical stress for slip.

101

Type VI (T > M,, curves k and I): As predicted by the Clausius-Clapeyron equation,
the critical stress to induce martensitic transformation increases linearly with the
temperature. The stress level reaches the ﬂow stress of the specimen and an unrecoverable
strain is introduced by dislocation glide.

Figure 3-64 schematically summarizes the six types of stress-strain curve in o-e-T
space based on the preceding discussion. The classiﬁcation is also made for other stress-

strain curves presented in Section 3.4.2.

3.4.4. Thermomechanical Behavior of NiTi Films

In this section, ﬁrst the dependence of critical stress to induce martensite with respect
to deformation temperature is examined for NiTi ﬁlms. Then the applicability of Clausius-
Clapeyron relation to a practical case is explored by thermodynamics analysis. Finally, the
plateau strain associated with stress-induced martensitic transformation is estimated from
measured thermoelastic values.

For ﬁlms H673, H703, H723, A2, A3, A5 and A8, the critical stresses to induce
martensite are plotted against the test temperature (given as adjusted deformation
temperature T - M,) as shown in Figure 3-65. The critical stresses are measured from their
stress-strain curves using the tangent-intercept method. All ﬁlms tend to give a linearity
suggested by the Clausius-Clapeyron equation in the austenite region except for ﬁlm A8
where the data deviate from linearity at higher stress. According to this linear relation, the
temperature approaches M, (= TJJO’M ) when tensile stress approaches zero as seen in Figure
3-64.

The Clausius-Clapeyron relation (Equation 1.18) links important parameters
associated with thermoelastic phase transformation. For the NiTi system, the measured
values and calculated ones using the Clausius-Clapeyron equation have been compared in
the literature [Miyazaki and Otsuka, 1984; Stachowiak and McCormick, 1988; Hou and

Grummon, 1995]. However, in their work, AH “"‘P "’M (= the heat released during the

102

phase transformation subjected to a stress, 0') was assumed to be equal to AH P 7” (= the
heat released during the stress-free phase transformation which can be conveniently
measured by DSC) in order to apply Clausius-Clapeyron formula. The following
discussion shows that the difference between AH 3"“? "’M and AH P "’M can be ignored in
such applications.

Starting from Planck-like formula61 [Perkins, 1975], Wollants, R008 and Delaey
[1980] have showed that if the difference in heat capacity of two phases is negligible,
AHWv P4” is a linear function of T. Based on Clausius-Clapeyron relation (Equation
1.18) a linear T -0' relation can be expected. In this work similar results can be obtained
through mathematical manipulation using J acobian notation.

In (T, 0') domain, the total differential dAS P-W can be expressed as follows:

P—rM P-rM
dASP—IM = (%—] dT +[i%—] (10' (3.2)
O’ T

Using the Jacobian notation [Mukherjee and Bieler, 1993], the partial differential teams in

Equation 3.2 can be converted to measurable parameters (for details refer to appendix A.2).

P M ACPaM P M
dAS " =——;,—dT+VAa 7 do (3.3)

Where ACﬁ-HM is the difference in iso-force heat capacity of the parent and martensitic

P-rM

phase and Aa is the difference in thermal expansion coefﬁcient of two phases. In

order to estimate the inﬂuence of T and 0011 transformation entropy, we substitute the

 

6‘ Planck was the ﬁrst to obtain the following formula called Planck equation for the temperature
coefﬁcient of the enthalpy function AH:
dAH AH AH (aw)
P

—=AC +———
dT P T AV arr

 

103

properties of NiTi into Equation 3.3, AC‘F’” z AC5“: -0.28 J/mol K,62 V = 1.64 x

105 m3/mol and AaP-W = 44 x 10-6 /K.

dASP'm = —0'—T28dT- 7.3 x 10-ll do (3.4)

For general application, T: 273-373 K and 0': 0-600 MPa. Within this T-O‘ window the
maximum deviation of ASP—m predicted by the Equation 3.4 would be 0.10 J/mol which
is only 0.9-1.1 % of AS P 7’” values for four NiTi alloys cited in Hedayat, Rechtien and
Mukherjee’s paper [1992].

It is evident from preceding discussion that for NiTi system the ASP—W is rather a
weak function of T and 0' in the range of consideration. This result suggests two things:

(1) It gives a theoretical explanation of the linearity of Clausius-Clapeyron formula, since

from Equation 1.14, a linear T-o relation can be expected if spa”

is constant. (2)
Because of AHMP—W/r: ASP-W (Equation 1.17), and ASP-’M e (hip-W)“0 =
AH P '7” / To, the value AH P ”M / To can be used for AH M", "’M / T in Clausius-Clapeyron
formula, where AH P “’M is the heat released during the stress-free phase transformation
which can be conveniently measured by differential scanning calorimetry (DSC), and To is
the temperature at which the parent and martensitic phases have the same chemical free

energy at zero stress. Equation 1.18 can thus be expressed by measurable values:

do AHP'W

_ = _____ 3.5
dT TOVeP‘W ( )

For ﬁlm H703, the critical stress to induce martensite and the stress at which the
reverse transformation begins, both determined by a tangent-intercept construction, are

plotted as a function of test temperature in Figure 3-66. The measured transformation

 

‘52 The values of C 5 and C 3’ , respectively, were determined to be 9.66 and 9.38 J mol"K'l from the DSC
curve of the ﬁlm H703 (Figure 3-45).

104

temperatures from the electrical resistivity measurement are indicated along the temperature
axis. Below M,, it is found that the yield stress declines slightly with temperature, reaching
a minimum near M,, In this range the deformation is caused by detwinning of martensite
variants. The stress shows a normal temperature dependence, i.e., it decrease with
increasing of temperature since detwinning is a thermally activated process [Miyazaki,
Kohiyama and Otsuka, 1991]. Above the M,, both the stress for martensite formation and
the reversion stress rise monotonically with temperature in a linear fashion in agreement

with predictions based on the Clausius-Clapeyron equation. The extrapolation of the two
lines intersect the temperature axis at around M, and A,.

We may estimate the plateau strain (6:qu

) using the measured values of H and
da/dT. It is worth to point out that the Clausius-Clapeyron relation is a thermodynamic
equation which describes equilibrium conditions, whereas the transformation in a practical
case like NiTi system deviates from the thermodynamic equilibrium as has been discussed
in Section 1.1.3. We input following data into the Equation 3.5, To = (M, + A,) / 2 =
283.5 K (Equation 1.2), V = 1.64 x 10:5 m3/mol, AHP-W = - AHM-W = - AHM-W =
-1599.0 J/mol (in this case R-phase is the parent phase of martensite, i.e., P = M in
superscript), and da/dT = 6.41 MPa/K which is obtained from the loading curve of Figure
3-66. The calculated value of plateau strain is 5.3 % which compares favorably with the
values (= 4.7-5.5 %) reported for bulk NiTi alloys [Miyazaki and Otsuka, 1984;

Stachowiak and McCormick, 1988].

3.4.5. Transformational Superelastisity

In this section, a stress-temperature diagram for predicting transformational
superelastisity is discussed ﬁrst. Then mechanical energy output and energy efﬁciency
associated with each superelastic cycle is explored.

Three types of yield phenomena are possible involve in a thermoelastic alloy. The

ﬁrst is attributed to the detwinning of unfavorably oriented matensitic twin variants into

105

favorable ones with respect to the stress axis. The yield stress declines slightly with
temperature as shown in Figure 3-66 at T< M,. The second one is related to stress-
induced martensitic transformation and the critical yield stress rises linearly with increasing
temperature as predicted by the Clausius-Clapeyron equation as shown in Figure 3-66 at T
> A,. The third one is a conventional yielding caused by dislocation glide in the parent
phase and martensite. As it will be addressed in following, whether or not a material
behaves superelastically essentially depends on the relative values of the stress to induce
martensite and the critical stress for slip.

Stress-temperature diagrams for ﬁlms H673, H723 and A3 can be plotted in Figure
3-67, 3-68 and 3-69, respectively, using the data obtained in the preceding sections. In the
diagrams the line 0;, represents the critical stress to induce martensite which was measured
from corresponding stress-strain curves using the tangent-intercept method as shown in
Figure 3-65. The line 0', is the onset stress for slip which was obtained from the yielding
points of ultimate elongation curves at 423 K (Figure 3-63), and by assuming that the
change of the critical stress for slip with the temperature was -0.2 MPa/K [Miyazaki,
Kohiyama and Otsuka, 1991]. The transformation temperatures A, and A, were measured
from the resistivity vs. temperature curves (Table 3-3 and Table 3-5).

In the diagrams, T-a conditions allowing superelasticity is conﬁned by three
boundaries:

(1) 0' > on, to stress induce the martensitic transformation,

(2) a< 0",, to prevent permanent strain, and

(3) T > A,, to eliminate the residual strain created in stress-induced martensitic
transformation after unloading.

It is evident that a greater 0,, a lower a}, and a lower A, tend to give a wider T-O'
window for superelasticity. As predicted by the stress-temperature diagrams, ﬁlms H673,
H723 and A3 all show superelasticity at T > A, (Figure 3-67, 3-68 and 3-69, respectively).

However, the superelasticity in ﬁlm H673 and especially in ﬁlm A3 are not ‘perfect’ as

106

exempliﬁed by a residual strain after unloading. There are perhaps two reasons making the
previous T-O' diagrams unable to predict a perfect superelasticity. First, a very limited
amount of permanent deformation, which may be large enough to jeopardize the perfection
of superelasticity, can occur in the linear region of stress-strain curve shown in Figure 3-
63. Second, some martensite may still be retained in the specimen after unloading as
pointed out by Miyazaki et al.[1984]. In order to predict perfect superelasticity, a proper
way to precisely detect the onset of the stress, where the slip or retained martensite start to
occur, is to measure the stress at which the ﬁrst dimensional change is observed when
repeating loading-unloading cycle with small increment of load. Miyazaki, Kohiyama and
Otsuka [1991] have reported the stresses for slip both in the parent phase and the martensite
for bulk NiTi alloys using such method.

In each superelastic cycle, the area under the loading curve is the work done per unit
volume on the specimen (Em), while the area under the unloading curve is the energy
density per unit volume which is stored and can be released as mechanical energy upon
unloading (E,,,). The area conﬁned by the loading-unloading loop represents the energy
density per unit volume (E,n -E,,,,) which is dissipated during the cycle.

Figure 3-70a shows the output of mechanical energy vs. adjusted deformation
temperature (T - A,) for NiTi ﬁlms. The output energy increases when deformation
temperature increases. This is mainly due to the fact that the critical stress to induce
martensite increases as temperature increases. The maximum output energy found in this
study was 1.7 x 107 J/m3 which was produced in ﬁlm H673 at 3.5 % total strain.‘53 As a
comparison, output energy per unit volume for various available microactuators are given
as follows: solid-liquid phase change: 4.7 x 106 J/m3 [Hale, Hoover and O’Neil, 1971],
therrnopneumatic actuator: 1.2 x 10‘5 J/m3 [Zdeblick et al. 1994], thermal expansion: 4.6 x

105 J/m3 [Krulevitch, et al. 1996], electromagnetic actuator: 2.8 x 104 J/m3 [Guckel, et al.

 

63 To avoid the fracture of specimen in grip areas, the total strain was limited to 3.5 % in this work.
Therefore, the possible maximum output energy could be greater.

107

1994], electrostatic actuator: 3.4 x 103 J/m3 [Fan, et al. 1994], and piezoelectric actuator:
1.8 x 102 J/m3 [Robbins, et al. 1991; Tjhen, et al. 1991]. Evidently, the NiTi thin ﬁlm
produced in this work displays the highest energy output. Figure 3-70b shows the energy
efﬁciency (E0... /E,,,) as a function of adjusted deformation temperature. For as-sputtered
ﬁlms (H673, H703 and H723), the energy efﬁciency is increases as temperature increases,
while for annealed ﬁlms no such relation appears. The large residual strain shown in
stress-strain curves of annealed ﬁlms reduced the energy efﬁciency. The energy efﬁciency

are varied from 47 % (ﬁlm A3) to 93 % (ﬁlm H723).

3.5. Fractography of N iTi Films

Fracture surfaces of specimens strained to fracture were examined by the SEM (the
ultimate elongation curves of ﬁlms H673, H723 and A3 are shown in the Figure 3-47).
Figure 3-71 shows the fractography of ﬁlm H623 which displays ﬁbrous characteristic
consisting of many microdimples. The fractography exhibits a typical transgranular mode I
(the opening mode) ductile failure. The ductile fracture occurs through three stages: First
small cavities nucleate at weak internal interfaces, such as second particle/matrix interfaces.
These cavities then expand by plastic deformation and ﬁnally coalesced by localized
necking of the material between adjacent cavities until completely separating specimen into
two parts. The force causing such fracture is the principal stress normal to fracture plane.
Microscopically, voids were formed through shear in the lattice planes. This is reason for
the fracture occurred at near grips of thin ﬁlm specimens. Since the maximal principal
stress reaches the maximum value at the edges near the ends of the gauge section as shown
in the contour map of maximal principal stress (Figure 2-18).

The fracture surface of the ﬁlm H673 is shown in Figure 3-72. The columnar grain

structure is clearly seen and the latitudinal dimension of grains (about 0.25 um) is

108

consistent with that observed by the TEM. This fractography also conﬁrms the columnar
grain structure produced by hot-substrate deposition.

Figure 3-73a and 3-73b, respectively, are fractography of ﬁlm H723 viewed from
surface side and substrate side. It seems that the fracture is mainly transgranular one but
with some characteristic of intergranular fracture because the fracture surface displayed
columnar grains. The external neck area in surface side of ﬁlm reveals a serial of valleys as
indicated by arrows (Figure 3-73a), but the neck in substrate side shows an even and ﬂat
morphology (Figure 3-73b). It is understandable from Hall-Petch relationship (Equation
1.25) since the extremely ﬁne (grain size = 25 nm) grains were formed in substrate side of
ﬁlm H723 and impeded plastic deformation. Even in the surface side of the ﬁlm, the free
surface (except the neck area) is quite smooth. This fact implies a very stable
rrricrostructure against the permanent deformation.

Figure 3-74 and 3-75, respectively, show SEM images of fracture surfaces of the
ﬁlms A3 and A5. Large cup-like features are perhaps due to a greater grain size and thus
greater extent of localized plastic necking. The rough free surface of the ﬁlm reﬂects that
considerable deformation had remained after ultimate stressing.

Figure 3-76 was taken from the ﬁlm A7. The fractography shows a ﬂat and straight
fracture surface with many small dimples. No neck is found in the edge of fracture surface
and the free surface of the ﬁlm is very smooth. These features are consistent with its very
limited plastic strain shown in the stress-strain curves (Figure 3-60), and can be explained

by the effect of a high density precipitates resisting dislocation movement.

CONCLUSIONS

Crystalline N iTi thin ﬁlms, which exhibits a well-deﬁned thermomechanical
behavior, have been successfully produced by one-step hot-substrate deposition. The
correlation between the microstructure of the ﬁlms and a key process parameter, substrate
temperature, has been systematically examined by TEM. The results have been combined
into a grain structure-deposition temperature diagram. The deformation characteristics,
including the critical stress to induce martensite and the ﬂow stress for slip have been
investigated using tensile tests. Based on obtained thermomechanical data, stress-
temperature diagrams have been constructed to predict superelasticity capabilities for NiTi
ﬁlms with diverse microstructures. The correlation between the rrricrostructure and
thermomechanical pr0perties has been clariﬁed. Detailed conclusions can be drawn as
follows:

(1) Compared with the conventional two-step N iTi ﬁlm process (sputter deposition
plus a crystallization anneal), the hot-substrate deposition technique has been proven to
provide advantages in terms of eliminating post-deposition crystallization, decreasing the
process temperature to form a crystalline structure (T = 623 K in the hot deposition vs. T >
773 K in the conventional process), and producing ﬁne-grained microstructure (D = 0.1-
0.3 um in the hot deposition vs. D > 1 pm in the conventional process).

(2) In hot-substrate deposition, NiTi grains started to nucleate at a substrate
temperature of 598 K and columnar grains formed at temperatures from 598 K to the upper
limit of this work, 773 K. The columnar-grained structure resulted from surface diffusion

Processes.

109

110

(3) Films deposited at lower temperature (TS 673 K) exhibited a strong (110),, ﬁber
texture. The texture vanished when deposition temperature was increased to 723 K. In
this case, extremely ﬁne crystallites (D = 25 nm) nucleated on an initially amorphous layer.

(4) The variation in Ni4Ti3 precipitate morphology with depth in the ﬁlm and with
deposition temperature, suggests that both surface and bulk diffusion for growth of Ni4Ti3
occurred during hot-substrate deposition at temperatures between 723 and 773 K.

(5) A unique NiTi/Ni3Ti2 dual phase structure, which cannot be easily produced by
either conventional two-step process or hot-substrate deposition, could be made by giving
hot-deposited ﬁlms an additional heat treatment.

(6) The resistivity-temperature curves could be classiﬁed into two categories: B2 H
R H M and B2 H M, except for some ill-deﬁned curves. Irrespective of the B2 grain size,
ﬁlms having ﬁnely dispersed precipitates in B2 grains displayed two-step phase
transformation behavior, whereas the ﬁlms with coarse precipitates showed single-step
transformations.

(7) Both small grain size and ﬁne precipitates were found to promote a perfect
transformational superelasticity. On the other hand, when the interspacing of precipitates
was less than 12 nm (in ﬁlms H623 and A7), the precipitates tended to impede stress-
induced martensite transformation. The combination of ﬁne B2 grains (0.25 pm) and
precipitates with moderate interspacing (3040 nm) in ﬁlms H703 and H723 provided
superior thermoelastic properties.

(8) The maximum output energy found in this study was 1.7 x 107 J/m3 which was
produced in ﬁlm H673 at 3.5 % total strain. The energy efﬁciency are varied from 47 %
(ﬁlm A3) to 93 % (ﬁlm H723).

111

It is challenge to realize a perfect superelasticity in NiTi thin ﬁlms which only have
micrometer dimensions. Precipitates and grain boundaries were found to play an important
role in resisting permanent deformation and thus promoting superelasticity. However, as
the interspacing of precipitates approached 12 nm, the ability of NiTi to form stress-
induced martensite tended to be reduced. It is interesting to know whether or not the B2
grain boundaries along (i.e., in precipitate-free NiTi ﬁlms) are capable of promoting perfect
superelasticity. If the answer is yes, then what is the grain size limit which allows the
formation of stress-induced martensite? These are questions remaining to be answered in a
future study.

Successfully depositing crystalline NiTi onto the polymeric substrates (Kapton) relied
on the hot-substrate deposition technique which can produce B2 crystals at a lower
temperature, and using a two-sided substrate heating device which can provide a uniform
substrate temperature and a reliable temperature reading. Combined with photolithography
process, a NiTi/polyimide microactuator prototype was fabricated. In appendix A.1 a

robust electrically—excitable shape-memory action move is demonstrated.

112

Table 1-1. Austenite-to-martcnsite lattice correspondences [after
Matsumoto et al., 1978].

 

 

 

 

 

 

 

 

 

 

 

 

 

Variant [1001M7 G [1100 no],
1 [100], [0111, [0111A
1' [100]A [oil]A [oil]A
2 1100].. 10111.. 10111..
2' 11001.. 10111.. 10111..
3 [010]A [101]A [101‘]A '
3' [010]A [To '1'], [loi]A :
4 [0101A [1011, [ioi]A l
4' [0101A [1011A [1011,
5 10011.. [1101. 11101.. i
5' [0011A [Tic]A [ 1 10]A l
6 1001].. [1'10], [1 101.. l
6' [001']A [l 1'0]A [ 1 10]A

 

 

 

113

Table 1-2. The 24 habit plane variants and related correspondence
variant-combinations. The habit plane indices were calculated from
the phenomenological crystallographic theory, and correspond to

the experimental data for solution-treated specimens [after

Miyazaki et al., 1978].

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Habit plane variant vgﬂzggzgtcgn Habit plane indices
1(+) 1-2 (-0.89, -0.22, 0.40) '
1(-) 1-2' (-0.89, -0.40, 0.22)
1'(+) 1'-2 (0.89, -0.40, 0.22)
1'(-) 1'-2' (0.89, -0.22, 0.40)
2(+) 2-1 (0.89, 0.22, 0.40)
2(-) 2-1' (0.89, 0.40, 0.22)
2'(+) 2'-l (-0.89, 0.40, 0.22)
2'(-) 2'-1' (-0.89, 0.22, 0.40) l
3(+) 3-4 (-0.40, 0.89, 0.22)
3(-) 3-4' (-0.22, 0.89, 0.40)
3'(+) 3'-4 (-0.22, -0.89, 0.40)
3'(-) 3’-4' (-0.40, -0.89, 0.22)
4(+) 4-3 (0.40, 0.89, 0.22)
4(-) 4-3' (0.22, 0.89, 0.40)
4'(+) 4'-3 (0.22, -0.89, 0.40)
4'(-) 4'-3' (0.40, -0.89, 0.22)
5(+) 5-6 (0.22, -0.40, 0.89)
5(-) 5-6' (0.40, -0.22, 0.89)
5'(+) 5’-6 (-0.40, 0.22, 0.89)
5'(-) 5'-6' (-0.22, 0.40, 0.89)
6(+) 6-5 (0.22, 0.40, 0.89)
6(-) 6-5' (0.40, 0.22, 0.89)
6'(+) 6'-5 (-0.40, -0.22, 0.89)
6'(-) 6'-5' (-0.22, -0.40, 0.89)

114

Table 2-1. Sputter-deposition conditions and resultant ﬁlm compositions.

The notation ‘H---’ is used in the present work where the ‘H’ signiﬁes
‘hot deposition’ and three digits signify the substrate temperature of
deposition in Kelvin. The substrate for all ﬁlms were quartz except

for ﬁlm H698 which was deposited on Kapton ﬁlm.

 

 

 

 

 

 

 

 

rant. Base Arorrg I " Code i " Code “ ” MW»

Film ness pressure pressure power potential rate comp. ,

(um) (x10“1 Pa) (Pa) (watt) (volt) (um/sec) (at.% Ni) 1
H573 5.0 2.11 0.67 200 582-559 0.62 51.5 I
H598 4.3 3.73 0.68 200 679-639 0.63 51.2
H623 6.7 2.96 0.67 150-200 638 0.69 52.3
H673 6. 3 1.47 0.67 200 556-522 0.71 51.0
H698 3.1 0.66 0.75 250 644-619 0.68 50.8
H703 7.1 0.61 0.75 250 471-460 1.25 50.8
H723 6.7 2.53 0.68 200 593-562 0.72 51.1
H773 4.9 4.00 0.84 200 538-524 0.79 51.4 J

 

 

 

 

 

 

 

115

Table 2-2. Heat-treatment conditions of annealed NiTi ﬁlms. For annealed NiTi
ﬁlms, the notation ‘A-’ is used where the ‘A’ represents ‘annealing’ and a digit
designates the detailed heat-treatment path as indicated in the table.

 
    
     
    
    
   
   
   
   

    

 

 

 

 

 

 

 

C?;Il.l‘)7:sl:lti))n temp .p(K) terrﬁrnenractatlllrgﬁK) Cooling method
A1 52.3 623 858, lb Furnace cooling
A2 51.0 673 858, lb Furnace cooling
A3 51.1 723 858, lb Furnace cooling
A4 51.1 723 813, lb Furnace cooling
A5 51.0 673 988, 1h Furnace cooling
A6 51.0 673 1123, 1h Furnace cooling
A7 51.0 673 1073, 2h Air cooling

 

  

 

 

 

 

 

 

116

Table 3-1. Summary of microstructures of as-deposited NiTi ﬁlms. In the table,
D32, 0,, and PS, respectively, designate the in-plane grain diameter of the B2,
the diameter of precipitates in longitudinal direction, and precipitate spacing.

Microstructure characteristics

 

 

 

 

 

 

 

 

 

 

 

l Film Location Reference
H573 Midplane Amorphous; Fig. 3-5
H598 Interface Amorphous; Fig. 3-6

' Midplane 1),, = 750.1 nm, (110) texture; Fig. 3-7

‘ D,,, = 12.2 nm, PS = 6.5 nm;

. Surface DB, = 94.0 nm, (110) texture; Fig. 3-8

1 Dppt = 6.3 nm, PS = 6.1 nm;

1 H623 Interface D32 = 86.4 nm, (110) texture; Fig. 3-9

1 ‘ 1),, = 12.5 nm, PS = 6.5 nm,

l Midplane D32 = 98.6 nm, (110) texture; Fig. 3-10

i D9,, = 9.4 nm, PS = 6.5 nm;

1 Surface 1),, = 101.6 nm, (110) texture; Fig. 3.11

l Dppt = 9.4 nm, PS = 5.2 run;

I H673 Interface 0,, = 237.5 nm, (110) texture; Fig. 3-12

3 0,, = 56.3 nm, PS = 30.1 nm;

I Midplane DB, = 221.1 nm, (110) texture; Fig. 3-13

' D”, = 18.8 nm, PS = 18.0 nm;

Surface DB, = 252.3 nm; Dppt = 15.6 nm, P8 = 15.4 nm; Fig. 3-14
H703 Midplane D3, = 235.2 nm; Dppt = 137.5 nm, PS = 30.7 nm; Fig. 3-3
H723 Interface DB, = 23.8 nm, random orientations; PS = Na; Fig. 3-16
Midplane D32 = 247.2 nm, random orientations; Fig. 3-17
D”, = 150.0 nm, PS = 41.5 nm;
Surface D32 = 286.1 nm; D”, = 62.5 nm, P8 = 25.8 nm; Fig. 3-18
H773 Interface DB, = 27.4 nm, random orientations; PS = n/a; Fig. 3-19
Midplane D3, = 271.3 nm, random orientations; Fig. 3-20
Dppt = 218.7 nm, PS = 66.4 nm;
Surface D3, = 318.8 nm; Dppt = 181.3 nm, PS = 85.6 nm; Fig. 3-21

117

Table 3-2. Summary of NiTi matrix and precipitate morphologies of annealed
NiTi ﬁlms. In the table, D32, Dptt and PS, respectively, designate the in-plane
grain diameter of the B2, the diameter of precipitates in longitudinal direction,

and precipitate spacing.

Microstructure characteristics

 

 

 

 

 

 

 

 

 

Film Location Reference

A1 Midplane DB, = 116.7 nm; Fig. 3-30
Dppt = 375.2 nm, PS = 258.8 nm;

A2 Midplane DB, = 207.8 nm; Fig. 3-31
Dppt = 350.5 nm, P8 = 877.8 nm;

A3 Midplane D32 = 232.5 nm; Fig. 3-32
Dppt = 475.0 nm, P8 = 1.3 um;

A4 Midplane D3, = 218.8 nm; Fig. 3-33
Dppt > 350 nm, PS = 1.1 um;

A5 Midplane DB, = 638.5 nm; Fig. 3-34
D,,, = 178.2 nm and D,,, = 635.0 nm,
PS = 152.9 nm;

A6 Midplane D32 = 3.7 pm; Fig. 3-35
D,,, = 962.5 nm and Dppt = 15.6 nm,
PS = 8.6 nm;

A7 Midplane 0,, = 1.8 ttm; Fig. 3-36
D,,, = 19.4 nm, PS = 11.7 nm; Fig. 3-37

A8 Midplane D3, = 1.9 um; Fig. 3-38

 

 

Dppt = 122.9 nm, P8 = 60.3 nm;

 

118

Table 3-3. Phase transformation temperatures of as-deposited NiTi ﬁlms.

composition
(at.% Ni)

 

51.5

 

 

 

 

 

 

 

 

 

 

 

 

 

119

Table 3-4. Transformation characteristics of the ﬁlm H703 obtained from
electrical resistivity and DSC measurements.

Temp. or heat ﬂow

Electical resistivity method

 

308 K

 

295 K

 

321 K

 

n/a

319K

 

263 K

 

231K

 

298.5 J/mol (2.8 J/g)

 

1897.5 J/mol (17.8 J/g)

 

1599.0 J/mol (15.0 J/g)

 

-3l9.8 J/mol (-3.0 J/g)

 

 

 

 

-437.1 J/mol (-4.l J/g)

120

Table 3-5. Phase transformation temperatures of as-deposited NiTi ﬁlms.

 

 

 

 

 

 

 

 

 

 

Film composition M, (K) M, (K) 'p (K) A, (K) A, (K)
(at.% Ni)
A1 52.3 254 282 310 307 314
A2 51.0 247 276 312 306 315
A3 5 l . 1 260 271 314 294 307
A4 51 . 1 n/a n/a n/a n/a n/a
A5 5 1 .0 267 282 323 319 329
A6 51 .0 184 248 308 278 315
A7 51.0 n/a 223 318 258 289
A8 51.0 228 255 322 318 327

 

 

 

 

121

 

(a) (b)

 

 

 

Lattice invariant shear Rigid body rotation

Figure 1-1. Schematic drawings of a two-variant matensitic transformation
[Chang, 1993]. (a) The lattices of the parent phase and the martensite. (b) The
Bain distortion, which converts the parent phase into the martensite, does not
yield an undistorted plane associated with the habit plane. (c) A lattice invariant
shear, through a reversible manner of twinning, produces an undistorted plane.
((1) The rigid body rotation, which brings the transformed lattice back, so that
the interface is in contact with the parent lattice.

122

(a) Single-crystal, single-interface

é

Martensite (%)
01
O

I

 

 

 

A
‘

 

 

""""""V'

 

 

 

 

       

 

 

 

 

 

 

A
o l 1 1 1 T :11- T1
9 o t _ -
Temperature -——> A M Interface
(b) Single-crystal, multiple-interface
M A - '
$3 100 _ t s ’M/M lilterface
E E ‘ G \ G
‘0 - i
5 50 g Q ‘~ I Q ‘-
t I
(U I
2 O I 1 1 1 :I 1 M
Me To Al
Temperature ——>
(c) Coarse-grained sample, multiple-interface
Grain boundaries

é

 

Martensite (%)
0|
0
l

 

Temperature ——>

(d) Fine-grained sample, multiple-interface

Mf As

.5
8
I
:1

       
   

 

Martensite (%)
01
O
l

A
‘

Me At

Temperature ———>

 

/

 

 

 

 

 

 

 

 

 

 

Figure 1-2. Schematic representation of the inﬂuence of frictional work
and stored elastic energy on thermoelastic martensitic transformation

[modiﬁed after Salzbrenner and Cohen 1979].

123

 

 

88 .9950

 

8..

(%) ellsueuew

 

All

.meoumctoﬁceb 83:83.: came—82505 :0 33:0 acme—o
e808 use 3.5.“ 3:285 .20 cop—once. 05 35853 $5396 omumEoeom .m-_ Semi

 

>995
ozone c9:

hAll

 

 

e.

 

cc, ,2
s u
...< 2

cocoa
.2285 :9...

 

 

 

 

 

8v

 

 

 

 

 

 

 

 

 

 

 

 

 

 

w
l O W
9
m, E
t W: B All
cow 6 ‘l
W
u
some 33
+ a
3
h All
on 38.82
.o m L\
a. a: w
s a m.
m.
we, 5 -8? W
cocoa

eases 26..

 

8..

(%) allsueuew

124

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MI
11
z 1
AA
0 a-b
ll
b
a
8
2
a e- d
9
Strain
O >

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 14. Upper figure: schematic drawing of shape-memory effect:
(a-c) the deformation is accomplished by detwinning or reorientation of
variants in martensite lattice. M is twinned martensiite and M' is deformed
martensite. Bottom ﬁgure: (a-c) the deformation is accomplished by

transformational twining in the austenite.

 

125

 

 

 

 

 

 

Variant 6: 0 Ni atoms
[100]... 41001132
[010] M _’ [1.10132 0 TI atoms
[001] M -> [1 10] 132
80: 0.3015 nml ’
3,2
(8)
Volume change
= +0.41 °/o
—-—>

   

(C)

. .. 0 Shuttle
Shear straln — 11.92 /o Shufﬂe proposed proposed by HS

 

Figure 1-5. The steps illustrating the transition from B2 to B 19' lattices
[modiﬁed from Hehemann and Sandrock, 1971]. (a) B2 lattices; (b-c) an
orthorhombic distortion; (c-d) a monoclinic shear; (e) two shufﬂe models.

126

K2 before shear K2 alter shear

 

Plane
of shear

 

 

 

 

K1

 

K1 : Twin plane

K2: Second undistorted plane

71,: Twining shear direction

11,: Intersection of K, and the plane of shear

s : Magnitude of twinning shear displacement

Figure 1-6. An illustration representing geometry relationships
between planes and directions in twinning [Reed-Hill, 1964].

127

 

(0.78 .039 .1148 1.2

2’(—)

 

Figure 1-7. The self-accommodation morphology of the monoclinic
martensite, and a model describing crystallographic relationships
between the martensite variants in the triangular morphology
[Miyazaki et al. 1989].

Rhombohedral Angle 0:

Electrical Resistivity

Heat Flow

(arbitrary units)

90.0
89.8
89.6
89.4
89.2

«— Exo o Endo—>

128

 

a Rhombohedral 7;” Cubic

 

I

Tia

1/3 (1 10)-type
1/3 (110)-type reﬂections reflections appear
look into precise positions—j ]

l L 1 I 1 l l l I l I

I

 
  

 

 

I)

  

;
l.
l

 

 

 

 

Temperature ——>

Figure 1-8. Schematic diagram showing physical property changes
associated with R-phase and martesitic transformation: (a) lattice
distortion, (b) electrical resistivity change, and (c) latent heat exchange.

129

o
9

V

 

Figure 1-9. Self-accommodating morphology and corresponding
schematic variants-combination of rhombohedral phase proposed
by Miyazaki and Wayman [1988].

Temperature °C

130

Weight Percent Nickel

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

16000 110! 210 , 39 r 410 T 510 I tip -, 710 ' 810 j 910' 100
670°C Ll}
1600
L
1400 1:
1 10°C
1200
1118°C
(Mi) .9
E
1000 o g.
942 c
882°C ..
, z
. at
1 r—
800‘ 7gioc
3-0m)
1 ....... 6. 3.0.; ........
60° VV VV I V ' I V I' V V V" I 'V ' V'I ' VVT Y VVV I
o 10 20 30 40 50 so 70 so 90 100
T1 Atomic Percent Nickel N,
(21)
I400
gun—- ’- .-— K‘- ~ x
f “' .V-V '— \ N K
1200- , r: \ \ \ \
V \ \ \
1025 +20- >V —
looo— 2’: /
o 800-
I-
3
‘3 625 It 20‘
a 600- ” . ‘
.E' T12 NH-ﬁ I, (~N13le]
l- . .
200- - u
. . . 9 N1 T1
leNH-mortensne B ’ t
45 ' $0 55

Ni,otomic percenl

(b)

Figure 1-10. (a) Ni-Ti equilibrium phase diagram proposed by Massalski
[1987]; (b) Ni-Ti phase diagram in the vicinity of stoichiometric
composition proposed by Wasilewski et al. [1971].

Aging Temperature ('0 l

131

 

i) ° 11'“ .TINIoTlrrﬂlre
I A TINI oTltrﬂluoThNIs ATINI oleﬂls
900$ o DTINIOTHNIa oTINIJ ITINIrI-TINI:_,
l
I
o
l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

' 0.1 r 10 100 1000 10000
Aging Time I hrs)

Figure 1-1 1. Isothermal time-temperature-transformation (IT I) diagram
for Ni52Ti48 alloy [Nishida, Wayman and Honma, 1986].

132

 

197357.2lrgw 9 Ms
a Tl-SOIlehNI /

Ms
I

 

 

his
Ms
I

c n-SOW/ #

d Ii-SI.OOI°I.NI

Resistivity

e Ti-Sl Bol‘leNl

llllllLlALJl
300 400

 

 

HJLthlrttrIrrLthrr

100 200
Temperature ( K )

 

Figure 1-12. The effect of Ni-concentration on the electrical resistivity
vs. temperature curves for NiTi alloys quenched from 1073 K [Nishida
and Honma, 1984]. The Ms temperature decreases rapidly with the

increase of nickel content.

133

Temperature (K)
200 250 300 350

2L 1 L L 1 I

f

150
l

L

 

(o)S.T.
Hr
l ”5

 

 

 

 

(9)20h

 

(1) 50h

I
As Pm 9:2(79)
sz(T=l.)

At

Electrical Resistivity (arbitrary units)

(9M00h

 

 

 

 

-150 -100 -50 0 50 100

Temperature (‘C)

Figure 1-13. Electrical resistivity vs. temperature curves [Wu, Lin and
Chou, 1989], of NiSITi49 alloy aged at 673 K after a 1073 K, 2 hour
solution-treatment for various times, showing the effects of aging on Ms
temperatures.

134

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(a) (b)
n n
no» 9, 8" gs Q 9 IO» .. G g l a 8 ‘0
3 a “ ‘ 8 ‘ a u t
g I " t g u
m ‘0 0
go 5» go 5»
AI d-Ieoum Cu d-GSW
o 20 3; ° 20 $
0 IO 1; o O S
A 5 1 6 5 x
I Yoeld I Yield
° ‘ ‘ 3 ‘ .0 ° 5 .b .5
ThackneSS/Glom Sue Thickness/Gram Sue
(c) (d)
'0 g ' O 8' 9 3 IO» . 0 . 2
5 Q l‘ g 0 ° 0 0
= m
m °° o
8 3* 3. ' °
3 ' E
8 3 '
8 ' 2 0°
205 05? o
O
Cu~l3m$Al d-40ur1 Fe d-ZSW
: f8: 0 upper yield stress
2: YIQ'd O lever weld stress
0 .6 2‘0 50 4?) 50 ° .2, 2‘0 3‘0 4‘0 so
Thickness/Glam Sue Thickness/Glam Sac

Figure 1-14. The effect of specimen thickness on the normalized flow
stress in polycrystalline (a) A], (b) Cu, (c) Cu—13Al (at.%), and ((1) Fe
[Miyazaki, Shibata and Fujita, 1979].

135

(a)

(b)

(C)

 

Figure 1-15. Three types of crystallographic relationships between the
lattice of the precipitate and the lattice of the matrix [Hirsch et al., 1977].
(a) coherent with negative misfit; (b) semi-coherent, and (c) incoherent.

 

136

Dislocation Precipitate

 

 

 

Dislocation
Precipitate _’ "’ '°°P 9
—-> —-> ——> —->
—> —-> —-> —->
Dislocation

(b)

“0
at? 9“ Precipitate

(C)

Figure 1-16. Interactions between dislocations and precipitates. (a) In lightly
aged alloys, the closely spaced stress ﬁelds make the dislocation unable to bend
between the particles. The dislocation has to move out of its slip plane by
climbing or cross-slipping. (b) In over-aged alloys, the particles are much wider
apart. The dislocation can bend around each precipitate and leaving dislocation
loops [Orowan, 1948]. (c) A schematic representation showing sheared N i3Al
particles by passing dislocations in a Ni-19% Cr-6% Al alloy after 2 %
deformation [based on Gleiter and Hombogen, 1965].

.305

3:93 8 80%»: £3 35¢? :35 0380>£ the 858280 05 3 roam—«2 mm gang—80v 2: 0-8
.ouatBE cummcotaﬁbcoba mo EoEo>oE 05 await €383 nouaEuEEab 0288.38 05. 3-3
.vouﬁamom :03 93 8658.88 mo 9.85258 98 wet—:35 couaccﬂmpab 05 20:3 3960 03:6 .3 m a 8
Rumba 295. mm a Soc 3:8 coggohou c3332 an @5862: $323828 Engage < .2; 25%"—

§ 5

 

 

 

 

  

 

 

055.580 . ctwocm
“ 25a 2332.. 3563......
" Baroda—.5
w 853 .32.:
m M 005.25 3N
25:93:. 93:23.: .
8:528 .. .. 352-53... .L.
w“ 8.
Swan overcome"; .83" m 531E
Ewan; 2: F .5..-.umcéc- P Quay. 353 53 motenﬁ
33.9 82. Eoéuwtué "a 2:- F gasps: F 53.75. 335.0 81o agile

Exam 292989 E:E_xm_2 55 .709: S 5. 8 955350 mctmmcm 96:. E332:

 

 

 

138

 

(a)

(it 1] [Cal.(°/o)]

/

 

 

(10.5 _ _ 40.5
o
(10.6)
8.0-- .
7-0“ (7.8)
6.0"

5.0“-

305—“Q” 46 «35)
[001)

 

(b)

Figure 1-18. The orientation dependence of recoverable tensile strain for
NiTi single crystals in BZ-to-B 19' transformation: (a) The values
predicted from the lattice deformation matrix [Saburi and Nenno, 1981].
(b) The measured values from solution-treated single crystals [Miyazaki
et al. 1984]. Numbers in parentheses indicate the measured results and
contour lines represent predicted values.

139

 

 

 

       
   
 

Strain (7.)
0
1s

<00!)
220' 240 260 ‘Fgaéo
,0 2‘ Temperature (K)

[III]

A

-04»

 

-06.

(b)

Figure 1-19. (a) The orientation dependence of recoverable strain for
B2-to-R transition at a temperature of (TR - 35) K. Solid circles indicate
the experimental data, whereas contour lines indicate calculated values;
(b) The temperature dependence of the recoverable strain for various
orientations. The recoverable strain in all directions increases gradually
as temperature decreases from TR [Miyazaki, Kimura and Otsuka, 1988].

 

140

.mva_ .0552 use «248%

£398 >02“ woancﬂz E gocmﬂeanm co uni 5an mo Soto 05. SV :32 Na 3 3383):
xoza 2E H3: Z E 85:28.55 gain—Boga :o @593 55:2 mo Soto 2t. AS AN.— 2sz

.xv z_<th
N . a . WIWIWJ

             

_.~n>.u
._ac_»

e==__n.~_. =_-_a

('dWl $33818 JIISNZI

°
9
n

anm .2" =:_u

uaov .-
n..a._

 

3; c _ S «m

 

 

N o N o
8. or. o. \
xmdom
N o N o N o
2: 8 0N ‘
x mdmm
. N o .N o N c
on. 8. om
3V

 

 

 

x $.93

 

 

AxUN

Azuv

 

o u
C)
N

(PdW) 559118 91151-131

§

oo~

 

E

141

 

 

 

   
 

 

 

 

Ground Cathode Negative
shield dark space glow plasma
. Sputtered
e' atoms :3:

' _' . N‘s ” +
Cathode _/ \ Anode
(target) ' 1 9 e'

E ‘. Primary
; = electrons
—-—_rl Substrates
M
E ’———/
E
.9
O
a.
Distance
(a)
Magnetic Plasma ring and
field lines

area of erosion
»>>§7~
a s an.
(13):. N

     
      

/N
N

Cathod
(target)

E x B drift
(b)

Figure 1-21. Schematic representation of (a)

/%

Erosion area

/\

 

Magnets

planar diode sputter

deposition [modified from Thornton, 1982], and (b) planar magnetron

[modiﬁed from Ahmed, 1987].

142

EQUILIBRIUM
TEMPE RAIU RE

r FF
LIQUID SUR ACE DI USION
AMORPHOUS

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SOLID
7- PHASE N'IORF'IOUS >- PHASE 7- PHASE
I . | |
' I
' I
' BU K ourru ION
I ; 5- PHASE L S
I
: 1 Ii
J— ‘l— a- PHASE

 

 

 

 

 

,-
SUBSTRATE iEMPE RATURE

Figure 1-22. A schematic illustration of the effect of substrate temperature

on phase formation for a hypothetical polymorphic material [Thomton,
1977].

143

 

 

SUBSTRATE
TEMPERATURE (T/Tm)

(a)

  

PRESSURE "“2 TEMPERATURE
(MICRONS) mm)

Figure 1-23. Structural zone models for physical vapor deposition
processes. (a) The model proposed by Movchan and Demchishin
[1969] for electron beam evaporations; (b) The model proposed by
Thornton [1974] for sputtered metal coatings.

 

 

(bl

o-tS' T-IZOK

 

 

 

 

 

 

 

 

 

 

Y T
N.
> lit-0.01 nm/s R-l nm/s DIV-r100 rim/s
)- 1.0 ’ e <
E 0.8-
2 .
5 0.0
E Ta 1,, Tc
1 1 l (d)
A l A L
300 000

 

400 5“)
$1857”?! TEWERATIRE T (Kl

Figure 1-24. A computer modeling of film growth using 2D hard spheres by
Muller [1985]. (a-c) show the inﬂuence of temperature on packing density of
ﬁlms. (d) displays that the temperature effects packing density while the
deposition rate changes the transition temperature of density-to-unity density.

145

(o)

 

(b)

 

 

(c)

 

 

 

 

Figure 1-25. Schematic microstructures and grain distributions for thin ﬁlms
undergoing secondary grain growth [Thompson, 1990]. (a) Microstructure
before the secondary grain growth occurs. (b) Intermediate stage in which the
grain sizes are bimodelly distributed. (c) After the secondary grain growth
microstructure shows crystallographic (ﬁber) texture.

146

/ Nylon turned cover

 

 
  
  
 
 
 
  

Brass cover
plate

\ .

I Membrane bias

Fiber optic with ,
pressure P1

lens formed in end
Kapton insulation ~—.-.

‘ Silicon valve substrate

.1 I L
Outlet Inlet P1 Outlet

(a)

Semiconductor
laser
Film heated Unheated film plastically Heated film returns
I 5 f0 150° C deforms (approx. 3%) to annealed shape
Laser heats one of '1' or '0' determined and set
the two actuators by which actuator is used

(b)

Figure 1-26. Schematic drawing of two devices [Johnson, 1992] using
the shape-memory effect of NiTi thin ﬁlms: (a) a microvalve, and (b) a

microactuator.

147

 

Figure 2-1. Photograph of magnetron sputter machine developed for this
study.

148

    

Seam £5 8m come—gov 88m? .532; coboawaﬁ mo Exams“. ocmﬁonuw .N-N 2=wE

 

 

 

 

 

 

 

 

 

 

 

0

E
4/ 955

an: $5.90“.

 

        

     

/ 520m

 

 

 

 

  

 

 

   

 

 

 
 

_
nEsa 862.5 \ @
o2m> 2:92... .
amHmEozomam moaaooohEmck
mmmE use ‘
magma Eazom>
95.0 \s
@589. .255
SEEREE.
09.38 5:33
.295 cozmcmms.
9:80
23% 5on z m M
582 08:35
\‘
9E6 693 M
.28 EoE\
233m .958 c
38:85

 
  

858 8ng

  

9.53 \

563 95000 _
_
_
_

_ .28; xomm
“0.50 ._< V

 

        
   

\

 

 

 

 

 

 

 

ILLI

 

  
  

 

 

 

Eoﬁom

sacs:

_o._Eoo
Bo: owns.

 

 

149

ISOTHERMAL WEIGHT LOSS
(Type HN Film. 25 um (1 mil))

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\b-lOO°C (752°F) Air
\~450°c (842°F) He-
M400°C (752°F) He
\ \1
‘b \Isomc (932°?) He
20 \V
\
|\ \x
30 .
V \ \
k \ \
40 | \‘
9" I \
WEIGHT I \
LOSS ‘
50 v
I I \
I l ‘
60 I I TSO-‘CIHPF‘lAir
| I
70 l g X
I I 575 C [1.067°F) He
I |
80
I |
I \ b550°C (1.0221’1r He
90 g ‘,
‘ 600°C(1.112°F)He
500°C.(932°F){\ir |
100 \ — aaO°C(‘l.022°l'-')Air
0 100‘ 200 300 400 500 600 700 000 900 1.000 1.100
TIME.Minutee

Figure 2-3. The weight loss characteristics of Kapton in air and helium
at a heating rate of 3 K/min [Product Bulletins, DuPont, 1993].

Heat Flow (arbitrary unit)

End0¢—> Exo

150

sputter deposited
from Nl51Tl49 target

sputter deposited
from Miss TisoCus

 

700 750 800 900

Temperature (K)

Figure 2-4. Differential scanning calorimetry (DSC) curves of
sputter-deposited binary and ternary NiTi ﬁlms at a heating rate
of 20 K/min showing the crystallization temperature.

151

 

 

HP 6010A DC
power supply

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I Installed in the
Quartz / vacuum chamber
lamp ,
® 6
‘__\\\\ JJJ/__L__i
,.__1Fj “ , u E Ly -1
Gold-plated spring
K J-type thermocouple
—£:é“v XY recorder
Kapton® I

 

 

 

k

 

 

 

—|I|I|1—1:—

J

12 V 3000 ohm

Figure 2-5. Experimental set-up for in situ observation of electrical
resistivity change in crystallization annealing of NiTi/Kapton laminate.

Electrical Resistivity (10'3 ohm cm)

Temperature (K)

1.6

1.4

1.2

1.0

0.8

0.6

800

600

400

200

152

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Time (minute)

t\\ Onset of crystallization: ,,,,,,,,,,,,,,,,, i
\ / sharp resrstrvrty drop
..... , /
--------------------------- iXmorphous: E E Crystalline phase:
negative temp. *“1 \ l“"" positive temp. coefi.
* """"""""""""" coeff. of resistivity ; E of resistivity
................. : /i
i {\w’ \
/‘ /l
/ //\
// \\ / \
/ \ / \
/ \l
30 60 90 120 150 180

Figure 2-6. The diagram showing the electrical resistivity of NiTi ﬁlm
as a function of temperature during crystallization annealing treatment

at a heating rate of 6.7 K/min.

Residual Gas Pressure (torr)

153

Temperature (’0)

100150 200 250 300 350 400 450 500 550

 

 

 

1.00E-05 l 1 1 1I l 1 l 1 l 1 l 1 l 1 l 1 [44
::::::::::::::::::t::::::::t:::::::j::::::::
CIIIIIIIIF ........ 1r ........ g ......... g ........ i
""""" r'PtotaI ill-j
""""" {Jill
"""tfl‘i ----- a --------- s ------- -.-a---'----i

'.-_--_.I_ - - - 1- _ -!--- ..... --
........ : E.- .. IE. I : .‘

: 1 1 i

1 1 1 1

1005-06 :::::::::t:::::::::c::::::::E:::::::::i:::::':::
::::::::2;:::::::2;::::::::::::::::::E::1:92;:
------ PN2(atomicmassunits: 28)"“"§',"”""

. : """"" : """" 3': """""
........ .1 .............1
‘90:9¢o.ioooo§" it

i 3 AA 5 A‘i

i :1 Mg; :

1.00E-07 -;-----;-'r ------ “1‘; ------ g -------- . ---------
:'3::::':F::::i§“*: ::::::.::::::::::::::::::1
2222222ZZEAZAI--IZ:§ZPcoz(at0micmassunits: 44)ii
A'm“? """" :r'P02(atomicmassunits: 32) "
......... r--------l.- ------r---------1--------4
o o 0' ' ' l o
......... i?-?-..-...:.. O .-..E-.-. .-.-:...-..-----

= E i E

1.00E_08 1 l 1 i 1 l 1 i 1 l 1 1' 1 l 1 L1 1 1

 

350 400 450 500 550 600 650 700 750 800 850

Temperature (K)

Figure 2-7. Residual gas pressure vs. temperature curves of Kapton
ﬁlm at a heat-up rate of 5 K/min.

154

Thermocouple

  
  
 

 

 

 

\ Heater
Sputter source I
J '
__J
Kaptonc
substrate Substrate
holder
(8)
Thermocouple
Shield
Sputter source
1.31“
' Kapton®
I 9“ substrate
__l 1
7,4,
Quartz lamp '\ Substrate
holder
1..J

 

 

(b)

Figure 2-8. Schematic drawings of substrate heating devices which
have been tested in this study. (a) A conventional substrate holder
with a heater situated under the substrate. (b) A heating device
consists of four 625 watt quartz lamps clustered around the sputter
source to irradiate the substrate.

155

 

 

          

 

.838 wagon 03533 topic?“ we 32> nope—98 2: ma??? 8:832: oneEonom .m-~ oSwE

     

2? 95mm:

 

 

  

6 .
eem " v,
_ _ _ v
d moasoooccoc _ n u u _
m _ _ _ _ .
ea \ m u u m m 3 .
_ _ . _ _ .
.520: u u _ _ u u _ _ _ .
_ _ _ _ _ _ /
6604 29.85 m n n u " H u u M H n a
J/ _ n n _ _ n " _.” _ _
u u u n n I; _ n n
_ _ _ _ _ _ _ u u _ _
. _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _
_ _ u " _ _ _ n n "
Ema / 11:11:28.
.2335 _ _ _ u n u u _ _
_
2:550 /7/ g . u u n u m
89m... m j _ n n n
3.58 actogam /, . _ “
eczema—2 mwugmnaw F A
[P

   

mega 2:53 B
822% 2.3 958:

   

225m

 

 

253 .6800

.28: xomm

uwuwwc E0.”—

156

ﬂ: NiTi film NiTi/Kapton®

Kapton® film microlaminate

i l/ Collodron layer Stick microlaminate

[ We Slide cover glass on cover glass

 

 

 

- Photoresist .
F___j Spin photoresist over

F j on microlaminate

 

 

o-ae
r! 1] Sftbk

illiiiiiiii UVsource

 

Expose

 

 

 

Develop photoresist

 

 

Post-bake

 

 

Etch NiTi film

 

Resolve photoresist
and glue

 

 

 

Microlaminate element

 

Figure 2-10. A patterning-etching process for fabrication of
NiTi/polyimide actuator elements.

157

7.6 um-thick polyimide

(Kapton®) substrate

/

 

 

f"—_'—__—___‘__‘_“_“

 

4(— 0.44 mm

3mm

 

 

 

 

 

Figure 2-11. Pattern for making eight NiTi/polyimide actuator

elements.

Substrate Temperature (K)

800

700

600

500

400

300

200

158

 

 

 

     

 

 

 

10E-2
Q 1
1“ j 5
r 1 Base ressure '
1/ p _
\\Q _ 2
” ‘\ Baking chamber Deposrtlng
‘\ and specimens / " i 5 105-3
" — 5
r- _‘ 2
\ Cooling
“0"" : 1015-4
- °-o- :
Substrate ~~~~ ‘0. a
temperature “~-~o~ : 5
““0 A
- 2
l l I» i 1 AV 1 i l i 1OE-5
O 2 4 24 30 48 50 52 54 56
T1me (hour)

Figure 2-12. A plot of substrate temperature and base pressure vs.
time for a typical deposition run (T = 670 K).

Base Pressure (Pa)

159

 

Resistance furnace

 

\VV‘V‘VVNWWVN‘\\\\\\NVV\\\NNN.

I’ll/Ilf/I/IIIII[I’ll/IIIIIII/

 

 

 

 

 

 

 

 

 

 

 

Quartz

chamber

 

To turbomolecular

vacuum pump

Figure 2-13. A quartz-tube vacuum furnace used for heat-treatment

of Ni i ﬁlms.

160

NIT i thin film Interface Free surface

/ \/

 

\\\ \\

 

 

 

 

 

Quartz substrate
Collodion Collodion
layer layer

,/

 

 

     

Electrolyte
twin-jet streams

i i i

i i 1
l l 1

Disk with the Disk with the Disk with the
interface midplane free surface
sample area sample area sample area

Figure 2-14. A drawing illustrating the preparation of TEM disks.
Three sampling locations along the depth of the NiTi thin ﬁlms could
be obtained by a masking technique.

161

The superlattice diffraction
in zero-order plane

M1»

002 012 T12

 

The bcc diffraction

in zero-order plane “ ‘ \ [110] zone

 

 

 

 

 

   

 

 

_. f\ t
/ 001 You 111

The superlattice .

diffraction in The DOC diffraction

one-order plane / 1n one-order plane

—q 35 Q-
000 010 110
(a)
Ring consisting of (01-1.), [110]
(011), (101) and (101) A Diffraction arcs
diffractions in one-order
plane
\ .
‘7 v ’ .
i
d Arc ~\ ( Image plane
‘\- ' ' (Ewald sphere)
ooo . I
Q \~
Arc -\

Ring c_qnsisting of (710) Tilting 8X15
and (110) diffractions

in one-order plane

(b)

Figure 2-15. (a) The reciprocal lattice of the BZ structure in [110] zone.
(b) The geometry for calculating the positions of diffracted arcs upon
tilting. Where p is the angle between tilting axis and the arcs in the
image plane, 0 is the tilting angle of the specimen, d is the distance
between the zero-order reciprocal rings and the higher-order reciprocal
rings and R is the radius of arcs in the image plane.

162

.mBBEomEB 259; an aim
a .5 Z 55 mates—«>0 a8 mama—Ema 52 0:28 we E5833: ozaﬁoaom dTN ocnwE

 

      

5&6on E__u_

925230 :95
9:5: oowtzm a

.3820 ﬁEeEEQScm

   

9:835 953: m_xm o__wco._.

              

        

 

8.32:8 2m. 26E

.0559: 36865

 

163

Grips

/ \

 

 

 

 

 

l
l
l E
<— _-...1... ..... .. .......... 1---}; ..... _. ....... it.
I 1-
!

 

 

 

 

 

 

 

Length = 6 mm E = 50 GPa
Width = 1.5 mm v = 0.3
Thickness = 6 pm

  

: A
I

No Xdisplacement Y
No Ydisplaoement X Same X displacement

Figure 2-17. Finite element grid for analysis of rectangular thin
ﬁlm tensile specimens.

 

dosoom owswm 05 mo menu 05 So: 3on a Esﬁvaﬁ 05 3:82
32% 39053 EECQE 05 85 :08 on :8 a .5882? 0:38 EE 55 we 32% Raﬁ—Em
12552: 2: wEBoﬁ min—a5 H5820 BEG 05 Set conﬁne 92: 58:8 < .wTN Emmi

333%

vofwomw.

IIIIIIIIIII gee.
III-ll... . .3....

vo+o©mv.
voicmnm.
370mm”
wore—ow.
vofwnmn.

2709:.

 

The number of Fractured
Tensile Specimens

165

 

 

 

5—

iL—d
4..

0 0.510 x
3..
2—
1..
o .

 

 

 

 

0.5 ~

1 1 1
N. 0". V.
O O O

005 -
0.1
015 ~
0.25 -
0.35 -
0.45 r

The Normalized x-Position oi the Fracture

Figure 2-19. A plot of the number of samples fractured versus the
normalized x-position where the fracture occurred. Each number
is collected from the interval of (x, x + 0.05).

166

920071171

 

Figure 3-1. TEM bright—ﬁeld (BF) image in (a) showing a completely
featureless structure, and the associated selected area diffraction pattern
(SADP) in (b) showing a broad, diffuse halo pattern which are normally
observed in the amorphous materials. The micrographs were taken from
the midplane of the film H573 at a tilt of 0 degree.

167

2©© nm

 

Figure 3-2. TEM BF image in (a) and the associated SADP in (b),
taken from the interface of the film H598 at a tilt of 0 degree, showing
the amorphous structure.

168

1

1",“.
_ this,
226112111111 1

5!

a. s:
f.

,‘.,

 

Figure 3-3. TEM BF image and the associated SADP taken from the
midplane of the ﬁlm H598 at a tilt of 0 degree. The SADP reveals a
single-crystal diffraction pattern in the [1 10]112 zone together with a
superimposed diffuse halo pattern scattered from the amorphous
background. This initial crystallite has an in—plane size greater than
1 um.

169

 

Figure 3-4. TEM BF images and the associated SADP’s taken from the
surface of the film H598. The BF image (a) and ((1), respectively, were
recorded at a tilt of 0 and 45 degrees. Elongated grain images in (d)
suggests a columnar grain structure. The SADP (b) and (c) were
recorded at a tilt of 0 and 30 degrees, respectively, showing a strong
(110) ﬁber-texture. The non-even intensities of the diffraction rings in
(b) reﬂect the restricted in-plane grain orientations.

170

 

Figure 3-4. (Contiuned) TEM BF images and the associated SADP’s taken
from the surface of the film H598. The BF image (a) and (d), respectively,
were recorded at a tilt of 0 and 45 degrees. Elongated grain images in (d)
suggests a columnar grain structure. The SADP (b) and (c) were recorded at
atilt of O and 30 degrees, respectively, showing a strong (110) ﬁber-texture.

The non-even intensities of the diffraction rings in (b) reﬂect the restricted
in-plane grain orientations.

171

 

Figure 3-5. TEM BF image and the associated SADP’s taken from the
interface of the film H623. The SADP (b) and (c) were recorded at a tilt
of 0 and 30 degrees, respectively, showing a strong (110) ﬁber—texture.

172

 

Figure 3-6. TEM BF images and the associated SADP’s taken from the
midplane of the film H623. The BF image (a) and (11), respectively, were
recorded at a tilt of 0 and 45 degrees. Elongated grain images in (11) implies
a columnar grain structure. The SADP (b) and (c) were recorded at a tilt of
0 and 30 degrees, respectively, showing a strong (110) ﬁber-texture.

173

 

Figure 3-6. (Contiuned) TEM BF images and the associated SADP’s
taken from the midplane of the ﬁlm H623. The BF image (a) and (d),
respectively, were recorded at a tilt of 0 and 45 degrees. Elongated grain
images in (d) implies a columnar grain structure. The SADP (b) and (c)
were recorded at a tilt of O and 30 degrees, respectively, showing a strong
(1 10) fiber-texture.

174

 

Figure 3-7. TEM BF image and the associated SADP’s taken from the
surface of the film H623. The SADP (b) and (c) were recorded at a tilt of
0 and 30 degrees, respectively, showing a strong (110) ﬁber—texture.

 

 

175

 

Figure 3-8. TEM BF image and the associated SADP’s taken from the
interface of the film H673. The lenticular precipitates in (a) can be
distinguished (indicated by arrow). The SADP (b) and (c) were recorded
at a tilt of 0 and 30 degrees, respectively, showing a strong (110)
fiber-texture.

 

 

176

 

Figure 3—9. TEM BF images and the associated SADP’s taken from the midplane
of the film H673. The BF image (a) and (d), respectively, were recorded at a tilt
of 0 and 45 degrees. A precipitate particle displays a perfect ‘butterﬂy’ shape of
the image contrast as indicated by the arrow. Again, elongated grain images in
(d) implies a colurrmar grain structure. The SADP (b) and (c) were recorded at a
tilt of 0 and 30 degrees, respectively, showing a strong (110) ﬁber-texture.

 

177

 

Figure 3-9. (Contiuned) TEM BF images and the associated SADP’s taken
from the midplane of the film H673. The BF image (a) and (d), respectively,
were recorded at a tilt of 0 and 45 degrees. A precipitate particle displays a
perfect ‘butterﬂy’ shape of the image contrast as indicated by the arrow.
Again, elongated grain images in (d) implies a columnar grain structure. The
SADP (b) and (c) were recorded at a tilt of O and 30 degrees, respectively,
showing a strong (110) fiber-texture.

178

0.; .0 ‘
‘s

V
.0
v»

JD 9 Q 11"]

.- ' 7 U
.Q' ,'
6
‘-1
.I

Figure 3-10. TEM BF image taken from the surface of the film H673.

 

179

 

(a)

 

 

 

 

—[111]

1‘ Precipitate

.m_____

—71_.__¥__

— Tensile
_strain fields

 

 

 

 

(b)

Figure 3-11. (a) The high magniﬁcation image of precipitate with ‘butterfly’
shaped contrast. The strain fields around Ni4T13 particles come from the
lattice mismatch between particles and B2 matrix. (b) The habit plane of
precipitate is {111}B2 and the maximum mismatch of -2.9 % is along the
normal to the precipitate disk. This minus value indicates a tensile strain
perpendicular to the precipitate disk. [modiﬁed from Kainuma and
Matsumoto, 1988]

 

180

 

Figure 3-12. TEM BF image and the associated SADP’s taken from the
interface of the film H723. Extremely ﬁne grains, 25 nm in diameter, were
nucleated on very thin initial amorphous layer. The SADP (b) and (c) were
recorded at a tilt of 0 and 30 degrees, respectively. The unchanged SADP's
upon tilting imply the absence of the grain texture.

181

 

Figure 3—13. TEM BF image and the associated SADP’s taken from the
midplane of the ﬁlm H723. The SADP (b) and (c) were recorded at a tilt
of O and 30 degrees, respectively. The unchanged SADP's upon tilting
display the random grain orientations.

182

 

Figure 3—14.TEM BF image taken from the surface of the film H723.
The precipitates show smaller size than that in the midplane. Three
ranges of precipitates can be seen in the low part of the micrograph.

183

 

Figure 3-15. The micrograph and associated SADP taken at the midplane
of the film H703 revealing crystalline B2 grains and transgranular
precipitates (indicated by arrows).

184

 

Figure 3-16. TEM BF image and the associated SADP’s taken from
the interface of the film H773. The SADP (b) and (c) were recorded
at a tilt of 0 and 30 degrees, respectively. The micrographs display
extremely ﬁne grains and the random grain orientations.

185

 

Figure 3-17. TEM BF image and the associated SADP’s taken from the
midplane of the ﬁlm H773. The SADP (b) and (c) were recorded at a tilt
of 0 and 30 degrees, respectively. The micrographs show coarse
precipitates and the random grain orientations.

186

 

Figure 3-18. TEM BF image taken from the surface of the film H773
showing the coarse grains and precipitates.

187

 

 

 

 

 

 

 

 

 

 

 

————-—>
O O
o o O
O .
a O O
O
O
O
\

 

 

 

 

 

 

Figure 3—19. Time-lapse sequence illustrating the development of the
Johnson—Mehl microstructure from the untransformed body [Mahin,

188

The surface region
is active since
surface diffusion
takes place

The bulk region
is frozen-in
since no bulk
diffusion occurs

 

 

 

substrate

 

; i i i Deposit flux

T Deposition rate

\ / ln-plane grain
growth rate

The shape of the columnar
grains is controlled by both
the grain growth rate and
the deposition rate

Figure 3-20. A schematic diagram illustrating the formation of columnar
grained structures. By means of superposing two-dimension grain growth
upon ﬁlm growth (increasing thickness), a columnar structure can be
formed as shown. The shape of the columnar grains is controlled by both

the in—plane grain growth rate and the deposition

rate .

189

.couﬁaoto 053-5

cobowoa o: 55 :5 .28?“ 08533 8 BEER 2 @8252 2a WEE» %o 283 Q: C on... :0;
.Baxoﬁconc 2: 3 $55 a Eamoa mueaﬁomEB 532 “a page“ mac—E 3v .mcoussto Eoueﬁ
>55. mEmDo 2: Ba 38% 05 .«o macs 88.5w use cog—BE 05 .«o .5323 5:95 8 26
“semi—mi Ho: 2d 23% mo momweoao 958:: was 38:; 2: mus—38388 93533 kiwi 2 A8
dengue 23x8 :o 36 53» 283-5 mo Soho 05 336% ESwEu 238028 < ._N-m 25mm

3V 3

      

      

  

  

9383
25:995..

Suzanna
msochE<

Ecomsun 5.5an
22on .8: 8F 5 85:25 585m

190

‘ (>234?

 

Figure 3-22. SADP sampled at the midplane of the film H673. The indices of
Ni4Ti3 phase were underlined for purpose of distinguishing them from the B2
reﬂections. The ring pattern in the figure also shows the (110) grain
orientation, since the (310)132 reﬂection having the radius of 64.3 mm absents.

191

 

Figure 3-23. Two SADP's, each including approximately one grain,
sampled at two areas of the midplane of the film H773. (a) [13-3132 //
[1-11]Ni4Ti3 zone, and (b) [-13-2]32 // [1-10]N14T13 zone.

192

 

800
E? 700
5
.2 600
S
0
N 500
an
“6
5;, 400
‘5 Surface
E
m
E 300
8
c_u 200
(L
<_'=

100

Interface
0

573 598 623 673 703 723 773 798
Deposition Temperature (K)

250

200

150

1 00
Surface

Interface

50

Diameter of Precipitates in
Longitudinal Direction (nm)

 

0
573 598 623 673 703 723 773 798

Deposition Temperature (K)

Figure 3-24. Diagrams showing the size changes of B2 grains
and precipitates as a function of deposition temperature in three
layers.

193

.m 35% do oeeeaee
03:33 a a caﬁomoueozsmm 2:: F2 :232 8m 3605 35883 < .mm-m Semi

2393th mugmnsw

Ash x: wwvd onto hmvd F36 mmvd mowd vmmd wumd Nmmd

 

0: wk QR 8n m8 new new mmo 8m 2m
_ _ _ _ _ _ _ _ _ .
8: ooamxmoownohw madame“
B .8 2:8 See 25
c. 9.302% 05 up 052
29595

 

 

 

 

 

 

 

 

 

  

 

€335 95:75 o M
w

v.0 N6 0 F
o .3 t m.
No I N M
to ‘ , m a
. . z m,
, (w

36:909.“. v

m cozﬂcoto Jn v, m
23x2 8F 5 5? 9:96 30:925.

88:9 5;» «.580

194

£sz .38 mEE coma Ea BEBE.» .38 mo 35 8 3:55 "0838.588?

53 25E sea a. e5 E 58 E 223:6 53 .E Seam 5 due .38 Deﬁne 388 e8
ecu—End 3:5 _ Z nmsoht @8353 on $8 mBSEUEQ owes .5 :Bwoubaw E .mEEm am 85:28
we 3:me 2e 2:: d comwom E 68% 32825 E negate Sara E a; Show be: wBSEEoE
0505023923»: .3 :omwoubzm E .Bauabm E £82925 bﬁuonow 2e :2on $5 an 3:893
255 53853386 038 8: can E. 28 _Z .8 gouache course 2: A :Ewom 5 93083238
SE :23 Z so gamete F can _ Z we Saute 05 managed wcouﬁcomoa oumﬁonom .©N-m oSmE

Al 2298th

 

   

 

   

     
   

E E $.33 5.936 8.986
S www-mmm m 5-85 83% e
. . ., . 875:5
moﬁEEou c2256 v==m , , moﬁEEov :Qwsgu momtqm :82:ch oz

 
   

$8558
counts 58

8335 .535ch oz

   

.: . «03558 57.2% 83.5 .

£3 08298 2:: + 82a .25 99% 23 + man 532 + :o 92:
235 mm .9355 885 mm 8:63.00 2.65 mm 5:83.00 30:995. w c <

 

 

 

5 85mm __ 85mm _ 8.9m

 

 

 

 

 

seintoruisomgw

195

 

Figure 3-27. TEM bright-ﬁeld image of ﬁlm A1 which was produced
by aging the film H623 at 858 K for 1 hour and furnace cooling.

 

Figure 3-28. (a) TEM bright-ﬁeld image and SADP’s of ﬁlm A2. (b) The
inserted picture taken at 45° specimen tilt revealing columnar grain
structure. (c), (d) and (e) are SADP's at three tilt angles 0“, 30° and 45",
respectively, displaying (110) texture. The axis of tilt is indicated in (d)
and (e).

197

 

Figure 3-28. (Continued) TEM bright-ﬁeld image and SADP’s of
film A2. (b) The inserted picture taken at 45° specimen tilt revealing
columnar grain structure. (0), (d) and (e) are SADP's at three tilt
angles 0°, 30° and 45°, respectively, displaying (110) texture. The axis
of tilt is indicated in (d) and (e).

198

m

4

2 GM) Lmt'n‘m ;

1;

'ﬁ

 

Figure 3-29. TEM bright-ﬁeld image and SADP of film A3 which was
produced by aging the film H723 at 858 K for 1 hour and furnace
cooling. The SADP in (b) was taken at 30° tilt showing random grain
orientation.

199

 

Figure 3-30. TEM bright-field image of film A4 which was produced
by aging the film H723 at 813 K for 1 hour and furnace cooling.

200

“goo mm)

-

was:
. 3‘

 

Figure 3-31. TEM image of the film A5 obtained by annealing film H673
at 988 K for 1 hour and furnace cooling. The size of grains have increased
during annealing. Obviously, the bulk diffusion for NiTi grain growth have
occurred.

201

:2 © © mm

 

Figure 3-32. TEM image shows the microstructure of film A6 produced
by annealing the ﬁlm H673 at 1123 K for 1 hour and furnace cooling. The
film is consist of large grains and lenticular precipitates. The precipitates
align along well-deﬁned crystallograghic planes of the matrix. Besides,
very ﬁne precipitate particles with strain ﬁeld contrast also can be noticed
in the area between the large precipitates.

202

 

Figure 3-33. TEM image of ﬁlm A7 obtained by annealing the ﬁlm
H673 at 1073 K for 2 hours and air cooling. The precipitation tends to
occur more rapidly at grain boundary areas since the grain boundary is
a favorable site for heterogeneous nucleation of precipitation.

203

 

Figure 3-34. Cross-section SEM micrograph of A7 displaying
coarse-grained microstructure. Deep etched surface morphology
reveals the crystallographic facets of the grains.

 

Figure 3-35. TEM image of ﬁlm A8 deposited at 1073 K, 2 hours and
673 K, 1 hour two—stage annealing.

.iao

 

Figure 3-36. BF image and two single-crystal SADP's taken from the large
precipitate plate in film A3. (a) An isolated precipitate plate in the edge of
the TEM foil was examined in favor of excluding diffraction spots from the
matrix. (b) The [0-10] zone of Ni3Ti2 phase. (c) The [110] zone of NiaTiz
phase.

Relative Intensity

206

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0‘3 #1 .s
8 E l;
_ E, 3 Z T: 298K
- $2 a
3‘5 ........ 3-5
N h -
4; 3- §i
‘ II N
‘0 u
eL’b
2 .2
t; E
2 z
E 5.
_ ..................... 6 _______________ ('5
[s
§ N-
‘ II II
I: 'b
M...“-
35 40 45 5° 55 60
26 (degree)

Figure 3-37. XRD curve of the ﬁlm AS recorded at the room temperature
(298 K). Besides (110) BZ peak, four peaks match the reﬂections of Ni4Ti3
phase at 20 angle between 35° and 60° within 2 % error.

207

 

Figure 3-38. BF images and single—crystal SADP's taken from the ﬁlm A6.
The index of large lenticular precipitates could be completed by assuming
that they possess the Ni4Ti3 structure. (b) and (d) are SADP's of zone
[0-12]Bz // [001]Ni4Ti3 and zone [230]}32 // [221]Ni4Ti3, respectively, taken
from indicated areas in (a) and (c). The underlined indices represent the
diffractions of precipitates.

Relative Intensity

208

 

 

 

 

 

 

 

 

 

 

 

 

N
CD
' 9 T: 353 K
C
5’2.
N
H
_ ‘0
.2 9
'7. E
_ E z
:3 7:
9‘ 5
. ...................... £1... ........ g,
- as
(O
as t.
_ N N
II II
‘0 ‘0
I I I I T I I I ﬁr j I 17 I I I I T T l l
35 4o 45 50 55 60

26 (degree)

Figure 3—39. XRD curve of the ﬁlm A7 recorded at 353 K showing (110)
B2 peak and two Ni4Ti3 phase peaks at 29 angle between 35° and 60°. The
reﬂections of Ni4Ti3 phase are very weak due to its ﬁne size.

Relative Intensity

 

209

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3N .9
a m I; ‘2
E R? E T— 353K
_ ____________________________________ 5.5.. ......... 2-5
n V " I:
i: "- E 8 -
.3 N A 1- :) ~9
_ i g :2 oi 0. E
,_ 0:) II N 2
<31 ~: 7°- .. r:
9}, co ‘/ ‘0 N
rs o
co ". V
s N U N“
- ll 8
N b _
- u ‘-
‘c II
A b
\ A
('3
b ‘5. E r;
1- v— -— n
_ Z ._
'5 \ / '7 A L3 T: 398 K
V o
- n v N z
— - Y-
3 i— A o m ;:
.—--~.: 5*: >3“
I.— A N O V a)
.2 .g : n ". m‘ i:-
.. Z ; '0 N h .3
:3 § 3 _ n (<31 2
“J g5 i ll g
_ ..................... 9" ......... i ....... F: b e
' i N -
§ ’ ll g
oi ‘c ‘9
,_
ll
‘0
I ‘T ﬁI T T I I I I IA I I I
45 50 55 60
26 (degree)

Figure 3-40. The XRD curves of the film A8, taken from the same specimen
at 353 K and the room temperature respectively. The curve (a) identiﬁes the
(110) BZ peak and ﬁve Ni4Ti3 peaks in 26 angle between 35° and 60°. In the
curve (b), the same set of the Ni4Ti3 reﬂections appears, however, the (110)
B2 peak splits into two NiTi R-phase peaks: (011) and (0-11) with a
separation about 0.4°.

210

 

 

 

 

 

 

. \‘ a "r ~~ .
" ', "a .‘ ' ,
I II 6 3 ~\ "'

gs ~ I, ~ I
-FHHH7 , . .
\ '-.~ o

f (Fig. 3-9) .. ,, ,. .
I, :~ . " z: Annealing

 

 

 

u .
n, .
1" g ‘Z: _.: 1.1 " I .
i: s ~ -.- r‘ n. ,-l. ~
,’ .~ ~‘\‘ ‘. “ \.,
\ 1" 0 ~‘.\ o _ ‘1 "
~', I “ r~ ~ \~’ ’ ‘
~ . .‘» '.,‘ '— ‘
I" . ~‘ , . -
. ..
- - . . ~ bl T
_-. .H I, ‘ ' I4 l3-:
~ - ~ \ ' .~ .
— I u. \ . r . .
.\ . - . .
~‘ \I ll ~ lII
‘ , I .u
.~ ,. \ ~
I \ l I ’a ‘\I ~

 

 

 

 

0.5 pm

 

 

 

I \\

 

v Film H723
(Fig. 3-12)

\ 9

legit

\ Annealing

 

If,

 

 

 

I

/ Ni4Ti3
\

 

 

Figure 3-41. Schematic drawings of in-plane views of various microstructure
formed in the post-deposition annealing.

211

1. Martensite H R-phase H Austenite

A M.

ng

 

 

 

 

.«2‘
.g
.9
0)
d)
a:
,8
§
LT!

d

( ) M,

A.
Temperature

2. R-phase H Austenite

l (3)

‘

 

 

 

 

- E

g 1% R i A

a Ta l

LU a? Ti
Temperature

3. Martensite H Austenite
4* (f) M A:

g 3: Ar A

3a M.

I.” m M8 ‘
Temperature

Figure 3-42. Schematic diagrams illustrating the classiﬁcation of electrical
resistivity vs. temperature curves for N iTi alloys. (a) Two-stage curve
involving Martensite H R-phase H Austenite phase transformations. (b, c) The
M H R and R H A curves decomposed from the curve (a). Curve ((1) represents
a two-stage curve with an incomplete austenite —> R-phase transition. (e) The
curve involving R-phase H Austenite transformation which is the same as the
curve (c). (f) The curve involving Martensite H Austenite transformation.

Electrical Resistivity (arbitary units)

212

 

V

a. Film H5‘}3 Ni51.5Ti48.5

\

 

b. Film H623 Ni52.3Ti47.7

The resistivity of film'H573 has
been compressed three times
with respect to other curves.

 

 

 

 

 

 

c. Film H673 Ni51.oTi49.o

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100

Figure 3-43. Electrical resistivity vs. temperature curves of as-sputtered

................................. / As TR
M, A,
d. Film H723 Ni51.1Ti4a.9 Mf/L—K
// I __
I I
/ A. Th
Ms A1
6. Film H773 Ni51.4'|'i48.6 ¢
Mr /
VII/4 A 5'
/ As 7h
L14111111111111111111111
150 200 250 300 350
Temperature (K)

N iTi ﬁlms deposited at various substrate temperatures.

213

 

Figure 3-44. Bright-field images of midplane of the ﬁlms H723 and
H673. The micrographs were recorded at 101 K, which is well
below their Ms temperatures. (a) The image of ﬁlm H723. (b) The
image of ﬁlm H673.

0.15

0.10

0.05

Heat Flow (W/g)

-0.05

-0.10

Electrical Resistivity
(arbitrary unit)

214

J/g

As
17.3 J/g 313.7 K
A.
310.2 K

b. Resistometry

M1

 

180 200 220 240 260 280 300 320 340 360 380 400

Temperature (K)

Figure 3-45. DSC scans (a) and electrical resistivity vs. temperature
measurements (b) on ﬁlm H703 having composition of Niso.sTi49.2. In (a)
the second heating scan from 223 to 373 K is indicated by double arrows.

Electrical Resistivity (arbitrary units)

 

f“'

 

 

 

 

 

215
a. Film A1: ’Nisaarim M
T3 = 623 K 5
Ta = 353 K, 1h, F.C. M! I:
.............. l j _
g I
M:
*“b. Film A2: Ni51.oTi49.o ------ i A-

 

Ts = 673 K
Ta = 858 K, 1h, F.C.

...........

\\

 

 

 

 

C. Film A3: Ni51.1Ti48.9
Ta = 723 K
Ta = 858 K, 1h, F.C.

......

 

 

 

........

 

 

A/
1111

 

 

 

A. 'r'R

 

 

 

 

100

150

200

Temperature (K)

Figure 346. Electrical resistivity vs. temperature curves of annealed NiTi

ﬁlms.

Electrical Resistivity (arbitrary units)

216

 

d. Film A5: :Ni51.oTi49.o
Ts = 623 K
Ta = 988 K, 1h, F.C.

 

e. Film A6: ‘Ni51.oTi49.o
Ts = 673 K

 

Ml

 

/

“.4

Ta=112§ K,1h,F.C. /
/

 

---- 1. Film A7::N151.oTi49.o
Ts = 673 K
Ta = 1073 K. 2h, ac.

 

9. Film A8: Ni51.oTi49.o
Ta = 673 K
Tai =1073 K, 2h.
Ta2 = 673 K, 1h, R.C.

 

 

 

 

 

 

 

 

 

250

Temperature (K)

Figure 3—46. (Continued) Electrical resistivity vs. temperature curves of

annealed NiTi ﬁlms.

217

 

.0523 Evans 05 E 5:3:88 access—Ba Emu—LEE,“ 05 .3 @823 803 Rama 05 5 350m

22» 55 5320 .22:8%2 2 e5 RE .20: .8 £2 85 as 3: .32 3 2.53 M 3.
an 86:05:. 835 8 8823. .328 05 .34” gamma 5 :Boﬁ mecca—2 Eczema—0-8635 9

@5383. .2 es. SE .90: 2:5 5a M 3. a Bass 8:8 Seamus». 23:2: .21” 25E

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

g :55 F o
J
a \ \
I \ . \ \
- .P . 4 MW
. 9 9
x ”we "K .mo no ma
2 EE \ mun: 5E \ E: SE
was. 8m V\ m
1\\ was. «3 1v
111)..
111111 "camera m w
x mm: «mm Eta b was. mmow ux‘wmwmkb an: mac? 1\
H x\
was. smuruvmﬁé .. ........

 

 

 

 

 

 

 

 

 

 

 

 

 

 

com

80

com

003

comp

COX.

(9ch) sseus

218

05:05:. «0 cough—80c came—o 05 DEE Ea couaﬁenequnb ozwcowaﬁézcoﬁm Mo 2333

2: :2: 63% E83 we cougceou acme—o 05 “PE ”mowﬁm 085 woe—Louie? 30am motno
2F .m< we“ mmbm £3: mEE e8 M mmm “a 83.8 8E3 grams—0 088:5 .me oesmﬁ

at sea _ o

mum: E_E

 

8v

com

com

08..

oov _.

(BdW) 539118

219

Film H623 TF3 = 332 K
Ni52.3Ti47.7

 

 

 

 

 

 

 

 

A 400 ~
E (d)

g ..
w _ Grain size = 99 nm
g 200 ppt size = 9 nm
a) - ppt spacing = 7 nm
303 K ' 1 1 Gauge length = 7.0 mm
(0) Width = 1.8 mm
Thickness = 6.7 pm

273 K L 1 L
(b)
3
9)
3,.
$6 243 K 1 1 1
’8»
(a)
213 K ' ' '
O 1 2 3
Strain (%)

Figure 3-49. Isothermal stress-strain curves of ﬁlm H623. The yielding due
to the detwinning of martensite variants and stress-induced martensitic
transformation did not occur at the stress level above 300 MPa. The thin ﬁlm
behaved as an ordinary metallic material.

220

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1000 F Film H673 M): 220 K
Ni51.o'l"l49.o Ms= 248 K
" 73 = 323 K
P A. = 292 K
800 (h) At = 309 K
E 600 ,:"‘:-‘ 1:!me
g .L -\ -.-:
g i-
g 400 ~ (9)
co Grain size = 221 nm
' ppt size = 19 nm
200 - ppt spacrng = 18 nm
Gauge length = 6.0 mm
l (f) Width = 1.6 mm
353 K ‘ ' ' Thickness = 6.3 um
\ 333 K ' L
V (A: < T< Md)
\ 313 K -
Ar
IV (As< T< A.) (d)
3
a
‘3

 

 

 

 

 

 

Strain (%)

Figure 3-50. Isothermal stress-strain curves of the ﬁlm H67 3.

221

 

Figure 3-51. TEM bright-ﬁeld images from deformed ﬁlm H673.
Compared with its original structure, it is seen that the microstructure
essentially remained unchanged.

222

 

 

 

 

 

800 '- x
P Niso.3Tl49.2 Ms: 259 K
1.: 60° 711 = 321 K
V A. = 308 K
33 400 ' (k)
92
a _
200 -
363 K ‘ Grain size = 235 nm

ppt size = 138 nm
1 ppt spacing = 31 nm

K / /
Gauge length = 5.5 mm
Width = 1.8 mm
333 K Thickness = 7.1 pm
V (A. < T< M,,) V/
318 K /. J 1")

\ A.

 

 

 

 

 

 

IV (A,< T< A.) 303 K ' ‘ ' (9)
As
\\ 288 K ' ' 1 (0
III (M.< T< A.) -> (a)
273 K
\3
it. 2535 . . 1d)
33.
6’ ll (M.< T< 114.1(6)

 

 

I(T<Mi) 228V
\ 213K \/1lél (a)
198K

Strain (%)

-

(b)

Figure 3-52. A series of isothermal stress-strain curves generated from a single
tensile specimen of the ﬁlm H703 at various test temperatures. Arrows indicate
the critical stresses for the ﬁrst and second yieldings.

223

Film H723 M. = 250 K

 

 

  

l' Ni51.1Ti43.9 M3: 266 K
800 7;; = 322 K
- A3 = 303 K
A: = 313 K
600
’n?
a.
.2, W
a) 400
m
93
a l )
200 9 Grain size = 247 nm
ppt size = 150 nm
ppt spacing = 42 nm
353 K

 

 

Gauge length = 5.5 mm

Width = 1.7 mm
ﬂ“) Thickness = 6.7 um
,\ 333 K
v (A, < T< M,,) \/(e
L— Ar —->313 K ﬂ

IV (As< T< A.)

; 293 K
\ Ill (M.< T< A.) W (d)
x 273 K '

3
360 M, (c)
3% II (M, < T< M.)
o

 

 

 

 

 

 

Strain (%)

Figure 3-53. Isothermal stress-strain curves of the ﬁlm H723.

 

Figure 3-54. TEM bright-field images from deformed film H723.
Just like its undeforrned structure, well-deﬁned precipitate particles
and grains without any dislocation can be seen.

225

 

Film A2 M, = 247 K

Ni51.oTi49.o M3: 276 K
Tn' = 312 K
A3: 306 K

Ag=315K

 

Grain size = 208 nm
ppt size = 351 nm
ppt spacing = 878 nm

Gauge length = 6.2 mm
Width = 1.3 mm
Thickness = 6.3 pm

 

 

500 r
E 400 -
g (C)
é) _
5 200 '
.. [.7 (b)
333 K ‘ ‘

\
\ v (A.< T< M,,) A ﬂ (8)
<3 :— 313fK l I l
3 IV (As< T< Ag)
09 '\ “W
301 K I I I

«a.
93, |||(M.<T<A.) o 1 2 3 4
Strain (%)

 

Figure 3-55. Isothermal stress-strain curves of the ﬁlm A2.

226

Film A3 M1: 260 K
Ni51.1Ti43.9 Ms= 271 K

 

@

Tn' = 314 K
A,3 = 294 K
A1 = 307 K
Grain size = 233 nm
ppt size = 475 nm
800 . ppt spacing = 1.3 pm
L Gauge length = 9.2 mm
Width = 1.5 mm
600 Thickness = 6.7 pm
E
E
8 400
2
a
200
348 K

 

Al
IV(A,<I<A.)303K0 1 2 3

Strain (%)

  

(a)

 
  

Figure 3-56. Isothermal stress-strain curves of the ﬁlm A3.

227

 

Figure 3-57. (a) and (b) TEM bright—ﬁeld images taken from the deformed
film A3. The absence of precipitates inside grains led the formation of
dislocations.

228

 

Film A5 M, = 267 K
Ni51.oTi49.o M5: 282 K
Tn = 329 K
A3: 319 K
A: = 329 K

 

 

Grain size = 639 nm
ppt size = 178 nm, 635 nm
ppt spacing = 153 nm

Gauge length = 6.6 mm

 

 

L Width = 1.4 mm

600 Thickness = 6.3 pm
‘75 ‘ (d)
g 400 -
g -
92

\ 353 K I I l 1
b)
v (A. < T< M ) / (

 

 

\ X A.

{3 IV(A,<T<A,) ,4,
318K '

9
09‘
$966 lll (M.< T< A.)

 

 

 

Strain (%)

Figure 3-58. Isothermal stress-strain curves of the ﬁlm A5.

229

200mm

 

Figure 3-59. TEM bright-ﬁeld images taken from the deformed ﬁlm A5.
(a) Slip bands which formed by a number of passage of dislocations
spread throughout entire grains. The slip bands in different grains tend to
align in a common direction which is the direction of maximum shear
stress. (b) High density dislocations can be seen.

1200 _

1000i

Stress (MPa)
0) m
8 8

.h
8

 

230

Film A7
Ni51.oTi49.o

Mi: n/a
Ms=223K

TF1 = 318 K
A5: 258 K
Ag=289 K

 

my;

 

(b)

Grain size = 1.8 pm
ppt size = 19 nm
ppt spacing = 12 nm

Gauge length = 6.0 mm
Width = 1.7 mm
Thickness = 6.3 pm

(a)

Figure 3-60. Isothermal stress-strain curves of the ﬁlm A7. The loading
curves does not show any transformation yielding due to the extremely

ﬁne Ni4Ti3 particles.

231

 

Film A8 M, = 228 K
Ni51.oTi49.o M3: 255 K
Tr=i = 322 K
A8: 318 K
A} = 327 K

 

 

Grain size = 1.9 pm
ppt size = 123 nm
ppt spacing = 60 nm

Gauge length = 7.3 mm

 

 

 

 

 

Width=1.5mrn
800 - Thickness=6.3pm
(c)
A 600 -
as
O. -
.5,
:3 400 _
g - (b)
200 -
\ l I
V(A,<T<Md)335K
\ A. (a)
\ IV(A.< RA.)
A A.—>318K ' ' '
9,) i
00% lll(M.<T<As)
I
9a 297K 1 ' ' '
o 1 2 3 4
Strain (%)

Figure 3—61. Isothermal stress-strain curves of the ﬁlm A8.

232

 

Figure 3-62. TEM bright-ﬁeld image from the deformed ﬁlm A7.
The microstructure was unchanged in stress-strain cycling.

yr‘

3
900 ‘r'r‘ﬂ

3'.

 

Figure 3—63. TEM bright-ﬁeld image taken from the deformed ﬁlm A8
The microstructure was unchanged in stress-strain cycling.

234

VI (T> Md)

Md ﬂlv (A.< R A.)

A A. m (M.< T< A5)

Stress

 

923°

 

09$“; II (M, < T< M3)
9 MSW

 

|(T< M,)

 

Strain

Figure 3-64. A schematic a-e-T diagram showing six types of
stress-strain behavior depending on the deformation temperature
relative to the transformation temperatures in polycrystalline NiTi

thin ﬁlms.

235

0.2

ONF

cow

cm

2 m: L .eaeaeﬂ cozméeoo Sea?

on

on

em

om

o¢

on

ON

.mEE _ .E Z e8 833
2380958 counscéon 835.85 .m> 0:23.86 2: 0265 8 8325 .825 .mw-m 8:3"—

 

 

 

 

 

 

 

\

\¢.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.Io-an-Joac.‘.-

 

---------------------------

 

 

 

 

x mmm u m: .9. 5E a
x «mm "a: .m< SE n.
x Em u i .2 can. a
x QR "a: .m< EE 0
x 8mm»: «NE at o
.x mmm 1: .85 EEi
v. mama»: «5: can. 9

 

 

 

 

 

 

 

 

 

 

8..

com

com

com

com

com

com

(new) eusuauaw eonpur 0: ssens leonuo

Yield Stress (MPa)

800

700

600

500

400

300

200

100

236

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ 0 Yield stress on loading y
0 Reversion stress /
t.
do
* — = 6.41MPa/K «
_ ............................................. dT “/ /
//
r /0/ /
i o / /
_o 0 0 Q. ......... 0 /
_ M. Ai/ A. /A,
l i 1 ¢ i L 1 ¢ ¢; 1 l
180 200 220 240 260 280 300 320 340 360 380

Temperature (K)

Figure 3-66. Tensile yield stress and reversion stress as a function of
test temperature for ﬁlm H703.

Stress (MPa)

237

 

 

 

 

 

1200 _ Critical stress to slip in 82 (,s
I— as—
1000 - o
Superelasticity 00°?
_ $4“
each o°°
Shape-memory 0b
' effect ‘9\
600 +- ‘oé
L <9.
(9:?
400 _ / o
_ Film H673
200 -

 

 

220 240 260 280 300 320 340 360 380 400 420

Temperature (K)

Figure 3-67. Stress-temperature diagram for ﬁlm H673.

Stress (MPa)

238

 

 

 

 

 

1200 -
1000 _¥ Critical stress to slip in 82

r
800 "

r Shape-memory
600 _ effect
400 r
200 _ Film H723

__ M. MB A. Al Md

0 l l 1 i ll 11 i 1 l ‘1 1

 

 

220 240 260 280

300 320 340 360 380 400 420
Temperature (K)

Figure 3-68. Stress-temperature diagram for film H723.

Stress (MPa)

1200

1000

800

600

400

200

239

 

 

 

 

 

 

 

* _ Critical stress to Silp in 82
Shape-memory

' effect

_ him. A: Ai Aid

 

220 240 260 280 300 320 340 360 380 400 420
Temperature (K)

Figure 3-69. Stress-temperature diagram for ﬁlm A3.

240

 

 

 

 

 

 

 

 

 

 

 

18 g E
(a) 0
A16 <>
“£14
. 1
“’2 12 Q ‘
L31o ‘
- ° l e
5 8 a: CQ
2 i
LU 6 $51
‘5
9 4
3
0 2
o

 

 

 

 

 

 

 

 

 

 

-20-10 010 20 30 40 50 60
Adjusted Deformation Temperature, T- A (K)

0 Film H673, A1 = 309 K
A Film H703, A1 = 308 K
0 Film H723, A1 = 313 K
0 Film A2, A1: 315 K
0 Film A3, A1 = 307 K
0 Film A5, A1 = 329 K
A Film A8, A1: 327 K

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100 E
,3. (b) 4}
a: A <>‘ 4’
$- 75 ‘ a
.1 i o
.. O
12‘ 50 Ml. a ......
.92
.2 ............................................

E 2
§ 25 ---------------
a: .........
C
LU

o

 

 

 

 

 

 

 

-20-100102030405060
Adjusted Defamation Temperature, T- A (K)

Figure 3-70. (a) Output of mechanical energy vs. adjusted deformation
temperature for NiTi ﬁlms. (b) Energy efﬁciency vs. adjusted deformation
temperature curves.

241

 

Figure 3-71. SEM image showing the fractography of film H623.
The fracture surface consisting of many microdimples exhibits
characteristics of mode I (the opening mode) ductile failure. The
fracture is a transgranular one since no feature of columnar grain
structure can be seen in the fracture surface.

242

 

Figure 3-72. SEM image showing the fracture surface of the ﬁlm
H673. The columnar structure is clearly seen and its grain size
(abut 0.25 pm) is consistent with that observed by TEM.

243

.mwoBonE use

can =o>o £523 “in 88393 E 35 :5 £305 .3 @2865 mm 93:? mo atom a wage EE 2: mo 02m
88:5. 5 «Ba SEQ: 3585 2F .37. 03533. Set woke; 885m 830an 05 mm 39 can .3: 835m
58m @030? Beta 838a 05 mm 03 .mmbm EE mo Fmﬁmoﬁﬁm 05 mEBonm mowed: 2mm .mb-m Rama

 

244

 

Figure 3-74. SEM image showing the fracture surface of film A3.
The fracture is mainly transgrainular type but it is also inﬂuenced by
equiaxial grain structure which was formed in 858 K, 1 hour annealing
(i.e., recrystallization) process.

245

 

Figure 3-75. The fracture surface of ﬁlm A5 shows large cup-like
feature which is due to a greater grain size. Evidently, the combination
of large grains and coarse precipitates posses poor structure stability.

246

 

Figure 3-76. The SEM image showing the fracture surface of the ﬁlm
A7. The fractography shows a ﬂat and straight fracture surface with
many small dimples. No neck is found in the edge of fracture surface.
The free surface of the ﬁlm is very smooth.

APPENDICES

APPENDICES

A.1. Electrically-Excitable Shape-Memory Actuator

The X-ray diffraction spectrum of NiTi ﬁlm sputter deposited at 698 K on a 7.6 pm
thick Kapton substrate (ﬁlm H698) shows a fully crystallized NiTi structure (Figure A-l).
The recovery temperatures A, and A; were 269 K and 293 K respectively which were
determined by observing the recovery of a deformed shape during a slow warm up from
the liquid nitrogen (LN2) temperature (104 K). The prototype microactuator displayed an
electrical impedance of 30 ohms and was deformed into an approximately 100 um radius
hairpin bend at the LN2 temperature as shown in Figure A-2a. With the device still
immersed in cold nitrogen gas, the application of approximately 3 volts dc induced a rapid
and complete recovery of the bending strain as shown in Figure A-2b. The effect was
repeatable for a number of cycles, leaving a ﬂat sheet with no evident creasing or
delamination.

The NiTi ﬁlm was also sputtered onto the quartz substrate at a similar temperature of
703 K (ﬁlm H703). Free-standing NiTi ﬁlm was prepared for evaluation. The bright-ﬁeld
micrograph and associated SADP of the ﬁlm H703 shown in Figure 3-15 reveal crystalline
BZ grains and transgranular precipitates (the microstructure has been detailed in Section
3.1). A similar bending test was recorded as seen in Figure A-3. A NiTi strap could be
sharply bent in near the LN2 temperature (Figure A-3a). When the LN2 ﬂask was removed
the ﬁlm restored its original ﬂat shape (Figure A-3b and c).

247

Relative Intensity

248

 

 

 

d=2.134, (110) 32

......

 

 

2.346, (2-21) Ni4Ti3

.5
L
F

 

 

 

 

 

 

 

 

 

 

35 4O 45 50 55
26 (degree)

Figure A-l. X-ray diffraction spectra of NiTi ﬁlm on Kapton substrate
sputter deposited at 698 K showing that the NiTi was fully crystallized
during the deposition.

 

249

 

Figure A-2. Photograph demonstrating electrically-excitable NiTi/polyimide
actuator. The prototype microactuator displayed an electrical impedance of
30 ohm. (a) The actuator was deformed into an approximately 100 um radius
hairpin bend at the LN2 temperature. (b) With the device still immersed in
cold nitrogen gas, the application of approximately 3 volts dc induced a rapid
and complete recovery of the bending strain.

 

Figure A-3. A series of photographs showing a bending test on a
free-standing NiTi strip (ﬁlm H703). (a) The ﬁlm could be sharply
bent in the LN2 temperature. (b-c) When the LN2 ﬂask was removed,
the film restored its original ﬂat shape.

 

25 l
A.2. Thermodynamic Analysis by J acobian Notation
Jacobian notation is a powerful tool for deriving relations between thermodynamic
variables. The basic J acobian equations Al-Al9 are presented as follows [Mukherjee and
Bieler, 1993]:

Deﬁnition and properties of J acobian notation:

(%H% A“
L4. 3] = -[B, A] (A2)
[A, A] = 0 (A3)
[A, B] dC + [3, C] dA + [C, A] dB = 0 (A4)
[A, B] [C, x] + [B, C] [A, x] + [C, A] [3, x] = 0 (A5)
[A, x] = b [B, x] + c [C, x] (A6)
where dA = b dB + c dC.
Fundamental equations:
dU = TdS - PdV w, x] = T [5, x]- P [V, x] (A7)
dH = TdS + VdP [11, x] = T [5, x]+ V [P, x] (A8)
(M = -SdT - PdV [A, x] = -s [T, x]- P [V, x] (A9)
d0 = -SdT + VdP [G, x] = -s [T, x]+ v [P, x] (A10)

From Maxwell’s equations:

[T.Sl = [P, V] (All)

 

252

If we substitute -F for P and V for L in above equations, the relations for FL-work

can be abtained.

 

 

 

Heat capacities:
were:
1—1
vertexes“

Volumetric thermal expansion coefﬁcient a:

 

 

= 1(93’.) = [V'P] (A16)
V 8T ,. V[T, P]
Linear thermal expansion coefﬁcient a:
a = L(§£) = i515. (A17)
1‘0 3T F LOI:T9F]
Isothermal compressibility [3:
l3 = _i(§_‘i) = .. [V’T] (A18)
V M T V[P,T]

Adiabatic compressibility ,Bs:

 

253

1 (9V [V,S]
B ___ _ —____ A1
3 V(3P), V[P,S] ( 9)

The derivation of Equation 3.3 is shown in the follows. From Equation A1 and A14,

 

 

 

 

2g ___[S.Gl=[S.F]=l(§_S_) =g_._ A20

aASP—iM Acre-1M

= A21
1 a l. . < >
From Equation A1, A11 and A17,

(2:2) = _[S’T] = __A[F’L] = [(211) = ALoa = Va (A22)

80' T [0!“ [F97] dT F

P-)M
(3A5 ] = VAa”"’" (A23)
do T

Substitute Equation A21 and A23 into Equation 3.2, we obtain the Equation 3.3.

 

BIBLIOGRAPHY

 

 

BIBLIOGRAPHY

Ahmed, N .A.G., Ion Plating Technology—Development and Applications, p56, John
Wiley & Sons, Ltd., (1987).

Airoldi, G., B. Rivolta and C. Turco, Proc. Int ’l Conf. on Martaensitic Transformations
(ICOMAT- 86), 691 (1986).

Ashby, M.F. and L.M. Brown, Phil. Mag., 8, 1649 (1963).

Baram, J. and M. Rosen, “Effect of Grain Size on the Acouatic Emission Generated during
the Plastic Deformation of Aluminium,” Phil. Mag., A44, 895 (1981).

Bastin, 6F. and GD. Rieck, “Diffusion in the Titanium-Nickel System—-II. Calculations
of Chemical and Intrinsic Coefﬁcients,” Metall. Trans, 5, 1827-1831 (1974).

Bhattacharya, K. and RV. Kohn, “Symmetry, Texture and the Recoverable Strain of
Shape-Memory Polycrystals,” Acta Metall., 44, 529-542 (1996).

Brophy, J .H., R.M. Rose and J. Wulff, The Structure and Properties of Materials, Vol. 2.
John Wiley & Sons, Inc., New York (1964).

Buckel, W., “Some Remarks on Crystal Growth in Thin Films,” Structure and Properties
of Thin Films, eds. C.A. Neugebauer, J .B. Newkirk and D.A. Vermilyea, John
Wiley & Sons, Inc., New York, 53-57 (1959).

Buehler, W.J. and FE. Wang, “A Summary of Recent Research on the Nitinol Alloys and
Their Potential Application in Ocean Engineering,” Ocean Eng, 1, 105-120 (1968).

Bunshah, RF. in Deposition Technologies for Films and Coatings, eds. R.F. Bunshsh,
et al. Noyes Publications, New Jersey, 139 (1982).

Busch, J.D., A.D. Johnson, C.H. Lee and D.A. Stevenson, “Shape-Memory Properties in
Ni-Ti Sputter-Deposited Film,” J. Appl. Phys., 68, 6224-6228 (1990).

Cahn, R.W., “Recovery and Recrystallization,” Physical Metallurgy, eds. R.W. Cahn and
P. Haasen, 1654 (1983).

Chang, F., Unpublished Data, Michigan State University (1998).

Chang, L., C.H. Simpson and D.S. Grummon, Mat. Res. Soc. Symp. Proc., 194, 301
(1989).

Chang, L., C.H. Simpson, D.S. Grummon, W. Pratt and R. LoLoee, “Structure and

Phase Transformation in Thermoelastic Ni()-,,,TiCu(,,) Thin Films Prepared by DC
Magnetron Sputtering,” Mat. Res. Soc. Symp. Proc., 187, 137-142 (1990).

254

 

 

255
Chang, L. and D.S. Grummon, “Elevation of Premartensitic Transformation Temperature
by (Ni+Cu)2Ti Precipitation in Sputtered Ni(1-,,)TiCu(,,) Thin Films,” Scripta Metall. ,
25, 2079-2084 (1991).
Chang, L. and D.S. Grummon, “Orthorhombic Martensite, Interrnetallic Precipitates and
Retained Austenite in Ti-Rich Ti(Ni+Cu) Sputtered Thin Films,” Mat. Res. Soc.
Symp. Proc., 246, 141-146 (1992).

Chang, L. and D.S. Grummon, “Structure Evolution in Sputtered Thin Films of Ti x(Ni,
Cu)1-,,—I: Diffusive Transformations,” Phil. Mag., A76, 163-189 (1997).

Chang, L. and D.S. Grummon, “Phae Transomations in Sputtered Thin Films of Ti ,,(Ni,
Cu)1-,,—II: Displacive Transformations,” Phil. Mag., A76, 191-219 (1997).

Chang, L., “Phase Transformations in Thin Films of TiNi Thermoelastic Alloys,” Ph.D.
Thesis, Michigan State University (1993).

Delaey, L., R.V. Krishnan and H. Tas, Review: “Thermoelasticity, Pseudoelasticity and
the Memory Effects Associated with Martensitic Transformations,” J. Mater. Sci., 9,
1521-1535 (1974).

Dodd, B. and Y. Bai, Ductile Fracture and Ductility, Academic Press, 1987.

Duerig, T.W. and R. Zadno, “An Engineer’s Perspective of Pseudoelasticity,” Engineering
Aspects of Shape Memory Alloys, eds. T.W. Duerig, K.N. Melton, D. Stockel and
CM. Wayman, Butterworth-Heinemann, London, 369-393 (1990).

Du Pont, Product Bulletin, 1993.

Du Pont, Product Bulletin, 1995.

Edmonds, K.R. and CM. Hwang, “Phase Transformations in Ternary TiNiX Alloys,”
Scripta Metall., 20, 733-737 (1986).

Enami, K., V.V. Martynov, T. Tomie, L.G. Khandros and S. Nenno, Trans. Jap. Inst.
Metals., 22, 357 (1981).

Eucken, S. and J. Hirsch, Mat. Sci. Forum, 56-58, 487 (1990).

Fan, L.S., S.J. Woodman, R.C. Moore, L. Crawforth, T.C. Reiley and M.A. Moser, in
Proc. Solid-State Sensor and Actuator Workshop, Hilton Head, SC, 38 (1994).

Filip, P. and K. Mazanec, “Inﬂuence of Work Hardening and Heat Treatment on the
Substructure and Deformation Behaviour of TiNi Shape Memory Alloys,” Scripta
Metall., 32, 1375 (1994).

Gil, F.J., J .M. Manero and LA. Planell, “Effect of Grain Size on the Martensitic
Transformation in NiTi Alloy,” J. Mater. Sci., 30, 2526-2530 (1995).

Gilbert, E.N., Ann. Math. Stat, 33, 958 (1962).

Gilman, J.J., “Deformation of Symmetric Zinc Bicrystals,” Acta Metall., 1, 426-427
(1953).

 

256

Gisser, K.R.C., J .D. Busch, A.D. Johnson and AB. Ellis, “Oriented Nickel-Titanium
Shape Memory Alloy Filmsprepared by Annealing During Deposition,” Appl. Phys.
Lett., 61, 1632-1634 (1992).

Gokin, K.K., Shape Memery Alloys, Gordon and Breach, Science Publishers, 1984.

G00, E., T. Duerig, K. Melton and R. Sinclair, “Mechanical Twinning in TisoNi47Fe3 and
Ti49Ni51 Alloys,” Acta Metall., 33, 1725 (1985).

Grummon, D.S., L. Hou, Z. Zhao and T.J. Pence, J. de Physique IV, Colloque C8
665-670 (1995).

Grummon, D.S., S. Nam and L. Chang, “Effect of Superelasticcally Deforming NiTi
Surface Microalloys on Fatigue Crack Nucleation in Copper,” Mat. Res. Soc. Symp.
Proc., 246, 259-264 (1992).

Guckel, H., T.R. Christenson, T. Earles, J. Klein, J.D. Zook, T. Ohnstein and M.
Kamowski, in Proc. Solid-State Sensor and Actuator Workshop, Hilton Head, SC,
49 (1994).

Gyobu, A., Y. Kawamura, H. Horikawa and T. Saburi, “Martensitic Transformations
in Sputter-Deposited Shape Memory Ti-Ni Films,” Materials Trans., Japan Inst.
Metals 37, 697-702 (1996).

Haasters, J ., G.V. Salis-Solio and G.Bensmann, “The Use of Ni-Ti as an Implant Material
in Orthopedics,” Engineering Aspects of Shape Memory Alloys, eds. T.W. Duerig,
K.N. Melton, D. Stockel and OM. Wayman, Butterworth-Heinemann, London,
426-444 (1990).

Hale, D.V., M.J. Hoover and M.J. O’Neil, NASA Contractor Rep., CR-61363, 1971
Hall, E.O., Proc. Phys. Soc., 64B, 747 (1951).

Hasiguchi, RR. and K. Iwasaki, “Interal Friction and Related Properties of the TiNi
Interrnetallic Compound,” J. Appl. Phys., 39, 2182-2186 (1968).

Hedayat, A., J. Rechtien and K. Mukherjee, “The Effect of Surface Constraint on the
Phase Transformation of N itinol,” J. Mat. Sci., 27, 5306-5314 (1992).

Hehemann, RF. and GD. Sandrock, “Relations between the Premartensitic Instability and
the Martensite Structure in TiNi,” Scripta Metall., 5, 801-806 (1971).

Heimendahl, M.V., Electron Microscopy of Materials, An Introduction, p116, New York,
NY. Academic Press, 1980.

Hirsch, P., A. Howie, R. Nicholson, D.W. Pashley and M.J. Whelan, Electron
Microscopy of Thin Crystals, Krieger Publishing Company, Malabar, Florida
( 1977).

Hirth, J .P., “The Inﬂuence of Grain Bounderies on Mechanical Properties,” Met. Trans.,
3, 3047-3067 (1972).

 

 

257

Honeycombe, R.W.K., The Plastic Deformation of Metals, William Clowes and Sons,
p223, p235 (1968).

Hook, RE. and J .P. Hirth, “The Deformation Behavior of Non-Isoaxial Bicrystals of Fe-
3% Si,” Acta Metall., 15, 535, 1099-1110 (1967).

Hou, L. and D.S. Grummon, ‘Transformational Superelasticity in Sputtered Titanium-
Nickel Thin Films,” Scripta Metall., 33, 989-995 (1995).

Hou, L., T.J. Pence and D.S. Grummon, “Structure and Thermal Stability in Titanium-
Nickel Thin Films Sputtered at Elevated Temperature on Organic and Polymeric
Substrates,” Mat. Res. Soc. Symp. Proc., 360, 369-374 (1995).

Hua, 82., CM. Su and M. Wuttig, “The Transformation Characteristics in Thin Film
SMA Heterostructures,” Mat. Res. Soc. Symp. Proc., 308, 33-37 (1993).

Hwang, C.M., M. Meichle, M.B. Salamon and CM. Wayman, “Transformation
Behaviour of A TisoNi47Fe3 Alloys, 1. Premartensitic Phenomena and the
Incommensurate Phase,” Phil. Mag., A47, 9-30 (1983).

Hwang, C.M., M. Meichle, M.B. Salamon and CM. Wayman, “Transformation
Behaviour of A TisoNi47Fe3 Alloys, 1]. Subsequent Premartensitic Behaviour and the
Commensurate Phase,” Phil. Mag., A47, 31-62 (1983).

Hwang, C.M., M.B. Salamon and CM. Wayman, “Transformation Behaviour of A
TisoNi47Fe3 Alloys, III. Martensitic Transformation,” Phil. Mag., A47, 177-191
(1983).

Ikuta, K., H. Fujita, M. Ikeda and S. Yamashita, “Crystallographic Analysis of TiNi
Shape Memory Alloy Thin Film for Microactuator,” Proc. of IEEE MEMS, Napa
Valley, CA, 38-39 (1990).

Ishida, A., A. Takei and S. Miyazaki, “Shape memory Thin Film of Ti-Ni Formed by
Sputtering,” Thin Solid Films, 228, 210-214 (1993).

Ishida, A., A. Takei, M. Sato and S. Miyazaki, “Shape Memory Behavior of Ti-Ni Thin

Films Annealed at Various Temperatures,” Mat. Res. Soc. Symp. Proc., 360, 381-
386 (1995).

Ishida, A., M. Sato, A. Takei, Y. Kase and S. Miyazaki, “Effect of Heat Treatment on
Shape Memory Behavior of Ti-Ni Thin Films,” J. de Physique Colloque, C8, 701-
705 (1995).

Ishida, A., M. Sato, A. Takei, K. Nomura and S. Miyazaki, “Effect of Aging on Shape
Memory Behavior of Ti-51.3 at. pct Ni Thin Films.” Metall. Mater. Trans., 27A,
3753-3759 (1996).

Jardine, A.P., “Cycling Times of Thin-Film NiTi on Si,” Mat. Res. Soc. Symp. Proc.,
246, 427-431 (1992).

Jean, R., and J. Duh, “The Thermal Cycling Effect on Ti-Ni-Cu Shape Memory Alloy,”
Scripta Metall., 32, 885-890 (1995).

Johnson, A.D., “Vacuum-Deposited TiN i Shape Memory Film: Characterization and

258

Applications in Microdevices,” J. Micromech. Microeng., 1, 34-41 (1991).

Johnson, A.D., J.D. Busch, C.A. Ray and CL. Sloan, Mat. Res. Soc. Symp. Proc.,
276, 151 (1992).

Johnson, W.A. and RF. Mehl, “Reaction Kinetics in Processes of Nucleation and
Growth,” Trans. Am. Inst. Min. Engrs., 135, 416-458 (1939).

Johnson, A.D., V.V. Martynov and RS Minners, Sputter deposition of high transition
temperature in Ti-Ni-Hf alloy thin ﬁlms, Abstracts of Proc. ICOMAT, Lausanne,
Switzerland (1995).

Kainuma, R. and M. Matsumoto, “Shape Memory Effect by Diffusional Transformation in
Ti-52% Ni Alloy,” Scripta Metall., 22, 457-478 (1988).

Kessler, H. and W. Pitsch, “On the Nature of the Martensite to Austenite Reverse
Transforrnatiom,” Acta Metall., 15, 401-405 (1967).

Kim, J .J ., P. Moine and D.A. Stevenson, “Crystallization Behavior of Amorphous Ni-Ti
Alloys Prepared by Sputter Deposition,” Scripta Metall., 20, 243-248 (1986).

Kim, Y.D. and C .M. Wayman, “Shape Recovery and Phase Transformation Behavior in
Ni-Al Alloys,” Metall Trans., 23A, 2981-2986 (1992).

Kim, T., Q. Su and M. Wuttig, “Thermo-Mechanical NimTisoISi Composite Thin Film
Switch,” Mat. Res. Soc. Symp. Proc., 360, 375-380 (1995).

Knowles, KM. and D.A. Smith, “The Crystallography of the Martensitic Transformation
in Equiatomic Nickel-Titanium,” Acta Metall., 29, 101 (1981).

Kocks, U.F., “Polyslip in Polycrystals,” Acta Metall., 6, 85-94 (1958).

Kocks, U.F., “The Relation between Polycrystal Deformation and Single-Crystal
Deformation,” Met. Trans., 1, 1121-1143 (1970).

Koskimaki, D., M.J. Marcinkowki and A.S. Sastri, Trans. Amer. Inst. Min. Engrs.,
245, 1883 (1969).

Krulevitch, P., A.P. Lee, P.B. Ramsey, J .C. Trevino, J. Hamilton and M.A. Northrup,

“Thin Film Shape Memory Alloy Microactuators,” J. Microelectromech. Sys., 5,
270-282 (1996).

Krulevitch, P., P.B. Ramsey, D.M. Makowiecki, A.P. Lee, M.A. Northrup and GO
Johnson, “Mixed-Sputter Deposition of Ni-Ti-Cu Shape Memory Films,” Thin Solid
Films, 274, 101-105 (1996).

Kudoh, Y., M. Tokonami, S. Miyazaki and K. Otsuka, “Crystal Structure of the
Martensite in Ti-49.2 at.% Ni Alloy Analyzed by the Single Crystal X-Ray
Diffraction Method, Acta Metall., 33, 2049 (1985).

Kuribayashi, K., T. Taniguchi, M. Yositake and S. Ogawa, “Reversible SMA Actuator for
Microsized Robot,” Proc. IEEE of MEMS, An Investigation of Micro Structures,
Sensers, Actuators, Machines and Robots, 217-221 (1990).

259
Kurt J. Lesker Company, Technical Tables, Vacuum Product, 1993.

Li, D.Y. and L.Q. Chen, “Selective Variant Growth of Coherent TinNiM Precipitate in A
TiN i Alloy under Applied Stresses,” Acta Mater, 45, 471-479 (1997).

Li, D.Y. and J .A. Szpunar, “The Slip System in Ordered Tini Alloy,” J. Mater. Sci. Lett.,
11, 510-511 (1992).

Lieberman, D.S., M.S. wechsler and TA. Read, “Cubic to Orthorhombic Diffusionless
Phase Change—Experimental and Theoretical Studies of Aqu,” J. appl. Phys., 26,
473 (1955).

Lin, H.C., S.K. Wu, T.S. Chou and HR Kao, “The Effects of Cold Rolling on the
Martensitic Transformation of an Equiatomic TiNi Alloy,” Acta Metall., 39, 2069-
2080 (1991).

Ling, HQ and R. Kaplow, “Macroscopic Length Changes during the B2 <—-> R and M -—>
B2 Transitions in Equiatomic Ni-Ti Alloys,” Metall. Trans., 11A, 77 (1981).

Ling, H.C. And R. Kaplow, “Stress-Induced Shape Changes and Shape Memory in the R
and Martensite Transformations in Equiatomic NiTi,” Metall. Trans., 12A, 2101
(1981).

Livingston, J .D. and B. Chalmers, “Multiple Slip in Bicrystal Deformation,” Acta Metall.,
5, 322-327 (1957).

Loloee, R., Private communication, (1995).

Mader, S., A.S. Nowick and H. Widmer, “Metastable Evaporated Thin Films of Cu-Ag
and Co-Au Alloys, 1. Occurrence and morphology of Phase,” Acta Metall., 15, 203-
214 (1967).

Madsen, 1.8. and A.P. J ardine, “The Effect of Cold-Working on the Crystallization
Temperature of Thin-Film Shape Memory TiNi,” Mat. Res. Soc. Symp. Proc.,
308, 45-49 (1993).

Mahin, K.W., K. Hanson and J .W. Morris, Jr, “Comparative Analysis of the Cellular
and J ohnson-Mehl Microstructures Through Computer Simulation,” Acta Metall.,
28, 443-453 (1980).

Martin, J .W., Micromechanisms in Particle-Hardened Alloys, Cambridge University
Press, 1980.

Massalski, T.B., Binary Alloy Phase Diagrams, American Metals Society, Metals Park,
Vol. 2, 1768 (1987).

Matsumoto, O., S. Miyazaki, K. Otsuka and H. Tamura, “Crystallography of Martensitic
Transformation in Ti-Ni Single Crystals,” Acta Metall., 35, 2137 (1987).

Melton, K.N., “Ni-Ti Based Shape Memory Alloys,” Engineering Aspects of Shape
Memory Alloys, eds. T.W. Duerig, K.N. Melton, D. Stockel and CM. Wayman,
Butterworth-Heinemann, London, 21-35 (1990).

 

 

260
Meyers, M.A. and K.K. Chawla, Mechanical Metallurgy, Prentice-Hall, Inc., 1984.
Michal, GM. and R. Sinclair, Acta Cryst, B37, 1803 (1981).
Mises, R.Von, Z. Ang. Math. Mech., 8, 161 (1928),

Miyazaki, S., T. Hashinaga, K. Yumikura, H. Horikawa, T. Ueki and A. Ishida, “Shape
Memory Characteristics of Sputter-Deposited Ti-Ni—Base Thin Films,” Smart Mater.,
SPIE 2441, 156-164 (1995).

Miyazaki, S., Y. Igo and K. Otsuka, “Effect of Thermal Cycling on the Transformation
Temperatures of Ti-Ni Alloys,” Acta Metall., 34, 2045-2051 (1986).

Miyazaki, S., T. Imai, Y. Igo and K. Otsuka, “Effect on Cyclic Deformation on the
Psedoelasticity Characteristics of Ti-Ni Alloys,” Metal. Trans., 17A, 115-120
(1986).

Miyazaki, S., S. Kimura and K. Otsuka, “Shape-Memory Effect and Pseudoelasticity
Associated with the R-Phase Transition in Ti-50.5 at.% Ni Single Crystals,” Phil.
Mag., A57, 467-478 (1988).

Miyazaki, S., S. Kimura, K. Otsuka and Y. Suzuki, “The Habit Plane and Transformation
Strains Associated with the Martensitic Transformation in Ti-Ni Single Crystals,”
Scripta Metall., 18, 883-888 (1984).

Miyazaki, S., S. Kimura, F.S. Takei, T. Miura, K. Otsuka and Y. Suzuki, “Shape
Memory Effect and Pseudoelasticity in a Ti-N i Single Crystal,” Scripta Metall., 17,
1057 (1983).

Miyazaki, S., Y. Kohiyama and K. Otsuka, “Plastic Deformation of Ti-Ni Alloys,” Proc.
Inter. Symp. on Interrnetallic Compounds, Structure and Mechanical Properties,
Sendai, Japan, JIM, 691-695 (1991).

Miyazaki, S., O. Matsumoto, K. Otsuka and Y. Suzuki, Abstract Bulletin, 95th Meeting
of the Japan Institute of Metals, Hiroshima, Japan, 143 (1983).

Miyazaki, S. and K. Nomura, “Development of Perfect Shape Memory Effect in Sputter-
Deposited Ti-Ni Thin Films,” Proc. MEMS-94, Oiso, Japan, 176-181 (1994).

Miyazaki, S., K. Nomura and H. Zhirong, “Shape Memory Effect and Superelasticty
Developed in Sputter-Deposited Ti-Ni Thin Films,” Proc. SMST, Paciﬁc Grove,
California, USA, 19-24 (1994).

Miyazaki, S., Y. Ohmi, K. Otsuka and Y. Suzuki, “Characteristics of Deformation and

Transformation Pseudoelasticity in Ti-Ni Alloys,” J. de Physique Colloque, C4,
255-260 (1982).

Miyazaki, S. and K. Otsuka, “Deformation and Transition Behavior Associated with the
R-Phase in Ti-Ni Alloys,” Metall. Trans., A17, 53-63 (1986).

Miyazaki, S. and K. Otsuka, “Mechanical Behaviour Associated with the Premartensitic
Rhombohedral-Phase Transition in a TisoNi47Fe3 Alloy,” Phil. Mag., A50,
393-408 (1988).

 

261

Miyazaki, S., K. Otsuka and Y. Suzuki, “Transformation Pseudelasticity and Deformation
Behavior in a Ti-50.6 (at.%) Ni Alloy,” Scripta Metall., 15, 287-292 (1981).

Miyazaki, S., K. Otsuka and CM. Wayman, “The Shape Memory Mechanism Associated
with the Martensitic Transformation in Ti-Ni Alloys——I. Self-Accommodation,”
Acta Metall., 37, 1873 (1989).

Miyazaki, S., K. Otsuka and CM. Wayman, “The Shape Memory Mechanism Associated
with the Martensitic Transformation in Ti-Ni Alloys—II. Variant Coalescence and
Shape Recovery,” Acta Metall., 37, 1885 (1989).

Miyazaki, S., K. Shibata and H. Fujita, “Effect of Specimen Thickness on Mechanical

Properties of Polycrystalline Aggregates with Various Grain Sizes,” Acta Metall.,
27, 855-862 (1979).

Miyazaki, S., Sugaya, Y. and K. Otuska, MRS International Meeting on Advanced
Materials, 9, 251 (1989).

Miyazaki, S. and CM. Wayman, “The R-phase Transition and Associated Shape Memory
Mechanism in Ti-Ni Single Crystals,” Acta Metall., 36, 181-192 (1988).

Moberly, W.J., J .D. Busch, A.D. Johnson and M.H. Berkson, “In situ HVEM of
Crystallization of Amorphous TiNi Thin Films,” Mat. Res. Soc. Symp. Proc., 230,
85-90 (1992).

Moberly, W.J., J .L. Proft, T.W. Duerig and R. Sinclair, “Deformation, Twinning and
Thermo-Mechanical Strengthening of TisoNi47Fe3,” Acta Metall., 38, 2601-2612
(1990).

Mori, H. and H. Fujita, in Proc. of Yamada Conﬁ 9 on Dislocations in Solids, Japan,
1984.

Motohashi, Y. K. Ohsawa, T. Hoshiya, Y. Okarnoto and M. Ohmori, “Grain Reﬁnement
of a Ti-Ni Shape Memory Alloy and its Inﬂuence on Phase Transformation
Characteristics,” J. Jpn. Inst. Met., 55,132-140 (1991).

Movchan, BA. and A.V. Demchishin, Phys. Met. Metallogr., 28, 83 (1969).

Mukherjee, K., and T. Bieler, Notes of the course MSM 851: Thermodynamics of
Materials, fall 1993, instructor T. Bieler, Michigan State University.

Mukherjee, K., S. Sircar and NB. Dahotre, “Thermal Effects Associated with Stress-
Induced Martensitic Transformation in a Ti-N i Alloy,” Mater. Sci. Eng, 74, 75-84
(1985).

Murr, L.E., “Some Observations of Grain Boundary Ledges and Ledges as Dislocation
Sources in Metals and Alloys,” Met. Trans., 6A, 505-515 (1975).

Nam, S., “Effect of Ion-Sputtered Thermoelastic NiTi Coatings on Fatigue Crack Initiation
in Copper,” Ph.D. Thesis, Michigan State University (1995).

Neugebauer, C.A., “Tensile Preperties of Thin, Evaporated Gold Films,” J. Appl. Phys.,
31, 1096-1101 (1960).

 

 

262

Nevitt, M.V., Trans. AIME, 218, 327 (1980).

Nishida, M. and T. Honma, “All-Round Shape Memory Effect in N i-Rich TiNi Alloys
Generated by Constrained Aging,” Scripta Metall., 18, 1293-1298 (1984).

Nishida, M. and CM. Wayman, “Phase Transformations in TizNi3 Precipitates Formed in
Aged Ti-52 at. pct Ni,” Met. Trans., 18A, 785-799 (1987).

Nishida, M. and CM. wayman, “Electron Microscopy Studies of Precipitation Processes
in Near-Equiatomic TiN i Shape Memory Alloys,” Mater. Sci. Eng, 93, 191-203
(1987).

Nishida, M., C.M. Wayman and T. Honma, “Electron Microscopy Studies of the All-
Round Shape Memory Effect in a Ti-51.0 (at.%) Ni Alloys,” Scripta Metall., 18,
1389-1394 (1984).

Nishida, M., C.M. Wayman and T. Honma, “Precipitation Processes in Near-Equiatomic
TiNi Shape Memory Alloys,” Metall. Trans., A17, 1505-1515 (1986).

Nishida, M., C.M. Wayman, R. Kainuma and T. Honma, “Further Electron Microscopy
Studies of The TinNiM Phase in an aged Ti-52 at.% Ni Shape Memory Alloy,”
Scripta Metall., 20, 899-904 (1986).

Nomura, K. and S. Miyazaki, “The Stability of Shape Memory Characteristics against
Cyclic Deformation in Ti-Ni Sputter-Deposited Thin Films,” Smart Mater., SPIE
2441, 149-155 (1995).

Orowan, E., Symposium on Internal Stresses in Metals, Institute of Metals, London,
p451,1948.

Ortin, J. and A. Planes, “Thermodynamic Analysis of Thermal Measurements in
Thermoelastic Martensitic Transformations,” Acta Metall., 36, 1873-1889 (1988).

Ortin, J. and A. Planes, “Thermodynamic and Hysteresis Behaviour of Thermoelastic
Martensitic Transformations,” Journal de Physique IV, C4, 13-23 (1991).

Otsuka, K., T. Sawamura and K. Shimizu, “Crystal Structure and Internal Defects of
Equiatomic TiNi Martensite,” Physica Status Solidi, 5, 457-470 (1971).

Otsuka, K. and K. Shimizu, “On the Crystallographic Reversibility of Martensitic
Transformations,” Scripta Metall., 11, 757 (1977).

Otsuka, K. and K. Shimizu, “Pseudoelasticity and Shape Memory Effects in Alloys,”
International Metals Reviews, 31, 93-114 (1986).

Otsuka, K., C.M. Wayman, K. Nakai, H. Sakarnoto and K. Shimizu, “Superelasticity
Effects and Stress-Induced Martensitic Transformations in Cu-Al-Ni Alloys,” Acta
Metall., 24, 207-226 (1976).

Palmer, J .E., Thompson, CV. and HI. Smith, “Grain Growth and Grain Size
Distributions in Thin Germanium Films,” J. Appl. Phys., 62, 2492-2497 (1987).

 

 

263

Pence, T.J. and Y. Ivshin, “Building Proper Behavior into Mathematical Models for Shape
Memory Alloys,” SPIE Proc. Mathematics and Control in Smart Structures, ed.
H.T. Banks, Vol. 2192 (1994).

Perkins, J ., The Principles of Chemical Equilibrium, p197-203, Cambridge University
Press (1966).

Petch, N.J., J. Iron Steel Inst, 173, 25 (1953).

Philip, T.V. and RA. Beck, “CsCl-Type Ordered Structures in Binary Alloys of Transition
Elements,” Trans. Am. Inst. Min. Engrs., 209, 1269-1271 (1957).

Quandt, E., C. Halene, H. Holleck, K. Feit, M. Kohl and P. Schlobmacher, “Sputter
Deposition of TiN i and TiNiPd Films Displaying the Two Way Shape Memory
Effect,” Proc. 8th Int. Conf. on Solid-State Sensors and Actuators, and Eurosensors
IX, Stockholm, Sweden, June 25-29, 202-205 (1995).

Robinson, C., “Shape Memory Alloys Find Expanding Applications,” InTech, March,
55-58 (1987).

Ray, C.A., C.L. Sloan, A.D. Johnson, J.D. Busch and ER. Petty, Mat. Res. Soc.
Symp. Proc., 276, 161 (1992).

Reed-Hill, R.E., Physical Metallurgy Principles, ed. W.W. Hagerty, D. Van Nostrand
Company, Inc., p250, p404, 1964.

Robbins, W.P., D.L. Polla, T. Tamagawa, D.E. Glumac and J .W. Judy, in Int.Conf.
Solid-State Sensor and Actuator Transducers ‘91, San Francisco, CA, 55 (1991).

Saburi, T. and S. Nenno, Proc. Int. Conf. on Solid-Solid Phase Transformations, eds
H.I. Aaronson, D.E. Laughlin, R.F. Sekerka and CM. Wayman, The Metallurgical
Society of AIME, Pittsburgh, p1455 (1981).

Saburi, T., S. Nenno and T. Fukuda, “Crystal Structure and Morphology of the Metastable
X phase in Shape Memory Ti-Ni Alloys,” J. Less Common Met., 125, 157-166
(1986).

Saburi, T., T. Tatsumi and S. Nenno, “Effects of Heat Treatment on Mechanical Behavior
of Ti-Ni Alloys,” J. de Physique Colloque, C4, 261-266 (1982).

Saburi, T., M. Yoshida and S. Nenno, “Deformation Behavior of Shape Memory Ti-Ni
Alloy Crystals,” Scripta Metall., 18, 363-366 (1984).

Salzbrenner, R.J. and M. Cohen, “On the Thermodynamics of Thermoelastion Martensitic
Transformations,” Acta Metall., 27, 739-748 (1979).

Schmid, E. and W. Boas, Plasticity of Crystals, Hughes (1950).
Schroeder, T.A., and CM. Wayman, Scripta Metall., 11, 225 (1977).

Shaw, J .A. and S. Kyriakides, “Thermomechanical Aspects of NiTi,” J. Mech. Phys.
Solids., 43, 1243-1281 (1995).

264
Shugo, Y., S. Hanada and T. Honma, Bull. Res. Inst. Min. Met., 41, 23 (1985).

Smidth, F.A., “Use of Ion Beam Assisted Deposition to Modify the Micro-Structure and
Properties of Thin Films,” International Materials Reviews, 35, 61-128 (1990).

Smith, C.S. ASM Seminar, Metal Interfaces, 65 (1952).
Smith, M.C., Principles of Physical Metallurgy, Harper & Brothers, 191, (1956).

Srolovitz, D.J., “Grain Growth Phenomena in Films: A Monte Carlo Approach,” J. Vac.
Sci. Technol., A4, 2925 (1986).

Stachowiak, GB. and P.G. McCormick, “Shape Memory Behaviour Associated with
the R and Martensiyic Transformations in a NiTi Alloy,” Acta Metall., 36, 291-297
(1988).

Swalin, R.A., Thermodynamics of Solid, p88 (1961).

Tang, L. and G. Thomas, “Microstructure and Texture Evolution of Cr Thin Films with
Thickness,” J. Appl. Phys., 74, 5025-5032 (1993).

Taylor, A. and R.W. Floyd, J. Inst. Metals., 80, 577-587 (1952).
Taylor, G.I., J. Inst. Metals., 62, 307 (1938).

Thompson, C.V., “Grain growth in Thin Films,” Ann. Rev. Mater. Sci., 20, 245-268
(1990).

Thompson, C.V., “Grain Growth in Polycrystalline Thin Films,” Mat. Res. Soc. Symp.
Proc., 343, 3-12 (1994).

Thompson, CV. and H.I. Smith, Appl. Phys. Lett., “Surface-Energy-Driven Secondary
Grain Growth in Ultrathin (< 100 nm) Films of Silicon,” 44, 603-605 (1984).

Thornton, J.A., J. Vac. Sci. Technol., 11, 666 (1974).

Thornton, J .A., “High Rate Thick Film Growth,” Ann. Rev. Mater. Sci., 7, 239-260
(1977).

Thornton, J .A., in Deposition Technologies for Films and Coatings, eds. R.F. Bunshsh,
et al. Noyes Publications, New Jersey, 190 (1982).

Tjhen, W., T. Tamagawa, C.P. Ye, C.C. Hsueh, P. Schiller and D.L. Polla, in Proc.
IEEE Conf Micro Electro Mechanical Systems, MEMS 1991, Nara, Japan, 114
(1991).

Tong, HO and CM. Wayman, “Characteristic Temperatures and Other Properties of
Thermoelastic Martensites,” Acta Metall., 22, 887-896 (1974).

Van Humbeeck, J., R. Stalmans, M. Chandrasekaran and L. Delaey, “On the Stability of
Shape Memory Alloys,” Engineering Aspects of Shape Memory Alloys eds T.W.
Duerig, K.N. Melton, D. Stockel and CM. Wayman, Butterworth-Heinemann,
London, 96 (1990).

265

Van Loo, F.J.J., G.F. Bastin and A.J.H. Leenen, “Phase Relations in the Ternary Ti-Ni-
Cu System at 800 and 870°C,” J. Less Common Met., 57, 111-121 (1978).

Walker, J .A., K.J. Gabrieland and M. Mehregany, Sensor and Actuators, A21, 24
(1990).

Walles, B., L. Chang and D.S. Grummon, “Residual Stress, Adhesion and Crystallization
of Ion-Sputtered and IBED Processed NiTi ﬁlms,” Mat. Res. Soc. Symp. Proc.,
246, 349-354 (1992).

Wang, F.E., W.J. Buehler and SJ. Pichart, “Crystal Structure and a Unique ‘martensitic’
Trandforrnation of TiNi,” J. Appl. Phys., 36, 3232-3239 (1965).

Wasilewski, R.J., S.R. Butler, J .E. Hanlon and D. Worden, Metall. Trans., 2, 229
(1971).

Wayman, C.M., “Phase Transformations, Nondiffusive,” Physical Metallurgy, eds.
R.W. Cahn and P. Haasen, 1032-1074 (1983).

Wayman, CM. and H.C. Tong, “On the Equilibrium Temperature in Thermoelastic
Martensitic Transformations,” Scripta Metall., 11, 341-343 (1977).

Wolf, RH. and AH. Heuer, “TiNi (Shape Memory) ﬁlms on Silicon for MEMS
Applications”, J. Microelectromechanical Systems, 4, 206-212 (1995).

Wollants, P., J .R. Roos and L. Delaey, “On the Stress-Dependence of the Latent Heat of
Transformation as Related to the Efﬁciency of a Work Performing Cycle of a
Memory Engine,” Scripta Metall., 14, 1217-1223 (1980).

Wollants, P., J .R. Roos and L. Delaey, “Thermally- and Stress-Induced Thermoelastic
Martensitic Transformations in the Reference Frame of Equilibrium
Thermodynamics,” Progress in Materials Science, 37, 227 (1993).

Wollants, P., M. DeBonte and JR. Roos, Z. Metallkd., 70, 113 (1979).

Wong, C.C., H.I. Smith and CV. Thompson, “Surface-Energy-Driven Secondary Grain
Growth in Thin Au Films,” Appl. Phys. Lett., 48, 335-337 (1986).

Wu, S.K., H.C. Lin and TS. Chou, “Transformation Temperatures of Martensite and
Premartensite in an Aged Ti49N151 Alloy,” Scripta Metall., 23, 2043-2048 (1989).

Wu, S.K., H.C. Lin and TS Chou, “A Study of Electrical Resistivity, Internal Friction
and Shear Modulus on an Aged Ti49Ni51 Alloy,” Acta Metall., 38, 95-102 (1990).

Xie, C.Y., L.C. Zhaoand and T.C. Lei, “Effect of Precipitates on the Electrical
Resistivity-Temperature Curves in an Aged Ti-51.8 at.% Ni Shape Memory Alloy,”
23, 2131-2136 (1989).

 

Zadno, GR. and T.W. Duerig, “Linear Superelasticity in Cold-Worked Ni-Ti,”
Engineering Aspects of Shape Memory Alloys, eds. T.W. Duerig, K.N. Melton, D.
and CM. Wayman, Butterworth-Heinemann, London, 414-419 (1990).

266

Zdeblick, M.J., R. Anderson, J. Jankowski, B. Kline-Schoder, L.Christe1, R. Miles and
W. Weber, in Proc. Solid-State Sensor and Actuator Workshop, Hilton Head, SC,
251 (1994).

Zhang, J., D.S. Grummon, “Stress Evolution during Crystallization and Isothermal
Annealing of Titanium-Nickel on (100) Si,” Mat. Res. Soc. Symp. Proc., Y,
Advanced Materials for Smart Systems, Boston MA, 459, 451-457 (1997).

Zhang, J ., “Stresses and Stress Relaxation during Annealing and Phase Transformation of
NiTi Films on Si,” Master’s Thesis, Michigan State University (1998).

Zhao, Z, “Development of Stress in NiTi Thin Films,” Master’s Thesis, Michigan State
University (1996).

 

"I11111111111111ES