~ _. .365».

u. .55
v

-.v

4
‘$," ’I . . thh -
~v l .

c:
v

v... r.>..:.\; ..
.3451" ..II a

.4 76“!)
23.2.1

«w. .. ‘
a}. N...» .5
us. . .

I:

.pn.uva~
.- o \‘v
v; ‘

.5105: h A
A...» r
33‘}

'47:.
all. 7.-
.1

Nu u .
7.4.}!!! 1.: n.

y 4 .1. 3.9
Nd. .tim ;

 

‘ 11:3..v:
I ain't... I

EOE}!!!
.31 opivlo...

 

THE-’86

NIVERSITY LIBRARIES

iii‘l “\lll‘ll iaiiii

3W
Iii

 

 

 

 

|\\\l\\”‘\l\\

This is to certify that the
dissertation entitled
Phase Transformations in Thin Films of

TiNi Thermoelastic Alloys

presented by

Liuwen Chang

has been accepted towards fulfillment
of the requirements for

Ph.D. degreein Materials Science

 

 

    

 

Major professor

Date 4///?3

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY
MlchIgan State
UnIversIty

 

 

L

._——.'

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution
omens-o:

 

PHASE TRANSFORMATIONS IN THIN FILMS OF TiNi
THERMOELASTIC ALLOYS

By

Liuwcn Chang

A DISSERTATION

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

DOCTOROFPHIIDSOPHY

DcpmtmcntofMata'ialsScicnocandMechanics

1993

ABSTRACT

PHASE TRANSFORMATIONS IN THIN FILMS OF TiN i THERMOELASTIC
ALLOYS

By

Liuwen Chang

The objective of this research is to investigate the microstructure and thamoelastic
transformation characteristics in sputter-deposited TiNi thin films with emphasis on the
relationship between processing parameters, structru'e, and properties as they compared
with similar bulk alloys. Ti,‘(NiCu)1.x films were fabricated by triode magnetron
sputtering from either a single TisoNi45Cu5 alloy target or alternating TisoNi45Cu5 and
pine titanium targets. Energy dispersive X-ray microanalyses showed the resulting films
with Ti-layer to alloy-layer thickness ratio of 0, 1/16 and 1/9 had titanium content of
approxime 47.4, 49.2 and 51.0 at% respectively. The 2.6 % titanium depletion found
in the films deposited from the alloy target resulted from preferential resputtering of Ti.
The sputtering yield ratio, Yrs/Tm, in amorphous TiN i alloys varied from 1.75 to 9 with
the bombarding particle energies of 500 eV to 50 eV, according to results of ion beam
asdneddeposifionsurdy.1hea&spunaedbhmandwrnaqfilmswaeamphouswhh
subsn'ate temperature T. < 500 K. The crystallization behavior of Ti(NiCu) films was
studied by differential scanning calorimetry and X—ray diffraction. Results showed the
amorphous films underwent polymorphic reaction with crystallization temperatures of
about 40 K lower than those of binary alloys. The microstructures and thermoelastic
transformation characteristics of annealed Ti(NiCu) films were examined using
transmission electron microscopy, X—ray diffraction, differential scanning colarimetry and
electrical resistivity measurement Generally speaking, the microsu'ucnues of the Ti(NiCu)
filmmmaledu823and923Kcoimidednddrthepredicdmmadefiommeexnapolanon

of high temperature ternary phase diagram. However, the features of thermoelastic
transformations were complicate. Two kinds of two-step transformations, an B2 H R H
momclinicmartensitemmoneandaBZ H orthorhombicmartcnsite H My; one, were
first found in a 47.4 at%-Ti film and a 51.0 %-Ti film, with about 5 at% Cu, respectively.
On the other hand, a near equiatomic Ti(NiCu) film (49.2 at%-Ti) underwent a simple 82-
to-monoclinic martensitic transformation. In addition, retained austenite were found in
49.2 %-Ti and 51.0 %-Ti films, and its volume fraction directly corresponded to the
vacuum condition dining anneal, i.e., lower vacuum resulted in more retained austenite.
Detailed crystallographic data and self-accommodation mechanism for orthorhombic
mrtensite were obtained by electron microscopy.

ACKNOWLEDGMENTS

I am deeply indebted to Dr. David S. Grumman, my major professor, for his
adviceinmandconnnuoussuppatdrmughoutdrecomseofthisreseuch

Sincaeappreciadoniseandedwmyguidancecomineemembas:Dr.Kafinadr
MWDr.WilliamP.PrattJr.andDr.MartinA.Crimp.

lwouldliketogivespecialthankstoRezaloloeeandVivion ShullinDepartment
of Physics and Astronomy, Dr. Karen L. Klomparens in Center of Eleonon Optics,
MichaelJ.RicthompositeMawrialsCenwruweuastr.RainmginDepuunem
of Chemistry at Michigan State University, and Dr. John Mansfield at the University of

Michiganfcrdreirassisnnceinaccessingvariousexpaimentalinsnummm.

Appreciation is also expressed to the National Science Foundation (under Grant
#M888821755)anddrchrdMotorCompanyforthefinancial supportofthisresearch.

Finally, I am very grateful to my wife Tzuei-Shiang for her understanding during
afldrosedayswhenitseemedlwomdradrasmyinmylabaatcrythanathome.

TABLE or CONTENTS

LIST OF TABLES
LIST OF FIGURES

1.

INTRODUCTION

2. LITERATURBREVIEW

3.

2.1. General Aspects of Martensitic Transfornmion
2.1.1. Thermoelastic Transformation, Shape Memory Effect and
Superelastic Effect
2.1.2. Crystallography of Martensitic Transformation

2.2. Diffusional Transformations in Near-Equiatomic TiNi and Ti(NiCu)

2. 3. Thermoelastic Transformations rn TiNi-Based Alloys
2.3.1. Thermoelastic Transformations rn TiNi Alloys
2. 3. 2. The EffectsofCopperAddition on Martensitic Transformation

2.4. Crystallization Behavior of Ti-Ni Alloys
2.5. Thin Film Processing
2.51 ThinFilmDeposition by Sputtering
2. 5. 2. Ion Beam Assisted Deposition
2.6. Cm'rent Research on TiNi Thin Films
EXPERIMENTAL METHODS

3.1. Materials

3. 2. Deposition Equipment
3.2.1. Magnetron Sputter-deposition Apparatus
3. 2. 2. Ion Beam Assisted Deposition Apparatus
3 2. 3. Modified IBAD Arrangement

3.3. Deposition Procedures
3.3.1. Magnetron Sputter-Deposition
3.3.2. Ion Beam Assisted Deposition
3.3.3. Modified IBAD

3. 4. Oystallization Behavior of As-Sputtered Ti(NiCu) Thin Films
3.4.1. Crystallization Temperature and Enthalpy
3. 4.2. X—ray Diffraction

3.5. Heat Treatment
3.6. Composition Analyses

§:

00‘ OI Ur H g:

3.7. Thickness Measurement

3.8. Microstrucun'e Evaluation
3.8.1. X-ray Diffraction
3.8.2. Scanning Electron Microscopy
3.8.3. Transmission Electron Microscopy

3.9. Martensitic Transformations in Ti(NiCu) Thin Films
3.9.1. Differential Scanning Calorimetry
3.9.2. Electrical Resistivity Measmement
3.9.3. X-ray Diffraction
3.9.4. Transmission Electron Microscopy

4. RESULTS AND DISCUSSION

4.1. Fabrication of TiNi Thin Films
4.1.1. Structural Characterization of Sputter-Deposited Films
4.1.2. Composition of Sputter-Deposited Films
4.1.3. Crystallization Behavior of Ti(NiCu) Thin Films

4.2. Microstructure Evaluation
4.2.1. SL Films
4.2.2. PML-16 Films
4.2.3. PML-9 Films
4.2.4. Fm'ther Discussion

4.3. Thermoelastic Transformation Characteristics

.1. Transformations in SL Films

.2. Transformations in PML-16 Films

.3. Transformations in PML-9 Films

.4. Additional Discussion: Thin Films and TEM Foils versus
Bulk Alloys

5 . (DNCLUSIONS
BIBLIOGRAPHY

4.3
4.3
4.3
4.3

233

Tabb
3—1:
3-2:

4-1:

4-2:
4-3:

4-5:

4-7:
4-8:

LIST OF TABLES

Deposition parameters for magnetron sputter-deposited films.

Depositionparametersforionbeamspumrionbeamassisteddeposited
film fabricated for direct TEM observation.

Depositionparametersforionbeamspumrionbeamassisteddeposited
films. A

Comparison of d-spacing values of the precipitates shown in Figure 4-5 and
the TiN phase [129].

Compositions oftarget alloy and nugnetron sputter-deposited films.

DSC data for magneuon sputter-deposited Ti(NiOr) films.

Transformation data for PML-16 filrm.

Crystallographic data for Type-I twinning.

Transformation data for PML-9 films.

Summary of transformation data for PML filrm.

Crystallographicdataforcrthorhombic martensitein thecubic basis.

vii

133

134

134

135
135
136
136
137
138
139
140

2—6:

2-7:

LIST OF FIGURES

Schematic drawing of the shape memory effect: (a) austenite single
crystal; (b) martensite consisting of two variants; (c) variant coalescence
on loading, and(d) martensite reverting to austenite on heating.

Two dimensional schematic drawing of the martensitic transformation: (a)
crystal structures of austenite and martensite respectively; (b) lattice
deformation transforming austenite lattice to martensite lattice; (c) lattice
invariant shear maintaining an undistorted habit plane; and (d) rigid body
rotation [maintaining a continuous interface.

Section through a martensite plate, showing the banded structure of
relativeamountsxoftwin2and(1-x)oftwin 1. Theplaneofpaperis
perpendicular to the twin planes [25].

Equilibrium phase diagram for Ti-Ni alloys [29].

Isothermal cross section through the Ti-Ni-O phase diagram at
1200 K [42].

Isothermal cross sections through the Ti-Ni-Cu phase diagram (a) at
1073 K and (b) at 1143 K [44].

Electrical resistance as a function of temperature showing a two-step
transformation in a Ti(NiFe) alloy [50].

(it run stereographic projection showing R-phase variants A, a, c and
D and twinning relations between them [53].

142

143

144

144

145

146

147

147

2-10:

2-11:

2-12:

2-13:
2-14:

3-3:

3—4:

3-5:

Self-accommodation morphology of R-phase (above) and corresponding
scherrntic variant-combination (bottom) [53]-

Self-accommodation morphology of monoclinic martensite (above), and a
sub-micro model depicting crystallographic relationships between variants
in the triangular morphology (bottom) [23]-

The dependence of the transformation temperature (M5) on titanium
cement in TiNi alloys [4].

Composition dependence of the crystallization temperature and the
activation energy for amorphous Ti-Ni alloys [75].

Theoretical calculation of the free energy diagram for Ti-Ni alloys at 235
K [781.151
Sputteringyieldofnickelasafunctionofionenergyandionmass [84].
Variation of sputtering yield with angle of incidence for 1 keV Ar"
incident on Ag, Ta, Ti and Al [85].

Structure diagram for thick films produced by sputtering [87].

Flow chart summarizing the experimental methods used in the present
study.

Schematic drawing of (a) the triode magnetron sputtering apparatus, and
(b) the substrate bolder assembly.
(a)Arrangementoftheionsoruces,thesubsn'atesandrelateddevicesfor
ionbeamassisteddeposition. Thedevicesshownwerecontainedina 14"
metalchamber. (b) Schematicdrawingofthesubstrateholder.

Modified arrangement of the ion scru'ces, the target, the faraday probes
and the substrates for ion beam assisted deposition.

Deposition procedures for (a) triode magnetron sputter-depodtion and (b)

148

149

150

151

152

153
154

155

156

158

160

161

3-6:

3-7:

3-8:

4-1:

4-2:

4-3:

4-5:

4-7:

4-8:

(a)Depositionratesandioncurrentdensitiesand(b) l/Aratiosplottedasa
function of the lateral displacement of the specimen from the sputtering
tarpt centerline.
Schematicdrawingofdreelecnicalreaisdvitymeastuementapparahts
SchemficdrawingofthecoolingstageattachedtoaRigakuX—ray
W.

Secondary electron micrograph of as-sputtered SL film showing the
fi'acture surface and the top surface.

X-ray diffraction patterns of as-sputtered SL, PML-9, PML-12, and
PML-16 films showing a broad first-order peak belonging to the
amorphous phase.

TEM results for Ti-Ni thin films: (a) ion-sputtered at Tmb=373 K: (b)
IBAD with IlA=0.33, Tuba-383 K; (c) IBAD with UNI-0.60, Tsub'423
K; (d) IBAD with llA=0.81, Tmb=403 K.

IBAD film irradiated with a 500 eV assist-beam and 1.08 I/A ratio,
showing extensive argon gas incorporation. .

IBAD with IIA=0.33, Tm=473 K: (a) bright field image; (b) diffraction
pattern; (c) dark field image formed with innermost diffraction ring and
(d)darkfieldimageformedwiththirddiffractionring.

Comparison between the calculated composition and the measured
composition of PML films.
'I‘hetotalfiactionofthefilmresputteredbyaSO,100and500eVassist-
beam as a function of VA ratio.

The change of titanium concentration of the IBAD processed Ti-Ni films
as a function of [IA ratio, relative to the expected composition of the film
intheabsenceofanassist—beam. Dataareshownfor50,100and500eV
beams.

163
164

165

166

167

168

169

169

170

171

171

4-9:

4-10:

4.11:

4-12:

4—13:

4-14:

4-15:

4-16:

4-17:

The change in titanium content for IBAD processed Ti-Ni films as a
functionofthetotalfiactionofthefilmresputteredbytheassist—beam.
relativetotheexpectedcompositionofthefilmintheabsenceofanassist—
beam. Dataareshownfor50,100and500eV beams.
Difi'erentialscanningcakrrimen'ydata: (a)fiee-standingSLfilm: (b)free-
standing PML-9 film; (c) PML-9 filmon Si (100) substrate and (d) PML-
18filmonCusubsn'ateataheatingrateof10Wnfin.

Plot of lots/1'1) versus the inverse of crystallization temperature yielding
the activation energies of crystallization for a number of magnetron
sputter-deposited fim

X-ray diffraction patterns of a number of magnetron sputter-deposited
filmsafteranisotlopicannealintheDSCcell.

Comparison of the crystallization temperatmes of ion sputter-deposited
and magnetron sputter-deposited films, and melt-quenched ribbons.
(Bothionsputta-depositedfilmsandribbonsarebinaryTi-Nialloys.)
Comparison of the activation energies of crystallization of magnetron
sputter-deposited films and melt-quenched ribbons.
Goes-sectionsecondaryelecnonmicrographfortheSLfilmannealedat
923Kforonehom'showing grain sizeandgrain morphology.

ln-plane TEM micrograph forthe SL film annealed at923 Kforone hom'
showing grain size and microstructure.
Electrondiffiactionpatternsacquiredfromthemauixphaseofthe923K
annealed SL films atroom temperature in (a) <111>32 and (b)<110>32
zone axes showing streaks in <110> and <112> reciprocal directions

respectively.

172

173

174

174

175

175

176

177

177

4-18:

4-19:

4-20:

4—21:

4-22:

4-23:

424:

4-25:

4-26:

AroomtemperatrnebrightfieldimageoftheSLfilmannealedat923 K
for one hour showing small precipitates with distorted Moire fiinges
roughly perpendicular to-the [100] 132 direction.
Convergentbeamdiffractionpatternstakenfiomsmallprecipitatesina
923 K annealed SL film. A tetragonal cell with a = 0.31 nm and
c-0.80nm is evident from the diffiaction patterns and the rotation
anglesbetweenthem. -
EnergydispersiveX-rayfluorescencespecn'aofmauixandprecipitmsof
aSLfilmannealedat923Kforfourhour.

A bright field microgaph of the (N iCu)2Ti precipitates taken almost
parallel to <1(X)>32 direction showing the rrrorphology and habit plane of
the precipitates. The misfit dislocation network (see enclosed rectangular
area) lies on the (001) precipitates interface.

(a) Bright field micrograph of the annealed SL film showing three
precipitate variants in matrix; (b) corresponding diffi'action pattern
showing [100] and [010] (NiCumi zone axes parallel to <100>32, and
(c-d) dark field images taken-fiom spot c and d in (b) respectively.

A bright field image of the annealed SL film showing gain boundary
precipitates. Theassociatcddiffiactionpattemisinan<112>zoneaxisof
TizNi phase.

X-ray diffraction patterns of the SL films annealed at 923 K for (a) 0.25
how; (13) l hounand(0)4h0m'8-
Electronrnicrogrrphshowingdregminsizeandmicrosnocnneotnpm,
16filmannealedat923Kforonehour.
EleononmicrographofthegainboundaryprecipitatesforaPML-16film
annealedat923Kforonehour. ‘I‘heassociateddiffiactionpatternisinan
<110> zone axis ofTigNi phase.

178

179

180

181

182

183

184

185

185

4-27:

4-28:

4-29:

4-30:

4-31:

4.32;

4-33:

4-34:

ElectronmicrographofaPmrl6filmannealedat923 Kforonehour
showing TizNi precipitates observed in the grain interior. The associated
diffraction patterncorresponds to two (111) twin-related <123> zones of
TizNi phase.

Electr'onmicrographofanfls-l6filmannealedat923 Kforonehour
showing blade-like precipitates. The corresponding diffiaction patterns
are in an <110>32 zone axis and a [011] precipitate zone axis, indexed
usinganorthorhombicunitcellwitha=0.441nm,b=0.882nmand
c -= 1.35 nm. -

(a) Bright field microgaph of a PML-16 film annealed at 923 K for one
hour showing misfit dislocations lying on the interface of Ni3Ti2 type
precipitates. (b)'I‘hecorresponding-diffi'actionpattem.takenfromarod-
like precipitate located between two Ni3Ti2 particles, corresponds to an
<110> zone axis of TizNi phase.
X—raydiffi'actionspecn'afortherr-l6filmsannealedat923Kfor(a)5
minutes;(b) 1hourand(c)6hours.
ElectronmicrographofaPltfls-9filmannealedat923Kforonehour
showingthegrainsiaeandmicrostrucnne.

Electronimicrograph ofaPML—9 film annealed at 323 x for one hour
showingtheg'ainsizeandmicrosn'uctlne.
BrightfieldmicrographofaPms-9filmannealedat923Kforonehom
showingthesizeandmorphologyofsecondphaseparticlesinthegrain
interiors.

Selectedareadifi'ractionpatternsin(a) [110132and(b) [llllmzoneaxes
showingthattheprecipitateshaveanfcclmitcellwitha-1.132nmwith
0me ll (100)32 and [0101m/l [010132.

186

187

187

188

189

189

190

190

4-35:

4—36:

4-37:

4-38:

4-39:

4—42:

4-43:

4-44:

3gdarkfie1delectlon micrograph showingstraincontrastaroundthe
Mpim
X—raydiffiactionspecn‘afortherr9filmaannealcdat923Kfor(a)5
minutesand(b)onehour.
Elecnicalresistivityasafimctionoftemperantrefor(a)thetargetalloy,
(b) the SLfilmannealedat923Kfor0.25hourand(c) the SLfilm
annealedat923Kforonehour.
Differentialscanningcalorimen'ydatafor(a)dretargetalloyand(b)the
SLfilmannealedat923Kforonehour.
X-raydiffractionspectrafortheSLfilmannealedat923Kfor0.25hour
acquiredatvarioustemperauue.

X-raydiffractionspectraforthe SLfilmannealedat923Kforonehour
acquiredatvarioustemperattne.

: TEMmicrographofanme—homannealedSLfilmtakenat255K. The

associated[012]32diffi'actionpatternshows well defined 1/3spotsinthe
lemdirection.
Electronmicrographofone-horn-annealedSLfilmtakenatZSSK. The
associateddiffraction patterned whichcan beidentifiedasan [112013
zoneaxisusingahexagonalunitcellwitha-O.738nmandc=0.532
nm.
Brightfieldimageofanone—hourannealedSLfilmtakenatle. The
correspondingdiffractionpatternshowing[100]zoneofmartensite.
Electron micrograph of anone-hour annealed SL film taken at 160K
showing small martensite variants. The associated [351] diffraction
patternsmddarkfiledmicrographshowing(112)twinrelationbetween

xiv

191

191

192

193

194

195

196

196

I97

198

4—45:

4-47.

449:

4—50:

4—51:

4-52:

4-53:

4-54:

(a) Bright field micrograph showing plate-like R-phase variants; (b)
associated diffraction pattern revealing the R-phase variants oriented
perpendicular or parallel to [110132 reciprocal direction and (c—d) dark
field images taken from spots c and d in (b) respectively showing
labyrinth—like R-phase domains.

Electrical resistivity as a function of temperature of a PML-16 film
annealedat923 Kforonehourinavacuumof2.6x10'3Pa.

Differential scanning calorimetry data for a PML-16 film annealed at
923 K for one horn in a vacuum of 2.6x10'3 Pa.

Differential scanning calorimetry data for a PML-16 film annealed at
823 K for one hour in a vacuum of 2.6x104 Pa.

Cold-stage X-ray diffraction patterns of a PML-16 film annealed at 923 K
in a vacuum of 2.6x104 Pa.

Cold-stage X—ray diffi'action patterns of the PML-16 film annealed at
823 K in a vacuum of 26le Pa.

Electron micrograph and corresponding [134313 diffraction pattern of
823 K annealed PML-l6 film taken at 215 K showing R—phase variants
with plate-like variants.

Electron microgaph and corresponding [023132 diffraction pattern of 823
K annealed PML-16 film taken at 215 K showing plate-like R-phase
variants. The [02313: diffraction patterns is superimposed with two (100)
twin-related [3145513 difi'raction patterns.

Electron micrograph and corresponding [022113 diffi'action pattern of
823 K annealed PML-16 film taken at 215 K showing labyrinth-like
variant-combination of the R-phase.

Electron micrograph of 823 K annealed film taken at 215 K showing
martensiteplatesinR—phasematlix. Tbecorrespondingdiffiactionpattern

XV

199

201

203

205

4-55:

4~56:

4—57:

4-58:

4—59:

4—63:

shows a [0101mm and a[011132pattern with 1/3 spotsin <lll>

. 11' . . V
Microsuucnneofmonoclinicnnrtensiteinannealedes-16filmstaken
at170K: (a) band-1ikevariantsand(b) zigzaggedvariants.

TEM micrographs ofself-accommodated monoclinic martensite taken at.

170 K: (a) bright field image; (b) [010] and [2T I] difi'action patterns; (c)
dark field image taken fiom (100)+(ori) spou labeled in (b); and (d)
[1101 and [101] diffraction patterns.

An electron micrograph of the 823 K annealed film illuminated by
converged electron beam showing zigzagged martensite plates retained in
the B2 matrix. The associated diffraction pattern are in [001]” and
[011132 zone axes.

An electron micrograph of the 823 K annealed film illuminated by
converged electron beam showing zigzagged martensite plates retained in
B2 matrix. The associated diffraction pattern shows two (111) twin-
related [3121mm zone, corresponding to two sets of martensite plates.
Schematic drawing showing the labyrinth-like morphology of R-phase.
Orientation dependence of the shape strain for Type-I twinning derived
from solution 1 listedinTable4-5.

: Electrical resistivity as a function of temperature of the PML-9 film

annealedat923 Kforone hourin avacuumof2.6x10'3 Pa.

Differential scanning calorimetry data for the PML-9 film annealed at
923 K for one hour in a vacuum of 2.6x10’3 Pa.
Cold-stageX-raydiffiactionpanernsofthePML-9filmannealedat923 K
in a vacuum of 2.6x103 Pa. 1
Cold—stageX-raydiffiactionpatternsoftllerr9filmannealedat923K
in a vacuum of 26le Pa.

210

211

212

212

213

214

4-65:

#66:

4-67:

4-70:
4-7 1:

4—72:

Cold-stageX-raydiffracfionpattansoftheWfilmannealedatflBK
inavacuumcf2.6x10'4Pa..

Retained austenite observed at 160 K: (a) bright field micrograph; (b)
corresponding [110] diffraction'pattern and (c) dark field image from
(mammothbcledinm '
Orthcrhombicnnrtensiteatambienttermerature: (a)darkfieldimage; (b)
ccrrespondingdiffi'action patterninthe [010]”..and [01 Unzzoneaxes;
(c) diffraction pattern in a [101]...» zone axis; ; and (d) crystallographic
naturecfthemartensiteplate. H .
mMnficmgraphsofselfaccomrodatedcrmorhanbicmartensitetakenat
rocmtemperature: (a) brightfieldmia'ograph; (b) [110] diffi'acticnpattern
taken from area B; (c) [110] diffraction patterns taken from area C; (d)
[101] diffraction pattern taken from area D; (e) dark field image taken
fi'omspotain(b)showingfourmartensitevariantsa,b,candd;(0dark
fieldimagetakenfi'omspotgin(d)athighermagnification showingthree
martensite variants e, g and f; and (g) schematic drawing showing the
variantdistributicn. ‘
Selfaccommodated orthorhombic martensite taken atroom temperature:
(a) bright field micrograph; (1)) corresponding diffraction pattern showing
tlnee(111)twin-related[101]pattemsand(c)darkfieldimagetakenfrom
spotain(b) showingthreennrtensitevariantsa,bandc.
OystaflographicnanneofthemartensitevariantsshowninFigmeW.
Electron nficrographandcorresponding [121] diffractionpattern showing
(111)twinsinathin-plate formoforthorhombic martensitetakenatroom
W
TEMmicrognphsofselfaccommodatedmhorhcmbicmartensitetakenat

room temperature; (a) bright field micrograph; (b) [100] difi'racticn pattern

215

216

217

219

223

4—73:

4-74:

4-75:

4-76:

4-77:

4—78

4-79:

taken from area B showing one set of (011) twins and (c) [100]
diffractionpatterntakenfiomareaC showinganothersetof(011)twins.
Electron micrograph taken at 123 K showing the monoclinic martensite
inherits (011) twin-related orthorhombic variants. The associated
diffraction pattern showing the electron beam is almost parallel to
[100]”...

Selfaccormnodatedmonoclinicmartensiteat 140K: (a)brightfieldimage;
(b) diffraction patterns of martensite variants a, b and c in [101] zone
axes; (c) dark field image shows variant a, b and c. Fine striations were
observed inside the martensite variants.

Self-accommodated monoclinic martensite at 123 K: (a) bright field
micrograph; (b) corresponding [110] diffraction patterns and (c)darltfield
imagetakenfiomspotcinm). 'l'hestreaksofdiffractionspotsin[001]
reciprocal directions result from (001) twins.

Electron micrograph and corresponding [110]...” diffraction pattern
takenat123 Kshowing(001)twinswithanaverage spacingof4nm.

(a) Untransformed 32 structure in which a fct cell is delineated. (b)
orthorhombic cell transformed from the fct cell. Principle axes (i'J'I')
areobtainedby45°rotationaboutthek axis.

Stereographic projection in [001]];2 direction showing the twinning
relationships between orthorhombic martensite variants.
Orientationdependenceoftheshapesnainfororthorhombicmartensite.

225

227

228

229

230

23 1
232

I. INTRODUCTION

About thirty years ago, Buehler and his coworkers in the U. S. Naval Ordnance
Laboratory developed a new series of engineering alloys based on the intermetallic
compound TiNi. These near-equiatomic TiN i alloys undergo a thermoelastic martensitic
transformation at near- or sub-ambient temperature, and possess a distinctive mechanical
memory ability which is known as the "shape memory effect" (8MB) [1,2].
Phenomenologically, when a TiNi alloy is mechanically deformed at below its
nansfannfionmmpaanne,mm«severeplasdcdefamafion(5~10%)cmbemoducedat
a relative low stress level, by twinning-detwinning mechanisms not relying on the
dislocation glide. On heating above its transformation temperature, however, the alloy can
spontaneouslyrecoveritsoriginal,pre—deformed shape. 'I‘henatureofSMEisillusn'atedin
a simplified two-variant schema in Figure 1-1. Figure 1-1(a) shows an single crystal of
TiNi alloy in its high temperature state (austenite). This single crystal transforms to two1
twin-related martensite variants on cooling below its transformation temperature, to
minimize the total strain energy associated with the transformation, as shown in
Figure 1-1(b). On loading, the two-variant configuration is rearranged, through the
movement of the variant boundaries, to release the stresses. The rearrangement of
macvuhnuresulmmmom-variamdominawdmngementmdashapechangeof
the material as shown in Fig. 1-1(c). However, both of the martensite variants will revert

back to the same austenite lattice orientation on heating, due to the special orientation

 

1Asirnplifiedtwovariantrnodelisusedtosimplifytheschematicrelaeaerltation. lnrealitthereanA
twh-rchtedvariartsddremonoclinicnnrtensificintheTmisystern. _

1

2
correspondence between austenite and each martensite variant, and hence the material

reverts to the original shape as shown in Fig. 1-1(d).

The application of stress can also induce the martensitic transformation within a
certain temperatme interval. The large anelastic strain (~5%) associated with the stress-
inducedtranaformationisrecoverableuponunloadingwhich giverisetothesuperelastic
effect (SE). Therefore, the superelastic effect is also known as an isothermal shape
memory effect.

Equiatomic titanium-nickel alloys also provide high ductility, superior fatigue
resistance and high damping capacity, which give rise to the best combination of
mechanical properties (compared to other shape memory alloys such as CuAlNi and
CuZnAl [3]) for engineering applications when the superelasticity effect or the shape
memory effect is required [4]. Consequently, TiNi alloys have been comercially used as
thermal actuators, tubing couplings and fasteners using shape memory, and in medical
instrument applications requiring superelasticity. Recently, TiNi thin films, which also
undergo reversible thermoelastic transformations, have attracted attention for potential
application asasuperelastic surfacecoatingstoimprove fatigueorerosionresistance, and
inmicroelectromechanicalsystemswhereSMEcanbeusedinacnlam [5,6]. Superelastic
flfinfihmueexpectedmsusminhighcychcsuainmgwimwtmmuhdngsevaedwslip
damage that enhances nucleation of fatigue cracks at free surfaces. On the other hand,
dummlasficdfinfilmsapphedmmcnnalmataiahcmdmserveaslmge—faoemlmg—
strokeacmatorsinminiannizeddevicesbasedontheshapememoryefi‘ect Thepurposeof
thepruentwakismmvesfigatemedetaikofmamodasficbehaviminTlNimmmmsin
anticipationoftheemergingimportanceofsuchapplications.

hmepasttwentyyeam,mnsiderableeffathasbeendevotedmmedevehpmentof
bulkTiNi afloygwhaeasdlinfilmexperimentsarerelativelyrecent. Oneofthechallengea
for thin film processes is close control of composition, since the transformation

temperatures for TiN i are extremely sensitive to titanium concentration (changing by as

3
muchaslprerat‘bTi). Then-ansformationbehaviorofTiNialloysisalsoafiectedby

their thermomechanical history. Thermomechanical treatments which have been well-
euabhshedfmbulkafloyaswhasmneahngmintamediawwmpaannesafiawldwak.
however,arenotabletoeasilytranslatetothinfilmfabrication. Inaddition,titaniumisa
sluggish diffuser which results in an amorphous phase for the vapor deposited films.
Subaequentannealing forcrystallization mayproduceurlfavorableresidual stress and/or
interface reaction problems. Finally, successful application will require a detailed
understanding of thermodynamics, kinetics and crystallography of thermoelastic
transformatiom in thin films.

In modern microelectronic technology, alloy thin films are often fabricated by
physical vapor deposition such as sputtering or evaporation. The quality of thin films.
described by properties such as residual stress-state, adhesion, density, topography, defect
density, texture, composition homogeneity, impurity concentration, etc., is determined by
processingparameters. Inadditiontothetwomostimportantparametersofbackground
pressmeandsubsmwmpaamenagedcpmfickbombudnnntofmegroudngfilmsis
also known to modify film properties. One ofthe applied techniques is called ion beam
assisted deposition (IBAD), which employs independent ion source to irradiate films
during deposition. Energetic ion bombardment is thought to enhance the mobility of
adatoms and near-surface atoms through energy transfer processes, and thus to change
nucleation and growth mechanisms. Energetic particle bombardment, which may result in
a composition drift in alloy or compound films, occurs in most sputtering operations
becausehighmagyspuming-gasamanmfleaedfiommespum-mrgmandmachme
substrate. 'l'hus,wheneitherconventional sputteringorIBADisused, anunderstandingof
dreefi‘ecfiofenagedcpuficlebombmdmentmmnnemdcomposifimmeimpomm.

'lheobjectsofthepresentstudyarebasicallytwofold: 'I'hefirstistofabricateTiNi
drinfihnawithdesiredcomposition,whichcanbehavedrermoelasticafly, andtogainbetter
understanding of compohtional and structural variations of TiNi thin films induced by

4
energeticparticlebomhardmentduringprocessing. 'I‘hesecondgoalistostudythephase

transformation characteristics of TiNi thin films. For reasons which will be detailed later,
an alloy system in which about 5 at% Cu is substituted forNi, has been chosen fordetailed
study. The present study has demonstrated that Ti(NiCu) thin films deposited by
magnetron sputtering can behave thermoelastically after annealing, and show certain aging
efl‘ectswhichdonotappearinsimilarbulkalloys. 'l‘hedifi‘erenceisattributedessentiallyto
thekineficcmdifimsofdeposifionwhichmsulwdmfilmswithmamphoussuucmm.
Wfimmwbseqmntwfid-maynaniufimrewfionandhemueamenhpossessed
microstructures characterized fine grain size and special precipitate distributions, neither of
which are easily attainable in materials made by traditional melt-solidification processes.
The ther'rrnelastic transformation characteristics of thin films were significantly affected by
the nature of these microstructlnes. The fine grain size also allowed observation of self-
accormnodation morphologies ofvarious martensites by TEM in whole grains.

Organization of the Dissertation. Following this general overview,
Chapter 2 provides a detailed review of phase equilibria and stability of near-equiatomic
TiNi and Ti(NiCu) alloys, thermoelastic transformations in TiNi and Ti(NiCu) alloys,
crystallization behavior of amorphous Ti-Ni alloys, thin film processing by sputtering and
ion beam assisted deposition, and results of recent research on TiNi thin films. The
experimental arrangements and procedures are described Chapter 3. Chapter 4 presents
resultsanddiscussion. ’l‘hesearearrangedwiththefirstsectioncoveringthecompositional
and structln'al characteristics of amorphous thin films and their crystallization behavior,
while second section includes the microstructure evaluation of annealed Ti(NiCu) films.
The third section provides both experimental observation and theoretical calculation of
various types of thermoelastic transformations found in annealed Ti(NiCu) films. Finally,
aconlparisonoleNialloybehaviorintheformofthinfilmsandinthebulkismadeinthe
lastsectionofOlapter4. ChapterSsummarizesthefindingsofthepresentstudyandlists
the detailed conclusions.

2. LITERATURE REVIEW

The pretentwotltwas undertaken aspartofa project entitled "stnrace Supcrelastic
Microalloying for Improved Fatigue Ferformance in. Ni, Ti, Fe and Cu Based Alloys",
supported by Nation Science Foundation under grant # MSS-8821755. One of the objects
of this study was investigation of microstructure and thermoelastic transformation
characuisdcsinspuua-depcsitedTlNithinfilmswith specialattentionpaidtothepotential
for thin films to behave superelastically. Bulk TiNi alloys with 5 to 10 at% copper
substituting for nickel have been found to have superior superelastic properties, that is,
smaller stress hysteresis, wider superelastic temperature range, and stable superelastic
characteristics under cyclic loading, without sacrifice of ductility and fatigue resistance as
compared with binary alloys [7]. In addition, transformation temperature of Ti(NiCu)
alloysarealsolesssensitivetothermalcyclingandTicontentvariafionthanthatofbinary
alloys [8]. 'l‘hischaracwrnmkesitpossibletocontrol then'ansformationtemperanlresmore
precisely in thin film fabrication. As a consequence, the Ti(NiCu) system, rather than
binary TiNi, has been chosen for the present study.

In this chapter, general aspects of the thermoelastic transformation, the shape
memory effect (SME), as well as the superelastic effect (SE), and the crystallography of
martensitic uansformadonsarefirstdescfibedtoprovideabackgroundforthediscussion
which follows. In the second section, the phase equilibria and phase stability of near-
equiatomic TiN i and Ti(NiCu) alloys are reviewed. The third section provides detailed
review of the thermoelastic transformation characteristics of TiNi and Ti(NiCu) alloys.
Since the as-sputtered films are usually amorphous, the crystallization behavior of

amorphous Ti-Ni alloys is described in the forth section. The fifth section provides a
5

6
nviewofmebasicspuuefingphemmenamecardafionbetwewprocessingpammetus
andmulfingfilmsfiuchmion-sohdintaacfimaandmhwdwakcnionbeamusiswd
deposition. Finally,thesixth sectionprovidesanup-to—datereviewofprogressonthe
investigationof‘l‘iNithinfilms.

2.1. General Aspects of Martensitic Transformation
2.1.1. Thermostatic Trattvormation, Shape Memory Efl'ect and Srtperelastic Efl'ect

Marwnsificuansfamafiomuediffusionlessphasemsformadonsinwhichamms
are cooperatively rearranged into a new crystal lattice without changing their nearest-
neighbor configuration. Since no long-range atomic movement is involved, the progress of
the martensitic transformation does not depend on time, but primarily depends on
temperature. Oncooling,theuansformationstartsatatemperatmedesignatedasMs,and
does not finish until a lower temperature, Mf. is reached. In the temperature interval
between M3 and Mf, the high temperature phase (austenite) and the low temperature phase
(martensite)coexist. Theprimarydrivingforceofthetransformationisthefreeenergy
difference between austenite and martensite. However, a strain energy is accumulawd in
the lattice during the transformation which resists further progress of the transformation
unless higher driving force is supplied by a subsequent decrease in temperature [9,10].

The martensitic transformations in some materials such as InTl [11], TiNi [l] and
Aqu [12] are reversible, that is, martensite reverts, without re4nucleation, directly back to
auswniteonbeating. Thestarttempennneofthereversetransformationisdesignatedas
As, andthefinish temperatureasAf. HereAsandAfarehigherthaanandMs
respectively, that is, there is a hysteresis in the transformation. The transformation
hyswredsuiginatesfiunmefiicfionaswcimedwimmemovememofmmnsiw-wstuuw
interfaces [13]. In fact, the austenite-martensite interface is glissile, and the reversible
msformafiminvdvesmeshrhlhgeofmuwnsitedomainsmthammmenucleafionof

new austenite crystals [14]. The reversible martensitic transformation is also called a

7
thermoelastictransfornmion. 'I‘hreerequiremenrsfortheiroccmrenceare:(l)smalllattice

deformation for the transformation; (2) martensite containing internal twins that can he
easilydetwinned; and(3)martensitehavinganorderedsu'ucmrethatisnotdestroyedby
slip (lining the transformation [10].

Almonmphysicalpropafiesofaustemeamdifimtfiomflloseofmartensitedm
to a significant change of crystal structlne during the thermoelastic transformation. One
typicalexampleistheyieldstrength. Theyield strengthof'l‘iNiausteniteisahouttwoto
six times higher than that of martensite [4]. The low yield strength of thermoelastic
nnrtensite is due to relatively high mobility of the twin boundaries between neighboring
variants. Electron microscopy study of a CuAlN i alloy conducted by Shimizu and
Otsuka [15] revealed that the twin-related variant boundaries are quite mobile on loading.
In0therwords,thec1itical shearstressfm'twinboundarymovementislowerthan thatfor
dislocation motion. Moreover, a single austenite grain usually converts into a twin-related
multi-variant configtn'ation in a thermoelastic transformation on cooling. The multi-variant
configuration is a result of a self-accommodation process through which the overall
transformation strain associated, with the lattice change is minimized. On loading, the
menshevuimtwhichcanyieklamaximumsheusuainabngflwbadingdimcfionadn
growattheexpenseoftheneighboringvariants. InTiNi, martensitecanbedeformedupto
5 to 10 ‘5, by thecoalescence of martensite variants, without activating dislocation slip
[4, 16]. The plastic strain resulting from the rearrangement of martensite variants at low
temperature (T < Ms) are recoverable on heating above Af, since each martensite variant
reverts to austenite in the original orientation. This is the shape memory effect (SME)
which was first found in TiNi [17]. The mechanism ofthe SME described above was first
proposed by de Lange and Zijderveld [[18], and then was modified by Otsuka et a1.
[19, 20].

The shape memory- alloys can have another unique mechanical property: the
superelastic effect (SB). Phenomenologically, in the superelastic effect, material suffering

large anelastic strain (140%) on loadingscan revert backtoits original shape upon
unloadinginacertaintemperature interval (Mg<Af<T<Mdl). Theoriginofthe
superelastic effect is a stress-induced martensitic transformation. In other words, the
applicafionofsuesshuanmilnefl'emmthemsfamafimuadecreaseintempaanne
[21]. Thiscanbeexplainedbyempbyingtheawsius-Clapeymnequadonwhichcmbe
wdmnas

 

-. A-ll
4‘; ”$417 (2.1.1)
0

where 0"” ( T) is the applied stress which can induce martensitic transformation at
temperature T, p is the density, 84'” is the strain corresponding to complete
transformation, AHA-N is the transformation enthalpy and T, is the temperature at which
the stress free austenite and martensite have the same free energy [22]. Accordingly,
0"“(T) equals zero at Mg, and increases monotonically with an increase of temperature.
Thetemperatureatwhich 0"“(17 equalsthccriticalstressforslipisdesignatedast. A
material cannot behave superelastically unless its Md temperature is higher than Af. The
MdandAftempaannesmusbwomeflleuppaboundandlowaboundfathesupuelasfic
effect. From a microscopic point of view, the applied stress induces the nucleation and
growth of certain martensite variants which can provide maximum transformation strain
alongthestressdirection toreleasetheelastic strain [23]. On unloading, sinceT>Af,the
stress-induced martensite variants become unstable and reverts to austenite at a stress
0“". The value of 04”" is lower than that of 0'4"” since the friction of austenite-

martensite interfaces postpones the reverse transformation.

2.1 .2 . Crystallography of Martensitic Transformation

The crystallographic origin of martensitic transformation has been described
successfully by a phenomenological crystallographic theory developed independently by
1m teutpaame it the upper bound for the superelastic effect

 

9
Bowles and Mackenzie [24] (BM theory) and by Wechsler, Lieberman and Read [25]

(WLR theory). The basic assumption of these theories is that the austenite-martensite
interface (habitplane) shouldbeoneofzelodistortion, inamacroscopic sense,tominimize
total strain energy, which has been proved by later experimental observation [26]. Under
dusecheumsmnceaahtdceinvafiantsheuandafigidbodymmfionuenecessaryfin
additiontoalatticestrainwhichdeformsthehightemperaturestructuretoanewcrystal
lattice (martensite) during the transformation (unless one of the principal transformation
strain vanishes)[20]. Theneedforalattice invariant shearandalattice rotation canbe
explained schematically as shown in Figure 2-1. Figme 2-1(a) shows the crystal structures
of nmtensite and austenite respectively. After a lattice deformation transforming austenite
to martensite, the interface does not satisfy the requirement of zero distortion (Fig. 2-1(b)).
A lattice invariant shear, as shown in Figure 2-l(c), is thus needed to maintain an
undistorted interface. The lattice invariant shear can be performed in a reversible manner
(by twinning) or in an irreversible one (by slip); the former is strictly the case for
thermoelastic martensites. Finally, a rigid body rotation is required to bring the
transformed volume back into contact with the parent crystal.

As shown in Figtne 2-l(d), there is plane strain associated with the transformation.
mmmmimmnmemmlsMenergy,memmensitecrysmlsmmermoehsficaHoys
will accommodate themselves through a twinning process which results in a multi-variant
combination, also known as the self-accommodation morphology. The ability of self-
accommodated martensite to be deformed by means of detwinning is a fundamental factor
underlying slap: armory properties.

The description of the phenomenological theory can also be expressed in
tmthematical form as follows

E = RPT (2.1.2)

10
whereE iscalledthetotaldistorfionmanix,andR,PandTaremanicesrepresendnga

lattice rotation, lattice invariant shear and lattice distortion, respectively [2712. Let us
cmsidadnumsfmmadonwbeoneinwhichasingkcrymlofwsteMeisuansfmmedm
a banded structure of twin-related martensite plates [25], as shown in Figure 2-2. A
maamcopicundistmmdhabhphnecanbeobminedifthemfiodmemicknessofmiml
and2hasacertaincriticalvalue. lnFig. 2-2,avectorr=OVlyingonthehabitplanein
the austenite single crystal becomes a broken vectors OA', A'B', B'C'...U'V' after
transformation. However, the summation ofthe vectors yields a new vector 1" having the
same length as the original vectorr. IfM) ansz are the matrices that describe the
lattice correspondence between austenite and variant 1 as well as variant 2 respectively,
i.e.,

r' =[(1-x)M1 +xM21r, (2.1.3)

thenthetotaldistortionmatrixE canbewrittenas
E = (140M, +xMz. (2.1.4)

ThematrixM; representsthedistortiontowhich variantiis subjected, andisin general
inlplrre. Animpmedistordmmauixisasymeuicmldcausesthemagrfinuieanddirecdm
ofavectorintheaustenitelatticetobechangedaftertransformation. IngeneraLanimpure
distorfionmatrixcanbedecomposedintoaptnedistortionmatrixT andarotationmatrix
¢,i.e.,

M1=¢1T1 andM2=¢sz (2.1.5) -

Thus
E=(hflMQ+flK2

 

23,3,P,and2‘ arernurisesratherthanwnsors. AccudingtothedefinhioninhnecalgebrnifXandY
uetwovectorspaceaeachvectorx ianmbemignedauniquevectrry inY. Theassignrnentis
canedamawilg(oruulformafionoropemtor)oinlalY,mdcmbeexprewdas

y-Fx
where! isthernqrpingmatrix. Incrystallographymmappingwhichdoesnotdistortsizeltdshapeof

objectsiscalledapointisometry,suchasronltionandinversion. lnthepreaancueonlyk isapoint
lsonletry. *

11

= (I-x)¢fl'1 + IQT; (2.1.6)

wherexrepresentsthevolumefractionofthemajortwinvariantz MatricesT; ande
canbeobtainedusingtheapproachproposedbyBaiM28]. Ontheotherhand.¢1 anddb
cannotbeobtainedatthemoment,howeva,therigidbodyrotation¢ between‘l‘winl
andTwiananbederivedsincethesetworegionshavetomaintainacoherenttwin
relationl-Ience

0;: O1 0' ' (2.1.7)
andthetotaldistortionmatrixEcanbewrittenas

E = ¢1[(I -x)T1 «1» W2] = (D11? (2.1.8)

F = (I-x)1'1 + I“;
=PT. (2.1.9)

lngeneralJ” isnonsymmenicandthereforerepresentsanimplnedistortionwhichcanbe
expressedastheproductofarotationnmrix ‘I’ andapuredistortionrmnixF,

F = ‘17,. (2.1.10)

Ethenbecomes

E = was, (2.1.11)

Moreover,arotationmatrixl" canbefoundwhichdiagonalizesthesymmetricmatrixfl,
sothat

F, = 1741'", (2.1.12)

wherng canbe writtenas

r, = o 2, 0 (2.1.13)
0 0 2,

Liebermnetal. [261provedthatthenecessaryandsufficientcondition foranundistorted
habit plane to exist is one of the principle distortion hi in F4 to be unity, that is, no
distortionalongoneoftheprincipleaxes. Thiscriterionallowsustoderivethenormalof
the habit plane based on the principle axis system. As a consequence, the related
pamnetetsordtenansrotmanoncanbeobtainedaaerthehabitplaneisround Inchapter
4, the WLR theory will be employed to calculate transformation strain of Type-I (111’)
twinning in monoclinic martensite, and crystallographic parameters, such as habit plane,
orientation relationship and transformation strain, of a B2-to-orthorhombic martensite

transformation.

2.2. Diffusional Transformations in Near-Equiatomic TiNi and Ti(NiCu)
Alloys

Copper-containing TiN i alloys are found to be attractive to the present work
because of the reduced sensitivity of Ms temperature on titanium content and superior
superelasticity. Therefore, the diffusional transformations and martensitic transformations
in both TiNi and Ti(NiCu) alloys will be reviewed in section 2.2 and 2.3 respectively.

The titanium-nickel phase diagrams available to date [29] (see Figme 2-3) show
stoichiometric fl-TiNi phase (with a B2 structure) at high temperatrue, with a much steeper
solvus boundary on 'I’l-rich side, and predict an eutectoid decomposition of TiNi to TizNi +
N13Ti at 903 :t 15 K. Though the existence of the eutectoid reaction has been supported
by several studies [30-33], many investigators [34-37] have argued against it. Wasilewski
et a1. [36] proposed a peritectoid reaction (TiNi + NigTi -> Ni3Tiz) in Ni-rich side at 898
:l: 20 K, but others [34, 35, 37] hold B-TiNi to be a stable phase with very limited
composition range below this temperature. To solve this controversy, Nishida et a1. [38]

13
Wk a study ofphase stability on Ti-50 at!» Ni and Ti 52 at% Ni alloys. They

concluded that neither the eutectoid nor peritectoid reactions occur, and the precipitation
sequence in the non-stoichiometric Ti-52 at% Ni alloy could be expressed as: TiNi -)
NiaTi33 + TiNi —-) NigTiz-l- TiNi —9 Ni3Ti + TiNi, where NiaTi3 and Ni3Tiz are
metastablephases. Atthesame time,Beyeretal. [40] alsostudiedtheprecipitationprocess
in N i-rich alloys and TiNi-Ni diffusion couples, and similar conclusions were drawn.
However, as pointed out by Gupta et a1. [33] the eutectoid decomposition is a sluggish
prmthatmightresultinarelativestableB-phase.

The TizNi precipitate has been widely observed in TiNi alloys, partially because of
the limited solubility of titanium ill TiNi. Moreover, the presence of oxygen and nitrogen
can enhance its precipitation [2]. Duwez and Taylor [30] concluded that the TizN i phase
hasafacecenteredcubic unitcellwith96atomsinit. Laterinvestigation [41] showedthat
oxygen is soluble interstitially in TizNi up to 15 at%. The resultant TiaNizox phase is
isomorphous with TizNi with a lattice parameter increasing slightly from 1.132 nm (TizNi)
to 1.134 nm ('l’hNizO). Oxygen was proposed to play a role as an electron acceptor in
TizNi, and hence stabilized this phase. Melton[4] pointed out. according to the Ni-Ti-O
phase diagram obtained at 1200 K by Chattopadhyay and Kleykamp [42] (see Figlne 2-4),
that contamination by oxygen in TiNi decreases the single phase range and can terminate
compositions within a three phase region consisting of TiNi, Ti4Ning and Ni3Ti.
Therefore, Ni3Ti can be present in a Ti-rich alloy and, more often, oxygen-containing
TizNi is found in Ni-rich alloys. The presence of TizNi precipitates is usually not
favorable because of deleterious effects on hot workability [4]. .

In Ni—rich alloys, the metastable Ni4T13 phase is coherent with the matrix in the
early stages of precipitation [43]. The crystal structure of the N14Ti3 phase is
rhombohedral with a = 0.6704 nm and a = 113.85° [39]. The orientation relationship

 

31hitwardetigmtedatal~hlanuphaaeinateotiginallieraune HowemJanrstusyconrhctedby
Sabtaietal.(39)]ItposeddmnmiglabemueappopdaemexpresthismaNi4fi3Mbaedm
thecr'ystalstrucale.

14
between 132 matrix and the Mini, phase is determined to be (110mm: II (321)” and

[1111143413 // [111132. Maximum internal stress fields are distributed in <111>nz
directions perpendicular to the interfaces, due to the largest mismatching of d-spacing
(2.9%) between precipitates and matrix. The internal stresses associated with the
precipitation of N14Ti3 were found to change the thermoelastic transformation behavior,
which will be described in the next section.

The titanium-nickel-copper :temary phase diagrams at 1143 and 1073 K were
explored by van Loo et a1. [44]and are shown in Figure 2—5. The substitution ofCu in B-
TiNiforNicanbeupto30at% m8(X)°Cwithoutdesnoyingthamoelasficcharactmisfics,
butthesolubilityoftitaniumdecreases with more than 3 % coppersubstitution. Copperis
also soluble in TizNi and NigTi to 6 % and 4 % at 800 °C respectively. Bricknell et a1. [45]
reportedthelatticeparameteroftheB2 austeniteincreasedslightlyfromO.302nmtoO.304
nm with copper addition up to 10%, and then leveled off. However, Pelton et al. [46]
undertook a microstructure study on an as-cast TigoNiaoCulo alloy and found (NiCu)2Ti
and Tiz(NiCu) phases co-existed with the B2-Ti(NiCu) phase. No further information
about phase stability of these ternaries is available at the moment. However, initial
experimentalresults [47] indicatethatthesephasescanalsoalwrthermoelastic behaviorin
tanary thin films.

2.3. Thermoelastic Transformations in TiNi-Based Alloys
2.3.1. Themloelastic Transformation in TiNi Alloys

Three kinds of thermoelastic transformations, B2 H R-phase (R), B2 H
monoclinic phase (MM) andR-phase H monoclinic phase, can occur in TiNi alloys. For
convenience in the following discussion, both the B2 H M and R H M transformations
arereferredtoa"martensitictransformation" whiletheB2 H Rtransformationiscalledthe
”R-phase” transformation. The R-phase transformation was first treated as a precursor

phenomenon to the martensitic transformation [48]. There are several indications in

15
association with this transition, such as a negative temperature coefficient of electrical

resistivity‘, softeningofdasdcmoduleaandappearanceofexmadiffraction spots. Later
investigation showed, however, that the R-phase transformation does not necessarily occur
prior to the martensitic transformation, and the former actually is not a precursor
phenomenon [49]. Hwang et a1. [50] studied the R-phase transformations in Ti(NiFe)
alloy, and interpreted it as an incommensumte—to-commensurate charge density wave
transition. AccadhlgmtheelecuicalresisfivitycmvesshowninFing-6,Hwangetal.
[50] interpretedthattheBZpuentphasefirstunderwemasecondmdermansifionman
incommensln'ate state associated with an abrupt resistivity increase and a rhombohedral
distortion that produces 1/3 (110) and (111) reflections. The onset temperatureofthe B2-
to-incommensurate phase transition was designated as T;'. The extra superlattice
reflections in fact were not located in the exact 1/3 positions (thus the phase
”incommenstu'ate"), and their intensities kept increasing on cooling. A subsequent first
order transition occurred involving a structure change from a distorted incommensurate B2
phase to" a rhombohedral one causing the extra reflections lock into precise 13 positions.
The onset temperature of the incommensurate-to-commensurate phase transformation T;
was defined to be the inflection point in the resistivity curve as labeled in Fig. 2-6. Further
decrease of temperature induced the martensitic transformation which resulted in a
resistivity decrease, starting at Mg. The reverse transformation were also thought to be a
two-stage transformation. On heating, the resistivity increased up to Ag where the claw
changedits slope. AtAf, theheatingcurvecrossedthecoolingcurve. HeretheAgandAf
points represented the onset and finish temperature of the martensite-to-R phase
transformation respectively. The followed R —r B2 transition finished at about Tr, due to
the small hysteresis (~1 K) ofthe B2 H R transition.

The crystal structure of the R-phase was given by 600 and Sinclair [51] as a
hexagonal lattice with a = 7.37 and c = 5.32 nm. The orientation relationship between B2

 

‘rbetempemuuecoetricientorelectriealretistivityc-dkg/drorbinatyrtNialloysasaociatedwiththe
.. f . is ”w.

16
and R can be expressed as (111)32l/ (00003 and <211>32 ll <21 TO>R. An enthalpy
associaladwiththetransformationwasalsoobtainedasOZkJ/mole.

Wu andWaymnn [52] proposedthatforu'R-phasevariantscan beformedfromthe
B2latticebyelongating myoneofthe<111$3gdimcfims Twinningrelationsbetween a.
phaaevarianttwithretpectto [1101323111[lmlnzpmmowbynmgefaL
[50]. MiyinkiandWayrmn [53] proposedthatthesetwotwinningsyswmsarecompound
twins and have exchangeable elements of K 1, m and K2, in. For example, the twinning
relationships between variant A (corresponding to (111) habit plane) and variant B
(corresponding to(111) habit plane) can be (011) twin or (100) twin, which is illustrated
by the (111) stereographic projection shown in Figure 2-7. The twinning elements,
therefore, are K, =- (011), m . [100], K; = (100), 112 = [011] and K, - (100),
m ' [OT Tl. K2 = (0T T). 112 3 [Too]-

Accadingtoopticalmicroscopeobservation, MiyizakiandWayman [53lproposed
a four-valiant combination for self-accommodation of the R-phase in which twelve variant
grains were divided by three (100) and three {110] twinning planes, as shown in
Figure 2-8.

On the other hand, the crystallographic characteristics ofmartensitic transformation
innearequiatomic TiNihasbeenaconfusingissueforalong time. Otsukaetal. [54] first
dedmeddleayndsnuenneofnnrtensitembeadimwdnwtypewMammchmcmh
cell forwhichthelatticeparameterisa=02889nm,b=0.4120nmc=0.4622nmand
B =- 96.8°. The orientation relationship between austenite and martensite was expressed as
(001)“ 6.5° from (101)32 and [1101M ll [111132. This monoclinic martensite was
proposed resulting from 1/6 (001)[010] planar shuffles on alternate (001)“ (or (101137)
planes. Detailed analysis carried out by several researchers [55-57] later found the
monoclinic martensitehasathreedimensional close-packed structlne,ratherthanalayered

17

(111) Type-I twinning [54, 58, 59] [011] Type-II twinning [60, 61] and (001)
compoundtwinning [62] haveallbeenobservedbyelectronmicroscopyinTiNialloysand
related ternaries. Theoretical calculations carried out by Knowles and Smith [60] showed
that only the (111) and the [011] twins can accomplish the lattice invariant shear.
Matsumoto et al.[63] studied the stress-induced martensitic transformation in a bulk Ti-
49.8 at% Ni single crystal and concluded that only Type-II [011] twinning is responsible
forthelatticeinvariant shear. Onthebasisofpriorresults. mm,m&3£Yaw
[23] deduced a three-variant self accommodation mechanism, as shown in Figure 2-9,
which forms a triangle bounded by three [754132 habit planes. There are three junction
planes between variants in the triangle and they are one Type-l (001) twinning plane and
two Type-II twinning planes respectively. The Type-l (T 11) twins thus were thought to
have been caused by a thin foil effect [63].

Martensitic transformation temperatures of TiNi are strongly dependent on
composition, particularly on the Ni-rich side [2], as shown in Figure 2-10. The Ms
temperature changes over 100 K/at% on the Ni-rich side, whereas the change is only about
30 K/at% on the opposite side. The transformation temperanues are also sensitive to
impmity content. A detailed study of the effect of oxygen content on the transformation
temperaturel64] showedthatdleMgtemperannedecreasedmonotonicanywidlanincrease
ofoxygencontentasthematerialswere annealedabove700K,whichwasexpressedas

Ms (K) = 351.4 - 92.63Xo (mole%) (2.3.1)

where X0 = 0.13 ~ 1.06 mole%. There are some other factors which affect the
transformation temperatures as following [65]:
(l) Agingafta'solution treatment
(2) Annealingattemperannesbelowtherecrystallizadontempetanmeafiacold
work.
(3) Thermalcycling.

18.
(4) Substitution of a third element (usually for Ni).

The first factor is effective for Ni-rich alloys in which the precipitation of Ni-rich
imermetallicphaseincreasestllematrixtitaniumconcenn'ationl38]. I-lowever,theinternal
stress fields associated with coherent precipitates, N iaTig, have a strong effect on
suppressing the Mg temperature. Similar results have been found when the rearranged
dislocations were introduced. Coherent precipitates and dislocations also make the
uansformationtemperatureintervalmf- Mg)1arger[65]. Proftetal. [66] foundthatthe
MstemperattneofmannealedbinaryanoydeaeasedaboutZOKdufingthefimten
thermalcycles. Theeffectofthermalcyclingwasatnibutedtodislocations introducedon
forward and reverse transformations. In addition, substitution of a third element. such as
Cr,AlorFe is alsoefl‘ectively suppresses theMgtemperature.

The R-phase transformation usually overlaps the martensitic transformation in a
solution-treated alloy. Some of the artificial efl‘eets mentioned above, however, can lower
the martensitic transformation temperatme relative to the R-phase and are summarized as
follow [67]:

(1) theinnodllcfionofmanangeddislocadonsproducedbyannedingafiercold
wa'k (r thamal cycling;

(2) ‘ the introduction of coherent precipitates by aging Ni-rich alloys;

(3) Addingthirdelements suchasFeorAl.

The R-phase transition start (Ty) temperanue of an alloy in which the dislocations and/or
precipitates are introduced is almost constant of about 300 to 330 K irrespective of
composition [68].

23.2.The Efects of Copper Addition on Martensitic Transformation
Copper-containing NiTi shape memory alloys were first investigated systematically
byMeltonandhisco—workers [8.69.70]. TheyconcludedthatthemajoreffectsofCu

substitution are to reduce the distortion needed to form martensite from austenite, and to

19
mducemedependenceofMgwmpdannemdnniumcmtent AlloyswithCuadditionsup

m253%mdagommensificnmsfamadonsimilarmdmaeobsavedinbinaryalbysand
theingwmperannesarenotgr'eadyaffectedbytheanrountofCuadded. However,later
study conducted by Tadaki and Wayman [71] showed the Cu content did change the
transformation behavior. Alloys with Cu content less than 3 at% undergo an R-phase
transitionpriortothemartensitictransitiononcoolingbutina19at%Cualloy,theB2to
R-phasetransformationwasnotobserved. Nounambiguousresultswereobtainedforthe
R-phase transformation inalloysorintermediate Cu content (3 at% to 19 at%). Low
tempmatmemarwnsiteinanoyswhaeCucontentislessthm8at%hadamonocfinic
lattice with common (111) transformation twins, but in the 19 at% Cu alloy an
a'tllorhombic martensite was found. This B2-to—orthorhombic transformation does not
require a lattice invariant shear according to the phenomenological theory, although both
untwinnedandtwinnedmartensitewereobservedinthisalloy.

Atthe sametime, Shugo et a1. [72] alsostudied theeffectsof Cuaddition. Mg
meannewurepawdmmcreaseshghdyfiom70°Cm90°Cwid1Cuwncennadonup
m30u%,whueasuaconsnnt(hlooncenuadmitdeaeasedwimoeaeasingTiconwnt
fromstoichiometry. ThedependenceofMgtemper'anmeonTiconcentrationinaSat%Cu
alloyisalmostthesameasinbinaryNiTi,i.e.,120errat%Ti,whileina10at%Cu
alloyitisonlyabout40errat%Ti. Orthorhombicmartensitewasalsoobservedinthe
alloyswithnltlretlran10%Cuaddition.

TheambiguityofmartensiteaysmlsmwnmemTTmiCuhnoysmainlymisesfimn
theexistenceofatwo-step transformationinthemediumCucontent (~10 at%) alloys.
This two-step transformation was first studied in a TisoNiaoCuto alloy by Shugo and
Rooms [73]. They concluded that there occurs first a BZ-to-orthorhombic, and
subsequently an orthorhombic-to-monoclinic transformation. The two transitions have a
tanpaanuemwrvflofabOmSO°densultintwonepsmaeaseofmsisdvityuwenu

two exothermal peaks on cooling [73a]. However, monoclinic martensite was also

20
observed in a 25 at% Cu alloy in the as-cast condition [7]. The monoclinic martensite

mamfamndmathorhombicmartensicafiermesamplewumnealedahightanpaanne
faahngpaidindiadngmepriamamalandprocessinghismrycmaffectmeayml

structureofmartensite.

2.4. Crystallization Behavior of Ti-Ni Alloys

Titanium-nickel is an easily amorphized alloy system. Though amorphous
TigNi(1-g) ribbons fabricated by melt spinning in the composition range 0.23 < x < 0.46
and 0.55 < x < 0.64 have been reported [74, 75], liquid quenched binary alloys are
crystalline. T'igNi(1-x) alloys made from physical vapor deposition with 0.46 < x < 0.55
are amorphous unless the substrate temperature is higher than approximately 823 K [76].
Thisdifl'erencecanbe attributedtothe highercoolingratein the vaporquench. However,
Buschow [75] pointed out that there is no correlation between thermal stability and the
gha-fandngabflitymTi-Nidbyssimeducrysmlhudonwmpaanmsmlyvaryflighfly
with respect to nickel content. Figure 2-11 shows the crystallization temperatures and
activation energies of amorphous Timing), ribbons versus their composition [75].

Thestabilityotmeamotphousauoyisduetothesmallfieeenergygapbetween
mphouunderystalhnephaaeaandmesluggishdifiusionordtanium. Alargenegative
enthalpy of mixing (~30 kJ/mole) in the amorphous phase of titanium-nickel results in
relative stability [77]. Figure 2-12 shows the calculated free energy diagram at 235 K,
deduced by Schwarz et al. [78], where the thick solid line and dotted lines represent the
free energies of amorphous phase and equilibrium intermetallic phases respectively.
Experimental results also indicate that the crystallization enthalpy of near-equiatomic TiNi
thin film is as low as 1.9 kJ/mole [79]. Moreover, the amorphous-to-crystalline
msfamafimcanbengudedasamamanyacdvawdprocesswhichneedsmecdlecdve
atomicrearrangementofbothnickelandtitaniumatoms, sinceallthecrystallinephases
involvedareorderedintermetalliccompounds. Thediffusionrateoftitaniumatomsinthe

21
aysulfimBthaseisapproxhnatelymmdaofmagmmdelowamanmuofnickeluoms

[80]. In the amorphous phase, there is a lack of data for the Ti-Ni system, but the
interdifl'usion constantandintrinsic diffusion constant of nickelin amorphous Zr-Ni have
beenmeasmed.Theresulmshowedmmmevalueofdwsetwodiffusimweffidenuwae
veryclose, indicating zirconiumisaslowdiffuserinamorphous phase [81]. Because of
the similarity of Ni—Ti and Ni-Zr systems, titanium is thought to be a relatively immobile
speciesinamorphousphase,too. Underthisch'cumstancethedifl'usionoftitaniumatoms
appeusmbethekineficcmsuaimonnuclcafionandgmwdld'aysmflhwphm

2.5. Thin Film Processing
2.5 .1 . Thin Film Deposition by Sputtering

General Features. Thin films can be fabricated by many methods, including
thermal evaporation, sputtering, ion plating and chemical vapor deposition. Among these
methods, sputtering deposition permits better control of composition and dimensions of the
deposited film and greater flexibility in the type of materials that may be deposited [82].
Sputteringisaphysicalprocessinwhich smfaceanear—surfaceatomsareejectedfiomthe
surface of a solid or a liquid due to the momentum exchange associated with energetic
particle bombardment. Generally speaking, in sputtering deposition, the bombarding
species(so-calledtheprojectiles)areinertgasionswithenergiesinarangeof500tho
50 keV. The average number of ejected atoms per incident ion is defined to be the
sputtering yield Y, which is a function of a number of variables, including the energy, the
mass,andtheincidentangleoftheprojectile,aswellasthemassandthesurfacebinding
energy of the ejected atom [83]. Figure 2-13 shows the energy dependence of the
sputtering yield of nickel [84]. The sputtering yield is also a function of the incident angle
ofthe projectile, with respect to the normal oftarget surface [85], as shown in Figtne 2-14.
The increase of sputtering yield follows a (case)! relation with the incident angle a and

22

reachesamaximumatO-65°~75°. 'Ihisisacriticalpointwhichaccountsfordifferences
in sputtering behavior for difierent target surface ma‘phologies.

Inmelowenagyreginnthespuueringmteisexuemdysensifivemdrevafiafionof
ion energies (15,) and energy of sputtering threshold (5,.) [s4]. Bohdanslty et at. [86]
derived a scaling law for low energy spqu in which the sputtering yield, r, is given
by

Y(E')=Q (erMersj)F(E') (15.1)

whereu, ansz arememasserotinadintedatomandtargetatomrespeeuvely,£. isthe
bonding energy and E' is a normalized. energy defined by E' =E,-/E,)., where E", is the
threshold energy of sputtering. In this equation Q isan experimentally determined yield
factorwhichcanbewlittenas

Q 2 0M2 [4&1er /(Ml +M2 >215”. (2.5.2)

where a is a material constant, and F is an universal yield energy curve given
appoximately by
F (5' )=s.5rtlo35' 114(1-1/5' )m, (2.5.3)

Asaresult,ahigherEu. resultsinalowerspumringyieldataconstantionenergy.
Thecondensafionpmcessofthespuneredatomscanbedividedinmfllreesteps
[8‘7]. Fnsthwidentammferldnedcenergymwbsuateorcoafinghtficeandbecome
looselybonded”adatoms”. Second,theadatomsdiffuseoverthesln'faceuntiltheyeither
become trapped at low energy site or are desorbed by evaporation or back-sputtering.
Finally, the incorporated atoms readjust their positions within the coating by bulk
diffusion. Atoms sputtered by ion bombardment have energies of about 10 to 40eV,
whichisommmoordasofmagrfimdegreatammthemermalenagyofatypicdmetal
latticeinterstitial. IioweverJhesputteredatcmsmaylosepartofmeirenergytocollisions
inflightfromthetargettothesubmate. Higherambientpressureresultsinmorecollisions

23
andthusdecreasestheenergyofthearrivingatoms. Clemensfoundthatanincreaseof

”snucunaldisada’ofTVNimulfihymwudimcdyrdawdmtheinaeaseofambient
presure(a'thedecreasecfadammmobility)[88]. 'l‘heter'm"su'uctln'edisorder"wasused
mdeaoribetlrelackoftexnneandpoorlayeringofnnllfilayaedTlNi.

Inaddition,adatommobilityisaffected by the surface homologous temperature
Tfl‘m,whereT..isthemeltingpointofthecoating. 'lheresultingmicrostructuresofthe
sputter-deposited films are therefore governed by two primary factors: the homologous
tempuanueT/Fmofthesubmwandwafingandtheamhientgaspressme. 'l'heyare
exprusedconciselyinasu'uctln'ezonediagram[87],asshowninFigureZ—ls. Atlow
filmWshadowingusimpkgeomuicmtaacdonbetweendurwghneuofme
growing sln'faceandtheincident angleofthe arriving atoms) resultsinpoor-qualityfilms
(calledZonelsu'uctln'e)withvoidedboundal-ies. Increasingambientpressureextendsthe
Zone 1 structure to higher temperature regime. At very high homologous temperatures,
bulkdiffusimenhancesthencrystauiufimpmcesswhichpmdweshrgewlumwm
equiaxedgrainsaone3structlne). Withlimitedatommobilitytoova'cometheshadowing
efiecgdensefihroussfiucnnesaonenmdcolumnarsnucnnesmneZMmfamedu
intermediatetemperatln'es. Computersimulafionresultsindicateddlanwithzemadatom
mobility,largervoidsformedastheangleofincidencewereincreased,eventhoughthe
substrate was perfectly smooth. However, a dense and smooth film was obtained at
normalincidenceassociatedwithaslightdegreeofadatommobilitylfl].

Sputtering of Alloy. When a virgin alloy target is subjected to energetic
pardckbombmhnenntheelemenmlcomposidmofthespumdfluxisualaflynmdw
sameasthatofthetargetsluface. Inothawmdacomponentsofwhichthealloyconsists
have different sputtering yields. This phenomenon is called preferential sputtering [90].
'lhe preferential removal of components fi'om sputter-target surface leads to the formation
of the so-called ”altered-layer” which is the region with composition different from the
bulk. At sufficiently low temperatures, where diffusion is negligible, the thiclmess of

24
altered-layerislimitedtodredepthinwhichdrecoflisioncascadeoccma Especially,stead-

moondidmscanheanaimdinwhichcasethespuuaedfluxandtheblflauoyue
constrainedtohavethesamecomposition. Whensteady-statehasbeenachievethefilm
oomporiuonwnldepmdminmracdonormerpunemdnuxwimthewmkinggaaandhy
diffuenddsdddngcoefieienuanddiffermfialre-spumdngatthesubsuate.

Triad: Splatering. For plasma-based sputtering. the triode source is known
toachievehighdepositionrateatlowpressme(5to$0Pa)andtargetvoltagebysupplying
elecuourmtheplannahomamermionicnlamentinaddidonmseeondmyelecuons
emittedfrom the target [91]. Therefore, the target voltage and target ion current are
independently controlled. A magnetic field is usually employed to minimize the radial
phsunbsaenbutdmprodwesadismrdond'mecmtdismmfimomdremguwn.
Simetheuiodesmnceisuwaflyopemedabwapresanenhespuuaeduomscmpass
through the plasma while preserving most of their kinetic energy. This ensures the
adatoms have higher mobility. Furthermore, the growing film is also subjected to
bunhmhnentbyenageficelecmenfinedfiomMenmdenagedcneunakreflecwd
hyrpuum-tmgetls71.rhermmmismeprimuysomoeotmhauuteheaung.mdmelam
canresultinvariouseffects,suchasresputteringanddensificationofthegmwingfilms,
which will bediscussed in section 2.5.2..

Ion Bean: Sputtering. Ion beam sputtering supplies some unique features,
comparing to the plasma-based sputtering techniques. Both the target and the substrate
may be locawd in places which provide greater isolation ofthe substrate from the ion
generationpmcess. lonbeamsputtelingpermitsanindependentconuoloftheenergyand
theclnrentdensityofions. Discreteionsourcesusuallysupplyionswithanarrowenergy
spreadofl-IOeV(Kaufmansource)to10-50eV(coldcathodepenningsource)andat
lowerpressure(0.01t00.l Pa) [92]. Fmdrermorenheincident angle of ions, withrespect
mmptnamahanddreimidentmgleofmespuwedamwimrespectmthesubmam

r , 25
normal, can be independently arranged. However, resputtering of the growing films by

energeticrenectedneuualsalsoocctusduetolowambientpressmem].

2.52. Ion mutated Deposition (IBAD)

loubeam'aaristeddeposition (IBAD)istheprocessinwhichagrowingfilmis
simultaneously bombarded with an independent controlled ion beam. Among various
plasrm and ion-based modification techniques, IBAD possesses advantages such as
dbwingaddckafihbbeanainedmmehhadhwtionimphnmfionmionnnxingand
permittingindependentcontroloftheprocessvariablea whichareimpracticalincomplex
pim- W.

Simultaneous ion bombarth during physical vapor deposition has been
observed to produce beneficial modifications in thin film properties, including improved
adhesion. densitification, modification of grain size and morphology as well as grain
orientation, control of residual stresses, and control of composition [94, 95]. Furthermore,
ion bombardment enhances the adatom mobility, providing the possibility of low-
temperature deposition by which the interdiffusion between film and substrate, and
significant amount of extrinsic stress, can be avoided However, some disadvantages are
associated with the employment of ion bombardment during deposition, including the
change in composition by preferential re-sputtering [96] and incorporation of the assist-
beam working gas [97].

Energetic ions bombarding a growing film can produce a variety of effects,
including sputtering, desorption. structure disordering, and enhanced diffusion. All these
effects result fiom two different energy loss processes: elastic collision and inelastic
collision. ElasficcoflisimsinvolvedrempulsiveCoubmbicfacebuweendresmficnmlei
andincomingions. T'hedisplacedatomsanddeflectedionsundergoadditicnalcollisionsin
whatiscalled acollision cascade. The incoming ioncanalsoinelastically interact with
aecuomminnermdoucrahelhandpmdmeiommmexcimuonortheelecuona The

V 26
mergyofthebombmdingpmficleismimpmantpammetamdetnnnnethepropmdonof
dreaetwoprocesses. Forlowenergy(50-10(X)eV)ionbeam,theenergylossprocessis
dominated by elastic collisions. These ' result in resputtering, incorporation of the
inpmgmgionspecieahmcevibrauondefmmanddisplwonmtofanfmeammsm}.

nunbudnremdfannfacewimenngencpudclesmmrenceamenrrcleanonandme
ealystagerofgrowthofthedeposits. Mminovl98ldepositedsilverfihnsonamorphous
cabonsubsuatesumcmtempmuue.1herendushowedthnthesilverishndsize
Wwimmauodawddeuuseinhlandnumbadensity,uacomequenceofme
concurrentArionirradiation. Theenergies-ofionsvariedfromltolOkeV. Pronounced
adatom-depleted zones were found. around large silver islands resulting from ion
bombardment Theaudrordrusconcludcdthatconctnrentionbombardmentofmegrowmg
filmsenhancedsurfacemobilitiesofbothadatomsandcrystallites. Ontheotherhand.a
mrdyofmedepoaifionofntunmumonadielecuicmbsnawwnmmmneoussnvm
ionbombardmentshowedanincreaseofislanddensityandadecreaseofislandsize[99].
Similar results were obtained by Lane andAnderson [100] which were interpreted using
memmpnonmatprefererdnnucleanonsiteswnemeaedbydeionbombmdmmt Itis
obvious that comment ion bombardment of a growing film influences nucleation and
growth mechanism However, no single model is yet able to explain all of the observed
irradiationeffects. ' '

Since the nucleation and growth mechanisms of a growing film are influenced by
the employment of ion bombardment, changes of the resultant film microstructure are
expected Microstrucnrre of thin silver film fabricated by dual ion gun sputter-deposition
and by thermal evaporation was studied by Huang er al. [101]. They found ion beam
spuuadepositedfilmsbombmdedbyArions‘showedmmhlesstexmmsmnlngamnu
andhigherdefectdensitiesincomparisontoanevaporatedfilm. Thesubstrateadhesionof
evaporated germanium films was also found to be improved substantially by using Ar ion
bombardment [102]. This improvement resulted from a annealing effect, caused by atomic

-. 27 V
rearrangement in bombardment-induced thermal spikes, which decreased the tensile

residualstressofthefilms. Fln'thermore,concmrentionbombardmentwasalsofoundto
exmddrecompodfionnngeinwhichtheremhantdeposinwueaysulhne[103]. Intlris
study, Ni-La alloy films were deposited by triode sputtering from a cylindrical Ni/La
was employed to accelerate ions. The unbiased films were amorphous at a La
cosmonationouatrborhigher. However,thefilmsdepositedatabiasvoltageof-l75
voltscontainedthecrystallinephaseatwidercompositionregionofOto 17at%La.
Mudn¢dl.[104]founddmionbombmdmentofthewapaawd2102filmswith6(flev
ArionsalZlDeVOfiumomtempmuneproducedcrysmlfinleozwithacubith
cell. Theevaporatedfilmswithoutionirradiationwereamorphous.
Thecomponentinan'afloyoracompoundwithlowersmfacebindingenergywin
be sputtered faster, causing a composition change. A simplified model of this preferential
resputtering for alloy was established by Haper and Gambino [96], assuming no gas
incorporation and unity sticking coefficient during deposition. The film composition
A,B(1.y)producedunderion bombardment was writtenintermsofthe target composition
AxB(1.g).thesputtering yield ratioYsYA/Yg,andthefractionresputteredxr:

ya = [a + (at2 + 41413 )mlflfl (2.5.4)

where a a (X;- +xAY - xA -1’Xr -1), fl - (X;- +Y -XrY -1), x4 is the atomic
compositionofelementA,aneristhefractionoftotalfilmresputteredbytheionfiuxat
the substrate. The experimental measurement in this study also found that the sputtering
yieldratioofekmentsinalloysmaydifferfromtheratioofelemental yields,andthatthe
former maybe strongly composition dependent. For example, the elemental yield ratio of
GdandCois0.37whi1etheyieldratiointhebinary alloyisinarangeof2t06.
Preferential sputtering becomes more pronounced at lower bombarding ion energy. Tarng
and Wehner [105] measlned the surface composition of a Constantan target (CusgNias)

afteritwassputteredbyArionswithvariojssenergies. TheAugerelectronspectmscopy
realltsshowedthatthesputteringyieldratioYafclemwasabout1.6ationenergiesof
above150vehereasY increaseddr'amaticallyfi'oml.6to7withadecreaseofion
ewgiesfmmlSOtho35eV.

Theincorpa'adonofinertgasionsissuenglydependentonionenergy[106]. In
biasrfspufiaingtheinatgascmtenLCg,wufomdemphicanymfonowaquadrafic
flmctionofsubstratebiasl98].

C,=KV¢,2 ' (2.5.5)

where V], is the bias voltage. Kornelsen [106] measured the sticking probability of various
inatgasesintungstenasafunctionofionenergy,andfoundthatitdropsrapidlyinthe
low energy regime. For argon, for example, the threshold of incorporation is about
100 eV, due to the high reflection coefficient.

Thin films deposited by physical vapor deposition, such as sputtering and
evaporation, possess a residual stress which can result in film buckeling, cracking or
delamination. Mostimportandy,theexistanceofresidualsnessinTrNifilmscanshiathe
Ms temperature, according to the Clausius-Clapeyron equation described in section 3.1.1.
Cln'rentionbombardmentofthegrowingfilmwasalsoknowntomodifyfilmstress
state[94]. Residual stresses in as-deposited and annealed (873 K. 15 min.) TiNi thin films
bombarded by Ar+ dming deposition were'studied by Walles [159]. The results showed
that the films deposited by ion beam sputtering without assist-beam bombardment had
residualstressofaboutSOMPaintensile. Aftermnealed,thesuessincreasedtoabout
380 MPa in tensile. On the other hand, the stress in the as-deposited IBAD films
decreasedtoOcreven becamecompressive monotonically with increasing I/Aratio. The
stressintheannealedIBADfilmsvariedwithI/Aratiointhesameway,however,tbeI/A
valuecorrespondingtozerostressshiftfi'om0.05intheas-depositedfilmst00.45inthe
annealed films.

. 29
Oneoftheefiecuofbombardmentoftheglowingfilmbyenergeficpardclesisto

enhancetheadawmmobilhy.1heenhancedmobihtycmcmuimmmamanyacdvawd
smfacedifiusionandimpactenhancedsurfacediffusion. Gilmoreetal.[107]eva1uatedthe
contributionofathermalspiketothethermallyactivatedeventsdlningdepositingAuon
NaClwitthOeVAr‘ionbombardment. Theyassumedionssuikeasln'facecreatinga
poimweofthumalenergandaehssicalheatuansfdequafionwuemployed The
remlushowedthnthhthermnspikehadhnleefiectonadamdiffusionsinceme
wmperanuepulsewaslinfitedtoaradiusofaboutlnmwithinSx10‘3second. The
numberofthermallyactivateddiffusioneventswasintheorderof10'5foreachimpinging
ion. Ontheotberhand,Rossnage1etaI.[108]calculatedthenumberofadatomjumpsof
Ta impurity on Cu surface induced by Ar ion impingement from experimental data and
foundthisnumbercanbeintheorderoflozto103perion,dependingonthesubstrate
temperatlneandionclrrentdensity.

2.6. Current Research on TiNi Thin Films

TiNi thin films with near-equiatomic overall compositions have been fabricated by
various methods Periodic multilayers of titanium and nickel deposited by dual-source
magnetron sputtering were studied by Clemens [109]. The as-sputtered films were
amorphous when the compositional wavelength was lea than 19 A. Films with larger
wavelength were composed of crystalline Ni and Ti. Annealing of the films with
wavelength of 200 A at temperatures between 423 K and 598 K produced amorphous
alloys. However, Auger depth profile for a sample annealed at 513 K for 22 hours still
showedconmositionperturbation. Highenergyionirradiationwasalsoemployedtomix
bilayered nickel and titanium films deposited by electron beam evaporation [110, 111].
Gaboriaud and Delage [110] reported that an amorphous TiNi layer was formed after
irradiating by 300keV Krions with doserange from2x10'" tonlO'16 ions/cmz. On the
other hand, crystalline TiNi with 82 structure was found in a bilayered sample after

30
bombardment by 55 keV Kr“ with-total ion dose of 5.5x10'1‘ ions/cm2 [111]. Although

themuldmesofionswaealmontheamembomcaaes,mecmrentdenfityofthehuer
experiment was about three to four'order ofmagninrde higher, which brought the substrate
more up to about 730 K.

'AmorpbousTiNi films with composition oanaNigg were alsoprepared by planar
magnetron sputtering [112]. Detailed processing parameters were not available.
Nevertheless, DSC results showed that the T‘iuNisg film has a crystallization temperatlne
of791Kwhenaheatingrateof10K/minwasused. Fln'therelectronmicroscopystudy
showeddrasmaflTigNiapardcleswaeformedindreamphousmauixuawmpaanne
(778K) which is lower than the start temperature of the crystallization exotherm. Films
isochronicallyannealedattemperaturefromMKupto873Kshowedafullycrystalline
suuculrewhichwasamixtlneoftheTigNiaphaseandtheR-phase. Afterannealingat873
K for seven days, the film was composed ofmainly TigNig phase and a small amount of
B2 TiNi. An amorphous T143Ni52 film was also reported to be deposited fiom a TisoNiso
target by R. F. magnetron sputtering [113]. The argon gas pressure during sputtering was
32mmdthesubsnatetemperannewasmaintainedatloomtempemtme. Thedepletion
of titanium may have resulted from the employment of substrate bias of -120 V during

Johnsonandhiscoworkershaveconductedresearchon shapememoryproperties
inTiNithinfilnnlsincethelate 1980s. TlNifilmsweredepositedfiomalloytargetsbyDC
magnetron sputtering. The sputtering condition were: PA, :- 0.08 Pa, 1 a: 0.5 A and
V a 450 V. The resulting films were amorphous for substrate temperatures of about 423
to 473 K dining deposition [114, 115]. No significant composition shift (< 0.5 at%)
between the targets and the resulting films were observed. DSC result for a 49.7 at% Ti
filmshowedthecrystallizationexothermlocatedatabout753K. AfterannealinginDSC
celeisfilmshowedatwo-stepthamoelasdcmsfamafionmbommngandcoofing
with transformation temperatures of about’100 K lower than those of the target alloy.

31
X-mydiffiacdonandebcuonmiaoscopyresmmdnwedmmdrepom-mnealedfilmnngm

beamixtureofprimarilydiaorderedBCCTiNiandfineprecipitatesoforderedBthase.
Thedepressimofuansfannfimtanpaauneswuamibuwdmfinegminsizeubomlum
in diameter) and oxygen contamination. Tensile tests carried out at two different
WT<MfandT>AfiwerealsoconductedtoevaluatetheSMBmdSE 'lhe
nmpledefamedulowwupaannebeganyieuingnsshfpamdprommedamcoveraue
su-ainof3‘lratl80ma. Moreover.alo-rtm-thicltfilmwasabletoliftao.4gobjecton
headngdemmsnadngarecovaysuessduptommwhichiscompuabkmthebulk
TiNi. On the other hand, a total of 2.5 ‘1) strain was recoverable upon unloading as
deformedathightenqrerature.

The influence of heat treatment and precipitation on phase transformations in
sputter-depositedfilmswerealsostudied [116]. Films annealed isothermallyat813 K
showedadeaeaseofMgtempaanneandamuepronwncedappeamnceofmeR-phase
n'ansitionwithanincreaseofannealingtime. Whenthefilmswereannealedisochronically
from room temperature up to various temperatures (813 K to 1073 K) within thirty
nfinmes,anhlaeaseofmalingtemperauneresulwdinsimflarmsldts,i.e.,adecreaseof
MgwmperanneandamorepronourrcedappearanceoftheR-phaseuansifion. Twotypes
ofprecipitateswereobservedinelectronmicloscope. Theprecipitatesformedinthegrain
bmmdafiuwaeidendfieduTizNi,anddnseappeafingmmegrainmtaiaswaeTigNi4.
No TigNi was foundin thefilms annealedat 813K for less than two hours. A small
amountofT‘izNi precipitates were foundin 8-homand16-hourannealedfilmsand alsoin
the873Kannealedfilms,andmuchlargerquantitiesofTigNiwereobservedinthe973K
and 1073Kannealedfilms. Theprecipitation of theTigNia, however,wasonlyobserved
in the 973 K and 1073 K annealed films respectively. Therefore, the decrease ofM,
temperatureinthe813Kannealedfilmswasatuibutedtothedepletionoftitaniuminthe
mauixduetotheprecipitationofTigNi. Theslightdecreaseosttemperaulreinthe873
Kannealedfilmthenwasexplainedbythesameargument. AfurtherdepressionoftheMg

32
WinniemxandlmKannealedfilmswasexplainedbytheeffectofintemal
stresaaaaociatedwiththeappearanceoftheTigNiaparticles.

Ahm-mgeTEMsnrdycmiedmubonhnmnlflrevealedthatdrecrynamzadon
rewfionpopeuedmenpidlymmeimickapmofmefiImindiafingdntduwfion
isgovernedbybulknucleation. Theaustenitegrainsizewasaboutlto3um. Atlower
mpaanne,mutensitewfianmfame¢butmenumberofvafianninasinglewsteniw
grainwaslimited. Somegrainsonlycontainedtwomartensitevariants.

Thick TiNi films which exhibited reversible shape memory efi'ect were made by
Kuribayashietal. [117]. Filmswith 10umthickness were deposited onto NaCl plates
andwereseparatedfromsubstratesafterward. Thefreestandingfilmswereannealedat
1073Kfor10minutesandthenwereconsu'ainedinaglasspipewith3.7mmindiameter
forasecondannealingat673Kforsixhours. Theannealedfilmwasafterwardcutina
horseshoe-er shape with each pad connecting to a electrical power source. Using a
recanguhrmveebcuicdcmrentmhemmefilmlmrseshoe,amvmibkbendingmodon
withamaximumfi'equencyofSszasdemonstrated.

AplanarfilmspringofTiNialloywasalsofabricatedbyWalkeretaI. [118,119].
T'rNithinfilmswasdepositedontoasiliconsubstratewhichhadbeencoatedwitha3um-
thickpolyimidefilm. ThethicknessofTiNifilmwasZum. Awet-etchwasemployedto
produceazigzagged spring pattern of TiNifilm. Theas-sputteredfilmswere givena
WmannealmGfiKformehomandyieldedcolumnarsmunewithgminsize
ofas-lum Afterwarddlefilmpatternwasreleasedusinganisouopicetchtoremovethe
polyimide spacer. Both the as-sputtered film pattern and annealed film pattern released
fromspacerexhibiwdsomecurling'duetoresidualsuess. Thecurledfilmsthenretln'ned
towudtheiraigimhumeleasedconfiglnaficnafieraresisfiveheafingwasappfied

nnrtaetou761argmddratthe"shapemnnmyefiect"obsavedinbomtmmphom
andcrystallineTiNithinfilmsintheaboveworkbyWalkeretal. [118] mightactually
resultfromthermalexpansioneffects. They‘depositedT'lNithinfilmsatvarious substrate

33
temperamesbyspuuaingandfoundthatcrysmninefilmswerepmducedonlyfm
substrate temperatures of exceeding 673 K. Films deposited at lower substrate
temper'aurreswereamorphousandwereannealedat673Kforonehour. No significant
crysulpakswereobservedinX-myspecuafamemneabdfilmsralthoughathamal
hysteresisappearedintheirelectricalresistivitycurves. Theas-sputteredamcrphousfilms
werealsoheatedby applyingelectricalcurrent.‘ Maximumcurrentdensityofabout
60(X)Alcm2 (i.e.,1.6x105W/cm1)wasappliedtotheamorphousfilmsbutthejouleheat
failed to uigga' the crysurllizaticn reaction. ‘ '

Comment and Prospect. The study of a sputter-deposited Tu9,7Nigo,3
filmcarriedout by Busch etal. [115] demonstrated the potential ofT'iNi films when shape
memory effect and superelastic effect is required. Constrained annealing, the processing
procedure which has been well-developed to generate reversible shape memory effect in
bulk alloys, were also successfully applied in thick TiNi films [117]. However, a full-
rcaledevelopmemofmermoelasncTrNidrmnlmsmhesonseverncrucialismea Oneis
thecontroloffilmcompositionandfilmstructureduringprocessing. T'iNithinfilmshas
bcenfabricatedprimarilybyionbeammixingandsputtering.Ikutaetal.[76]repa1edthe
lower-bound substrate temperature of 673 K to yield crystalline films by sputtering Rai
and Bhattacharya [113] showed that negative substrate biasing resulted in titanium
depletion in the as-sputtered film‘s. Though both studies supplied important information
about film fabrication, more systematic. studies “must be carried out to improve the
understandingoftherelationshipbetween processingparametersandfilmcompositionand
structure. .

Otherimpcrtantissuesarethethermalstability ofmicrostructure, thethermoelastic
transforamtion characteristics and associated mechanical properties in thin films.
Busch etal. [116] showed that high temperatme and/or long time annealing caused second
phase precipitation and transformation behavior variation. This is not the usual case for
bulkalloys. Thestabihtyofazmsteniteindrinfilnsdrereforemaybeaffectedbyertnnn

34
factorssuchasgascontaminationathightemperatlne. Moreover,theappeu'anceofTigNi4
precipitates at 973 K and 1073 it also challenges the TTT diagram [38], derived from a
Tiaras; bulk alloy, according to which the upper-bound temperature where the
precipitatescanappearisabout950K. Inaddition,’observationofulu'a-finegrainsizeand
limited number of crystallographic martensite variants within austenite grains, reputed by
Johnson [6], showed that thin films have pronounced differences in microstructure, as
comparedwithbulkalloys. .Thisindicatedthatthemechanicalproper'tiesofthinfilmsmay
alsobedifferentfromthose-ofbulkalloys. However,dedicatedandsystematic studieson
inthinfilmsarestillnotavailable. '

Walker et at. [118,119] demonstrated a deposition and patterning procedure which
could be accommodated to the present processing techniques used in silicon-based
microelectronics. Howeva,drela¢kofbasiccharacterizafiononas-spumredandannealed
films makes the results of this work dubious, since in both case, the films ought to be
amorphous according to the crystallizatibn data reported by Busch et al. [115] and
Kim et al. [112].

InmmaryflrepresentmsemchmnearequiamnncTrNiafloysinthefmmofdrin
films are still scattered and inconsistent. The present work is accordingly designed to
address on two major issues: The correlation between processing parameters and film
composition, and the corresponding microstructureand transformation characteristics of
masnctron sputter-deposited Ti(Nin) films

3. EXPERIMENTAL METHODS

The principal goals of the work are the study of the microstructure and
mnsfmfimchmuisdcsmTrNiminfilmwimemhasismmemhfimshiphetwem
processingparameters,su'uctlneandpropertiesastheycomparewith similarbulkalloys.
For this purpose, single layer and periodic multilayered Ti(NiCu) films were prepared by
triode magnetron sputtering. The as-sputtered structure, the crystallization behavior, the
post-annealing microstructure, and the thermoelastic transformation characteristics were
then studied using various methods. Ion beam assisted deposition experiments were
undertaken to develop a better understanding of the composition shift found in the
magnetron sputter-depositedfilms. Theeffectofion bombardmentonfilmstructlue was
also studied. A flow chart summarizing the experiments is provided in Fig. 3-1. This
chapter describes the method used to fabricate Ti(NiCu) and TiNi thin films, and the
characterization methods employed to study their composition, crystallization behavior, as-
sputtered and post-annealed microstructlne, and thermoelastic transformation behavior.

3.1. Materials

In this section, the specification of sputtering target materials, substrate materials
andrelatedmamialsusedinthepresentworkwillbeprovided.

The sputtering target alloys with composition of Tiso_osNi44,99Clla,95 and
Tigo,17Ni49,gg respectively were donated by a commercial vendor. Both alloys were cut
fromhotextrudedrod,62mmindiameter1. Theternaryalloywasthenmachinedintoa

 

imammmanemmmammmmavmmm
m.

35

36
munddisk,57mmindiametaandabGu7mmichnssfauiodemagneuonspuuaing.
Thebinaryalloywasfurtherrolledatabout12MKintoathinplate,85mmindiameter
and3rmnthickness,forionbeammumring. Bothtargetsurfaceswerefinalgroundusing
#6“) abrasive grit papers. In addition, a custom pure titanium target (99.91 at% Ti),
fabricated by Varian Associates Inc., was supplied by Professor William Pratt in the
Department of Physics and Astronomy at Michigan State University for magnetron
spumrhgandusedwithoutfmtherpreparation.

Inionbeamassisteddeposidon(IBAD)expaiments,dtaniumfoflsandwheswae
widely employed for parts, such as ion gun apertlnes, target shielding and composition
trimming wires, which directly suffered from sputter-beam irradiationz. All the foils and
WWfiomJohnmnMauheyCompany,haddiesamcomposifimof99.8at%
Ti with majority of impurity of about 0.102 at% oxygen, 0.031 at% iron and 0.(X)9 at%
nitrogen Aficrmachininthefimniumpartswaefirstcleanedinanetchanacmsisfingof
onepartHF,threepartl-INOgandfivepan‘erinnvolumeandthenbydistilledwater,
acetoneandmethanolsubsequently. Thesubstrateshutterandsubstrateholdersweremade
fromlllllseriesaluminumalloysheets. Allthealuminumpartswerecleanedinthesame
MummmummmmwithTrNipriormthelBADmmnconnd
straycontamination.

FivekindsofnnterialswereusedinuiodemagneuonspuueringandIBADasthe
substrates. Potassiumchloridewaferswerecutfromalmex12mmx76mmcleaved
single crystal, purchased from Harshaw/F’rltrol Partnership, and were employed to deposit
filmMnananissimelecnonmicmscopya'EM),anddifiaenddscanmngcahdmeny
(DSC) studies. Films for electrical resistivity measurements were made on fused silica
plates,lmmthick,whichwerecutintol2rmnxlmesquaresbyadiamondblade

 

2Evendroughconvergerngridswereernpolyedfa'tlresraltter-beunsancedivergaenceoftheionbeanrstill
exieted. Forexample,spuueringof8weVions,gareratedfiunadistanceof10cm withoutaperatlae,
undueedavisiblemakwithaboutlScmindiamemonthetitanium shielding. Therefore,apertlneswere
womandivagememdadlielfingpodtkneduhhdmemgu(m-cdbdmdueflimwu
usedtocontrolcoraatninuion.

37
wheel One-side polished (100) silicon single crystals and pure copper foils (99.9 at% Cu)

of 0.1 mm thickness, purchased from Virginia Semiconductor Inc. and EM Science
Company respectively, were also used as substrates in magnetron sputtering and IBAD.
F’mally,mieroscopeslides(25mmx76mm)wereusedintheIBADexperimentsforfilm

thicknessmeasurement.

3.2. Deposition Equipment
3.2.1 Magnetron Splitter-deposition Apparatus .

System Geometry. Magnetron sputtering was carried out in an apparatus
equipped with four Simard 383 triode magnetron sources. A schematic drawing of the
sputteringsystemisshowninFigure 3-2(a). Then-lode sourceassemblieswere situatedto
fire upward. and were held by side-mounted 8-inch knife-edge flanges with an angle of 90
degreesbetweenthem. Forntarget shutterplates,controlledbyarotaryfeedthroughonthe
bottomofthechamber, were situated3cmaboveeach uiodesourcewith twochimneyson
eachoneofthem. Oneofthechimneys was blocked by aluminum foils whileanotherhada
5 mm x 5 mm window. Substrates were loaded face-down on a rotary table which was
positioned 10 cm above the sputtering sources. The table allowed a maximum of eight
substrateholderstobeinstalled,eachofwhichcouldbeloadedwithtwo l2mmsquare
substrates. Each holder was self-shuttered while out of duty, as shown in Fig. 3-2(b). A
side-mounted wobble stick manipulator was employed to open or close the shutter by
rotating the shutter plate 90 degrees. High pressure cold nitrogen gas, passing through a
tubing filled with liquid nitrogen coaxially, was introduced into the central rod which held
the rotary table to cool the table and substrates; The motion of substrate table was
controlledby an 80386-based computerdriving a steppermotcr.

Plnlping Capacity and Gas Pressure Monitor. The system was pumped by a
rotary mechanical pump first, and then by an 8-inch cryopump (CIT Cryon 8) with
pumpingspeedof 1500l/sofair. Afterbakingforover24holn'satabout340K,the

. 38
chamber is able to attain a base pressure of 1.3x10‘ Pa consistently. Furthermore, a

Dycor M102 quadrupole mass spectrometer was attached to the chamber to measure the
partial pressures of residual gases. _

Thickness Monitors. Two quartz crystal thickness monitors (Temescal
PPM-3000), consist of quartz crystal sensors and monitor/controller units, were employed
tomeasuredepositionrateofeachsputtering source. Quartzcrystal was fabricatedintoa
diskwith 12mmindiameter,andhada6.0MHzresonantfrequency. Thequartzcrystal
sensors were held diagonally about 1 cm above the apparatuses on the substrate table,
without water-cooling, as shown in Fig. 3-2(a).

In principle, the resonant frequency, fq, of the quartz crystal sensor varies as a
functionofthemassoffilmdeposited [120]. Asaresulhthemassoffilmdepositedcanbe
expressed as

mf = (”(1 Pa / fife Z) (“W112 W 10.4 'fc ) ’fqm} (3-2-1)

where Nq is the frequency constant for AT-cut quartz crystal, pq is the density of quartz,
fq is the resonant frequency of uncoated quartz, fc is the resonant frequency of coated
quartzandZ istheacousticimpedanceratiowhichiswlittenas

Z = (pq Ha I pfuflm (3.2.2)

wherepf isthedensityofdeposited film,anduq anduf aretheshearmoduliofquartz
crystal and deposited film respectively. The depostied film thickness can be calculated
fromfilmmassandfilmdensity accordingly. Inthepresentwork,thedensityofalloyfilm
and Ti film are replaced by their bulk densities of 6.5 gm/cm3 and 4.5 gm/crn3 respectively
to calculate their deposition rates. The acoustic impedance of 14.03x105 dyne/cm3 for
titanium is used. Moreover, the acoustic impedance of amorphous Ti(NiQr) alloy film is
not available, and is replaced by the value for quartz crystal (8.83 dynel'cm3 ).

39
Theprimaryermrindeposifionratemeasluememresultsfiomthedifference

between of bulk density and film density. According to equation 3.2.1, film deposition

ratekgcanbewrittenas
Rd? = duff/M/dt
== (dnyl fit) I pf. (3.2.3)

Thevalueof(dnrfldt) isaconstantandthereforedkggcanbeexpressedas

AR“, 2 (I/pf) - (I/pb ) (3.2.4)
whereDb isthe density ofbulk material.

The shearmodulusofequiatomicTrNiwasrepcrtedbyBuehlerandWang [1] tobe
24.8 GPa. Accordingly, the acoustic impedance of bulk TiNi is calculated to be
12.7x105 dynclcm3. However, Li [121] pointed out that the ratio between elastic moduli
of amorphous and crystallized alloys is generally between 0.7 and 0.8. If this is the case
for Ti(NiCu) film deposited, the acoustic impedance is hence estimated to be about
9.5x105 dyne/cm3 (75 ‘5), which is only 7% larger than the value of quartz crystal used in
the present work. Moreover, calculation results shows that the film mass measured varies
less than 2 $63, as the 2 value changes from 0.5 to 1. As a result, it is reasonable to
conclude that the approach used in the present work results in a negligible effect on the
accuracy of film thickness measlnement.

Sputtering Sources and Work Gas. The Simard 383 somces were driven by
two power supplies: a Simard TS/Z power supply operated at V . 550 volts and
1:0.9amperesforsputtelingcathode, andaSimardPD/20000ne operatedat V260V
I = 24 amperes for thermionic emitter in the present work. The sputtering target was
soldered with indiumontoacopperheat sinkwhich wascooledbywater. Argon wasused
asaworkinggasforsputteling, whichwaspmifiedbypassingthroughacoldtrapfleptat

L

3rhedev'aannofmtincreaaeswidrtheincreaseow, {4%.

40
about 90 K) to remove water vapor, and a hot-titanium gas purifier (Hydrox Model 8301)

tomininn'zecontaminationwithreactivegases.

3.2.2. Ion beam Assisted Deposition (IBAD) Apparatus

System Geometry. The sputtering-deposition system, contained in a 14-
imhmudbefljar,wnsiswdofa85nnndianewrflbymrga,a3—chaufmann-typeion
gun (Ion Tech Model 3.0-15(X)-1(X)), a 5—cm cold cathode ion gun (Anatech IG-5C),
substrates and relative assemblies, as shown in Figure 3-3(a). The alloy target had a
nominal composition of Tiso,2Ni49,g, over which were stretched ll bundles of three
titanium wires, 0.25 mm in diameter, spaced at 6.3 mm intervals, bringing the average
effective target surface composition to Ti53Ni47. The target was equipped with liquid
nitrogencoolingandwassituatedat45degreeswithrespecttoboththeiongunandthe
substrate. Thesubstrate holderwasacircularplate, 76mindiameter, situatedcoaxially
on a rotary stage, as shown in Figure 3e3(b). Four 12 mm x 12 mm subsu'ates could be
loaded on the holder, and had a 90-degree angle between them. The substrates were
coveredbyanapatmelocatedo.5cmabovemesubsuaws,havinga24nmholeananged
toalbwedsubsuatesmbeexposedmthespuueredspeciesandionfluxoneuadme. A
aluminum shutter (coated with NiTi tocontrol stray contamination) was positioned 1.5 cm
above the substrate holder, and controlled by a central rotary fecdthrough. Titanium
apertures and shielding were used to control contamination by stray sputtering from
chamber structures.

Ion Source and Working Gas. The sputter target was irradiated with 800 eV
argon ions, having an incident angle of 45 degrees, using the 3-cm Kaufmann-type gun
fittedwithconvergingoptics. The3-cmiongunwasdrivenbyanlonTechMPS-3m0FC
powdersupply, running with adischarge voltage of55 voltsandan acceleratorcurrentof
2-3mAatabout4mvolts. TheassistedbeamwasgeneratedbytheS-cmcoldcathodeglm
mounted with its axis normal to the substrates. Argon gas, with purity of 99.996 ‘5,

. 41
supplytobothiongunswaspassthroughadryingcolumnandagaspm'ifyingcartridge
(OxiCJearDGP-250R1)toremovetracewawrandoxygen. Gas flowtothe 3-cmgunand
5-cm gun during sputtering brought the chamber pressure to 2.0x10'3 Pa and 2.6::10'2 Pa
respectively. Gasflowrateswerenotgenerallymeasured

Pumping Capacity and Gas Pressure Measta'ement. The l4—inch metal bell-
jarwaeequippedwi¢mAlcaml5400nnbomoleaflarpunqrhaMgapumphgspwdof
Wilsofair,andanAPDcryopumpwithpumpingspeedof12001/sofair. Thechamber
base pressure attained about 1.3x10‘5Paafterbakedfor4hours. Residual gaspartial
pressuresweremonitoredbyaDycorMA lOOquadrupolemassspectrometer.

IonCta’rentMeasurement. ' Theassistedbeamcurrentwasmeasured
by a faraday cup situated on the substrate shutter, as shown in Fig. 3-3(b). The faraday
cup, 9.5 mm in diameter and 5 mm height, was biased by a negative potential of40 volts
supplied by the MPS-3000 FC power supply. The ion current was displayed on a digital
meter, with resolution of 0.01 mA, on the power supply.

Substrate Heating. During deposition, substrate heating was
accomplished by radiation from a 600 W quartz lamp situated underneath the specimen
fixture, illuminating the substrate through an aperture in the specimen retaining plate.
Temperature was controlled by a powerstat and monitored with a fine gauge type-K
thermocouple loosely mounted on the back ofsubstrate.

3 .23 . Modflied IBAD Arrangement

' System Geometry. Amodified IBAD experimentwasdesignedto studythe
effect of varying ion/atom arrival ratio on film characteristics. The target position was
modifiedinaway thatthenormaloftargetdidnotintersectwiththeassist—beamaxis.
Instead,thetargetwasrotated about30degreeswithrespecttotheverticalfeedthroughon
whichthetargetwas situated,andstillkeptanangleof45degreeswiththel(aufmanngun

42
axis‘. Thesubstrawswerearrayedinamanner, showninFigure 3-4,whichexploitedthe

inherent assistcbeam current profile, together with the natural angular falloff of the
depodfimflmngameasymficmiaflonintheiawtanmnodmingdeposidm.
Ion Current Measurement . Assist-beam currents were measured with four
separatebiasedfaradaycups,9.5mmindiameter. Thefaradaycupsweresituatedona
substrate shutter which was fixed onto a rotary feedthrough, and could be swung into
positionsymmetricaltothespecimenlocationwithrespecttotheaxisoftheassist—beam
ngheimcmreotwasdrend'mplayedonadigihlmeterwifiresoludonofOJuA.

3.3. Deposition Procedures
3.3 .I . Triode DC Magnetron Sputtering Deposition

Thepreparafionanddeposifionprocedmesofthetanaryafloyfilmsaredesaibed
below. The relative deposition parameters of five deposition runs were tabulated in Table
3- 1 .

Substrate Cleaning . All the substrates, except potassium chloride, were
cleaned by acetone and methanol subsequently in ultrasonic cleaner for five minutes to
remove grease contamination prior to being installed. The substrate holders were cleaned
by an etchantofone partI-IF,fourpartI-INO3 and five part H20 in volume first to remove
previously deposited coatings, and then by water, and subsequently in acetone and
methanol in an ultrasonic cleaner.

Evacuation. The vacuum chamber was evacuated by cryopump for more
than48hourafterthepressure was broughtdownto0.l Pabyamechanicalpump. As
was mentioned in section 3.2.1., the chamber base pressure attained about 13le Pa
afterbaking. Theexactbasepressrneandthepardalpressmesofwatervapor,oxygengas
andnitrogengasofeach depositionrunarelistedin'l‘able 3-1.

 

mmmmwmmeaomummmm.

43
I’m-sputtering. Except the first deposition run (0 170), two sputtering

mmamfinniumoneanda'l‘isonsmumcmseonewm employedwdcrrosittbc
films. After attaining a chamber base pressure of about 13le Pa, argon gas was
inooducedandmechambapresmwumainnineduo.3Pabyadjusdngdresnokeof
the gate valve. Virgin alloy targets were pre-sputtered for 0.5 how at full power level to
anainswadysutespuueringconditionspriortoexposmeofthembsnates. Afterwml.
boththedtanirmnrgetanddreTKNiCuhfloynrgetwerepre-sputteredfa‘10minutesat
fullpowerlevelinthebeginningofeachdepositionrun. Sputteringparametershavebeen
describedinsection3.2.1.

DepositionRateMeasurement. Inthebeginningofeach experimentrun,the
whole substrate assemble (including substrate table, substrate table and fixture of quartz
crystal sensors) was first cooled by cold nitrogen gas to about 220 K. After each target
was pro-sputtered for at least 10 minutes, one of the thickness monitors was positioned
above the selected sputtering source by rotating the substrate table to measure the
depositionrate. Exposmeofthesensortothesputteringsotncewaslimitedtooneminute
mavoidunpaauneinaeaseofmequanzaysnlmowwmdwexamwmunesofme
quartzcrystalswerenotknown. Thedepositionrawswereassumedtobeconstantforall
filmsdepositedafterwardunlessthesom'ceswasshutoff. Typicallyafterthreesamples
weremade,thesubstratetemperaturewascloset0373Kandthepowerofbothsources
wereshutofi'foronetotwohours. Depositionrateofeachsourceweremeasuredagain
afwrpowerwasmstored. Thefluctuationofdepositionrateineachexperimentrunstayed
within +/- 0.1 A. The thicknesses of as-sputtered films were measured once by
profilometer, and the results indicated that the average deviation of true thickness and
desired thickness was about +/— 5%“.

Film Deposition. Prior to exposure of substrates the desired sputtering
mesmedeposiuonmmofeachmmcemnddesimdhyaafilmmicknesswaeenued

 

Mu-ssemunmmmmmmwwmmmmnmm

44
innacomputermdewrmimmefimemmalsfmwhichthesubstrawswaeexposedm

eachsouroe. Thesubsnateshuuerandsorneeshuuerswerethenopenedsubsequentlyto
deposit samples. Substrate temperature was monitored by thin gauge thermocouples
attachedtothebackofselectedsubstrates. Themhetratetempemnnewttnoteonrunthm
wasmaintainedbelow 373Kduringdeposition. Thefilmsdepositedfromsinglealloy
target are hereafter called single layer (SL) films, and those which deposited from
flunafingafloymguandfiMnilmmgetmecalledpaiodicmulfihyflWWJfilma For
andIeMfimmefinniumhyerwulnmthicknesswhaeasdntofauoyhyervuied
fiom9t018nm. Inthefollowingtext,theterminologyofPhfl.-”N”willbeusedto
represent the PML films with alloy-layer thickness of N am. For instance, PML-9 film
meansthatthefilmiscomposedofaltemateTi-layerandalloy—layerwiththicknessratioof
lmer nm. Detailed deposition procedures are shown schematically in Figure 3-5(a).

3 .3 .2. Ion Beam Assisted Deposition (IBAD)

In order to study the ion-solid interaction efi‘ects on film structure, thin films of
TiNi binary alloy were prepared on KCl substrates by ion beam assisted ion sputter-
deposition in a 14-inch metal bell-jar with systemarrangement described in section 3.2.2.
Two runs were performed with the experimental parameters listed in Table 3-2. Below are
description ofthedeposition prowdures. '

Evacuation. The chamber was first pumped by turbomolecular pump to
about P = 1.3x103 Pa. and then evacuated by both turbomolecular pump and cryopump.
Thebasedpressureis4.0x 10'5Paand2.0x10'5Paforrun#1andflrespectively.

IlARatio determination. 'Ion-to-atom arrivalratios were determinedfromthe
depositionrateMMndioncmrentdensitya). Themeasmementofioncmrentdensity
has been described in section 3.2.2. However, the deposition rate was not measured in-
situ for the following reason: Post, the deposition rate was estimated to be about 0.5
Alrec which was small compared to the resolution of the thickness monitor of 0.1 Nsec.

45
innddiuondteureorthicknestmommmaeuedmedinicultyotsunyoonmmimdon
control. As a consequence, a sample was made in separate run while sputtering with the
Kaufmangun-,runningwithadischargecurrentof0.6Aat55Vandanaccelerator
currentof2mAat450V,toproducean0.8keV,30mAbeam. Thesputterdepositionrate
was determined by dividing film thickness, measured by profilometer, by the total
depositiontime. Thaeforesputtaingmwsfordletestspecimenswaemaintainedatthe
predetetnunedlevelbykeepingthexnutrnnnngtmmmetemconsunt

TheI/Aratioswerethencalculatcdusingthefollowingequation:

1m. 1
am 1.6 x10”

0 J.
R" 10 pm -2-6.02x10”
Wm

 

I/A=

 

(3.3.1)

 

where HP) is the ion current, A(F) is surface area of faraday cup, pm),- is the bulk
density of TiNi and Wm is the molecular weight of TiNi (106.6).

Preconditioning. Sputtering target was pie-sputtered for 10 minutes with an
800 eV ion beam generated by 3-cm Kaufmann gun operated at the same conditions
described above. Meanwhile, the 5-cmiongun wasfiredwithemittercurrentof 120mA at
me 360voltsandbeamvoltageof500voltstoobtainastablecondition. Inaddition,
thembsuatewaspre-heatedbyaquartzlampmanaindredesimdtemperanne.

Film Deposition. Thin films for direct TEM observation were
deposited onto cleaved potassium chloride crystals. Prior to the opening of substrate
shutter,thebeamvoltageofS-cmiongunwasadjustedtodesiredvalueandtheioncurrent
was recorded. Deposition for approximately 0.35 hours produced final film thicknesses of
20-75 nm. The specimens were not rotated and the sputtered flux was thus incident at
roughly 45 degrees. Detailed procedures are shown schematically in Figure 3-5(b). Table
3-2 summarizes the deposition parameters for the various films, with references to results
shown in the next chapter.

46
IBADfilmsweredirectlyobservedinaJEOL lOOCXtransmissionscanning

electron microscope (STEM) operated with acceleration voltage of 1m kV. Samples
peparedforelectronmicroscopestudywillbedesclibedin section 3.8.3.

3.3.3.ModifledIBAD

Foreachofthedepositionruns,four 12-mm square cleaved potassium chloride
crysnlswaearrayedtogethawitha76mmx25mmmctanguluglussubsoatewhich
hadbeenfittedwithacopperfoil_mask,0.lmmthickmproduceastepequaltothefilm
thickness. Theghsssubsuateswerecleanedaccordingtothemethoddescribedinsecdon
3.3.1. The system was brought to a base pressure lower than 1.3x10'5 Pa, which
wmmpsnyingbypuddpresauesofwacrvapamxygengasandninogengasnhrhnd
inTable3-3foreachdepositionrun. Thedepositionprocedureswerebasicallythesameas
thosedescribedinsection3.3.2. Table3-4summariaesthedepositionparametersforthe
variousfilms.

IIARan'o determination. AsshowninTable 3-4,therewerethreecontro1
runs, without concurrent ion bombardment, carried out to measure the deposition rates.
Theion-bamcmentdensifieswaemeamnedindneelBADmnswimassist-bumenagy
of 50, 1(1) and 500 cV respectively. Figure 3-6 shows effective deposition rates, ion
currentdensityandion-to-atomarrival(llA)ratiorespectively,eachplottedasafimctionof
thelateralpositionofthesubstrates,forwhichtheaeropointcorrespondstoalocation
whichwascoaxialwiththesputteringtarget. AsecondX-axisscaleisshownwhichgives
the effective assist-beam incidence angle, 0, which is taken as zero for normal incidence
(ie.,foraposidononthecoldcathodegunaxis)andrisestoapproximately25degreesfor
the specimens located near the sputtering target centerline, as shown in Fig. 3-4. The
depodfionrammeaduaminedbydividingmeasdeposiwdfilmmicknessbymemnl
spuneringtime,areshowninFigme3-6(a)forthreesepamtecontrolruna Thedeposition
mmamflemdlemgulardependenceofthespuwedfluxfiomtheanoymgetand

47
theincreasedefi'ective subsu'ate-targetdistanceforthespecimensdisplacedfromthesputter

targetaxis. rhedrrmnticincteaseorthedepotidonrrteinconuol-lnmrertutedrromthe
useofcmvagentgridsofKaufmanngmwhibdivagentgridswaeempbyedinwnnoL
2 and control-3 runs. The sputtering-beam current of 65 mA used in control-3 run was
higherthanthatincontrol—Zrun(50mA),whichresultedinadifi‘erenceinthedeposifion
rates. In addition, ion current density curves are also shown in Figure 3-6(a) for three
separateIBADruns. Thesecurvesreflecttheradialcurrentdensitydistributionintheion
beam, which depends on the ion energy and total ion beam current. As a result, the
experimental arrangement shown in Figure 34 can generate a systematic variation in the
VA ratio faseveral specimensinasingle deposition run.

Thecalculated I/Aratioswerethus determinedflomtheunassisteddepositionrates
and the incident ion current as measured during deposition. These values are seen in
Figure 3-6(b) to vary from 0.19 to 1.13 for the 500 eV IBAD run, and from 0.17 toO.57
forthe 100 ev IBAD run, and0.08 too.14 forthe IBAD run at50eV. Becauseofthe
fallofi'ofassist-beamcmrentwithdeviationfromtheaxiallocation,andbecauseofthe
increaseinthesputterdepositionfiuxasthespecimenlocationapproachedtheaxisofthe
spuuerurgetlughervnluecot‘ognverisetolowervAmdos.

Modified IBAD runs were Conducted to fabricate thick films for composition
analysis and sputtering rate measurement, which will be described in section 3.6 and 3.7
respectively.

3.4. Crystallization Behavior of As-Sputtered Ti(NiCu) Thin Film
3 .4 .1 . Crystallization Temperature and Entllalpy

The crystallization characteristics of as-sputtered films, including free-standing SL
films, PML-18 films on Co substrates, PML-9 films on Si substrates, and free-standing
PML-9 films, were studied by DuPont 910 differential scanning calorimeter (DSC), using
drynitrogen, withaflowrateoi50c.c.lmin,aspurging gas. Specimensloadedin sealed

4s
aluminumpanswereheatedwithconstantrateson,10,30and50K/min.respcctively.
fromambienttemperatureto823K. Thesamplewasweightedbyanelectronicbalance,
withmaxinurmresolufionofo.lmg,priortobeensealedinthealunfinumpan. Sincethe
weightofthefilmsonCuandSi substratescouldnotbemeasureddirectly,thewhole
sample (film plus substrate) was weighed first. Afterward the film surface area was
estimated from subm weight (W3), density (1);) and thickness (is) as follows:

A = W3/(pgx t5) 3 (W, + Wp / (p51 ts) (3.4.1)
where (Wf +Ws) is the overall sample weight. The weight of film (W1) was then
computed as

Wf 5 pf! (A x tfl (3.4.2)

wherethepf isfilmdensitywhichwasreplacedbythebulkdensityof65 g/cm3inthe
calculation and if the desired film thickness, estimated from the deposition rate and total
deposition time. In the present case, the thickness of copper foil and silicon wafer was
measuredbycaliperstobeOJZmmand0.26mmrespectively. Errorscanarisefi'omthe
use of bulk density, the roughness of the sample edges and the film thicknesses which
were not checked by profilometer.

. Thecrystanizafimtemperammwasrefermdtothepeakposifionofdlefirstsharp
exotherm in the DSC trace, and the transformation enthalpy was the integral under the
whole exotherm peak minus background. The activation energies of crystallization of the
ternary films were then calculated by using Kissinger's model [122]. Here it is assumed
the that the tenurerature of maximum deflection, Tp, in differential thermal analysis (DTA)
co'mcideswithdletemperauneatwhichthereacfionmteisatamaximum. Formostsolid-
solid reactions, the rate equation can be expressed as

dx_ _ . __E_
E-Aa x)¢xp( m.) (3.4.3)

49
Ifthetemperaturerisesataconstantrateofs=tfl'ldt,thenbydifferentationofequation

3.4.3,

 

dt( dt dt RT“ An(l x) at” RT». (3.4.4)
Atr=r,, %(—:)=0 and

An approximation of n(1-x),"'1 = I was used by Kissinger and the activation energy E
waswrittenas ‘

dun-17’3-

-—-_,L (3.2.6)
(1(7) )

This method has been widely employed in the analysis of the kinetics of crystallization

reaction,andhasbeenprovedtobevalidforbothDTAandDSC[123,124].

3.4.2. X-ray Difliraction

Theaystalsoucnnesofas-spunaedandisochmnanyannealedfilmswaesuldied
by X-raydiffractiononaScintagXRD 2000diffractrometeroperatedatV=40kilovolts
andls30mAusingCuKaradiation. Theas-sputtered samples were scannedfromtwo
thetaof30°to60°inarateof0.4°lmin. Theannealedfilmswerescannedfi'omtwothetaof
37° to 49°, since the strongest peak of most ternary phases, such as 32 TiNi, T'iz(NiCu),
Ti(NiCuh, and TieoNisa,5Cu3,5, are located in this range. The scanning rate was
0.15°/min.

50
3.5. Heat Treatment

Theas-sputtaedfilmswaegivenacrysnlfizafionand/orhomogeninfionmnealat
varioustemperatures,mainly823Kand923K,ineithaaquartz—tubevacuumflnnaceora
14" stainless steel vacuum chamber. The tube furnace was pumped by a 15-cm diffusion
pumwhichmaintainedapresslneofabout26x10'3Padminganneal. Sampleswereheld
in an alumina crucible situated in the center of the heating zoom. Temperature was
mfiuedbyaTypeKmamocouplewhosedpwassinntedinsidemeheafingmneabom
1 cm above the tube center. The temperature was maintained within +/- 2° K during
annealing. Ondreodrerhanddlememlchamberwasabletoauainabetterbasepressrneof
1.3x10'5Pa and increased to about 2.6x104Paduring the anneal. A Sillwatt quartz-
halogenlampwasemployedastheheatsourcetoannealsamplesatadistanceof30mm,
and was manually controlled by a powerstat. A 0.1 mm-thick stainless steel foil with a
thin-gauge Type-K thermocouple spot-welded on the back was positioned next to the
sampletomeaslnetheannealingtemperature. Thefluctuationoftheannealingtemperature
waslimitedwithin-l-l-4°K. Afterward.sampleswerecooledinvacuuminarateofabout
20%

3.6. Composition Analyses

Elemental composition of the films was determined by energy dispersive X-ray
microanalysisinaJEOL l(X)(D(scanningtransmissionelectronmicroscope(ST‘EM)anda
Hitachi S-2500 scanning electron microscope (SEM) respectively. Thin filrrrs prepared by
electrochemical polishing, which will be described in 3.8.3, were used in the STEM,
operated at 100 kV in scanning transmission mode, to collect spectra. A specimen was
loadcdinagraphite holderandtilted 30degreestowardthedetectortoobtainahighX—ray
take-off angle. Two standards, a TigoJNiepp alloy and the sputter-target

(risossNimgcmp, ), were used to determine the intensity ratios of Ni, Ti and Cu.

51
Shwednminfilmqiuimmsadsfifimefilmmosidmwucflculawdaccudingm
tlreQifi-Inimermethod[l60]whichcanbeexpessedas:

M=Mxxu=flxlflx£fl (3.6.1)
c,(u) mu) mu) I,(s) c,(s)
212‘)...ng =flflx141‘lx9efl (3.6.2)

qua-mu) “ mu) us) C,(s)

g+q+q=1 on»

where C; is the concentration of element i , I.- the intensity of the characteristic x-ray line
of element i , C(u) the composition of film and C(s) the composition of standard
respectively. The term K t,- varies with operating voltage, but is independent of sample
thickness and composition as the intensities are measured simultaneously. Thus, the values
ong obtainedfiomstandardscanbeusedtodeterminethefilmcompositions.
Composition analysis carried out in the SEM-EDS system employed a different
approach. Films with thickness of 5 pm on Si or glass substrates were illuminated by the
electronbeamwithanenergyoflStomkeV. SincenosilicOncharacteristicradiation was
observed. the substrate was taken to have no interaction with electrons, and the spectra
could be thus treated by a ZAF correction. A program called ZAF-4, supplied by Link
analytical Inc. was used to analyze all the spectra. which were collected under carefully
controlledconditions. Theopticsofelectronbeamandthegeometry betweenlens, sample
and X-ray detector were rigorously consistent during data collection. The LaBg electron
somcehencehastobestabilizedforatleast1.5hourpriortotakinganyspectra. Alltlre
spectra were collected at the same working distance (15 mm) , magnification (1000 x),
tiltingangle,deadtimepercentageandtimeinterval. Theelectricalparametersofthelenes
remained untouched throughout the whole process to yield a consistent electron intensity.
The sample image was thus focused by adjusting the height of sample to maintain a
consistent working distance. The Kit peak acquired from a pure nickel standard was

.52
employedtobethecalibrationspectrumwhichwascollectedevery 15to20minutes. Prue

copper,nickelandaTiso,1Ni49_9alloywereemployedtobethestandardsforCrnNiand
Ti respectively, and the composition of target alloy was also calculated to evaluate the
accuracyofthismethod. Thebulkmaterialswerelastpolishedbyusingo.05umalumina
powder to yield mirra-like surfaces and then cleaned by acetone and methanol. The
specimern,hrcludingsnnduds,waeloadedonagraphitestagewithoutdlfing
Thecomposifionofspm-mgetalbydeterminedbytheZAFmedrodis:502at%
Ti,45.7at%Niand4.1at%Cu. Theresult showsthatthemeasured titaniumcontent of
thetargetalloyisingoodagreementuddrtherealvalue‘whereasdlenickelcontentisabout
0.7 at% higher, and the copper content is 0.85 at% lower than the real values. Larger
deviationsofthecopperandnickelconcentrationcanbeascribedtotheoverlapofCuKa

andNiKflcharacteristicpeaks,which resultsinanoverestimate ofone element(Niinthe
presentcase)andunderestimateofanother.

Both methods mentioned above have been used to examine the composition of
IBADfilmsdepositedontocopperfoils. The 100eVand500eV IBADfilmsandtheir
conuolfilmswereanalyzedbytheCliff-LorimermethodcarfiedoutintheJEOL 100CX
operatedat40kV,sincethisEDSsystem wastheone availableatthattime. The50eV
lBADfilmsandtheassociatedcontrolsarnplewerelaterstudiedintheI-I-ZSOOSBM.using
ZAFmethod.

3.7. Thickness Measurement

The IBAD films and the associated control films deposited onto glass substrates
were used to determine the deposition rate by measuring the total sample thickness. A
Dektak II profilometer was employed to scan across a 3 mm wide step, produced from

 

‘lhe composition of sputter-target alloy (TisomN'ruggCuago) was supplied by the donor, us. Tinitol
gammy. lioweva,mfinduinfumafionwmwaihbleMwhathdniqmwaempbyedbrhflmhe
compo-don . .

53 .
covering a copper foil during deposition, and then yield the thickness profile. The

scanningdistancewasSmmwiththelowestspeedoflmm/min.

as. Microstructure Evaluation
3.8.1. X-ray Dioraction

The phase constitrrents of annealed Ti(NiCu) films were studied by using the same
ScintagXRDZWdim'actometerdescribedinsection3AJ. Thesampleswerescannedin
astep-scanmodefor0.05°perstepandwithadwellof15 secondsforeach step.

3.8.2. Scanning Electron Microscopy (SEM)

Themorphologyofboththeas-sputteredandannealedTimiOr) filmsproducedby
magnetron sputter-deposition were studied by SEM. The as-sputtered films deposited on
silicon substrates were bent to fracture to provide surfaces transverse to the film plane for
obsavadoninalfimchiS-Mscanningdecnonmiaoscopeopaatedatwkv.

To observe the cross-section of the annealed films, free-standing films were first
mounted in acrylic, and then polished by using 0.05 um alumina powder. The polished
samples were chemically etched in a solution consisting of 20 % H2804 and 80 %
methanol, and finally coated with gold (~10 nm). The samples were also observed in a
Hitachi S-25(X)scanningelectronmicroscopeoperatedat20kv.

3.8.3. Transmission Electron Microscopy (TEM)

Thin IBAD films deposited onto KC] substrates were first broken into 2~3 mm
squares,andthenfloatedontodeir>nizedwaterwiththefilmupward. Afterthefilmflake
separatedfromKClandfioatedonwater,a100mcshfoldinggridwasusedtoliftit. After
being dried in air, these samples were studied using a Hitachi H-800 transmission electron
microscopefl‘EM)operatedat200kV,oraJEOL 1(X)CXscanningtransmission electron
microscope (STEM) at 1m kV.

54
The TEM samples of the magnetron sputter-deposited films were prepared as

follows: Annealedfiee—standingfilmswaecutinto3mmsquaresusingasurgicalblade.
andthensandwichedbetweentwoCugridswichmmxlmmslot. Afterwardthefilrm
waethimedbydecoopohshingusmgmetwinjetpohsheropmtedatV-BOVoltand
I-90mAatambienttempentme.usingmelecuolyteconsisnngof5%paehloricacid
and95‘lrawticacid.

Themiaosmwnneofthefieesundingfilmswasthenenminedbyusingailitachi
H-8lDT‘EMoperatedat200kaithadouble-tiltholder. Gystalstructlneinvestigationof
precipitatephasesintheSLfilmswascarriedoutusingconvergentbeamdiffractionona
JEOL 2000 FX microscope in the Electron Optical Laboratory at University of Michigan,
and refined elemental analysis of precipitates was done on a Vacuum Generators 501KB
dedicatedSTEM.

3.9. Martensitic Transformations in Ti(NiCu) Thin Films

3.9.1. Differential Scanning Calorimetry (DSC) .
Mutensiteuansfamadmtemperaunesandenflnlpiesofdreisodramaflyanneabd

films were determined by DSC, which was described in section 2.1, with 5 Wu heating

andcoolingrates. Thetransformationstartandfinishtemperamres were assignedtothose

correspondingto5 % and95 %, respectively, oftotal heatreleasedorabsorbeddulingthe

exo- or endothermal reaction.

39:. ElectricalResistivity Measurement-
Themsfamafiontemperatmesofthefilmsonglasssubsuatesweresnldiedby
four-probe resistivity method, using '5 K/min heating and cooling rates. The resistivity
measurement apparatus is shown schematically in Figure 3-7. Samples for electrical
resistivity measurement were deposited on fused silica substrates overlaid with an
alunintunmaslttoproduceazmmwidesuipwithnveattachmentpadstorelecuodetand
a thermocouple, as shown in Fig.3-7.’ The sample was situated in the middle of a 135le

55
mmtestnrbewhichwaswrappedbyheatingwireandpositionedinadewarfiask Thin

gaugecopperwuesandaT-typethamocoupbwaesoldaedwithpmeindiumonmfilm
pads. Abatterysuppliedaconstant¢mrentof0.4mAdnoughthecirouit,whiledrevoltage
drop(AV,)acmssa3nnnlongfilmsuipwasrecadedbybothaXYrecaderanda8086
basedcomputer.'ForaS-ttmthickfilm,AV3valuewasabout0.2mAandthenoiselevel
waslessthan0.01mA. Thetypical electrical resistivity was estimatedtobeabout
zoo tin-cm. Steady cooling and heating rates were accompanied by adjusting the flow
nedconmuogengaaandmemtpmotpowermpplywhichuccnnectedmtheheaung

wire.

3 .93 . X -ray Difl'raction '

A Rigaku RU-ZwB X—raydiffracuometer, operatedat V= 50kV and]: loom.
equipped with a Rigaku CN 2351B 1cold stage was employed to study the crystal structure
ofthe isothermally annealed Ti(NiCu) films at various temperature from 323 K to 80 K.
The whole cold stage resides in a sealed vacuum chamber, as shown schematically in
Figure 3-8. The sample holder with a build-in heater was mounted on a cold finger which
is connected to a liquid nitrogen dewar. Sample temperature was controlled by adjusting
thecurrentofheawrandthestrokeoftheliquidnitrogen stopper. Sampleswereirladiated
byCuKaradiationinascanrateof1°lminfromtwothetaangleof35toSOdegrees. All
the sanrples were first heated up to 323 K. and then cooled subsequently to desired
temperatures to collect spectra. The temperature fluctuation during scanning was
maintained within +/- 1 K.

3.9.4. Transmission Electron MicrosCopy (TEM)

The martensite phase, which was stable at ambient temperature, was studied using
the Hitachi 800 TEM, operated at 200 kV, with a double-tilt stage. In addition, the JEOL
100 CX STEM with a Gatan 626 single-tilt cooling stage was employed to study the
thermoelastic transformations in-situ at low temperature. The sample temperature was

r 56
detectedbyasilicondiodemountednearthesampleholder. A30Wheaterwasmounted

inside the supporting rod, cooled by liquid nitrogen, to control the sample temperature.
Theaccuracyofthetemperaturereadingwasnotsuppliedbythevendor.

4. Resorts Alto Drscussrou

Theexperimental findingspresentedinthischapteraredivided intotluee sections.
each containing both results and discussion ofindependent topics. Section 4.1 reports on
memhfimbetweenprocessingpanmaasandnnaosuucuneasmnucomposiflonof
the as-sputtercd TiNi films. the crystallization behavior or the magnetron sputter-
deposited films is also presented in this section. The microstructures of the annealed
Ti(NiCu) films, fabricated by magnetron sputter-deposition, are described in section 4.2.
Section 4.3 presents the thermoelastic transformation characteristics of the annealed
Ti(NiCu) films. Section 4.4 discusses the resultant microstructure and transformation
behavior of the magnetron sputter-deposited films in comparison with bulk alloys of
thermoelastic TiNi.

4.1. Fabrication of TiNi Thin Film

Section 4.1.1 describes the structure of the as-sputtered Ti(NiCu) and TiNi films,
fabricated using the triode magnetron sputtering apparatus described in section 3.2.1, and
the dual ion beam sputter-deposition system described in section 3.2.2. The compositional
analysis results of the magnetron sputter-deposited Ti(NiCu) films are reported in
section 4.1.2. The composition variation found in the deposited films is rationalized by
resultsfiomaseriesoleADexperimentsaccmdingtomethodsdesclibedin section 3.2.3
and 3.3.3. Section 4.1.3 reports on the crystallization behavior of the magnetron sputter-

deposited Ti(NiCu) films.

57

58
4.1.1. Structural Characterization ofSputter-Deposited Films

Morphology and Structure of As-Sputtered Ti(NiCu) Films . An as-sputtered
SLfilmdepositedonasificonsubsnatewasbentwfiacnuemdobservedbyscmning
electron microscopy (SEM). Themicrograph showninFigure 4-1 reveals thatthetop
srnfaceofthefilmissmoothandthatthefiacuuesurfaceisfieefiomvoids. Noindication
oftheexistenceofaaonestructure,whichisacommonfeatureofthesputta-deposited
films,isfound. I.ong,wavyslipband8.W0PW8inroughlythesamedirection.were
observedtoextendfromthefractmesurface. Thelackofwell—definedslipplanesindicates
thattlrefilmisamorphous. Themorphologyofdreas-sputteredPWfilmsshowedsimilar
features. X-raydiffractionspectraofas-sputtered SL,PMLo9,PML-12ande.-l6films
showanigme4-2vaifydmtmesefilmswaeamorpbousinmeas-spuwedcondifion
umdicandbymebroadfirstordapeaksbcawduatwodremweofapproximamlyfl
degrees.

Thedenfityandsmoothnessofmefilmsfabficatedbyuiodemagneuonspumfing
canbeattributedtoseveralfactors. Allfilmsweredepositedatlowworkingplessure
(0.33 Pa) which allowed the adatoms to have higher average kinetic energy, promoting
surfacediffusionandthusthedensityandsmoothness[125]. Inaddition, elevated surface
temperature1 resulted from heatof fusion deposited by the adatoms and from energy
nansfumdbybombudnentwimenageficpamcksfimludmgebcuonaionsandneuufls
This further decreases the sticking probability of inert gas and increases the surface
diffusionrateofadatoms [126]. Also, bombardment by high energy neutrals reflected
fromtargetdmingdepositionmayfirrtherdensifiedthegrowingfilms[94]. Thelackof

 

IAlthoughthesubeuateemperaturetweremaintninedbelowmx.i.e.,r/r.,-o.zss.intertacerercfion
mdlmdimrfimbuwwnfdeSi(lw)mbsmwaecommauyobsavedaMmsflwdmabmding
whichissosumgmafiacuneocanedmosflymsidednsificmwhendnfihnwaspededofi. Thereal

mmefihnanfwedmingdeposidmduefaenmybelmeKhighadmdntmeed
hmdebmkfidedmemmmebadsofmwmhMVehukamddmmwdiffiumm
Ni-Si,T'i-Siandl~liTi-Siattemperaturesabove473Kresultsiafamatioaofsilisidesmdamorphous
phnes.

. 59
shmpmfwefaceummeyowingampmfimmmenamlmcidunemgleofme

depositionfluxminimiaedtheself-shadowingeffect.

TEM Observation for IBAD TiNi Thin Films . In the present study, it
wasalsoofintereuwdetaminewhethalBADcouldmduceaysulfiudonofgrowing
filmsatmoderatesubstratetemperatrues(3m-500K) [104]. Figure4-3 shows avarietyof
TEM results for the ion-sputtered film (Fig. 4-3(a)) and forvarious IBAD processed films
in Figure 4~3(b) through (d). The ion-sputtered film (Fig. 4~3(a)) is fully amorphous, as
indicatedbythedifi'usefirstorder(d~0.215nm)andsecondorder(d~0.125nm)halos,
and thecomplete lack ofcontrast inthe bright field image. Figure 4-3(b) through 4-3(d)
showhightfieldhnagesanddiffiacfimpamsoleADfilmsprowswdwimmaudng
ion-to—atomarrivalratiosandassist-beamenergies. ThediffiactionpatterasoftheIBAD
filmsbasicallycontaintwodiffractionhaloswhichhavethesamed-valuesasthoseahown
in Fig. 4-3(a), indicating that the structures are mainly amorphous. However, some
clnngesinsuucunearecvidentasdreion-to-atomarrivalratioincreases Athirddiffracton
halo (d~0.280 nm) adjacentto the zeroorderspot appearsin the diffraction patterns
showanig.4-3(c)and(d). Theappearanceofadditionalfaintdifi'ractionhalocanbe
explained by the enhacement of short-range ordering [127]. In addition, nano-scale
mottled contrast in the microstructures, shown in Fig 4-3(b) through (c), was observed.
Similar structure changes, reported by Murri er al. [128],were found in RF sputter-
depositedGaAsfilmaandwemamibutedmtheuansifionoftheschhnefiommndom
networktoonehavingnano—scaleordering. Theyalsoreportedthattheaveragediameter,
D,ofthe'agglomerates'hasaweakdependenceonprowssingparameters. TheD-value
increasedfrom3to9nmwithadecreaseinambientpressurefrom9PatoO.4Pa,which
resultedinmoreenergeticparticlebombardment.

In the present work, at higher energy (500 eV) and high ion-to-atom arrival ratio
(1.08), inert gas incorporation become quite marked, as shown in Figure 4-4. Extensive
porosity,associatedwithugongasbubbhs,isapparent. Thediffractionpatternandbright

60
fieldimageofmeIBADfilszshownmfigme4-5Jamedwithasubsuatewmpmme

of 473 K and a 100 eV assist-beam (giving an I/A ratio of 0.33), show different features
fiom the others. The bright field image displays very fine particles, on the order of
5-10 mo in diameter, which have been nucleated in the film during deposition. The
conespoodingdiffiacdmpauancmninsavafiuyoffingswhichconfimmeprewnceof
a crystalline phase. Two dark field images, made from the brightest of the rings in
Fig. 4-5(b), unambiguously associate the particles with the ring system. Table 4-1
tabulatesd-spacingsofalltheringsappearedondreaiginalnegative. Allcanbeassociated
with the d-values for the TiN phase, having the NaCl structure with a = 0.4224 nm
[129].

Electron micrographs and their associated diffraction patterns shown in Fig. 4.3
indicatemuminaeaseofsubsuawwmperauueandUAmdocanmcreaseflwdegreeof
short-range ordering. The structural relaxation of metallic glasses on heating has been
widelysnrdiedandmeobsavedphenomeMamdividedmmchmgesmmpobgicdshat-
range order (TSRO) and compositional short-range order (CSRO) respectively [130, 131].
Becauseofthe largenegativeheatofmixing forTi-Ni alloys, theoccurrenceofCSROis
ratherlikelyandhasbeenreportedinbothmeltquenchedandmechanically alloyedT'r-Ni
[78, 131]. This means that the quenched-in as-sputteled structure has higher free energy
thanthe'relaxedone,duetothehigherconfigurational entropy. However, therelaxationof
quenched-in structure is a thermally activated process which can be performed by
increasingumpletemperannetoabout400to700Kl78, 131]. Theevidenceofshort
nngeadaingobsavedmtheIBADfilnnsuggesmthndreionbombudmentcmenhmce
the thermally activated processes, such as diffusion, so that adatoms and/or near-surface
atoms can rearrange themselves into a lower energy configuration. In fact, the
crystallization of amorphous alloys requires the same kind of atomic motion in which,
however, many more atoms are involved. Therefore, it is apparent that low energy ion
inadiationofthegrowingfilmcanenhancethermalrelaxation.

61
Oren[132]proposedthudwthamalrehxadonprocessleadsamorphousafloysm

experience an irreversible change ( i.e., TSRO) which increases the activation energy for
crystallintion by a factor [1+(d lnSg)/(d W )l, where Sc is the configuratimal entropy.
Experimental results on the crystallintion temperatures of vapor-deposited and melt-
quenched amorphous nickel [1331 supported the proposed theory. However, as pointed
out by Buschow [75], compositional short-range ordering may decrease the activation
energy, through a change of kinetic path. Schwarz etal. [1341 has also suggested that
lower crystallization temperatures observed in mechanically alloyed powder, with
conmositionclosetoT‘iNia,isduetotheexistenceofCSRO.

hthepreaentsmdy,thecmehfimofdrethamalstabifityofTT-Nifilmsandtheh
atomic structure has not been established yet. Therefore, an overall survey is needed to
understand the effect of chemical short range ordering on the crystallization behavior of
amorphousTi-Nifilms. IngerleraLitwasnotfoundtobepossibletocrystallizegrowing
filmswithanionassist-beam. However,aswillbeseenlawr,IBADexperimentsprovided
useful information about compositioml effects of ballistic impingement.

The TiN precipitates observed in the present study may have resulted from
contamination of gas supply lines to the ion sources, and/or fiom high nitrogen base
pressure TiN precipitates were found in directly N" implanted TiNi alloy with the dose
fluency in a range of 1013 ion/cm2 to 1017 ion/cm2 by popoola et al. [135]. Thick TiN
coatings were also made by Kant et al. [136] to deposit Ti with concurrent argon ion
bmbatdmentinanitrogenatmosphere. OptimumdepositionconditionswereaNztoTi
ratioof5andAr+toTiratioof0.3-0.4atAr*energyof500eV. Ion bombardmentwas
foundminuusethennfweniuogencomennadonandmcmvatphysicaflyabsabedNZ
moleculetoachemisorbednitrogenatomll37]. Inthepresentstudy,thebasepressureof
the experimental run which yielded the TiN containing film was 40 ttPa. The nitrogen
partialpressureisthereforeestimatedtobeabout 1 uPa. Theimpingementrateofnitrogen
is then estimated to be about 1012 particle/cmzlsec [138], which yields a Ng/I‘i ratio of

62

about 0.01. Since the assist-beam energy (100 eV) and the UA ratio of0.4 are very close
mtheopdmlmmndinonsdescfibedabovemeappearameofmmysnlhnefmparddes
is understandable. However, TiN precipitates were found only in one IBAD film which
wuthentnumplcmadcmmatexpaimeomlnminwhichthteemhsuatcswaeloadcd
Thesecondandthethitdsamplesarethoseshowninfig.4-3(b)and(c)respectively. In
casemarpplyfinaofthebnsmncesqumiMTlehaseshwlddsoappeamd
intheothertwosamples,sincetheywerebombardedbyicnswithsimilarenergies(100-
MeWanddosefluency (IlefO.4-0.6). Underthesecircumstances, theformationof
T'inhaseinthislBADfilmmay haveresultedfromhigh initial ninogenpartialpresstu'e,
orfiomaeffectivepru'geofthesupplylineaftertheinidaldepositiolh

4.1.2. Composition of Splitter-Deposited Thin Films

Compositions of Magnetron-Sputtered Ti(NiCu) Thin Films. Compositions
of single layer (SL) and periodic multilayered (PML) Ti(NiCu) thin films, fabricated by
triode magnetron sputtering, are tabulated in Table 4-2. Compared to the composition of
the target alloy, the titanium content in the SL film decreased about 2.6 at%, while the
nickeleontentinelused 1.5 at% audtheeoppereontentincleased 1.1 at%, accordingtothe
ZAFresults. Shlcetherestfltforfllemrgetanoyindicatesthatdlecoppercontemhasbeen
underestimated, therealcoppercontentinthe SLfilmsshouldbehigher. Theanalysisfor
thin foils by the Cliff-Lorimer method for the SLtfilm yielded similar value for titanium
contentandabout 1 at%increaseinNiandl at%decreaseinCulespectively,ascompared
tomemtetults ‘

The additioncftitanium multilayers broughtthecverall titaniumconcentraticn in the
PML-16 and mm films to 49.2 at% and 51.0 at % respectively. On the assumption that
therealdepositionrate,drnfldt,ofeach spuuaingsomceisconstanathecompositionsof
the PML films'can be estimated in the following way: The real mass (per unit area)
depositedofeach layer,rn,canbeexpressedintermsofthedesiredthicltness,1‘,the

63
deposition rate Rap determined by thickness monitor in equation 3.2.3, and the real

depositicnratetbn/dt asfollows
motley =(Tauoy/RddalloflHdmoncy Id!)
= Talley Mallow (4.2.1)

"in = U'n/Rdopfl'i))(dmt ldt)
= Tn Mi) (4.2.2)

where p. is the bulk density employed to determine the deposition rate in equation 3.2.3.
The weightpercentage ofTi, Ni and Cu in alloy-layeris obtained fromthecomposition of
theSLfilm,determinedbytheZAFmethod,as42.l %Ti,50.7%Niand7.2%Cu
respectively. The Ti content in the PML films are thus calculated and plotted versus Ti-
layer thickness ratios in Figure 4-6. The measured Ti contents are about 0.6 at% lower
than the expected one.

Before further discussion, it should be pointed out that using bulk density in
measurement ofdeposition rate may produce films with thickness greater than the desired
valuez, but has no effect on the deviation of composition described above. Hence, the
0.6 at% deviation may have resulted from two factors. First, the titanium concentration
deuminedbytheZAFmethodwasestimatedtohavea :1: 0.2at% errorwhichwas labeled
in Fig. 4-6. Second, the real deposition rate, drnf/ dt, might have suffered small
fluctuations during operation. The :l: 5 % deviation of resultant film thickness, found in
the present magnetron sputtering system as mentioned in section 3.3.1, could be attributed
primarilytothefluctuationoftherealdepositionrate. The'l'iconcentrationsinthePML
films were thus re-calculated assuming fluctuation occured in the alloy source only and

 

ZWWfihnsleMymdmxmmmWymeanoy-hyakm
Vapor phase deposited titanium films are usually polycrystalline. However, the X-ray diffraction study on
multilayered Ti-Ni thin films, conducted by Clemens [109], showed that both Ti-layer and Ni-layer le
mphcusasdrethicknessofeachlayerwaslessthanfivemonolayers. lnthepuentmtlreT‘i-layu’
consistsdabcudneetofuumonolammdthmcmbeueatedaslnaphmn.

64
plottedinFig.4-6. Thedeviationresultingfmmthefluctuationcftherealdepositicnrate

increaseswiththeincleaseofT‘i-layerthicknessratio. However,inthepresentcase,an
average 1: 0.2at%enoronthetitaniumcontentiscalculatedwitha :l: 5 ‘1) fluctuationof
deposition flux.

Severe titanium depletion, relative to the sputter-target compodtion, was found in
the SL films. Since the sputtering target was preconditioned by sputtering for 30 minutes
prior to deposition, and was cooled by water, the surface composition of the sputtering
target was expected to reach a steady-state. As mentioned in section 2.5.1, the sputtering
fluxandthesputtu-targetalloyhavethesamecompositiononcethesteady—stateisattained.
Inotherwords,thefilmcomposition shouldbethesame asthatofthe sputter-targetalloy,
unless the growing films were bombarded by energetic species, resulting in preferential
resputtering. and/or the alloy components have different sticking coefficients [90].

Preferential resputtering of the growing film is due to components of the film
having different sputtering yields, and usually results from direct ion irradiation on biased
substrate and/or bombardment by energetic neutrals reflected from sputter-target. At low
ambient pressure, these neutrals suffer few collisions during transit, and arrive at the
substrate with high energy. In the present case, the films were deposited at a relatively low
pressure (0.33 Pa, corresponding toa mean free path value of 2.5 cm) at which preferential
resputtering is expected. though-substrates was kept neutral. A Ni depletion3, produced by
preferential resputtering, is expected according to published sputtering-yield data of
elemental Ni and Ti which results in a yield ratio of Yn/ YM- ~ 0.5 [84]. However, the Ti
lossfmndmmeSLfilmindicawsmatmespumuingpropufiesofNiandTimamcrphous
Ti—Ni alloys and in pure elements may be different. The IBAD experiments described
belowhaveproducedadditionalinformationinthisregard.

Splitter Yield Ratio Analyses. Ion beam assisted deposition (IBAD)
experiments were designed accordingly to gain better understanding of the sputtering

 

3Fadmplkhy,haethedbcusdonwuhmiwdhaTi-Nitwocomponemsymimdamme
conmlicaetemuyone.

behavior of Ti and Ni in amorphous Ti-Ni alloys. The analysis of Harper and
Gambino [96], as described in section 2.5.2, is employed. The film composition is thus
written as

yr.- -la+(a2+4xnplmvzp , (2.5.4)

where a a (x, ”7" - x1; -. rx, .- l), p s (x, + Y - x,r - 1 ). Therefore, the
sputteringyieldratio, Y,atcertainion energy,canbeobtainedby measuringanumberof
thickness fractions of film resputtered (X, ), and film compositions (an ) both of which
vary with the ion-to-atom arrival (l/A) ratio. ”A dual ion beam system for the IBAD
experiment has been described in section 3.2.3 and shown in Figure 3-4, which exploited
the inherent assist-beam current profile, together with the natural angular falloff of the
depodficnflumbgeneruteasystennficvmiafioninmeUAmfiodmingdeposidon Figtu'e
3-6(b) give an example of the variation in the UA ratio with respect to the substrate
position.

Thetotal fiactionofthefilmrcsputtcrcdbythcassist-beam,X,,can bedetermined
by comparing the film thicknesses of ion sputtered films with the thickness ofthe IBAD
film at the corresponding substrate location, according to:

x =53le (4.1.2)

inwhichtmsandtmparethefilmthicknessesfortheionbeamsputteredfilmwidloutan
assist-beam, and the ion beam assisted deposit thicknesses, respectively. Results of this

analysis, in which Xr is plotted as a function of the ion-to-atom (I/A) arrival ratio, are
showninFigure4—7. FortheSOOeVassist-beamenergy,theexpectedtrendofincreasing
X, faimmasingUAmfioisproducedandmecmveshowstheexpecwdtendencywwmd
X,--0asthellAratiogoestozero. TheresputteredfractionforlOOeVIBADisseento
initiallydecleasefromahighinitialvalue,reachingaminimumatl/A=0.4,andthento
begin to rise slightly at VA 2 0.55. As the assist-beam energy is lowered to 50 eV, the

 

66
resputtered fraction increases slightly with the increase of VA value, levels off and then

dropstoabout0.02atllA-=0.l3. (AseccndaetofX-axesarealsoattachedtotheseplots
to demonstrate the relations between the assist-beam incident angle, 0, and the sputtered
fraction Kr.)

ThetitaniumdepktionisplottedinFigme4-8asafrmctionofflAratio. Theresults
showageneraluendoffiMumdeplefiminmelBADprocessedfilmswhichhraeases
withincreasingofl/Aratio. Asisexpected,500eVassist—beamenergyyieldsthelargest
amountchi-depletion. Thedataforthe50eVandthe100eVlBADfilmscan,however,
befittedwithasinglecurve,revealingthationswithenergyof50eVand100eV
respectivelycanproduce similarsmormtoftitaniumdepletion inTiNifilms.

In Figtue 4-9, the change in titanium content from the control value is plotted
againstX, for50,100and500eVionbeamassisteddepositions. Amonotonicdecrease
in titanium content of the films is evident for the higher assist-beam energy (500 eV),
whereasthe ImeVdatadisuibuteinaoppositetrend acrosstheextrapolationofthe500
eV data. The titanium depletion for the 50 eV assist-beam energy first decrease
monotonically withanincreaseofxr,andthendeviatetowardthesametretldastlre
100 eV data shown.

Knowledgeofthevariation offilmcomposition withthetotalfractionresputtered
by an ”etching” beam may be used to determine the relative sputter yields of the
components of the film. Equation 2.5.4. may be recallled as
yr.- =- [a + (a2 + 4xnfl VIII/25 where a = (X, + xnl’ - x1; - YXr - I), and
p = (x, + Y - x,l' - 1). In the present context, 11'.‘ represents the Ti content of the
films prepared without concurrent ion irradiation, rather than the sputtering target
composition. The latter was found to be 0.485 for 500 and 100 eV IBAD films, and0.535
for 50 eV IBAD films respectively. Using values of X, and x1.- from a linear least squares
fit to the 500 eV data results in Y= 1’1le”,- = 1.75. The data for the 100 eV assist-beam
energydonotlendthemselvestoacurvefittoestimatethevalueon,thoughthedata

67
clusteraroundavalueconsiswntwiththehiglrenergyresult. Ontheotherhand,afraetion

ormewevmwhiohcaretpoodsmlowavAuuo(orhighaasnswd-beumincidmt
angle),canbetteatedinthesamemannertoyieldanothervalueon=YnlYm -9. With
theinereasechAndo(a'medeereaseof0),thedautmdmdefleamwmdhighamdd
AdramaticincreaseofthesputteringyieldratioYsl’z-all’m isfoundatlowassist-
beamenergy. Theassisted-beamenergiesofSOeVusedinthepresentstudyarecloseto
tllesputter'ingthresholdener'giesofabout5t050erormostofmaterials[l39]. lnthe
bwenagyregimemespmwdngnteisexnemelysensifivemmevafiadmofionenagia
(5,)rndenergyotsputteringthreshold(5,,.)[s4]. Bohdanskyetal.[86]derivedascaling
lawrorlowenergyspuueringinwhichthesputteringyicld.r isgiv'enby
Y(E')=Q(M1,M2,Eg)F(E') (4.2.3)

whereM) ansz arethemassesofilradiatedatomandtargetatomrespectivelyfig isthe
bonding energyandE'isanormalizedenergydefined by E'IEdEfi,WhflC Eu. isthe
threshold energy of sputtering. In this equation Q is an experimentally determined yield
factorwhichcanbewrittenas

Q a 0M2I4M1M2 KM! +412915”. (4.2.4)

where a is a material constant, and F is an universal yield energy curve given
approximately by
F (E' ) = 85x10'3E' ”4(1-1/3' )7”, (4.2.5)

Asaresldhahigherlid. resultsinalowersputteringyieldataconstantionenergy,andthe
sputtering yield ratio deduced at higher energy hence may no longer hold. Tarng and
Wehner[105]foundthatthespuneringyieldrafioofcopperandniekelinaOlssNiasalloy
showedaplmwncedinmeasewithadeaeaseofbombardingimenagyinduhwenagy
regime (<150eV). They notedthatthesputtering thresholdofcopperisl‘l eV whichis
smaller than that of nickel (21 eV). Similarly, the sputtering yield ratios derived from

68
Fig. 4-9 for 50 eV and 500 eV assisted-beam energies imply that nickel has a higher

tinesholdenergythantitanium. Theinereaseofsputteringyieldratiocanalsoexplainhigh
dtaniumdepledonmtefortheSOeVassist-beamerugyshowninFigu.

However, the observed behavior of the resputtering rate found at low energy
regimecannotbeentirelycxplainedbythechangeofspuneringyieldmtio. Onepossibility
isthahulowenergy,dleangulardependenceofmerespuuedngmtebecomesthe
dominantefi‘ect. Forthepresentexperimmtalaetup,anincleasingI/Aratiocorrespondsm
adecreasingineidenceangleoftheassist-beam. lntherangeofOtoZSdegreesapplicable
www.mmgle-depmdenceofmespuuuingmfmfispunuedwhhlkeVAr‘is
quite pronounced [85], increasing by a factorof about 25% over that range. For light ions
such as H+,D+ andHe+( 1 to 8 keV), this angular dependence ofsputteringrate of
nickel, reported by Bay and Bohdansky [140], is even greater, increasing about 50 % in
therangeofOtoZSdegrees. Consideringanelasticcollisionbetweenanenergeticparticle
andanammwifllwoimdalewgymmiwdbymewnsavadonofenagyandmcmamm,
theenergytransferredbyalkeVH+isinsimilaramountasthatofa50eVargonion.
Theobaavafimofmmgulndependemeofspuuaingyieldfmhghtionsnnndoned
aboveshouldbeabletoapplyonlowenergy,heavyionbcmbardment. Thus,itmaybe
specuhtedmmmemcidencemgleefiectdomimtesmemspuwedngmuinwrmediate
mgleawhaeastheUAmdodependmeebeginsmexertemuolforspecimnbubuw-
namflmist-bumimidemeangles(cmespondingwhighUAranos),andhighmgla(m
which IIA ratio approachs zero). Furthermore, Yamamura and Bohdansky [141] proposed
mumemmsholdenergyEmdecreaseswimmeincreaseofmeionMdencemglewim
respecttoncrmalincidence. Asdiscussedabovethesputteringrateisextremelysensitive
to the energy of incident ions B,- when E; is close to E“. Under such circumstances, a
strongerangulareffectisexpectedwiththedecreaseofionenergies. The50eVdata
showninFig.4-2indicatesthattheangularefi‘ectissostlongthattheresputteringrate
decreases only when the UA ratio approachs zero, coinciding with the argument The

69
incidenceangleeffectmayalsoexplainwhytheImeVdatadonotappeartotendina

simplerashiootowardxnottl/Aao.

Asisthecaseforthedependenceofresputteringratio,X,,onl/Aratioandonion
incidencemgletheccmpositiondependencecnxp for50eVand100eVdatashowsa
demeaaemTiruptlna'ingmtewiminmeasingtmdfiacdonrespunemdMgA-m. 'lhe
datasuggestthattherelatlve sputteryieldsofNiandTiarealsodependentonassist—beam
incidenceangle. AsaccnsequencetheratiocfSpuw'inzyicldsYnlYm,hasamaximum
valueassputteredbyanassist-beamwithanormalincidentangle,andthendecrease
graduallywiththeincreaseofincidentangle. Inotherwords,thesamea’mountoftitanium
depletion can be produced with lower total resputtering rate. Again, the results can be
mfionafimdbymnsidaingmemmdependenceofmespumingmreshddenagiesand
the difference of Eu. (Ti) and E“. (Ni). According to the model proposed by
Bohdanakyetal.[86],anorderofmagnitudechangeofthesputteringratecanresultfrom
achangeofener'gyparameterE'sEglEu. from2t08. For50eVassist-beamenergy,the
energyparameterfornickelatnormalincidence angle maybeclosetounity but thatfor
titanium may be two or three times higher, which results in an high value of Yfl/Yui. The
vflmofspunaingyieldmfiomwdecmasesadmmmaeaseofionmcidencemghaince
YnhasaweakerangulardependencethanYm. Theangulardependenceofsputteringyield
ratio isweaker for 100 eV assist-beam energy due to a larger value of energy parameters
for both titanium and nickel. The tendency of the decrease in T‘i resputtering rate with
increasing total fraction resputtered then becomes more obvious at larger ion incidence
angles as shown in Fig. 4-9.

According to the IBAD results, the titanium depletion in the magnetron sputter-
depodmdfilmsismfiomhndbydwpmfaenfialmspumefingoffibymeenagencneumls
reflectedfromthesputter-target. Vossen [142]reportedthatthebombardingionshavea
high possibility ofbeing neutralized priortoimpactingthe target srn'face. Theymaythen
be reflected as energetic neutrals without being influenced by the electric field over the

70
target surface. In the present case, low ambient pressure of0.33 Pa, corresponding to a

memfieepahvalueof2.5cm,dmingdeposidonauowsafiacfionofreflecwdneunahm
presavenatchoftheirinidalkinedcenergydminguansittothesubsuatea Therefore,the
bombardmentoftheenergeficneuualsresultsinpreferenfialrespunering. However,the
reflectedneufiahhaveawideenergyspecfiumwhichmakehdiffimntmesfimatethe
incidentflrutcftheaebombardingparticles.
ThemdoforsputtuingyieklofpmedtaniumandpmenickeLY=YfiYmisab0n
0.5inmoderateionenagyregimes(5(llto1WeV)acccrgingtothepublisheddata[84],
whichisapproximatelytheinverseoftheratioobtainedin'l'i-Nialloyfilmsinthepresent
study. A similar result was deduced for Gd-Co system [96]. Moreover, the elemental
sputteringyieldofcopperisslightlyhigherthanthatofnickel. Theenrichmentofcopper
inmeSLfilmsnemsmcoincidewimtheempincalmhdmdeducedfiomTi-Nisystemin
thepreaentstudy.
Asacmclusion,MmladonofthespuMngyieldsofcopper,nickelanddMum
internm'yamphousalloyscanbeexplessedas

Y1: > Y"; > Ya.

in low and intermdiate energy regimes. The sputtering yield ratio of Ti and Ni in
amorphous'l‘iNialloysincreasefrom1.75at5meVofAr+toabout9at50eVofAr+.
Thedmmtmdeplefimofmedeposiwdfilmsobsavedinmepmsentsmdymayflsolesuh
from different sticking coefficients of Ni and Ti. However, the absence of experimental
data limits further discussion. ,

4.1 .3 . Crystallization Behavior of Ti(NiCu) Thin Films

The crystallization behavior of the magnetron sputter-deposited Ti(NiCu) thin films
was studied by differential scanning calorimetry (DSC). Figure 4-10 shows the DSC
traces of various magnetron sputter-deposited films subjected to heating at a rate of 10
Klminto7‘l3K. rhecxothermalpealtappearingatsroundtsoltincachcurvercpreucnts

71
theamcrphous-to-erystallinereaction,withanenthalpyinarangeofl.3t02.3kJ/mole.
PricrmaynaHindcmmecrtwobmadexodramnocawdbetweenwOandmOKwae
alsoobaerved. ThisphenomenonhasbeenrepcrtedinT‘i-Niamorphousalloysprcduced
bymechanicalallcyinlnss l43],andwasattlibutedmathermalrelaxaticnprocesswhich
resultedinashort-rangeorderedstmcture. Anotherbrcadexothermalpeakoverlappingthe
crystallizationpeakintheDSCtraceofther. ISfilmsprobablyresultedfrcmthe
oxidationofthecoppusubsWusedfortheaespecimens. Asmallsidepeakisslso
observedintheu'acescfthePML-9'filmsatabout755K. Thenatureofthissmallpeakis

Table4—3hsmdleaystalfimfiontempaannes(Tc)anddecarespcndingmmalpies
forDSCscansconductedatvariousheatingrates,s. Theinereaseinheatingrateraisesthe
erysnnizadonwmperannebutdepressestheassociatedenthalpy. Thefilmsshowedscme
changeincolor,indicatingareactionwiththegasremaininginthealuminumpanorwith
thepurginggas. TheZK/minf'llmssufferedseriousoxidationwhereastheSOKlmin
annealed films remained mirror bright over most of their Stu-face area. Therefore, the
decreaseofenthalpycanbeatnibutedtothedecreaseofeffectivemasswhichwasstill
amuphous(ie.,nmreacmd)pricrmmachingmeonsetofdecrysmlfiufimreacdon The
motors/T,2 vemusthemva'secrysmflizadcntempmmTc'1[122],ushowninfigme
+11,yieflsmewfivafionenagiesofcrymlhudmfmmeSLfilmsanddefilnnas
485 kJ/mole and 390 +l- 6 kJ/mole respectively.

Figure 4-12 shows the X-ray diffraction patterns for the isochronally annealed
films. Nopeaksotherthan(110)nzcouldbeidentifiedattwothetaanglebetween37and
49degreeemvealingthumemajuproductmmeammphouem-aysumnereacdonwu
orderedBZNiTiinallthefilnnstudied.

The crystallization temperature Tc, of binary ion sputter-deposited films [144],
tanuymagneumswnadepodwdfilmsfiomdepresentsmdymndjmmfaencehfa

, . 72
binaryalloyribbonsobtainedbyBuschow‘US]areeachplottedasafunctionoftitanium

concentration and are shown in Figure 4-13. 1', for binary ribbons decreases
mmmnicanywimimmasemMumcmtentinthecomposifionmngeoHSafiTisnd
65at%Ti. Thedatarrombinaryionspuuer-depositedtilmstmlandternmymagneuon
sputter-deposited films from this study follow this trend also, except that the overall
aysmnindontemperannesofmemmyfilmsinmiscomposidonmngeueabomwK
lower. Figme4-14showsacompariscnofacfivafionenergiesbetweenternmythinfilms
ofthetlesentstudyandbinaryribbcnsobtainedbyBuschcw [75]. Theactivationenergies
famysmlfimdmalwincmasewimdecmasmgdmmumcontenhandmoscofdleternafies
are about 150 kJ/mole lower than those of the binaries. The reduction ofcrystallization
tanperanneandacfivadmuwrgyofmaryminfilmsmaybeamibuwdmmeeffeaofdw
third element addition which depresses the melting temperature. Schwarz et al. [78]
pointed out that for compositions far from those near deep eutectics, the ratio of
crystallization temperature to melting point, Tc [1}... is approximately equal to 0.5.
ConsideringthefactthatthemeltingpointofTiCuisonly1255K(328Klowerthanthat
of TiNi), and only a two-phase region lies between TiNi and TiCu, the melting point of
Ti(NiCu) isexpectedtodecrease with theincreaseofcopperaddition. Thusthechalues
decrease, too. As a result, the addition of Cu substituting for Ni in TiNi alloys can
substantially decrease their crystallization tenmerature.

X-ray diffraction results (Fig. 4-12) show 32 Ti(NiCu) is the only resultant phase
aftercrystallizationinalloftheternaryfilmsstudied. Thisfaetindicatesthecrystalliution
process is a polymorphic reaction in which a single phase with the same composition is
formed [145]. In other words, the polymorphic reaction needs only single jumps of atoms

 

4ThecryrtallhetiouWac)tuntlaedmreraereens]reteedmmraeotszmm-l.
However.thercvaiuerretertoaheatingratcotloxruia-lornbeyieldodbyurmgtherpprosimte
Wmehtodas

Tc( 10) - [bl-50) ' T450)! I W10)
which It. been d'ncussed in reference [160].

73
across the interface into their lattice sites [145]. A similar result was found by

Battezati etal. [77] for the crystallization process of amorphous TisoNigo powders after
mechanicalalloying. Thepolymcrphicreacticnoccmedonlyinconcenu'ationrangesnear
the single phase region [145]. However, this crystalline phase is not necessarily the
thermodynamicallystablephaseatthecrystallizationtemperahne. Thisfaetindicatesthat
regardlessofwhetheraeutectoidreaction (TiNi -) TizNi +T‘lNi3) existsornot, in an
amorphouruloywithoompositioudeviatiogshghuyhomstoichiomctry, asuper-saturated
singlephasesolidsolutionofBZT’rNicanstillform. Inbulkmaterial,thesteepsolidus
lineinTi—richsideoftheBZTlNiregionmakesdieprecipitationofdleTizNiphasedming
solidificationandbcmogenizationvirtuallyunavoidable.

4.2. Microatrueture Evaluation

Thksecticnpreaentstheresultsofthemierostructurestudy formagnetron sputter-
depodted films with overall composition or TursNitssc-tat (8L film). TusaNhtsCusn
(PML-16 film) and Ti51,oNi44,4Ctu,5 (PML-9 film) respectively. A discussion follows the
experimental results for each independent topic presented.

42.1.SLFilms

TheweraflconpcsidcnofmeSLfilmsbcatedinatwo-phaseregionccnsisfingof
aBthaaeanda(NiCu)gTiphase,accordingtotheTiNiCuternaryphasediagr-amat
1073K[44]. However,aTi-rich TizNitypephasewas foundtoprecipitateinthegrain
boundaries. Thissecfionrepcrtsdieresultsofadetailedsmdyonthemicmsnucnueoftbe
annealedSLfilms.

SL Films: General Featw'es. The general features of grain size, grain
morphologyandnucmsnrwuueoftheamlealedSLfilmsueshowninFigme4~15and
416, respectively. Figure 4-15 is a cross-section secondary electron micrograph for a SL
filmannealedat923Kforonehour. Thegminsareequiaxedtluoughoutwithavengesize
of 1 pm. Figure 4-16 is an in-plane transmission electron micrograph showing similar
grainsizeandmcrphology. Numeroussmallprecipitates,about10nmtol(X)nminsize,
aredistributedinthegrainboundariesandwithinthegrains. Electrondiffractionpattems
nkenumcmwmpaanue,showninFlgme4-l7,mvealthndnmauixphaseiscubicwim
a CsCl-type stacking order, referred to as the 32 structure. The lattice parameter is
0.301nm. Diflusesfieakswaeobservedin<110>and<211>reciprocaldirectiona The
appeuanceofsuchdiffusesueakshasbeenmpmwdasaprwmphenomenmofmek-
phasetransitionl48].

SLFilms: Precipitares in the Grain Interiors. The precipitates in the grain
interiorsproduceddistortedMoirefringes,asshowninFigme4-18,.whichareallroughly
perpendiculartoIlOO] directionintheBZ lattice, indicatingthataconsistent orientation

74

75
relationexismdbetweentheparentlati'weandtheprecipitates. Thedistortionofthefringes
findramggestsdrepruarceoflocalsnessfieldsusociatedwimprecipimtea Sincethe
diffiacmdspmsgenaamdbythepredpiuwsuevayclosemdnmsnixspommeaysnl
surnnneofmepredpimwusnldiedbyusingcmvagembeamdiffiacfimmchniqlwm
avoid ambiguity. Figure 4-l9isaseries of convergentbeamelectrondiffractionpatterns
flomtheprecipitatephasewhichhavebeenindexedtoatetragonalcellwitha80.3lnm
andc=0.80nm. Thesizeoftheconvergedelecn'onspotisaboutSOnmindiameter,
whichissdnmobigmgaerawdiflncdonmfmmadmccmingfiomthepredpimteabne.
Thehigha-aderlauemnehneswhichshwldappeuimidedemmfitwdanddiflraemd
spotshavebeenobscmedbyinmrferencefromthematlixlattice. Therefore,noatomic
symmenyinformationisavailable.

X-ray fluorescence microanalysis indicates that the precipitate phase is enriched in
copperccmparedtothemanix,asisapparentfromdatashowninFigme4-20. Usingthe
overall (matrix + precipitates) spectrum, shown in Fig. 4-20(a), as a standard yields a
compositionestimatefortheprecipitatephaseofT'ioggNngCuom, whichcorrespondsto
the (NiCuhTi phase. The lattice parameters, reported by van Loo et al. [44] for this
ternaryphase,ofa -O.308-0.314nmandc=0.792-0.800nmalsoagreewellwith
the present results. Thus, the second phase is here identified as (N iCu)zTi. The
conipositionofB2matrixisalsocalculatedfrcmthespectruminFigrn'e4-20(b)as
Ti47,5Ni43,5Cu3,g, using the same standard, which indicates no significant titanium
emichment in the matrix due to the precipitation of the (nickel+copper)-rich phase.

Figure4-21showsbrightfieldimagetakenappoximatelypuallelm<lm>indre
BZlattice,demonmfingdietypicalmorphologyofdleprecipitatcs The(NiCu)zTiphase
appearsasathinrectangularplatewithahabitplaneparalleltooneofthe[1001mplanea
Theiravaagesizeislmnminmelcngimdinaldhecfionandwnmindncknesa Amisfit
dislocaticnnetworkcanbediscernedinthe[100]minterfaces. Thisdislocationnetwork
indicatesthattheinterfaceissemicoherent. figme4-22(a)showsthebrightfieldimagein

76
whichthreevariantsoftheprecipitateintheBZmatrixareobserved. Thecorresponding

diffracticnpatwrnsshowninFigtue4-22(b)areinan[100]maoneaxis, whichisparallel
to two [111)] (NiCuhTi zones. The (010)32 and ((1)1)32 are found to parallel to each ofthe
(001) precipitate planes respectively. The dark field images of two precipitate variants,
taken from spots 4: and d in Figure 4-22(b), are shown in Figure 4-22(c) and 4-22(d)
respectively, illustrating the habit plane of the precipitate phase is (mlkflmh‘rll 1001;;
plane. Consequently, the precipitates oriented themselves in such a way that their three
principle lattice axes are parallel to thou of the matrix.

In summary, the crystal structure and lattice parameters of the (NiCu)2Ti phase
observed in this (Ni+Cu)-richfilmcoincidewiththeresultsreportedby vanLooetal. [44].
Furthermore, the present study provides the first reported data on the morphology, the
habit plane and the interfacial structure of the (NiCu)2Ti particles, and their orientation
relauon with 32 parentphasein theirearly stage cfprecipitaticn.

SL Films: Grain Boundary Precrbitates. TEM study showed that the grain
boundary precipitates have two different morphologies: an equiaxed one and a thin-plate
one, as shown in Figure 4-23. The rectangular plates are accordingly (NiCuhTi
precipitates. Analysis of the associated diffraction pattern (see Fig. 4-23) reveals that it
containsasetofdiffractionspotsgeneratedbytheequiaxedparticles. Thesespotscanbe
indexedina [112] soneaxisofaTizNi phase, havingafcc unitcellwithlatticeparameter
of a :- 1.132 nm. The average size of the TizNi particles is approximately 20 nm in
diameer,andtheywereonlyfoundinthegrainbomldaties.

Later investigation indicated these grain boundary precipitates are most likely
T‘u(NiCu)20, phase. TigNi and Tig(NiCu)20, have the same crystal structure, which is
the Fe3W3C type, with lattice parameters of 1.132 nm and 1.133 nm respectively [41],
whicharedifficulttodistinguish byelectrondiffraction. Thepartial substitution ofCu for
Nimayftn'theralterthelatticeparameters. Similargrainboundaryprecipitatesinannealed
TiNi thin films have been analyzed by Busch et al. [116] using windowless EDS dector.

77
However, no unambiguous results were yielded, since oxygen was found in both the

precipitaesandtheadjacentueu(withnoprecipitates). Asaresult, the terminologyTizNi
will be use in the following text to represent either Tig(NiCu) or Tr.(NiCu)zo,, unless
otherwise noted

SL Films: Therrml Pluse Stability. X-ray diffraction was employed to study
thcthermalstabilityortheazphase. Figure4—24arediffractionpatternsfromfilms

annealedat923K,for0.25 hour,onehour,and4hoursrespectively. Thebroadpeakat
two-theta of 41.2 degrees consisted of two partly overlapped peaks of (1 10)(N;c..)1-n at

41.15° and (333M500 plus (115mm) at 41.4°. The 013ch»; peak at 44.9° also
partlyoverlappedwiththe(440)-n1m)peakat45.28°. Thepatterrlshasbeen normalized
suchthatthe (110)32peaksinallthepatternshavethe sameintensity. Theintensityofthe
precipitate peaks, hence, represent the relative volume fraction of each precipitate in the
matrix. It is obvious that the volume fraction of both (NiCu)2Ti and Ti2(NiCu) phases
increases with annealing time.

In summary, the annealed SL films with composition of Ti47,4Ni45,5Cu5,1
containedtwokindsofprecipitatesintheBZ matrix: T'heequiaxed'l‘izNiparticlesappeared
in the grain boundaries with diameter of about 20 nm, and the (NiCuhTi phase was
distributed in the grain boundaries and throughout the grain interiors. The (NiCuhTi
precipitates have the morphology of a thin retangular plate whose plate-plane is the
(001)(mc..)21;/[100]32 habit plane. In addition, the principle lattice axes of these
precipimws'welefoundtoparalleltothoseoftheB2matlix.

ThenextsectionpresentsthemicrosnucmreoftheannealedPMbl6filmwith
overall commsition of Ti493Ni44_gCug,o. The volume fraction of the second phase
precipitatesismuch lowerascomparedtothatoftheSLfilms. However,the82grain
boundarieswerealsodecoratedbyTigNitypeprecipitates.

 

78
4.2.2. PML-I6 Films

PML-16 Films: Geneml Featta’es. Figure 4-25 shows the microstructure of
thePML-l6filmsannealedat923 KforonehouratP=2.6x104 Pa. Thegrain
morphologyissimilartothatoftheSLfilms. TheaveragegrainsizeisaboutZumJnd
thegrainboundariesaredecoratedwithsmallequiaxedprecipitates. Onlyfewscattered
precipiuwswaeobsavedinmegrfinintaimandmofthemwemassodawdwim
dislocations. The microstructure of the films, annealed at 823 K for one hour at
P=2.6x10'4Pa, havesimilarfeatures, however, with fewer precipitatesin both the grain

PML-I6 Film: Grain Boundary Precipitates. Figure 4-26 shows a bright
field image, and its corresponding diffraction pattern, of the grain boundary precipitates,
takenfromafilmannealedat823Kforonehour. Thegrainboundaryprecipitatesare
about30nmindiameter. Thediffractionpatternisatypical<110>zoneofface-centered
cubicmucmmandthelatticeparameteriscalculatedtobe1.13am,whichisingocd
agreementwiththepubilshedvahieforthefccTizNiphase[41]. Althoughaboutonethird
oftheprecipinwsshownmmeimageshmemeamesetofdiffracdonspotamspedfic

PML-16 Fibltt.FrecipitatesintheGrain Interiors. In the PML-16 films
annealed at 923 K for one hour, a small number1 ofTizNi particles with a diameter of
abomlmnmwaefoundinmegr'ainmtaioraofienasscciatedwimdislocadons Figure
4.27habdghtacldnuaographmdhscooeepondingdimacumpauernottwo(lll)twm
related TizNi particles. Nospecifiecrientationrelationbetweentheprecipitatesandthe
matrixwasfound.

Bhdehkeplecipintesdisuibuwdinmegrainmmriasofdlefilmsannealedawfi
Kand823Kforonehom,showeddistortedMoir6fiingesinthewfieside. Thefringes

 

lNoqmndmdveenimteofdievdmnefiacfionofdreTizNiplusewumadeumemanalL However,
theTizNiprecipitateswerefoundmlyinpartoftheaustaitegrainsinalimitedamountk 10
paucwgrain).

. 79
sreperpendicularto<110>mdirectionswithaspacingof7.0nm. Figure4-28isabright

fieldinngeanddwcamspondingdifimcnmpanansofdnprecipimmkenfianafilm
annealedat923Kforonehour. Thedifi'racdonpamrnsccnsistsofa[110]mzonepanern
andalOlllmpamfiomtheWcmwhichismduedufingmehtdcepamnem
of NiaTiz phase as: a a 0.441 nm, b 30.882 and c =- 1.35 nm [38]. As shown in the
distractioopauerns, the (01?) and (200) precipitate planes are parallel to (001) and (T10)
32 planesrespectively within :l:1 degree. The spacingofMoire fringes. calculated from
parallel (110).: and (200mm, would be 6.8 nm which is close to the measured value of
7.0 nm. Therefore, me orientation relationship between 32 matrix and the N13Ti2 type
phase is expressed as: [Twins/110111”. and (110m // (200)”. The precipitate blade is
about 150nmwideand5mto 1000nminthelongitudinaldirection whichisparallelto
<110>” direction. Dislocations are observed on the interface roughly parallel to the
(11%: and then extended into the matrix. These interface dislocations can be observed
more clearly in Figure 4-29(a) which is a bright field image showing two parallel N13T12
precipitates. Two sets of dislocations lying in the interfaces intersect with each other to
form a network, showing the interface is semi-coherent. Figure 4-29(b) is the associated
diffraction patterns which contained an <110> fcc none generated by a particle located
between these two NigTig precipitates, indicating it is a T'izNi phase.
Theblade—likeparticlesobservedin thePML-Mfilmsannealedatboth 823 K and
923 K is first treated as a ternary phase, TiaoNisg,5Cu3,5, based on the ternary phase
diagram at 1073 K shown in Fig. 2-5(a). The ThoNisg,5Cu3,5 phase is an
thermodynamically stable phase, which was first reported by van Loo et al. [44], at 1073 K
but not at1143 K. However, TisoNiggJCugg was reported to have a tetragonal unit cell
with a a 0.44 nm and c a 1.32 nm [44], whereas the blade-like particles in the present
studyhaveancrthcrhombieunitcell. ATizNig phasewithsimilarstructurewasalso
foundinbinaryalloyby Wasilewskietal. [36] toprecipitateatatemperatmebelow 1000
K. This phase was proposed to be a metastable phase by Nishida et al. [38], having a

80
tetragonallatticewiththesamelatticepmmeterasthatoftheTnoNisuCugjphaseat

temperaturesofabove400K. Thetetragonalphasewasrepcrtedtoundergoatwo—step
thermoelastic transformation at temperatures between 323 to 400 K in a sequence of
enagonalt-torthcrhombic Hmonoclinic[l46]. Theorthcrhombicphasehasalattice
parameterofa=0.44nm,b=0.88nmandc s1.32nm,andthemonoclinicphase
inherits its lattice parameter with 7 . 89.3°. Therefore, it is possible that the
TiaoNi55,5Cu3,5 phase observed in the present study has also undergone a similar
transformationfrcmtetragonal toorthorhombiccrmonoclinic synunetryabove ambient
temperature. 0ndreotherhand,thesimilarityofthemomtemperanirex-raydiffiaction
patterns ofthe TizNig phase andmeTuoNi“5Cu35 phase, pointed out by van Looetal.
[44], may indicate that the crystal unit cell of the latter is orthorhombic rather than
tetragonal. However,aftn'therstudyisnecessarytoclarifytheambiguity.

The TizNi3 particles have been reported elsowhere [36, 38] to have a plate-like
morphology in the early stage of precipitation. An orientation relationship between
monoclinic TizNig and 32 TiNi was also deduced to be [llllnz ll [501mm, and
(110)32 l/ (010)752Nt3 by Nishida et al. [38]. This orientation relationship is different
from that between TuoNisg,5Cu3,5 and 32 matrix, found in the) present study to be
['l'loluz II [011mm, and (lime: l/ (200mm,, though the morphologies of the TizNig
phase and the TiaoNisssfilss Phase are similar.

PML-16 Films: Thermal Phase Stability. Normalized X-ray diffraction
patternsshowninFigure4-301evealthatthefilmannealedat923Kfor5minutesat
P =- 2.6x10" Pa has no indication of second phase precipitation. After one hour
annealing,peaksflomboththeTizNiphaseandtheNigTizphssecanbeidentified. Peaks
for N13T12 precipitates are barely recognized, illustrating a very small volume fraction
existed,whichccincideswitlitheT‘EMobservaticn. Thediffracticnpatternofthefilm
annealed for six hours shows apparent increase of the volume fraction of NigTig phase
relative to the precipitation of TizNi.

81
In summary, themicrostructure study for annealed PML-16mm showed that the

32 grain boundaries were decorated with small TizNi precipitates (about 30 nm in
diameter). In the grain interiors, a small number of T‘izNi and a T'lsoNisg,5Cu3,5
precipitateswereobserved. TheTizNiprecipitateswereabcut lmnmindiameterandno
specific orientation relation with 32 matrix. On the other hand, the TiaoNigg5Cu3,5
precipitates with a thin blade-like morphology have an orthorhombic unit of
a=0.441nm, b=0.882nm andc=1.35nm, rather than a tetragonal one according
to published data [38]. An orientation relationship between the T‘isoNisg,5Cu3,5
precipitatesandtheBthasewerefotmdas:[110132//[011]ppt and (110),” // (200)“...

ThemicmsfirrctmeofdreannealedPhflflfilmawithanoveraflcomposifionof
TimNimCurg, is presented in the following section. Only TigNi type precipitates
would be expected to appear in this T'i-rich film.

42.3.PML-9Films

PML-9 Films: General Features. Figure 4-31 shows the microstructure of
thefree-standingPML-9films,annealedat923KforonehomatP=2.6x10'3Pa.ina
brightfieldTEMimage. TheaveragegrainsizeisaboutZumandlargenmnbersofsecond
phanpuficlesappeumafinedispersionmmughoutmegr'ainmtaimswimanavemge
diameterofaboutl5nm. Aprecipitate—freezoneisapparentadjacenttothegrain
butnduieewldchuediemselvesdecoratedwithpmcipitatepardcbe Theparticlesizeand
density of the precipitates observed in the film annealed at 923 K for one hour at
Ps2.6x104Pa are essentially the same as those shown in Fig. 4-32. However, the
microstructureofthefilmsannealedatSZBKforonehomatP-ZfixlO‘Pa,shownin
Figure 4-32, has slightly different features. The grain boundaries are still decorated with
nnaflequiaxedpardcles,whmeasmoremmonekindofpredpimesisobsavedmme
graininteliors. Theequiaxedprecipitatesdisuibutedinthegraininteriorshaveasizeof

82
onlyaboutSnmindiameter. 'lheotherprecipitamaretoosmalltobeidentifiedwithout

ambiguity.

PML-9 Films: TizNi Precipitates. In Figure 4-33 the particles in the grain
interiors are shown at high magnification and display a roughly equiaxed, though
sometimesfacetedshape,andaverageabout20nmindiamewr. Plane,undistorted Moire
fringes,perpendicularto[110] Bz,withaspacingof3.16nm,arealsopresentonalarge
fraction of the precipitates. indicating a strong particle-matrix crystallographic orientation
relationship. Electron difiraction patterns in [110] and [111] directions shown in Figure
4-34 reveal that the small particles have a face centered cubic unit cell with a lattice
parameterri1.132mwhichcanbeassociatedwimthe’l‘izNiphaseMl]. 'lhediffraction
patternsshowthatthethreeprinciplelatticeaxesoftheparticlesandtheBZmatrixare
parallel, i.e.. that film“; [I [100]” and (010)“; // (010)32. The expected spacing of the
Whingesfunndfiompmalkuumpptanduwmwaneswascakuhwdaslwum.
which is in a good agreement with the measured value of 3.16 nm. Large d-spacing
difference ( 6.5 ‘5 of [440)”; and {110132) and the planar Moire fringes indicated no
coberencybetweentbemanixandptecipitamlattices. However,acloaeexaminationofa
TEMmicrographshowninFig. 4—33revealstheexistenceofstrain field whichresultsin
lobe-likecontrastaroundthefineprecipitatesasindicatedbyarrow. A3gweakbeam
micrograph, taken from the same sample, is shown in Figure +35 in which the strain
contrastcanalsobeidentified. ConsiderationofthelatticeparametersofBZmatrixand
TizNi phase reveals that volume misfit strain may exist in the early stage of precipitation
sincetheT'eriphasecontainhigherratioofthelarge'I'iatom[147]. 'l‘hevolumemisfitis
calculatedtobe9.7% accordingly,basedonthelatticeparametersofthe82phaaeandtbe
TizNi phase. Additional electron diffraction study also showed that the precipitates
distributedonthegrainboundariesarethesameTizNiphase.butwithnoapparent
orientationrelationshiptotheBnginsoneidrersideoftheboundary. Moreover,thefine
equiaxed precipitates found in the PML-9 film annealed at 823 K for one hour at

 

83
p a 26le Pa were also TizNi phase with the same orientation relationship to the 32

nurtrix as mentioned above.

PML-9 Films: Thermal Phase Stability. Figln'e 4-36 ‘ are normalized X-ray
spectraoffilmsannealedat923 Kforfive minutesandonehourrespectively. A weak
peakbcawdutwotheuofflfisdiscanedinthespecmfamefivenunutemneabd
filmindicatingthattheTizNiphasebeginstoform. Afteronehom'annealingmomore
meeflenfifiedhaddiMmBZfiNiandTuNLwimidingwimmeTEMmsult

Awefl-defhledorientationrehdonshipbetweenthepmentBthaseandaTigNi
precipitatephaseisdeducedforthefirsttimeinthepresentstudy. Apossiblereasonthat
this orientation relationship has not been previously identified in bulk TiNi alloys in some
previously studies [32, 33] is described as follows: According to the Ti-Ni phase
diagram [29] shown in Fig. 2-3, the steep solidus line on Ti-rich side restricts the
formation of a single phase (32) solid solution with over-saturated titanium. In other
words, in bulk Ti-rich alloys, the TizNi phase should have already precipitated after
solidification and/or homogenization. The processing procedures employed by
Gupta et al. [33] in which, for instance, the Ti-rich alloys. were hot rolled and then
homogenized at 1173K forone hour prior to annealed at 1173K forone hour would
destiny the orientation relationship between the 'l'izNi particles and 32 matrix (ifit existed)
by deformation and subsequent recrystallization of 32 phase. In the present case,
however, a. ploymorphic reaction at about 750 K allows the multilayered mm films
crystallize to a single 32 phase, with super-saturated in titanium. The X-ray diffraction
study showed that no TizNi peaks were detected in the PML-9 film after isothermal
annealing at 923 K for 5 minutes (Fig. 4—37), confirming that most of the 'l'izNi precipitates
nucleated after the crystallization of the 32 phase. As a result, the TizNi particles
precipitating from this over-saturated matrix can maintain well-defined orientation
relationship with the 32 parent phase.

. 84
humymnfccflzNiphasewasfoundtoprecipimeinawidedispersionindle
mnealedPML-9films. Theequiaxedprecipitatesinthegraininteriorwere 5-15 nmin
diameter, depending on the annealing temperature, and had a well-defined orientation
relationship to the 32 parent phase, this is: [1mln2m ll [100132 and (010mm // (010)32.

42.4. Further Discussion
Akeyisutefadteenploymtotunniummtudhyastoinaeaseuunituncontem
inBZmuixiswhemeranmthedrivingfaceofcheuucdhomgeninfimcancompbte
with‘that ofprecipitation. Both ofthem depend on the same kinetic process: diffusion.
Thechenucnmioncoenidemmeqnamcazrmmbyhasheendemedhyamn
andRieck[37]tobe '

b (anzlsec) = 0.0020 exp(-E um (4.2. 1)

wha'eE =- 142000 joulelmole, which yields 15 = 1.84x1011 oral/sec and D = 1.94x1012
emz/sec at 923 and 823 K. respectively. The study also revealed that the intrinsic diffusion
coefficientofNi is aboutanorderofmagnitude higherthan thatof'l‘i. InamorphousTiNi
alloy, titanium atoms were also found to be a slow diffuser [81]. Accordingly, the
approximate diffusion distance, (D: )1”, of titanium at 923 K and 823 K for one hour are
estimated by using 01,-: 0.115 to be 814 nm and 264 nm respectively. Both values are
much larger than the layer thicknesses of 9 and 16 nm. Therefore, it is reasonable to
conclude that the composition in the 32 matrix of the annealed PML films are
homogeneous throughout.

In addition to the chemical homogeneity of the matrix phase, the true titanium
comentintbemanix,mm«dlmdlewmflcomposifimoftheMfilmsafummahng
mustbeaddressed. Anextremecasewillbethatsegregatedtitaniumatomsinthe'l‘ilayer
results in extensive precipitation of Ti-rich phases in the early stage of annealing. If these
Ti-rich precipitates remain stable tlu'oughout the anneal, the titanium content in the 32

85
matrix will not be increased effectively by the addition ofthe Ti layer. In fact, TizNi

particleswerefoundinthegraininteriorsofthel’ma-Mfilm. Thelackoforientation
nhfimflfipbaweenBZmauixanddwpudchaarggemdmmeymayhavenudeamdnear
the’l‘ilayerspriatothecrystallintionreaction. Asimplifiedcalculationwasconductedto
estimatethedensityofthe'l’eriparticlestsumingthatallthetitaniumatomsinthe
titaniumlayerswereconsumedintheprecipitationreaction. Theresultshowsthatabout
eighty TigNi particles with the observed size of (100 run)3 should be found in an
lumxlttmeJumvolume. Thisnumberisatleastoneu'derofmagnitudelarger
flundlatfoundindlerrl6filmsaccadingtothepresentTBMobservadon. 'l‘hisfact
indicatesthatlessthan10%oftitaniumaddedexistsintheTi2Niparticleswhichformed
prior to crystallization. Results of X-ray diffraction for the PML-16 and PML-9 films
showninFig.4-30and4w36,respectively,confirmthatnopeaksfromTi7Niphasecanbe
identified after annealing at 923 K for 5 minutes. Therefore, it is reasonable to conclude
thattheadditionofperiodicTilayers effectivelyincreasesthetitanium contentinthe32
matrixphase.

The extensive observation of TizNi type precipitates in grain boundaries of TiNi
films was also reported by Busch et al. [116], for samples that were either annealed
isothermally at 813 K for various lengths of time, or annealed isochronically up to
temperatures between 813 K and 1073 K. The volume fraction of grain boundary
precipiuteswufoundmincmasewithmincrcaseinbothannedingwmpemmeand
annealing time. In their study, Ti3Ni4 precipitates distributed in the grain interiors were
also observed in conjunction with the appearance of TizNi in the films isochronically
annealed to 973 K and 1073 K respectively. They concluded that the grain boundary
precipitates were TizNi instead of rump” according to EDS analyses. However, these
filmswereannealedinsealedquartznrbeswhichwereevacuatedtoapresslueofabout

0.1Papriortoanneal. Thisvacuumconditioncouldcauseseriouscontaminationfrom
reactivegasessuchasoxygenandnitrogen. Furthermore, 32 'l'lNiisathermodynamically

86
stable phase at a temperature above 923 K, which has been confirmed by many

studies [30-36]. Thaefaemeprecipitationoffipltliinafi-richfilmathighwmperauues
shouldnotbeassociatedwithanappearanceofaNi—richphase,whichhasbeen
demonstratedinthepresentcaseofPMLSlfilms. Onthecontrary,theTizNiphasewould
notbepresrunedtoappearinaNi—richfilmduring hightemperatureannealing. The
appeamnceofTizNiuemosdyinmegrainbomdafiesisparfiwhdysuspiciouaand
suggests oxygen contamination.

Since oxygen has been found to stabilize the TizNi phase [41], one possibility is
that oxygen diffused along the grain boundaries during annealing, and enhanced the
precipitation of Twizo, phase. The X-ray diffraction pattern in Fig. 4-31 also reveals
mmnmiwmtmmpmlofimmmmy stageofannealing. The
concomitant depletion of titanium atoms in the matrix thus enhanced the further
precipitation of TizNig thereafter. Therefore, the grain boundary precipitates extensively
observed in the (NiCu)-rich films in the present study are concluded to be the 'I"14Ni20x
phase. Ontheotherhand, theTizNi typeprecipitatesobservedin thePML-9filmsarea
mixture of TizNi and TiaNizox. In the grain boundaries, most of precipitates are
TittNigO," whereas the particles in the grain interiors is expected to be TizNi due to the

super-saturation of titanium.

4.3. Thermoelastic Transformation Characteristics

ThedlemnelasficnansformadonbehaviorofthefimiCu)thinfilmssuldiedin
section4.2ispresentedinthissection. Crystallographicdataonthemartensiteswillalso
bereportedanddiscussed. V '

4.3.1. Traltp‘onaations in SL Films

Tbethermoelastic transformation behaviorofthesnnealedsLtilmswithovetall
compositionof’l‘iauNiauCuu isdescribedin thefollowing section. The transformation
temperatures, transformation sequences and crystal structures of the transformation
products were studied by electrical resistivity measurement, differential scanning
calorimetry and X-ray diffraction. Transmission electron microscopy was employed to
study the microstructure of the martensitic phase ill-Sills.

SL Films: Electrical Resistivity and DSC Results. The electrical resistivity
clu'vesforthetsrgetalloyandtheSLfilms, annealedat923Kfor0.25 hourandforone
hour, respectively, at P = 2.6x10'3 Pa, are shown in Figure 4-37. The electrical
resistivity of sputter-target alloy has a negative temperature coefficient, i.e.,
Cr: ng/dl‘<0, during the thermoelastic transformation. which has been reported for
Ti(NiCu) ternary alloys with more than 5 at% Cu addition [148]. On cooling, the
appeuanceofmartensiteisaccompaniedbyanabruptincreaseinresisdvity beginningat
315 K. leveling off at 292 K, and falling again after the transformation has apparently
finished. Onheafingmemsisfivityhlcreaseceasesmmkindicadngmestanofnva'se
transformation, after which the resistivity drops sharply until 335 K is reached. The
nominal M3, Mg, A, and A: temperatures for the sputter-target material are accordingly
determinedtobe315K.292K,308Kand335Krespectively.

The resistivity curves for the film annealed at 923 K for 0.25 hour also shows a
negative coefficient at sub-ambient temperature, accompanied by an one-degree hysteresis.
l-lowever,anotherresistivityvariation associatedwithalargerhysteresisisfoundatlower

87

88
temperature (< 210 K). This second-stage variation of hysteresis is not found in the

sputter-target alloy, and may be associated with an unidentified thermoelastic
transformation.

For the one-hour annealed film. me two-stage change of resistivity is more
pronounced. Theonsettenmaatmeofthesecond-stagevariafiononcoolingisabotuZOK
lowuthmthuintheOJS-hmnmnealedfilmwhaeuthemvasevafiafimofmsisfivity
onheatingstartsattemperanlrewhichisaboutaoxhigher. 'I‘llatis,therwo-stage
vuiadonofresisfivityintbeonehmnannealedfilmshuhrgahystaesesofabom 10K
(firststage)and50K(secondstage),comparedtothoseofabout1Kand30Kinthe
0.25-hour annealed filml.

IngcnaaLboththeOJS-holnannealedfilmanddleone-hourannefledfilmshow
behavior which is similar to that reported for TiNi and Ti(NiFe) alloys having well-
separated R-phase and martensitic transformations [67], as shown in Fig 2-6.
Accordingly, theresistivity curves foreach ofthe annealed SLfilmsareinterpretedina
way which has been used for the two-step transformation described in section 2.3.1. The
tanpmuueofmemidflresisfivityfisemmofingisasmcimedwimdwR-phasemnsidm
andisdesignatedsz. 'lhefinalu'ansformationtothemonoclinicmartensitecoincideswith
the onset temperature of the resistivity decrease, and is designated M3. Finally, Mf is
determinedineachcaseasthetemperatureatwhichtheheatingandcoolingcurves
converge at low temperatures, which is below the range which our apparatus could attain.
Austenite transformation temperatures are conversely determined from the beating
resistivity curves. In this case, As is assigned to the onset of abrupt resistivity slope

 

1Thebystaesisofthesecond-stagevuiationofresistivityknutwell-defiued. 'l‘henrrrnbasdlowaabove
mefiunamrghesfimafimbymmfingdwawwidhbuwemdwheafinguflcodhgcm
21nthepresentstudy,theR-phasetransitionisusedtorepresentacombinationofthesecond32-to-
incommensurate phase transition and the first order imcommensurate-to—commensurate phase
transformation. 'lheonsettemperauneoftheR-phaseumsition TpthuscorresponrhtotlntofdleBZ-b-
wmmn'hrefwlsmmhichhubemexplahdhmw.

89
changeonheating,andAftothetopoftheresistivitypeak3. ThCTr,Mg, Mf,AsandAf

ternperaunesareaccordinglydeter'mined.andruelabeledinFig.4-38.

Differendalscanningcalaimenywudaoemployedtosnrdydrenansformadon
behavior. Figure 4-38 shows DSC traces of target alloy for reference, and SL film
annealedat923KforonehouratP-2.6x10'3Pa. Singleexothermalandendothermal
peabwaeobsavedinnrgadloywithavaagemnsfamafimenthalpyoflmlllmole.
TbecodingscanfathemehommnealedSLfilmshowssevaenoisewhhonebmadand
ill-definedpeaklocatedinbetweenZ‘l8Kand239K. 'l‘hisnoisybackgroundisacomrnon
fume in cooling scans, especially for the films with a weight of several milligrams.
However,theheatingscanshowstwoendothermalpeakswith temperatrue rangesthat
cmespmflrelafiwlyweummeAsandAfwmpannnesasdgmdmdwheadngresisdvity
curve. TheatdnlpiesofdlebwwmpaannepeakandhighwmperauneoneueUSJ/molc
and 125 J/mole. respectively. RepeatoftheDSC scans shows thattheheatofthelow-
temperatru'e peak varies with temperature to which the sample was cooled. On the
www.memgh-wmpmmeexmhamhasanhfiveconmntvdueofenthalpywhichis
close to published value for the R—phase transition [51].

SL Films:X-Ray Difi'raction Resuln. X-ray diffraction patterns from SL
filmsarulealedat923KforO.25hourandonehourrespectivelyatP=2.6xlO'3Pa,were
acquiredatvarioustemperatruesasshowninFigme4-39andFigme4-40. InFig.4-39,
onlypeaksfromBZaustenite,andthe(NiCu)2Tiphaseappearat297K.thatis,abovethe
nominalAftemperatureof283K. 'l‘hebroadpeakat200f41°mayresultfromthe
overlapping of the (110) (NiCuhTipeakandthe (333) TizNipeak. Asthefilmwas
cooled.the(110)32peakbecamebroader. 'l‘hisisoneofthepreclu'soryphenomenaofthe
R-phasetransformationHB, 149]. 'lhepatterntakenat239Kshowsthatasidepeakstarts
tosplitfromtheleftshoulderofthe(110)32peahindicatingthattheauswniteu‘ansforms

 

3’l'lrewayalrlefirretheAppoiu'lthesamemtlminreference[50],bratheAfternper'mrre'urbsigntetl
nubmdmmgmwwummdmmmmmgm

90
fully tothe R-phasell49]. ‘l‘herbombobedraldistortioncontinueson coolingafterthell-

phasemfidmfllusuawdbymeinauseofdletwo-theumglebaweenmespfitpeaka
carespondingto(1122)gpeakand(0330)gpeakrespectively. Thisresultcoincideswith
thereportot‘IJngetal.[49]. VeryweakpeaksarediscernedattwothetaofSTand39°at
Whelowl93K,whichmayindicatematavaysmaflamountofmuwnsitehas
formed. InFigure4-40forthefi1mannealedat923Kforonehour,splittingof(110)nz
peskstartsattemperatrueslightlyaboveZflK. Inaddition,moremartenritewasformed
attheternperauuebelowl‘l3K. Consequently,the0.25—bowandone-hourannealedfilms
have similarR—phase transition temperambutthemartensitic transformationinthe
former is more sluggish. The increase of the intensities of (110) and (013) precipitate
peaksresultedfi'omdleoverlappingofpeaksfiemR-phaseandmartensite.

RecdfingdnresisfivitydatauldDSCdamhisckardmdleinaeaseofredsfivity
betweenabout280Kand240KoncoofingcarespondsmdleR-phaseuansformafion
whichalsoresultsinabroadexothermintheDSCu'ace. Thefurtherdecleaseofresistivity
atabout190Kthusisassociatedwiththemartensitictransformation. Onheating,anR-
phase-to—austenite transition follows the reverse transformation from martensite to R-phase.
producing two separate DSC endotherms. Accordingly, the low-temperatlue peak,
corresponding to the abrupt increase of- resistivity, represents the martensite-to R-phase
msfamadmandthehigh-tanpemnneonechuectaiumemaseR-phaseuansidm
which results in a decrease of resistivity. As a result. the SL films undergo separate
32 HRandR HMu'ansfornntionswhichhasnotbeenpreviouslyreportedinanyCu
containingTiNialloys.

SL Films: Transmission Electron Microscopy Results. ‘ Room temperature
elecuondiffiaedonpanamaequiredfiomrheSLfilmsannealeduflBKfamemu
P226x103PamdicammamesnucnueueprimarflyconmosedofdeBZWwim
no martensite or R-phase present. However, diffuse streaks appear along certain
crystallographic directions, such as <110>” and 61b”, as shown in Figure 4-18.

91

AtawmpaemonSSKdnbrightfiddimageandimcorrespondingdiffiacdon
patterninFigure4-41showwell-defined ll3spotsinthe<321>32 directions indicating
dramastaafloftheR-phasemnsfmmafimwucoupletendlismmbmmum
martensitewasyetpresent. Nocontrastchangewasobservedineitherbrightfieldadark
fieldimagesasthesamplewastilted. Figure4-42showsabrightfieldimageandits
correspondingdiffractionpatterntaken along the <112>nz direction. 'I‘hediffraction
patterns can be indexed with a hexagonal unit cell, having a - 0.738 nm and
€80.532nm.uproposedbyGooandSinclairISllnndarefmmdtobeinan<ll§0>
zone axis. The orientation relationship between 132 austenite and R-phase is accordingly
expressed as (111),” // (0001)]; and [2111321/[21'1'01m also coinciding with the result
forTi(NiFe) alloyreportedbyGooandSinclair[51].

At 1601((31Kbelowtheonsettemperanneoftheresistivitydeelineobserved
druingmohng)dwspechmnprodwedme[100]mumnsimdifiracdonpaumanddnfim
martensiteplatesshowninFigureA—43. Themartensitevariantsareaboutlwnminsiae
with dense transformation defects. These dense defects indicate that martensite has a
monoclinic unit cell, which require twinning as lattice invariant shear during
transformation. Electron diffraction study revealed that neighboring martensite variants
usuallyhada'l‘ype—I(lll)twinningrelationship. Figlrre4-445howsanexarrlpleinwhicb
the bright field micrograph contains various fine martensite variants oriented in different
d'uections. Theassociated diffraction pattern contains two [321] martensite zoneaxes
whichare(111)twin—related. Onesetofvariantscorrespondstothistwilmingrelationis
showninthedarkfieldmicrograpbtakenfroma(012)uspot. Thehabitplanebetween
neighboringvariantscoincideswiththetwinplane. (Anetfortforfurtheranalyseother
mrtenfitevmimminmesamemstemwmwashnfiwdbydwnngk-dhholdaandfine
variantsize.)

rhea-phaseissnumepredominantphsseindtesmlmmooxmtdsomeplate-
fikevuianmwaediwanedinmeR-phaseasdnsampkwasuienwdincauindnecdons

92
ThesiuoftheR-phasevariantsissimflartothatoftbemartensitevariantmbuttheformer

canbeidentifiedaccordingtotlrelartkoffineinmnaldefects. Atypicalexampleisshown
in Figure 4-46. The bright field micrograph in Fig. “5(a) shows two sets of plate-like
variantswith boundariesorientedperpendiculartoeacbother. Fig. 4-45(c-d) arethedark
field images taken from diffraction spots c and 4 respectively labeled in Fig. 4-45(b).
They show the detailed structure of two labyrinth-like R-phase domains which are
composed of two sets of plate-like variants with their boundaries roughly parallel and
perpendicular to <110>” reciprocal direction respectively. Additional diffraction
observation revealed the presence of the R-phase at temperatures as low as 120 K,
indiadngthemamndficnansfamadmfinishtanpuammmgisbebwmistanpamme.
SL Films: Trandomration Sequences and Temperatures. Near equiatomic TiN i
alloyswith4t08 at%coppersubstitutionforNihavebeengenerallythoughttoundergoa
32 H M transformation [7, 71, 72]. Only in alloys with less than 3 at% Cu were
32 (-) R transition observed by Tadaki and Wayman [71]. However, no reports have
shownthatthecoppercontainingalloyscouldundergoseparateBZ —-> R and R -r M
transformations after aging or thermomechanical treatment,as has been widely observed in
binary TiNi alloys. In the present study, the thermoelastic transformation in the SL films
mooolhgisidendfiedmbeatwo-swpuansfamadonmwhichBstteruwundergoesm
R—phase transition priorto amartensitic transformation. This two-step transformation was
also reported in several ternaries including TiNiFe [50], TiNiAl [150] and TiNiCr [151].
Thechangeoftransformation behaviorin theternarieswasattributedtodifferenceoffree
energy changes between the R-phase and martensite [65]. However, this is apparently not
the case in the Ti(NiCu) alloys, because the two-step transition is not a general feature of
the thermoelastic transformation. According to the EDS analyses, the titanium content in
32 matrix of the SL films is close to the overall titanium content of 47.4 at %, indicating
that the precipitation of 'I"14(NiCu)zOx in the grain boundaries had offset the titanium
emichmentduetotheprecipitationofmiCuh'l'iphase. UnderdesecimumstarrceatheMg

93
meratluewouldbeestimawdlnltobeonlySOK.incontrasttothemeasuredvalueof

about 200 K. Both the existence of a two-step transformation, and the relatively high
nansfarmdontempaauuesfamchlowfimniumcontenuuebdievedmbemlawdmdle
precipitation of semi-coherent (NiCuhTi particles, and a detailed discussion will be given
below.

Meet 0f (NiCuhTi Precipitates on Traniomration Behavior in SL Films. First,
tbeEDSanalysesshowninFig.4-21revealsthattheprecipitationofthemiCuhTiphase
decreasesthecoppercontentintheBZmatlixfromal at%to3.8at%. Tadakiand
Wayman [71] mentioned that the temperature coefficient of resistivity, C1 = ng/tfl‘, has a
positive value associated with martensitic transformation in binary alloys and ternaries
containing less than 5 at% Cu, whereas it has a negative coeflicient in ternary alloys with
more than 5 at% Cu addition. In the present study, the SL films having a positive Cr
valuecoincides withthecopperdepletionintheBZmatrixdetectedby EDS. Tadakiand
WaymanalsosuggestedthattbeadditionofcopperinTlNisuppressedtheBZ -) R
transformation [71]. Accordingly, lowering the copper content in the 32 matrix of the SL
filmsisexpectedtoreleasetheconstraintofthe32 -) Rtransformation.

However,thedecreaseofthecoppercontentitselfdoesnotnecessarilyresultina
well-separated two-step transformation. 'I‘heeffectofprecipitatesmustalsobeconsidered.
As mentioned in Section 2.3.2., a two-step transformation in Ni-rich binary alloys can
result from the appearance of coherent or semi—coherent precipitates of Ti3Ni4, or the
existence of rearranged dislocations“. An orientation relationship between TigNia particles
and the austenite, and with the martensite, has been deduced to be
[11 11113164” [111132 // [101]m[43]. Consideration of the lattice parameters of the
32 phase and TigNis phase indicates that the (111) precipitate plane and (111) austenite
plane have the largest d—spacing mismatch of about 2.7% [43]. This implies that a
maximumtensibsuessexisminmemauixdongadhecfimpapaldicularmtheflll)

 

‘Thuemmageddisbadoumhfiemnneahagaimamedianemme,pecededbycold
dam .

94

habit plane. Since both [111132 and [101]” are the most extensible directions5 during
the R-phase and martensitic transformations respectively, this tensile internal stress along
<111>32 would seem to either both transformations [43]. However, the formation of
either R-phase ormartensite variants generates a back stress around the precipitate, which
wndstoresistcondnuedforwuduansformadon. Heretbebacksuessistheelasticstress
mededmmaintfintbehtficewhaemybuweenprwipimandmeresmfingphaseafta
transformation. Consequently. larger lattice deformation associated with the martensitic
transformation generates higher back stress.

The calculation of elongation along [Hung and [TOUW during the R-phase and
martensitic transformations yields values of about 1 % and 10.5 % respectively. This
mggeurMflR-Waaeuamfmmafimiaflmaeusflymducedbyimanalsuessmd
less interfered with by back stress than the martensitic transformation. In fact, the
martensitic transformation is suppressed by the back stress in the early stage of
precipitation, according to the results reported by Wu and Lin [152], and Nishida and
Honma [68]. Recent 'I'EM study carried out by Xie et al. [61] showed that fine, coherent
TI3NI4 precipitates do not provide nucleation sites for the martensite, but coarse, semi-
coherent TigNi4 particles do favor nucleation of martensite at precipitate interfaces, but
constrain it from further growth. In other words, the Ms temperature is elevated at the
expenseofdlewideningofdlenansformafionemperauneintervalNg-Mf).

In the present study, semi-coherent (NiCu)2Ti precipitates were observed in the SL
film annealed at 923 K for one horn. As demonstrated by TEM observation, lattice
coherency only exists in the [001 }/(001) habit plane of the 32 phase and (NiOuhTi phase.
Further calculation reveals that the d-spacing of (100)32 is about 2.7 ‘5 smaller that the
(100) and (010) spacing in (NiCuhTi. This mismatch results in dilatation of the matrix

 

meuficedefanudmnwdedmamuaufmwhuicemhm
willbestrainedinordertotrmsformtoalatticevectorinmartensite. Acca'dinslyrthereisonelattice
vecuxwillexperiencedrelm'gesttensilemainduringdleuamformadon, basedonthespeciallattice
ammmmanmmmammnemmbum

95
latticewhichcanprodlrceanelongationalongfun<lll>mofasmuchas2%,which'u

smallerthanthatproducedbytheTigNiaprecipitates. However,theseinternalstresses
favor four R-pbase variants. In other words, the back stress associated with the
uansformafioncanbereducedbydleformafionofself-accommodatedR-phaae. Infact,
nH-wcmdawdmhkeR-pmmwimabomwmmeereobsaved
intheone—hourannealedSLfilms,asshowninFig.4-43. Thereforeitissuggestedhere
mummdsuainspodwedbysenu-mhaentmiCuhfipredpimammmisemeRs
phaseuansfumadonwmpaanuebyprovidingapopuhdmoffavorabkhuaogeneom
nucleationsiws.
BoththeSLfilmsannealedat923Kfor0.25homandonehournespectively,
yield a similar T; temperature of about 280 K. This coincides with the observation,
reported by Nishida andHonma [68], which showed that IhCTr temperature is relatively
insensitivetotheannealingtime. However,thevalueong-inthepresentstudyis20n40
degreeslowerthanthosereportedinthebinaryalloys. ThelowerTr-value,infact,is
expectedbecausethemiCuhTi precipitates produce less tensilestrain(about2 %) along
<111>32directionsversusthe2.7fiproducedbythe'l'lgNisphaseinbinaryalloys.
hisappuentthemthudweffectofthesemicohemntmiCuhTipncipimteson
martensitic transformation maybesimilartothatofthesemicoherent 'l'i3Ni4. Xieetal.
[61] foundthatcoarseTbN‘uparticlesprovidednucleation sitesformartensitevariantsand
thenelevatedtheMgtemperatme. Inthepresentstudy,TEMobservationshowsthattbe
variantsizeofthemartensiteissimilartoorsmallerthantbeparticlesiaeoftbe(NiOu)2Ti
phase. Mm,uflikedlesingleR-phasevmiantfounduoundtbeTi3Ni4min
[43], multiple R-phase variants formedpriortothe martensitic transformation may assist
thesdf-acconumdafimprecessofmemmensiwvafianmmdfunhadecmasetheback
stresseffect. AsaresulutheMgtemperatrnesoftheSLfilmsannealedat923Kareabout
150Khigherthantheexpectedvaluebasedonitsnnuixdtaniumcontents.

Therevaseuansfa'mafionintheonZ-6hommnealedSLfilmswualsoatwo-step
transfamation, according to the DSC and resistivity data. The coarse (NiCuhTi
precipimseemmuabihnethemuwnsiteandpostponedwmaseuansfmmadon Asa
remhmeagwnmeranueoftheonehomsnnealedsmlmisaboutsoxbighermanthn
ofthe0.25—hourannealedfilm. Postponedreversetransformationsalsoresultsinalarger
hymeas(1lx)ofthea-pbasenanndoumtbeonebommnealednlmverntsmstoflx
intheo2s-bourannerlednlm

In conclusion, the precipitation of the serm'coherent (NiCubTi phase in this
Tmmwtmmmwdme matrixcoppercontentandproduced internal tensile
sneuesbothofwhichpromotedmea-phasemsfmmauonandresultedmstwo-nep

transformation.

4.3.2. Trand'onttatlous in PML-I6 Films

The thermoelastic transformation behavior of the annealed PML-16 films with
overall composition of Ti49,2Ni44,gCug_o is now described. According to the
microstructure study reported in section 4.2, no (NiCuhTi precipitates appeared in the
annealedPML-l6films. lnfactthegmininteriorsofthesefilnuwererelativelyfreefrom
precipitates. Different transformation behaviorofthe annealedPML-l6 films is therefore
expected.

PML-16 Films: Electrical Resistivity and DSC Resulm. Figure 4-46 shows
electrical resistivity curves from a PML-16mm annealed at 923 K for one hour at
p =- 2.6x10'3 Pa. These curves are qualitatively similar to that ofthe sputter-target alloy,
except that the transformation temperatures were substantially depressed. The
transformation temperatures, Mg, Mf, Ag and Af, therefore, are determined according to
the methoddescribed in section 4.3.1. to be 287 x, 228 K, 238 K and 297 K respectively,
which yieldan average transformationtemperatureintervalof59 K, and ahysteresis of
10 K.

90
Figme4-47showsDSCrendtsforafreestandingPhfl’16filmJnnealedat923K

fa'onehouratP-Z.6x10'3l’a.showingresultswhicharealsosimilartothoseforthe
sputter-target alloy. BothAfandMg temperatruesareabout 12 K lowerthan those
deanu'nedfromresistivitymeastuemcnt. 'l‘hewidthofendo-andexothermalpeaksinthe
calonmeerscannwhiehgivemeunnsfamsdouempaammmtnvnarmeonlysbmu
20K.wherustheuansformadmhynaesisof9Kissinularwthnobtainedfiomthe
resistivitymessurements. 'l'heaveragetransformationenthalpyobtainedfromtheintegral
ofbothpeaksis685 ”mole. 'AdditionalDSC surveys were performed forthefilms
annealed under various conditions, and results are tabulated in Table 4.4. The film
annealedat923Kfor6hoursinavacuumof2.6x10'3Payieldedalowertransformation
endulpy of 427 J/mole without a change in transformation temperatures. On the other
hand,thefilmsannealedatP=2.6x104Paforonehouryieldsu'ansformationentbalpies
of 1113 J/mole (923Kannealed)and854.llmole (823Kannea1ed), respectively, whicharc
close to the value of 1076 Jlmole for the sputter-target alloy.

Itisworthnotingthatthcfilmannealedat823KforonehouratP-2.6xlO"Pa
basalngamsformafionwmpaamninterval,about35K,compamdmthatof9Kfor
the923Kannealedfi1ms. TheDSCexo-andendotbermsofthe823Kannealedfilmare
shoammrigme4-48whichmveulmntwoesomermalpeaksappenonoooungbetween
242 K and 202 K. Pronounced background noise makes the determination of
transformationtemperaturesmoredifficult. Thefirstcoolingpeaklulsanenthalpyofabout
lSOJ/mole whichisclosetopublishedvaluesfortheR—phase transition [51]. Inaddition,
asbmtldanobsavedmdtemghaenmmnnesideofthehcaungendmhermmdicanng
thatitconsistsoftwooverlappedpeaks. ‘l‘hetransformationsequenceofl’mrmfilms
annealed at 823 K for one hour at P = 2.6x10-4 Pa is, therefore, identified to be
32 -) R —>Moncoolingand32-)Mandlor32-iR—)Monheating.

mus Films: X-Ray Divination Results. , Cold-stage X—ray diffraction studies
forthefilmsannealedat923Kand823K,respectively,foronehorn'atP=2.6x10'4Pa

98
wereundertakentoverifytheDSCresults. Figure4-49istheX-raydiffractionpatternsof

thefilmannealedat923Ktakenatvar-iouswmperattues. Onlypeaksfrom32austenite
andmndphaseprecipinwsueobsuvedumxindicadngmmutensimexisu. At
251K,bdowdefcmpaauuedetaunnedbyDSC,monochnicmartensite(Mu)peaks
mnbeideuunedmdmepeakmendtyoftllomdeneasesabmusosndicadngmnme
324Wtransformationisonlyabouthalfwaydone. Themartensitictransformationis
foundtofinishuabGu201K,whichisabout50KlowerthandmdetanunedbyDSC.
Figme4-50showresulmfathefihmanneabdu823fishoudngthummamndtepuks
canbeidendfiedyetat249K,thoughme(110)ggpeakintensitydeausesabom25‘5.
Thedecreaseofthe(110)32peakintensityisassociatedwithaslightpeakbreadening
whichisknowntobeapreclnsayphenomenonoftheR-phaseu'ansformation. At198K.
odymn'wnsiwpeaksamobservedshowingmewholeuansfamafionsfinishuaabove
198K,whichisveryclosetothertemperntureof202KdeterminedbyDSC. Asa
mmmecdd-uagex-mydifi’mdmdanfmmefilmsmneabdusflKcoincidu
with the DSC results very well. However, the Mf temperature determined by X-ray
diffiaedmindrefilmsannededathisabGuSOwaerthanthuobninedfiemthe
DSCresults.

PML-I6 Films: Transmission Electron Microscopy Results. Coldestage TEM
observationshowsthatthera-l6filmsannealedat923Kforonehoru'undergoa
132 -t M... transformation on cooling. No R-phase was found in the films. However,
thesampleannealedat823KforonehorrratP=2.6x10'4Pashowsatransformation
sequenceof32—iR-tMuoncooling. ’I‘hecrystallographicnatureoftbemonoclinic
martensites observed in all the films is essentially the same regardless the prior
transformation sequence. Consequently, the majority of the following results are taken
fromasampleannealedat823KforonehorrratP=2.6x104Paunlessnotedotherwise.

BoththeR-phaseandmartensiteformedfirstatathickpordonofthesampleon
wohngandmemnsfamadmtempemnneswaefwndmbeafumdonoffoflmichless

99
inarangefromOtoabout300.nm,andthetemperaturesreportedbelowmaynot

correspond well to the resistivity and DSC data. Figure 4.51 shows a bright field
micrograpbanditseonerpotrdingdinraetionprttemtaltenatzlslc Thebrightfieldimage
shows three R-phase domains, labeled A, a and c, with plate-like substructures, formed
indreaustenitegrain. Tbecorrespondingdim'actionpatterntakenfromdomainA canbe
indexedtothen§431g zone, using thehexagonalunitcellproposedby Gooand Sinclair
[51], or to the [012132 stone with in spots along <123>32 reciprocal lattice vectors. The
phte-likesubstructuresindomainA areabout30nminwidthandelongatedacrossthe
variantsinadirectionpanlleltodlelllmmrecipecallatticevector. Thesubstructuresin
domain a have the same size and morphology, but orient perpendicular to the [100132
reciprocal lattice vector. Figure 4-52 shows a bright field micrograph and its
corresponding diffraction pattern acquired along [023132 direction at 215 1c The bright
fieldimageconfimsdlattheplue-hkedomaimampufllelwthe<100>agreciprocd
directionandareinclinedwithrespecttothebcamdirection Traceanalysesindicatethat
the plate-like domins are parallel to (110)32 or [100le planes. As a consequence, they
maybethe(100)or(110)typetwinssimilartothosewhichhavebeenobservedin
Ti(NiFe) alloys by Hwang et al. [153]. The diffraction patterns reveal that neighboring
domainsarealso(100)twin-related. , '

A labyrinth-like substructure is also observed as shown in Figure 4.53. The
usocnmddifnacnonpannnisinatofzilgmneaxiswhichngeneatednommeuea
havingthedarkercontrastlevel. Thetracedirectionsofthesubstructureboundariesarein
approximately [113m and [83' 541p reciprocal vector directions respectively. It is also
foundthatthe [022T1g zone axis isonly11.75° fromthenearest<01T>m direction As
the [0'2'2'1'1p zone axis is rotated toward the [01‘1'132 direction, [2134]. and [8534];
were found to align with [100132 and [011132 reciprocal vector directions within :l:1
degree. Therefore, the substructure boundaries are thought to be (room and (011).;

100
planes respectively, coinciding with the above observations shown in Fig 4~51 and

Fig. 4-52.

Monoclinic martensiteplateswetealsofoundat215Kas ShOWflillFiglll’C4-55.
11seasaociateddiffractionpattemshows[110]ggpatterncnntainingl/3 typereflections
superimposed with [010].“ pattern, indicating that the nurtensiw is consuming the R-
phaae instead of 32 austenic. .

Asthetemperaturewasloweredto l70Kia-siuaalltheR-phaseand82austenite
mnsfmmedmmuwunemceptinvayhnntedminueaneartheedgeofmehole. Figure
4,-55 shows the general features ofthetmrtensite variants and how they accommodated
eachother. InFigure4-55(a)themartensiteformedband—likevariantsinawidthofwto
Mammatheoriginalaustenite grain. Finetransformationdefectswithaspacingof
about 20 um can be observedinside the band-like variants. Figure 4-55(b) shows the
zigzag variant arrangement in which two variants with planar transformation defects formed
predominantly in the original austenite grain, accompanyed by some scattered
third-variants. In both case no well-defined martensite variant boundaries are found.
Moreover, cases in which one variant dominated the whole austenite grain was rarely
found.

Type-I (Ill) transformation twins werethe mostcommon transformation defects
observed in the monoclinic martensite. Figure 4-S6(a) shows an electron micrograph of
zigzagged martensite plates formed in the austenite grain. The diffraction pattern in
Fig. 4-so(b) shows superimposed [010] and [271'] martensite zone axes. A dark field
image taken from (IMWHOIDMM spots labeled in Figure 4-56(b) is shown in
Figure 4-56(c) in which two sets of martensite plates, labeled a and b respectively, are
obsavednmpaoflneumeinta'facesasshownintheencbscdmcmfirmingthaflny
have different crystallographic orientations. As the sample was tilted along the [001]
reciprocal axisby 335°, adiffraction pattern containing two (11 T) twin-related (1101......
zones and one [101]mono zone was acquired as shown in Figure 4-56(d). Dark field

. 101
mhoscopyprwesmatthenlompamscmrespondwhuimnmflypualklphwsaand

c. respectively, in which (1 l '1') twinning acts as the lattice invariant shear. Moreover,
anflysisofthediffracumpanamindicammuthehaimnnnyandvadcanyuhnwd
plawsaandb showninFrg.4-56(c)arealso(lll)twin—related. ('lhesingletiltholder
hosedhmhamtdyofdeaynulographicmienmordefmthsaofmumnteplaea)
Martensitewasfoundtorevettsbacktoaustenitewhentheelecuonbeamwas
convergedtoheat'lthefoil. Aninterefingfeanueiathanbeforerhereveraeuanaformarion
wuconrpbmmofmeremainingmuwnsitephwswaethoaewhichwuesimmdu
tbecenteroftheausenitegrains Figlne4-57showsasetofleaf-likemuwnsiteplawslefi
inthecenterofanaustenitegrain. 'lhesemartensiteleavesareaboutZOOnminwidthand
SMnminthebngimdinaldirecdon,andamsdnaccommoduedinazigzagform. The
associated diffraction patterns contains a [001] martensite zone and a [011] BZ zone
without 1/3 spots. The reverse transformation is therefore believed to be an one-step
u'ansitionwhichbypassestheR-phase. Wenthegrainboundariesseemstoconstrain
the nucleation of martensite plates, since the last remained martensite upon heating is
usuallythefirstonewhich nucleatesoncooling [l4]. Figlu'e 4-58 showsthatanothersets
ofthin-leaf martensite plates in a neighboring grain. The corresponding [3 T2] diffraction
patummvealdlatdretwosasofmanensitephteshaveanallflwinmhdmship.
Trmdonnan'ou Behavior of the PML-I6 Films. DSC traces of all the PML-l6
filmsannealedat923Kshowthatonlyoneexo-andoneendothermalpeak wasfoundin
cooling and heating scans respectively. This feature coincides with the TEM observation
thatnoR-phaseappearedduringthetransformation. However,tltetilmartnealedat8231<
for one hour at P = 2.6x104 Pa shows a two-step transformation on cooling and an
overlappeduansformationonheating,accordingtobothDSCandTEMdata. Moreover,
flnMgmmunesofflle823Kmmeabdfilmsue40w50degreeslowerthmmemgof
the923Kannealedfilms. Therealreasonforthisdifferenceisnotpresentlylmown,since
theyhawdmilmwmposifimmdnnaosumuneaexceptdwfewasecondphasepuficbs

102
wereobservedinthe823Kannealedfilms. Althoughlowtransformationwmperaturescan

beamibuwdmhwaduniumcmmresulfingfiunfabricaummahighavdune
mfiooffi4((NiCu)20gwithrespecttofiyfii3inthe823Kannealedfilm8.thelowtitanium
cmtentalonecannotaccountforthetwo-steptnnsfmtion.

Microstnrctrrres tithe R-phase. The self-accommodation morphologies of
theR-phase foundintherrlé'filmareessendanydifferentfiomthosereportedfor
bulk binary alloys [53], as shown in Fig. 2-8. The phenomenological theory calculation
carried out by Miyizaki and Wayman [53] showed that the R-phase transition does not
require a lattice invariant shear. Therefore, the llOO} or [110} twin-related plate-like
substructures shown in Fig. 4»52 through 4-54 are R-phase variant grains rather than the
lattice invariant twins. R-phase variants with similar morphology (which was called a
needle-like domain) have been observed in Ti(NiFe) alloys [153] and in TiNi alloys [61].
However, dresevafiantplateswaerepa'tedtohaveanaveragethicknessfaabout450mn
[153lvwhichismorethantentimeslargerthanthatof30nmfoundinthepresentstudy.
Moreover, no well-defined self-accommodation morphology of these plate variants was
foundinthinfoilsmadefrombulkfimiFehnd’TlNialloysinTEMlol, 153].

In the present work, two plate-like R-phase variants combined in a single domain.
Withinasingleaustenitegraimasmanyastlueesuchdomainswaeobserved(Fig.4-51),
with plate-like variants oriented in different directions. The habit planes between variants
are [1001mm [110}32twin planes. By using thedistortion manicesderivedby Miyizaki
andWayman[531,thetotaldistortionmatrixE iscalculatedforthistwo-variantdomainas
fOIbW ,

E = [[2 (EA + 53)
1.0000 0.0000 0.0000

= 0.0000 1.0000 - 0.0047 (4.3.1)
0.0000 - 0.0047 1.0000

103

whereEAandEgarethedistortion matricesofvariantA (correspondingtod'll) habit
plane) and variantB (corresponding to (111) habit plane) respectively. This result shows
thudremmdiswrfimmauixofatwo-vafimtself-accommdafioncombmafimconuim
odytwonmewohnrdafivdymafldreUcompmmsratflthrdiauthuthem
variant combination can effectively reduce the total strain energy associated with the
transformation. Moreover,dleeaimnceofnalhipleself-acconmodafioncombinafionsina
singkarnwnimgrdnmayfmflrannninnaethemndimusoeiawdwithdlek-
phasetransformation. Anodserpossibilityisthattheretwovariantcombinationshsving
differentcrystal orientations, nucleatedatthegrainboundaricsandthen extended intothe
grain interiors on cooling to form the multiple-combination configuration. However, as
mendmedabovedregrainbounduiesseemswconsnainmenucleadmofmuwndte
plates. This may also be true for the nucleation of the R—phase. However, further
observation is suggested toclarify the origin of the multiple-combination configuration. As
a result. it is reasonable to conclude that the two-variant domain can act as a basic self-
accomodationunitoftheR—phase.

1helabydnth-fikeR—phasedomainsobservedindre8LfilmasshowninFrgA—44
haveoneoftheirboundariesparallelto(110)gzplane. ‘I‘hisfactindicatesthattwosetsof
domainsare(110) twin-related. lfeach'setofdomaincontainsonlyoneR-phasevariant.
meodrasetofdaminbmmdmiesshouldbflmflmphncsmdingwdwminrdadms
explainedin Fig. 2-7. Otherwise, they canbe (011)gzor(101)32. Theformercasehas
been demonstrated in Fig. 4—53 in which a set of R-phase variants grow in two directions
andaccommndatewithmodwrsetofvarimmdlmughtwoconjugatetwinsmfmmme
labyrinth-likecombinationwhichisillusu'atedinfigme4—S9. Intheupperpartofthesame
picture, another variant combination can be observed, here the plate-like variants are
maintained Thisindicatesthatatwo-variantcombinntionisabasic way fortheR-phaae
varianutoself-acconnnodate.

1

Microstntctttres of Monoclinic Mar-21412 . The martensite variants observed
' intheSLfilmshaveadiameteroflessthen100nm,whichismuchsmallerthanthoseof
about 1 trmindiamewr.observedinthe PML-l6 films. TheTEM study conducted by
Xie et al. [61] showed that the coarse Ti3Ni4 precipitates favor of the nucleation of
martensite variants, but constrain the martensite variants from further growth.
Consequently, themartensite variantshadasimilarsizeto thatofthe precipitates. In the
present study, (NiCuhTi precipitates are proposed to play a similar role with respect to
thermoelasdcnansfamationastheriaNiaprecipiuteshavebeenshowntodo. Therefore,
stmllmartensitevariantsizeobservedin-the SLfilms,asshowninFiglne4-43and4—44,
istobeexpectedon the basis ofthe size ofthe (NiCu)2_Ti precipitates found. On the other
hand, without the interference of the semicoherent precipitates, martensite variants
observedintherr16filmsgrowcontinuouslyandattainalargu'size.

Type-ral?) twinning wastheonly twinningmodefoundinbothSLandPML-w
fihnstogivethelatticeinvariantshearandself-accommodation. Thecalallationconducted
by Knowles and Smith [60], using phenomenological theory, showed that both Type-I
(lli) twinning and Type-n [011] twinning can perform the lattice invariant shear in TiNi
alloys. Mostofthecrystallographicdatameasmedfrombulksinglecrystals showedgood
agreement with this calculation using the latter as the lattice invariant shear. However,
Type I (Hi) twinning has been widely observed in TiNi alloys andrelated ternaries in
TEM studies [54, 59]. Matsumoto er al. [63] proposed that the appearance of Type-I
(ll'f)twinswascausedbyathinfoileffect. lfthisisthecase,thecrystallographicdata
basedonTypeII[011]twinningacquiredfrombulkalloysmaynolongerbeofvaluefor
applicationtothinfilms. .

Table 4-5 is the crystallographic data deduced by Knowles and Smith [60] for
(ll'f) twinning as the lattice invariant shear. The phenomenological theory calculation
yields two sets of solutions, labeled Solution-l and Solution-2 respectively. The ”total
distortionmatrix”,E,asdescribedinsection 2.1. hencecanbeexpressedas

103

a - (1.x) hr, +XM2, (4.3.2)
whaexisthethicknessmdoofthemajatwinfi‘acfiomandulanduzuethedimfim
matricestowhichthetwotwinregions,1and2,aresubjected,aslistedinTable4-S.
Undameucimumsnmtheumformanonsuainu,canbeevuuamdbycnnsidaauou
effigz-lwhichindicatesthndemndtofthemsfamanononaumtvecnrnormnto
thehabitplanert. Fromthefiguref,

at-rt'm ' . (4.3.3)

wherert'sErr. Thedhectionofshearandthen'ansformationshearsu'ainaretherefore
obtainedas

at, = u' - (3' or ) is (4.3.4)

as, = I n' - (n' '1!) rs l (4.3.5)

The results for habit-plane. variants l and 2 are listed in Table 4-5, revealing that the
transformation strain ofabout0.ll which is smallerthan thatof0.13reportedfor'l‘ype II
twins [63]. This result predicts that NiTi alloys may display less maximum shape strain
whenappliedmmefamofammmbecwwofmisdiffaencemmeprefamdtwinning
mode. The orientation dependence of the transformation strain is also calculated and
plottedinFigtue4»60forsolution l. 'I‘heresultforsolutionZisalmostthesameasthatof
solution 1 becausethe sheardirection fortheformer (rug)isroughly paralleltoru andm
isalmostparallel tong. Thedirection which yieldsthermximumSchmidfactorisroughly
in [5 35132, as shown in Figure 460, which is slightly different from that in {251132 for
Type-II twinning [154]. However, in both case the <100>32 directions are less favorable
fortheapplicationsrequiringshapemmoryefi'ectandsuperelasdcity. Thislmowledgecan
be important in improving both shape memory effect and superelastic effect in
polycrystalline TiNi films, if a texture is developed during deposition or during
crystallization annealing.

106

433. Trwonnarloar in PML-9 Films

Thissectiondescribesthetbermoelastictransformatiou behavioroftheannealed
m9 films with overall composition of Ti51,oNi44,4Cu4,5. A wide dispersion of the
TizNipecipitateswerefoundinthegraininteriorsofthis'I‘i-richfilm,asmentionedin
aection4.2.3. Anorthorhombicmartensiteisfoundtobeanintermediatephasedtuingthe
martensificuansformafioninwhichmemonocfinicphaseissdnmefindpmducton
cooling. Theapparanceoftheorthorhombicphasechangesdiemicrostrncnueofthe
monoclinicphase,ascomparedtothatdescribedinsection4.3.2.

PML-9 Films: EIectricaIResr’stiviry andDSC Results. Martensite transformation
characteristics for the films, annealed at 923 K for one hour at P226x10'3 Pa, are
shown in Figure 4-61 and Figure 4.62, respectively. Electrical resistivity data are plotted
inFigure4-61 whichrevenluansformauonbehaviorsimilarinnanuetothatofthesputter-
tar'getalloy.asshowninFig.4-37. 'IhenominalM..Mf,A,andAt-temperattuesarethus
determined to be 305 K. 225K. 247K and 322K respectively, giving transition temperature
intervalsofBOKoncoolingand75Konheating whichareaboutfourtimeslargerthan
that of the sputter-target alloy. However, the transformation hysteresis of 17 K is
essentiallythesame. Figure4—62 showsdatafromDSCscansforthefilmannealedunder
the same conditions which indicate that the martensitic transition begins at 297 x on
coohngmdathehnishtempcehuemheadngis312Kmabmh9degreesbelowthe
corresponding transitions obtained from resistivity measurements. The widths ofendo—
andexothermalpeaksinthecalorimeterscansareonlyabout 12Kelvins.whichisthesame
amuseofthesputter-target alloy,.andtheaverage transformation enthalpyobtainedfrom
the integral of both peaks is 482 ”male. The DSC curves also. yielded similar
transformation hysteresis of 15 K. (Table 4-6 summarizes the apparent transformation
temperatures and properties associated with DSC and electrical resistivity measurements.
In addition. rim-9 films annealed at ”823 K and .923 K, respectively, for one hour at

107
P-2.6x104Pa,werestudiedbyDSC,andtheresultsarealsotabulatedinTable4-6.

1hefilnnannealedat923KwithP=26x103Paat923KwithP=2.6x104Paandat
823 K with P - 2.6x10" Pa, respectively; have similar transformation temperatures (as
determinedbyDSC),butwithdifi’erententhalpies. The823Kannealedfilmyieldsan
avu'agenansformadonenthalpytudceasgreatuthuofthefilmmnealedu923xn
P-2.6x10-3Pa. ForthcfilmsannealedatlhesametempmhneOZBKLhighervacuum
resultsinahigherheatoftransformation. "

”(1,79 Filnrs: x-ray Diaractien Resultr. As was mentioned, the cold-stage
X-ray diffraction results for the PML-16 films annealed at 923 K for one hour at
P - 2.6x10'4Pa yield a Mf temperature which was about 50 K lower than that
deunnnedfiomDSC. Cold-stageX-raydifi'ractionpatternswerealsotakenforthera-
9 films annealed under various conditions, and are shown in Figures 4-63 through
Figure4-65. Figlue4-63showsdiffractionpatternsforafilmannealedat923Kforone
houratPa2.6x10'3Pa. Only'l‘igNi andB2peaks‘appearat323K,thatis.abovethe
nominalAfeumeratlneasdeterminedbytheDSCdata. TheBlelOIpeakisnotpresent
at143Kandhasthusvanishedatsometemperaturebelow222K. However,martensite
mdamwniteueseenbcoexinatmxandmmdthoughmeDSCdammuflindicate
that the martensite transition had finished at 285 K, The low temperature (T a 143K)
X-ray spectrum also shows that the martensite has a monoclinic unit cell with
a=0.292nm,b=0.411nm,c=0.462nm,andfl=96°. Examinationofthespectra
takenat268Kand222Krevealsashiftof(110)32peaktowardhighertwo—thetavalue.
However, this shift must be caused by mechanical drifi dining cooling, since a similar
amountofshiftisalsoobservableinthe(002)leadpeak. AnabnormalincreaseinX-ray
intensityatatwo-tltetaanglearourtd42.5°isalsoobservedatzosxwhichmayresultfrom

 

6halalnnbpealrshowainthespectrathroughontisfrotnathinleadplateonwlticltthefilrnswere

108
dwappeuanceofme(020)aduhonfiicpak7.1hisinauwofimensityknmcwudby
anoverlapof(110)32and(m0)..m.peaks,becauseasimilarsituationdoesnot occur
between(110)32and(11-1)mpealts,thoughinthe‘latterpair,thepeaksaremoreclosely
spaced. 1hemgulardifl’aencebuweenfln(ll0)nzpcaklnddre(OQO)amahombicpeak
islargerthanexpected. However,theexactpositionofthe(020)orthorhombicpeakis
diffictfltmdetaminefiomthespecmmduewtheinterferenceofthe(110)ngand
(020)....ka

Figme4~64and4~65uetheX-raydifi’ractionpanernsfordtefilmsannealedat
923Kandszsltrespectively,foronehour.atalowerpresstueofP:-2.oxlo41>a(and
hence, presumably, under. conditions less prone to oxidation). In Figure 4-64, the
diffi'acfionpauannkenu323KshowsmesamefeaunesumainFig.456,whaeasthe
intensity of (110)32pealtat221 Kis solowthatit is barely distinguishable. Inother
words, the film annealed at lower pressure acquired a higher Mf temperature. The
coexismweoftheormmmbicandmonocfinicpeaksuzmxmdicawsdiefilmwu
undergoing an orthorhombic-to—monoclinic transformation in this interval.

’I'hediffractionpattemsofthe823Kannealedfilm(atP-2.6x104Pa)isshown
inFigure4-66. ‘IheinwnsitiesoffigNipeaksaresignificantlylowerthanthoseinthe
patterrnof923Kannealedfilms. AbroadpeakappearingatflOKisprobablycontainsa
snongathahombicpeahamodaatemonochnicpeakandaweakBZme.1Iussnong
aduhombicpeakindicatesdmahrgefiacdmofausterutehadflmadymsfmmedw
mthahombicmutensite,whaeasdwsecondsmgeofnansidmwassdflinprogresa Asa
consequence, the relations of the 'true' PML-9 Mf ternperannes determined by X-ray
diffractioncanbeexpreasedas:

Mf(szsltatlo‘Ps)>Mf(923xatlo"ra)>Mf(923ltrtlo"ra)o

 

71kmfluhanicphnisfoundmbeminwmedimephaxofmeBZ-w-mflnkmahe
transfumafion.haviaglatticepamneterofn-O.291mb-OAZSarnandc:0.340-,mdwillbe
Whinfolbwirusubaections.

109

Pigmet—oouauanummonelecuonnuaographandacarespondingdiffracnon
patterntakenat 160 Kelvinsfromasample annealedat923 Kforonehourat
P-2.6x10'3Pa. ThoughthisishelowthenominaletemperamreobtainedfromDSC
utdresisuvitymeumemeummaduuaunmitediffracumspouinthelllolrnneun
couldstillbeobserved. Adarkfieldhnagefiommemmspmuindicatedbyarrowin
the diffraction pattern, shows that-this retained austenite is distributed between the
Myriam Theaustenindcnninsaredisuibutedthroughouttheoriginalaumnite
gra'mandhavediametaoflOtoanm.

Retained Austenite in the FML Films. The apparent transformation
wmpuauuesofdlePWfilmsusociawdwimthevmiommeasmemenmmcludingwld-
mgeX-nydiffiacdeSQandresisfifiqcmvauesummarizedinTable4-7. Allthe
Pmfilmsdwwsomemtainedmsteniteatmewmpaanuebebwdterpoimdetamined
byDSC. lbwevamflthereninedaustenitedoesfinanymnsfamtomutensiteatlower
tempa'atures. InodrerwadadieBZaustenic-to-martensiteuansformationocclusintwo
stages. Inthefirstsmymncoolingmfiacdonofflreausteniteapparenflymsfamsto
martensite within a short temperature interval of about 20 K. It is this 'normal'
transformationwhoseexothermappearsin‘theDSCscans. Thefollowingstepisamcre
sluggishprocesswhichproceeds'overaninmalofcoolingwhichcanbeaslargeas100
it This slower ( or so called 'microscopic") transformation probably produces osc
peaks which are too broad to be readily discernible. On heating, the 'microscopic'
transformation occurs first, followed the 'na'mal' transition. The enthalpies recorded from
the DSC exo- and endotherm are therefore represent only that involving the first stage
transformation. Thusthelowu'ansformadonendialpygenerallyassociawswidllow'uue'
vaalue.( the Mftemperature. determined by eitherresistivity curvesorcold-stage X-ray
drffiactron‘ ' ). Inaddidonflemminedaustenitecanalsoexplainwhytheelecuicdmsisdvity

 

WW'W‘MWNWMWWJMMWN
maturiticmsfa'amtionistnosrnalltobeidenlifiedbyDSC.

110
cmesumanyyieldannwhlowaMfandAgvduesdthoughdreMgandAftempmnnes

coincidewiththeDSCdatarelatively well.

lasitu TitMobservationinthePWlbfilmsannealedattBKfororrehoruat
P-2.6x10'4Paindicatedthatthecentralregionsoftheaustenite grainshadhigher
transformationt‘emperattues. ‘I‘hishasbeenshowninFig4—57and4-58inwhichthe
heuinedmumdephuwaehcawdinmemmgionofdteaumnicgraimafume
samplewasheatedbyconvergingthe electronbeam. Accordingtotheopticalmicroscopy
study of the martensitic transformation in a Ag55Cd45 alloy carried out by Tong and
Waynnn[14],memanensiwphmspmducedfirstmdlefmwarduansfmdm(amtemw
-tmartensite)isthelasttoundergothereversetransformation. Accordingly,micrograph
sltown'inFig4-57and4-58indicatedmartensiteineachaustenitegrainstartedtonucleate
atthecenterofthegrainoncooling. However, study ofmartensite nucleationinaAqu
alloy, conducted by Ferraglio‘ and Multherjee [155], found that the nucleation of the
mamoehsficmmensimwashemrogenwusandgrainbomtdafieswaemeofmefavaim
heterogeneous nucleation sites. Insitu 'I'EMobservationsinaNi-ricthNialloywerealso
reputedmnshowedmemmennehmappeuingprefamuanyagrainboundmieamdn
the matrix-inclusion interfaces [61]. As a consequence, the absence of the 'retained'
marter’rsiteappearinginornearthegrainboundariesinthel’hfls-mfilmscannotbe
exphimdunlessdieneu-bwndmymgionhasabwermnsfamadmmaaunesmm
centralregion. ' _

On the basis of the in situ TEM observation in the PML-16 films, a possible
exphmfimforapudmofmeausteniteuansfanfingwmartensiteubwawmpmuues
invavuunmumdepleumagrainboundanesanheenufacesduemmeptecipimuonof
Tumor Examining the data listed in Table 447, we find that both the PML-9 and
PML-16filmsannealedat923 Katapressureof 2.6x10'3Pahavethelowest'true' Mf
temperatures and enthalpy values, respectively. On the contrary, the film annealed at
823KatP-2.6x10'4Pahavethehighesttruervalues. Flu'thermore,X-raydiffraction

. 111
results, showninFig.4—30 forthe?ML-16filmsandinFig.4—36forPML-9films

respectively, reveal that increase of volume fraction of the TigN i type phase (including
TiaNi30,)correspondstodecreaseofthetruervalueinthePMLfilms. Shugoetal.
[filhsmpandmumeaddifimofoxygeninTlNidbysdecmsedmeiruansfmdm
mebyfanimmeTuNthfidewhichdeplaedminBZmauk
Aneaikrrepafilfildnindicandmmefimitypepecipiummbembifiaedby
nitrogen. Wmhfmmafimoffi-richprecipitatesingr'ainbmmdmiesandatfiee
surfaces of the PML-16 films substantially decreases the titanium content in the
surrounding area, for which the Ms temperature is subsequently depressed. The situation
maybeworsemthePMIr9filmsbecauseoftheprecipitadonoftheleNiphaseinthe
grainimaiawhichreudninfmthacomposiumpaunbadomespechflysimemis
aslowdifi'userintheBZmatrix. 'I'heelectronmicrographshowninFigA—éérevealsthat
maflauswnimdomainswimasiuoflo-lmnmmedisuibutedthmughuutheaiginfl
austenitegraininthePML-Slfilms. Itisthusproposedthatafractionofaustenitenot
having suffered a change of composition transforms to martensite in the first stage
('normal') transition, and the remaining austenite, which has different transformation
temperature due to a lower titanium content, transformd to martensite at a lower
tenureTature. Accordingtotheabovediscussion.thePML—9film.annealedat9231(at
P-2.6x10'3 Paisexpectedtohavethelargest amountofretainedausteniteandthe
lowea'uue'Mfmpmmwhichcoincideswidltheexperimentalreslms
Furthermore, the precipitation of the TigNi/TiaNigox particles produced
compressive stress fields in the matrix which may further constrain the martensitic
transformation on cooling. Hedayat et al. [157] reported that the residual stresses
umciatedwithmannededcarboncoadnngrNiinuodwedasmfaceconndntwhich
postponedtheforwardtransformationoncooling. 'I'hatis,aportionofalloyclosetothe
TiNi/Cinterfacehadlowertransformationtemperatlues. Asaconsequence.theretained

112
austeniteobsavedinmepresentsmdymaynmhfiomefieascombinMgfltedmium
depletionandthestressconstraint.

. MphwomonofreuinedwsteniteinbulkfiNimoyshasnmbcenprevioudy
Wmmpanoncmmumtedbyoaygendedwingmuumulyismnuano
besignificant. However,datarecenuyreportedinmalnubynuschetnl.lllolshowed
MincmasinganneafingtanperanueandmnealingdnnnsulwdinadecmseofMg
WMaWdWWMAT-Mg-Mfdmto
theprecipitationof’l‘izNiinthegrainboundarieslllG]. hammereresultsuntkrline
diehnportanceofstringentvacuumcondifionforannealinngNidlinfilm.

PML-9 Films: Transmission Electron Microscopy Results TEM study showed
thatasmallfractionoftheBZausteniteinaPms-9sampleannealedat923Kforonehour
upsaolePahadureadyumsfanedtomamnsiteuambientmrmmbudere
isnoindicationofthepresenceoftheR-phaseorlfldiffractionspots. Interestingly,the
ebcumdimacdmpammkenumsxindicatemmismomwmpaammsheh
orthorhombicratherthanthemoreusual monoclinic. Figure 4~67isatypical example.
showingadarkfieldimageofmartensite plates which have developedintheazmauix,
andacorresponding diffractionpatternsintbe [010W0111m direction. “101]...
diffraction pattern taken from martensite nearby is also shown in Fig. 4-67(c). The angle
between (100) and (001) martensite planes is90° which proves themartensite is not
monoclinic. The lattice parameters of orthorhombic cell, calculated from electron
diffractionpattemsinFig.4-67(b),usingthematrixspotsasacalibrationstandard.are
080.291 nm, b=0.425 nm and c=0.450 nm. The orientation relationship between
thisorthorhombic martensiteandBZ austenitecanbeinterpretedas [0101mm // [011132
and (202),...onearparalleltoaingz. Themartensite platesareauoriented roughly
parallel to (3 52m . However, close examination shows that the plates, which have
thickrresaesoflStoSOnmdonotextendinasimplelateralfashion. Instead,theygrow
zigzaggedly, first along (3 52m and then (fills: planes, alternating every 100 to 300

113
nm,asillustratedinFigure4—67(d). Asaresult.theaustenite—martensiteinterfacesare

composedof(3§2)32 and (2mm steps.

Whenthesamplewasdippedinliquidniuogenandthenwarmedbacktoroom
Wthemnnebecamefunyamahombic-msmnsiw,uobservedinthe
electronmicroscope. Themorphologyofa'thahombicmarmitedifi'eredfromthatofthe
monocfiniennrmsiwobmcdintheSLandpmzlofimmummuansfmdon
defecnnchunvinniugandsacldngfndnwaefoundinndetbemmensitevarimu.
This untwinned martensite has been previously reported by Tadaki etal. [71] and by
Moberly and Melton [7] in orthorhombic martensites. Furthermore, the orthorhombic
vtiamshaveaavaagenzewhidlislessmanonewnthofdreauswrucgrainslze.

The small variant size allows an observation of the self-accommodation
mphologyoftheuthahombicmuwnsitembemrdataken,whichhasnmmviously
beenpossibleinbulkmaterialswithlargegrainsize. Amulti-variantcombinationisshown
inFigure4—68. ThebrightfieldmicrographinFig.4-68revealsthesizeoforiginal
austenitegrainofabout3.5|.tm. 'I‘hediffractionpatternsshowninFixJ-“(bhhrough
468(d) were taken from different regions, labeled 3, C and D respectively in
Fig.4-68(a),atthesametiltingangle. 'I'hedifiractionpatternstakenfromregionB andC
contain two (111)...“m twin-related [110].“m zones as shown in Fig. 468(b) and (c)
respectively. The angle between the [1T0] and [001] reciprocal vectors is 90 degrees
whichagainconfirmsthemartensiteisorthorhombic. Inaddidontllediffr‘actionpatternin
fig.468(d)mvedsthnmgionbconminsmmemmsiwmhavingacommm
axisalongthellOlh...direction,witha+/.1120“anglebetweenthem. Figure moist
dukfieldhnageshowingfllatmgimflcmminstwoseuofvafiannwhichaccommodate
themselves with respect to the (111).“... twin plane. In region C the variant
parallelogram having the darkest tone and the brightest tone are variant c andd,
respectively, which correspond to the diffraction patterns shown in Fig. 468(0). These

two variants also have a junction plane on the (111)”;m twin plane. However, the

114
neighboringvariantswithintermediatecontrasthavedifi'esent, butunidentifiedorientations.
Inotherwords, areaC consists'oftwosetsoftwo—variantcmbinations. Thedarkfield
imageinWWflsnkenuhighmagnificaMdbwsclosaennfinafimofthednee-
variant combination. These three variants are twin related relative to three (111).“m
planes, with 120° anglebetween them, andarereferredtovariante,f andg respectively.
Mamtnmmmncvmmmm‘mnMcgninwimtdimof
only about 3.5 pm.

In addition to the predominant multi-variant self-accommodation morphology, a
two-variant combination was also sometimes observed. Figure 4-69(a) thrbugh (c) show a
bright field’ image, its corresponding diffraction pattern, and a dark field image,
respectively. The diffraction pattern in Fig. 469(b) is similar to that in Fig. 4-68(e), and
mveakumdeongindausmnitegminwnmimukastuueemnsitemiamsrefaredm
variant a, b and c respectively. Figure 469(c) is a dark field image taken from the
Immune spot of variant a showing the variant distribution and morphology. Most grains
of variant 0 and variant 5 have a shape of parallelogram bounded by two (111),“,
planes, and two (131)...“lo planes or two (101% planes, whereas grains of variant c
andneighboringvariantsofo andb areboundedbytwosetsofflllmplanes. Figure
470musuatestheaystauograplucnanneofdlesennnensitepmalldognm. Asshownin
Fig 4-70, variants n, b andc accommodated themselves through three (111),“, twin
planes. Ontheotherhandthemajorityofvariantsaandb formaseIf-accommodation
pattern which is divided by parallel (111W twin planes into several divisions. Each
division contains interchanged variants a and 5 having junction planes of
(101),¢J(131)m.

Occasionally, however, further variants in the morphology of orthorhombic
martensitewereobservedinthePMIJfilms. Figure4-71 isaelectmnmicrographandits
corresponding diffraction pattern of(lll)...._, twins found in a thin-plate formsirm'larto
the internally-mined monoclinic martensite in binary alloy, with about 10 to 20 nm

115
thickness. Thesphericalparticles showninthebrightfieldmicrographaretheTTzNi

precipitates. tnaddidon,(011)....twinnedmutensiteispresentandshowninrigtue4-
72. The bright field micrograph shows that two martensite variants form thick plates
acrosnheaustenitegninintheupperleftpart(regiona). Fig.4-72(b)istheassociated
difhacnnnpamnkennomregionashommnthesethickpntesne(ou)mtwia-
related Thewedge-hhevariantrshowuinthelowernghtpart(tegionC)ue(01T).,..,
twin-relatedscccrdingtomediffractionpanernshowninlrrgtemc). Sincevariantst
andc,uhbeledinthemicmgraphanddifiracfionpams,amnmminrdatedwith
respecttoeither(011),....,or(01'i),,,.., p1anes,Thevariantc which is thoughttobe
formed afterward, hence must taper before reaching variant b, and then results in the
Mae-likens:
Asatesamplewascooiedweubelowthememperauuedetanunedbyresisdvity
measmementmouofuteoriginnuthahombicvnimtmmphologywassunmainuined
thottghlaerTEMstudyshowedthatthemartensitewasmonoclinic. However,therelative
size of the diverse variants did change, i.e., there occurred some intervariant boundary
migration. Pigme4—73isaelecuonmicrographandassociateddiffractionpanemofan
aigimlumcdnyainnkenfiomadimcflmappoximatelypuaflelwllmbnlfim
showing(011)twin-relatedvariantsinaself-accommodationunit. Figure4-74showsan
elecuonnnaographsnddifhncuonpanansnkcnatlwnofdtmemomhnicmiann
whichsnnrenindteaccomdationmmyitologyformedatorthcrhombicstate. However,
mmennesuucnuesinpnteukefmmsneobsavednndethevuimmseeenclosedne:
in the micrograph), indicating that an internal change occurs during the orthorhombic-to-
monoclinictransformation. Figure4-75‘showstheelectronmicrographtitwomonoclinic
martensitevariants. ttscnrrespondingdiffractionpntterucontainstwo(lli).,.,twin-
related [110]...“o zone. Each set of patterns shows streaksin [001]...” reciprocal
direction. Mmeaksuenmsymmeuicdwithrespemwdtemaiandweakand
elongatedspotscanbeidentifiedinthestreaks,asindicated by arrows. Thereforethe

. 116
streakdo not result from stacking-faults, but from thin (001) internal twins. Furthermore,

themonocfinicvarhnnshowmirreguluphtemaphologywhichisdifiaentfiomdle
wendesnedpmmpuallelognmsofmemthmhombicvmimndescnbedmrig.m
through 4.72, indicating rearrangement of variant boundaries may occur during the
transformation Ahighmagnificationmicrographanditsassociateddiffractionshownin
Figme4-76revealthese(ml)twmshaveaspacingofonly4nm. FurtherTEMstudy
slmwsthnmePMIAfilmhavingdifieremmneahngcondidomJeJnnealednmx
for'onehomatP-2.6x10'3Paandat823KforonehouratP-2.6x10"l’l.havethe
samefeaunesofthemartensitemicrostructmethatareshownabove.

' Tranvbnnatiou properties ofthe mus Film. X-ray diffraction and TEM
results show that the PML-9 films, with about 5 at % copper content, undergo
thermoelastictransformationsinasequenceofBZt-tMot-tMu. Theappearanceof
athmhonrbicmmnsinummermediatephasedrmngmemamoeusucuansfamaumin
PML-9mm demonsuatestltecornplexityoffliisalloysysm Ternaryauoyswithabout
lOu‘ltCuaddidonwaereportedmundergoatwostepmsformafionwhichwinbe
referredtoaBZHMoHMuuansformationl7,73,7Ba]. IncreaseofOucontenttolS
at%orhigher resultsinaBZ HMO transformation [7, 73]. However, themartensite
mnueandthemsfmmauonsequmceinmdemeandhighcoppaeonminmgauoys
werefoundtobeprocessdependent. AspointedoutbyTsujiandNomuaU3a],annealing
aftercoldworkcanstabilizetheorthorhombicphaseatlowtemperature. Ontheother
hand,mu-castafloywith25u%Cuaddifionyieldedanmnocfinicmartensiteon
cooling[7].

Severalpossiblemasonsmayconnibutetodleappemanceofthemhorhombic
phaaeinthisS‘b-Cufilms. Fnstanenrichmentofcopperinthematrixmayresultfrom
the precipitation ot’TigNi-type Phases. ifthe solubility ofcopper in TizNi decreases at
lowertemperature. However,theincreaseofcopperinthematrixisestimatedtobeless

117 - .
TizNi [44], which seems insufficient for a Bl

than2at‘b,forzemsolubilityofcoppain
thHMutransition.

Amdrapossibflhyisdratacmrpressivesuessfiddresulnfiomdieprecipimdmof
TuNiphasewhichmaysuppteuthe’R—phaseandtheBZHMumsfmdonsrdadve
totheBZthtramfamation. ,Fromamermodynamicpointotview,meadditionot
mppainTtNiin'mbsdmfingforNihasmetendencymdecreasethefieeenergyofme
adrahombicphaseandminauseflnfieeenagyofR-phanrdadvewmuofmomcfinic
one[‘7]. luvuiadonofmerdadvefieeenagylevdsamgmesethreephasesresuluin
a change of transformation sequence from 32 H R H MM to B2 H MM to
82 H Mo HMMandfinallytoBZ HMowithaninereaseofcoppercontent.

Inaddifiontocoppercontenuthesuessisanothervafiableindtedtermoehsdc
system,aswasmentionedinsection2.l. 'l‘hereportontheabilityofannealingaftercold
wakmstabifiudnmhainmbicphasdnalcoimideswidrdnobsavadmmutheback
stress of dislocations can effectively suppress the 32 —) MM martensite transformation
[63]. Furthermore, a compressive stress is not favorable to R-phase transformation
[52, 67]. As a result, the existence of the compressive stress fields around the TizN i
precipitates are expected to increase the free energies of the R-phase and monoclinic
martensimuidresultsinaBZHMoHMMtransformation.

Crystallographic Calculation of the 82 HOnhorhonrbt'c Trarm‘bmratiou. The
orthorhombic martensite found in the present study is transformed from austenite without
mvdveunmdahmcemmtshwmoincidingwimmeobsavafionmpatedbyhdald
andWayman [71],andbyMoberlyandMelton[7]. Theuntwinned nnrtensite indicates
thatanundistorted habitplane already exists betweentheBZ austeniteandorthorhombic
martensite without the assistance of lattice invariant shear. According to the
phenomenological theory derived by Wechsler, Lieberman and Read (WLR theory) [25],
meneceasuyandsuffidedeonfaaphneofmdismrdmmexistismuomofme
threepr'inciplelatticedistortiorts(m)isunity,andnj>1andm<lrespectively. Aswas

. 118
point out by Wechsler er al. [25], the smallest lattice distortion for a 132 austenite to

transform homogeneously to an orthorhombic phase is a Rain distortion such as is
illustratedinFigure4-77. InFigme4—77,af.c.t.cellisdelineatedinanuntransformed82
structure. Thecubic axis ayatemiaa,j,rt),anddreprincipie axes (l',j',k')forthe
orthorhombiccellisobtainedby45°rotationaboutthek axis. 'l‘hematrixwhichdeacribe
thisdisurfionwinbediagonalinthe(i',f,k')axissyswmandisexpressedu

,- b .
HAM) 0. 0
0 8 c
r o ""72? o . (4.3.6)
0 0 "1(1)
. a, .

 

 

whereaoisthelatticeparameterofausteniteanda,b,andcarethelatticeparametersofme
orthorhombicphase. Indrepresentcamdreprinciplelatticedistortionsarecalculatedtobe
111-0.966, 11220.995andn3: 1.07, respectively, accordingtothelattice parameters
determinedbyelectrondiffraction. 'l‘hesevaluesaresimilartothosecalculatedbyTadaki
andWayman [71] ina 19 at% Cu a11oy'(n,eo.9so, 112:0.991 andms 1.05). Asa
consequence, the principle distortions may be regarded roughly as me 1, 112 ~1 and
n3>l,andthusmeettherequirementsoftheWLRtheory. Atheoreticalcalculationwas
nndmtakmondemmmpdonmmbafianwhichyietdaadistmdonmanixexpretaed

as
l. 0000 0 0
I": 0 1.0687 0 . (4.3.7)
0 0 0.9659

Sinceallthevectorslyinginthehabitplanekeepthesamelengthaftertransformatioruthat

is,

119
(r' )2 - (1"? )2 - (r )3. (43-8)

arelationbetweenr'=(x',y',z')andr =(x,y,z) iswrittenas
x"+y"+z'? = (111211 + m2 + 113121)

=£+y+fl «a»
which yields
y/z = :l: K (4.3.10)
where
I-n’
K=i '57-'17 =i'0.8508. (4.3.11)

Two vectors in the habit plane are arbitrarily chosen to be (0, 4(2),“) and (1:2, {22.22)
andtheircrossproductgivesthedirectionofmehabitplanenormaltobe

o o
rt: 2 1 = 0.7787 . (4.3.12)
JI+K :tK $06274

Amfionmauixwhichuansfmthedefmmadonmauixmaxesafigmdabngmecubic

axescanbewrittenas

I .
73 0
I

73 0. (4.3.13)

0 I

I":

°ci~ct~

 

 

I
1

Therefore, the normal of the habit plane in the austenite system is written as
n = [0.5506, 0.5506, +I-0.6274], which is perpendicular to [110]” and is 3.6° from
[111]or[llT]nz,depuidingondtesignovaaluechosen.

Though no lattice invariant 81103301011111!“ in the austenite—to-orthorhombic
mnsformadon,arigidbodymtadonmaybesdflnecessarymmaintainthelatdce
coherency. Bycxaminmgtheiatticediatmtionmanixr,itiaronndthatonrymevector,
[111)].0nthehsbitplaneisnotmtatedafterdistcrtion. Inotherworduhellmlvectorisa
mtationaairr..whichirexpmttedaauiolmvectorinthecnhicryatcm Consequently,
atiltangleof-2.52°isfoundbyexaminingtheanglebetweenavectnrrsrrxr. andthe
vectorr'ur'r. Aanglebetweenthe(lm)32and(100)planeoforthorhombicphase
rhownmrigmwmmeamedmheaéwhichiamnoodameemmtwimmemtcnmd
value. mmmmuscanbewdttcninmofdiccflculfledfiltlnfleOu

x,’(I-cos9)+cos0 x,x2(1-cos0)-x,sin0 x,x,(1-cos0)+x,sin0
¢= x,x,(I-cosa)+x,sin0 x,’(I-cos0)+cos0 x,x,(I-cos0)-x,sin0
x,x,(1 —cos0)—x,sin0 x,x,(1-cosO)+x,siu0 x,’(I—cos0)+cas0

0.9995 - 4.834 x 10" 0.03111
:- -4.834 x104 0.9995 0.03111 . (4.3.14)
-0.03111 -0.03111 0.9990

'l‘hetotaldistnrtionnntriinsthuscalculatedas

E I 017' '1'”‘
1.026 0.02631 0.02998
I 0.02631 1.026 0.02998 (4.3.15)
-0.03278 -0. 03278 0.9626

A vector in the direction It will thus be deformed during transformation, and the
deformtiondirectionandtheammtofshapestrainisyieldedby
0.5305

as = n' - u = Err - rs = 0. 0.5305 . (4.3.16)
—0.6611

121

Thetansfamadonsuainintheauswnite-m-adnhombicuansfanndonismfla
compared thatof 11 to 13 % in the austenite-to—monoclinic martensite transformation,
whichisundasundabbsincediefawudadiahombic-m-nnmcflnicmd'mmadmmay
contributenadditionalfractionofthetntalshapesuain.

Habit Plans and WW afdre Ortharlranrbr'c Mannie. 'lhe
mmmhphnefoundfiomexpuimmmlobsavadonisazignggedphne
composedolelllmand(322)32planeswhichisinrelativelygoodagreementwiththat
of a [0.5506, 0.5506, 0.6274)” (or roughly a (877)”) plane from theoretical
calculations. Sabtn'ieral.[158]carriedoutaniusim 'I'EMstudyontheearlystagesofthe
B2 HMotransformation. 'I'hehabitplanewasdeterminedusingtraceanalysistobea
{334132 plane. They proposed that this [334132 habit plane was composed of [211132
and[011]32stepswith[lelnzstepsizetobellA. Inthepresentstudy,thehabitplane
isobaerveddirectlyin'l‘EMtobecomposedolell]32and (322)32stepswithamuch
largerstepsizeofabout 1mm3wnm.

In the present study, the self-accommodation morphology of the orthorhombic
martensite in TiNi was observed by TEM, for the first time. According to theoretical
cahdafimdtaeuetwdvehabh-phmvafimmdaivingfiomsixhtdcecamspondewes
between 32 austeniteandorthorhombic martensiteasshowninTable 4-8. Thesevariants
canbedividedintofourgroups,ineachoneofwhichthevariantsshareacommon
<111>32pole,andhavea[111]mtwiningrelation between them. Theaverageshape
snahrmauixfoerup-Iiscalcuhtedasanexampletobe

To. £1/3(T1+T3+T3) ‘

1.005, 0.010 —0.010
= 0.010 1.005 -0.010’ . (4.3.17)

-0.010 -0.010 1.005

Themanhahowahommediagonarandshemanamcomponmmuevcrymhindicadng
that this three-variant combination can accomplish efficient for self-accommodation. In

122
facLTEMmdysesalsorevealthuthedtreevafiamcombinadonisoneofthemost

frequentmorphologiesobaervedintheorthorhombicmartensite. Furtheranalysisofthe
diffractionpatterninFig.4-68indicatesthatvariantsa ands aswellasc andd,
appearing as two-variant combinations, have common [111132 and [111132 poles
respectively,ifthevariantss,g andf showninFig.4—68(f)haveacommonpolein
[IIThgdirectiom 1nthiacaac,thcthirdvariantinoroup-nandmaaahownin
Table 4.3, corresponding to[111]m and [111132 polesrespcctively, may not he able to
form in the thin foil sample. Moreover, the average shape strain of these two-variant
combinationanteatrocatcntatedas

0.994 0.030 0.001
= 0.001 1.026 0.001 (4.3.18)
-0.001 0.030 0.994

which is also small compared those found inbinary alloys, using Type-II twinning as the
lattice invariant shear[23]. Therefore, the combination in which three variants join
themselves through three (111).“Io planes is thoughtto be the basic mechanism for self-
accormnodationinorthorhombic martensite. Thetwo-variantcombination thenmaybea
special we in thin unifies
Thetadnningrehdonsbetwearmamnsitevafimucmalsohedemonsuawdusing
stereographic projection shown in Figure 4-78, in which solid circles represent the
calculated habit planes. Variant 1, for example, is found to have Type-l (111)."... (or
(110)“) twinning relation with variant 2 and 3; Type-I (011).“... (or (100)”) with variant
7and 10; andcormormd(010).hwithvariant4,asindicated. 'lhereforeitisnotdifficult
tounderstandthatthe(011)..mtypeself-accommodation morphology showninFig.4-73

123 ‘
isbasicaflyatwo-vafiamcombinadonmchuvuiantland7aland10demonanated

above. Thetotalu'ansformationsu'ainmaa'ixiscalculatedacoordinglyas

Tar -= “2 (Tr +710)

1.026 0 0
- 0 1.026 0.030 . (4.3.19)
0 -—0.033 0.993

Theresultindicawsthamuwnsitevmianmaccommodadngthennelvesdnough (011)type
twinningcaneffectively eliminate shearcomponents, butthe diagonal componentsretmin.
This can explain why this kind of self-accommodation morphology is less commonly
observed intheTEMfoils. '

'lheorientationdependenceofshapesn'ainfa'orthorhomhic martensiteiscalculated
andisshowninFigure4r79. Themaximumshapestrainisyieldedalongadirectionclose
to [110132 for variant 1, whereas both [100132 and [111132 are less favorable. This result
is obviously different from the orientation dependence of shape strain for monoclinic
martensite in which the favorite direction is along either [5 35132 for Type-I twinning or
[221132 for Type-ll twinning. Moreover, the monoclinic martensite in the PML-9 films,
transformed from orthorhombic phase, is found to retain feattu'es of the orthorhombic
variant morphology, and contains (001),...“ transformation twins instead of (111) or
[011] twins. AsaresulnthemonocfinicmarwnsiteindlePMIr9filmsisexpectedtohave
slighdydifferentmechanicalpropertiesfiomthatindlePMLMfilms.

43.4. Additional Discussion: Thin Films and TEM Foils versus Bulk Alloys.
Theismeinmgmdtowedladlediermehsdcuansfmmadmsindlin-filmmdbulk

anoyshavemesannchm'actaisdcsiscrmialinmedevebpmentoleNidunfilms Inthis

section, the thermoelastic transformation characteristics in the Ti(NiCu) films described in

_ 124
thehstthreesecdonsisfmtherdiscussedbasedmthecompafiwnwimbulkafloys

' TWomationsinSLFiInrs. I‘heannealedSLfilmawithcompositionof
TuuNmCugg, underwent two well-separated 32 H R and R H Mu transformations
whichwerefoundinTi(NiCu)alloysforthefirsttime. Asdiscussedinsection4.3.l,this
two-nepuansfamadonisdnfiluinnannemdlatfmmdinagedNi-richbinuydloya The
teammamistwostepnansfamafimhasnmbeenplevioudyobmedinmryflbys
canbeattributedtothelimitedtitaniumsolubility(<1at%)in32phase,havingmorethan
3 at% Cu substituting for Ni, at higher temperature. This indicates that it is more difficult
for bulk ternary alloy to yield a microstructure with fine, coherent or semi-coherent
(NiCu)gTi precipitates distributed uniformly in the matrix. On the other hand. the
polymmphicmyanhradonmacdonmthesmlmsmowsasinglenzphaaemfmmprior
to the decomposition reaction. Therefore, it is proper to conclude that the two-step
mnsfmdminwrnuyanoyfilmsnndufiomspeciflprweuingcharactaisdcsinthin
films, i.e., amorphous -) 32 H 32 + (NiCu)gTi, rather than the ”thin film
characteristics”itself.

Transfomraa'ans in PML-16 Films. In general, the transformation behavior
ofmePM1’16filmscoincideswimthatobsavedinthetanuybulkafloywimsimilu
composition. However, the twinning system which responds to the lattice invariant shear
inmonoclinicmtntensitehaslongbeenunclear. ‘I‘hetr'aceanalysesfor'l‘rNisingleclystals
conducted by several researchers [23, 63] concluded that only Type-II [011] twinning
could act as lattice invariant shear. I’Therefore, theType-I (11 T) twinning which has been
widelyobservedin'I'EMwastreatedasa'thinfoileffect'[63].

IntheprewntwahType-I(111)anasalsoafiequemuansfumadmdefectin
monoclinic martensite observed in TEM. However, TEM foils, about0.1 pm thick, are
still 'thin' when compared to the TiNi films of l to 5 11m thickness, deposited for
applicadononsurfacesuperelaSficcoatings,orforuseasmicroacmators. Itisnotknown

125
yetthatwhetherthelatticeinvariantshearinaSumthickfilmsinvolvesType-Itwinning.
Type-ntwinning,orhoth. Thus,improvementoforn'understandingonthedependenceof
different twinning systems on dimensionality ln monoclinic martensite must await more
thorough'study.

Traitstbrnradom in PML-9 Films. The orthorhombic martensite observed
mrEMfmnmnmdeecwdmmeangmanlnnofsumthiclmeubyx-
raydifiraction. 'I'hustheappearanceoforthorhomhicmarwnsiteinthers-9filmswith
1owCucontentdidnotresultfmmthe'thinfoilenects'. Asdiscussedinsection4.3.3,the
32 HMoHMMmightbecausedbyaninternaISH‘éssconsu‘aintwhichsuppressedthe
32 HMMtlansformationrelativetotheBZ H Moone. Furthermore,theproductphase
of martensitic transformation in Cu-containing butt alloys was also found to be strongly
dependent on processing parameters [7, 73a]. Accordingly, the same argument cited for
duchmgeofmsfamadmbehaviaintheSLfilmscanheusedindwcaseofdlePMlfi
films. 1hatis,a32HMoHMMuansformationresultedfiomtheuniquemiaosmlcnne
found in the low-Cu ram-9 films, rather than their dimensionality.

In addition, the basic self-accommodation morphology of orthorhombic nmtensite
inwhichtlueeadjacentvariantssharedacommon[llllmpoleshouldbeageneralcasein
both thin-film and bulk alloys, since the theoretical calculations yield a total distortion
matrixthatisnumericallyclosetounity. 0ntheotherhand,thetotaldistortionmat1ixfor
the two-variant morphology which involved (011) twinned variants produces the diagonal
components which are not unity; only two out of three shear components vanish. Thus,
the two-variant morphology may exist in thin foils only whose two-dimensional
characteristicscantoleratethisunaccommodateddilatation. Howthickcan'thin'beinthis
context? Itcannotbesaidfromthelimiteddataavailable.

5. CONCLUSIONS

Thepreaentsnldyhubeenconcanedwithtwomajorissueszthedevelopmentof
reliable processing techniques for fabrication of Ti(NiCu) thin films with controlled
compoddomandammoughundastmdingofmephaseuansfamadmcharacwrisdcsof
these films as compared to bulk materials with similar cormosition. Since substantial
dMumdepledmwasohservedinfilmsdeposiwdbyuiodemagneuonspuuaingfioma
TigoNiuCusafloymrgeaasuamyhasbeendevdopedmcompemawthedeumlmshy
addingdrinfinnhmhyusfiodioaflmwfamamuldhyuedmnnedufingdeposim
The extremely thin and closely-spaced Ti-layers (1 nm layers spread at 9 to 16 nm
mmdsmimnnmdmeefiecfivediffusimdisnnoemqtmedfamktehomgeniufim.
Although interdiffusion during the crystallization anneal can create transient regions with
composition favoring precipitation of Ti-rich crystalline phases (such as TigNi) their
thickness would not exceed about 2 nm, which is small compared to a conservative
estimate of the size of a critical GP-mne nucleus (~10 nm). Nucleation of Ti-rich
chhaseswouldthusbedifficuhfiombofiakineficandadramodymnficpdm
ofview,sincehtaaldifiusionoffiwouldberequiredforwhichthaeisnopardcular
driving force. Furthermore, titaniumisknownasaslowdiffuserinhothTiNiandB-Ti.
Results from DSC and electrical resistivity studies showed an increase in the martensite
nansfamadmtempaaunewiminaeasingmerauTicontentwnfimingdlathedumum
contentioBthasehadheenefiwdvelyincreasednhniamTiwuloumdeZmanix
byprecipitadonofTi—fichintermetalhcsdmingthecrystalhzationanneal. Itisthusshown

126

127
that film composition can be controlled by adjusting the ri-layer/alloy-layer thickness ratio,

accordingtoaserm-eumificalrelationshipohtainedbyexperiment

AdditionalanalysesofthesputteringyieldratioofTiandNiinamorphousTrNi
alloys were performed by concurrent argon-ion irradiation of the ion sputter-deposited
films. Theresulunvealedthuthedmniumdeplefimobservedinhothniodemagneuon
mmwmmmsmpfimflymmwwmfidmmsd
titanium by bombardment of energetic neutrals reflected from the sputter-target surface.
TheprefaendflmspudaingofdMumwasenhancedwhenmehomhardingimenugies
werecloaetothesputteringthresholdsofNiandTi. Moreover,thesputtesingyie|dratio.

=Yr/Ymakobecannmesensifivemsmaflvafiadmsinbeamincidewemgleinthue
energyregimes. Theemploymentofconcmrentionbomhardmenttoimprovefilmqualities
isthusseentoproducenon-trivialcomposition-contlolprohlems.

Annealed Ti,((NiCu)1.x films, with copper content of approximately 5 at%, showed
avuidyofmicrostucunesmdmamodasdcuansfmmadonchmuctaisfiawhichhawna
been observed in bulk alloys having similar composition. The as-fabricated films were
amorphous, and underwent a polymorphic reaction during crystallization to form a 32
phase with a supersaturation of Ni or Ti. Such supersannations would be difficult to
achievebyingotmetallurgy,asthesolubilityoftitaniuminthe32phaseisverylimited
(<1at%) at high temperature. During subsequent annealing (beyond that required for
ayaflfimdm),secondphaseprecipimmsfamedinawideandmifmmdispasiom,in
contrast to the case for ingot-metallurgy alloys where troublesome grain-boundary
precipitation is often observed. The precipitates generally showed specific lattice
orientationsandlorlatticecoherencyrelationships withthe32matrix. Asaresulnthe
mamoehsdcumsfumadmswaeafiectedbyinmalsuessesreafldngfiomthehtfice
misfn strains between the precipitates and the 32 matrix.

128

Self-accommodation morphologies for R-phase, monoclinic martensites and
orthorhombic martensites were studied by TEM. Pairs of plate-like R-phase variants
alternatedtoform self-accommodateddomainswithalamellarstructme. Unlikethefour-
vuhntcombimdonobservedinbulleNianoyadlemmldismrdonmauixfametwo-
variantcomplexhadtwonon-zeroshearcomponents, whichmaybeallowedonlyinthe
low-dimensional geometry of thin films. Similarly, this low-dimensional geometry may
alsoberesponsibleforthe appearanceofmi) lattice invariant twinning, the large variant
mandmem-definedseH-wcommdadmmuphdogyobsavedhaefadlenmmcfimc
martensite. Accordingly,themechanicalpropertiesofthemonocfinicmanensiteinthin
filmsmaybeexpectedtobedifferentflomthoseofbulkalloys. Ontheotherhand,the
majority of orthorhombic martensite variants were found to form as parallelograms
boundedby [llllntwinplanea Threevariants, which sharedacommon [111)mpole,
formedwell-definedself-accommodation unitswhosetotaldistortionmatlixisnumerically
close to unity. The observation of this three-variant self-accommodation mechanism is a
first step toward a more thorough understanding of the shape-memory properties of
orthorhombic martensite.

hsummary,thepresentsuldyhasdemonsuawddm11(NiCu)minfilmsfahricawd
by physical vapor deposition techniques can behave thermoelastically after annealing. and
showcertainagingeffectswhichdonotappearinsimilarbulkalloys. Thedifferenceis
amibuwdeuenfiaflymthekineficcondidmsofdeposidonwhichmnwdinfihmwhhm
amorphous structure. The films, after subsequent solid-state crystallization reaction and
heat treatment, possessed microstructures characterized extremely fine grain size and
spedflprecipimwdisuibudmanddwrofwhichueusflyauainabkmmsmdeby
traditional melt-solidification processes. The thamoelastic transformation characteristics of
dannlmsweresignincandyaffectedbythenanneofthesemicrosmtcnnea Thefinegrain

129
size also allowed, for the first time, observation of self-accommodation morphologies of

various martensites by TEM in whole grains. The results offer valuable information for
interpretation of the shape-memory properties of thin films. It should he pointed out,
however, that the self-accommodation morphologies observed here may also have been
infllwncedbythebw-dimensionalconsuainminTEMfoflawhichwaeomatwoadas
ofmagninldemimayetmandlespumuedfilmssuxfiedbynsisdmwyandcabfimwy.

(1) Thin films deposited from a T150,05Ni44,99Cu4,95 target by triode DC
magnetron sputtering yielded a composition of T1474N1465C06J at an argon ion energy of
550 volts. Concurrent ion-bombardment of the growing TiN i films, with ion energies of
50 eV, 100 eV and 500 eV, verified that titanium atoms were removed preferentially by the
energetic particles in amorphous Ti-Ni alloys. The sputtering yield ratio of 500 eV ion
bombardment was Y = Yn/Ym = 1.75. The value of 50 eV ion bombardment was found
to be 1’ 2 9.

(2) The addition of 1 nm titanium layerforevery9 and 16 nm alloy layerincreased
theovetalltitaniumconcennationinthenlmstowzatsandsr.0at%respective1y,which
were about 0.6 at% lower than the calculated values. Microstructure in films annealed at
823 Kand923Kcoincidedre1ativelywell withthepredictedone,accordingtothephase
diagram obtained at 1073 K [44]. This revealed that the multilayered structure can be a
reliable method for fabrication of TiNi films with controlled composition.

(3) No evidence of Ti-Ni crystallization was found with 100 to 500 eV concurrent
argon ion bombardment, at ion-to-atom arrival (I/A) ratios of0.33 to 1.08, and at substrate
temperatureofbelow 500K. However, shortrangeorderinginthefilmswasenhancedas
suhsn'atempfl‘lml'BtionenergandI/Aratioinaeased.

130

(4)Tinarticleswithabout5nmindiameterformedinawidedispersioninan
IBAD film, with 1G) eV assist-beam energy and IIA ratio of 1.08, when the nitrogen
partial pressure approached 10'6 Pa.

(5) The amorphous-to—crystalline transformation in 1i,(NiCu),., films with
titanium concentrations in the range of 47.4 at% to 51.0 at % was found to be a
polymorphicreactionwithaBZproductphase. 'I‘hecrystallizationtemperaturesofthese
films,ataheatingrateoflOK/min,increasedmonotonicallyfrom749l(to7641(aa'l‘i
contentdecreased from 51.0 at% to’47.4 at%. This trend coincides with an extrapolation
ofthedataobtainedfrom'l‘iNi alloyribbons [75],however, thecrystallizationwmperatlnes
offlwwrnaryfilmswueabomleowaascmnpuedmthoseofthebinarydloyswith
me same titaniumcontent. All theternary filmswith compositions between 51.0at% and
47.4 at% Ti released approximately 2.2 :1: 0.1kJ/mole during crystallization

(6)Alltheannealedfilmshadaustenitegrainswhichwere1toZumindiameter.
'lhegrainboundariesweredecoratedwithfigNitypepr-ecipitates.

(7) Annealed films with composition of T147AN14650K1 contained precipitates of
(NiCuhTiphaserpeafinginthegrainbomldafiesmdinthegrainintefias. Specific
orientation relationship between the (NiCu)2Ti precipitates and the 32 matrix was first
deduced to be: mammoth-n ll <100>32 and (comments-n II 1001 las. rho (Niomri
phase has a thin-plate morphology with a semi-coherent (001) plate surface. The films
mneabdu923Kmdawenttwosepamtedrermelasficuansfamadonanabefaefomd
in Cu-containing alloys, which consisted of a 32 H R-phase transformation and a R-
phase H monoclinic-martensite transformation. The change in transformation behavior
infiisfilmwuclodymhwdmmeprecnceofmiCuhTipreciphaws,whichmsuhedm
copperdepletion,andformationofinMnalstressinthe32matrix.

(8) Films with a compofltion of T149,2Ni44,gCug,o were relatively free from
precipitates. Only a small amount OfTuoN1565Cu35 and TizNi precipitates were found in
the grain interiors. The TigoNigg,5Cu3,5 phase was found to have an orientation

131
relationship with the 32 phase, which could be expressed as: [0111”]! [01 um and

(200)”. ~1° away from (011m. This was different from those between the TigNig
precipitates and the 32 phase found in Ni-rich TiNi alloys, though they have similar crystal
structures. In addition, electron diffraction showed the T140N1565Cll35 phase had an
orthorhombic unitcellratherthanatetragonalonereportedbyvanlmetal. [44]. The
filmsannealedat923Kforonehomunderwentaregular32 Hmonoclinic-martensite
transformation in which Type-I (11 '1') twinning, observed by TEM, was the lattice
invariantshear. Ontheotherhand,R-plmsewereobservedinthefilmsannealedat823 K.
The self-accommodation morphology of R-phase observed in this (N i+Cu)-rich film was
different from that found in the bulk materials. Two sets of plate-like R-phase variants
were arranged alternately through [10013; and/or {110132 twins to form a accommodated
domain. An austenite grain usually contained as many as three sets of two-variant
domains.
(9)3ansivepredpiufionoffimTrgNipardcleswufoundmtheg'ainmteriasof
the Ti-rich PML films having a composition of Tisr ”Ni“ A0114 ,5. The TizNi precipitates
were equiaxed, and were first found to have a well-defended orientation relationship with
132 matrix: <100>Ti2Ni // (100)32 and {0101mm // (010m. Thermoelastic
transformstionsinthesefilmsproceededbya two-steptransformationsinvolvingtheinitial
formafionofmmhorhombicmanensiwpfiormmsformafionmdlemonocmphase.
This two-step transformation has previously been observed only in bulk ternary alloys
having more than 9 at% copper addition. The orientation relationship between 32 and
orthorhombic phases was: [010]...“ // [01 1132 and (zozlaasll (211m, which coincides
with a previous prediction made by Saburi et al. [158]. The habit plane was for the first
timedirecflyobservedbyTEMtobeazigmggedplanemomposedofalmate (322132“
(112)” steps, each of which were about, 100 to 300 nm in length. 30th theoretical
calculation and TEM observation indicated that a three-variant combination could be the
most stable and the most frequent self-accommodation morphology. Here, three variants,

132
sharing acommon <lll>32 pole, are boundedby three (111).“.o twinplanes witha i:

120°anglebetweenthem. Fmthermore,austenitegrainsweresometimefoundtoconsistof
two band-like variants arranged alternately through (011) twin planes. Theoretical
calculations showed that the shear strain associated with the transformation could be
elinn'natedeffectivelybythisanangement. 'l'hemonoclinicmartensitedescendedfromthis
structurecaninherittheoriginal orthorhombic variants, butwillform (001)Wtwinsas
the lattice invariant shear.

Finally, successful application for TiNi thin films as micro-actuators and
superelastic surface coatings relies on their superior shape memory or superelastic
properties. In the present study, the films with slightly different titanium content have
shown a wide variety of transformation characteristics, which offered an excellent chance
to study the correlation between transformation behavior and corresponding mechanical
properties for bath applications and academic interest. Further research will required to
establish precise stress—strain-temperature relations, and relate them to the transformation
characteristics, allowing desired mechanical properties to be made in a predictable and
mproduciuemanna.1henechmucalpropadesofmesefilmsmayalwbeafiecwdbyme

unique microstructures observed, such as the extensive distribution of grain boundary

predpimtesmddwwidedispasimofvafimpreciphatesindlegraininwfias.

133

4883.22 2.3222 13.330 @538 .3 35.85 as? Beams—fiat

 

 

 

_ 2 2 a 2 N .2 252.2 9562 .25.." zm
as on as as ea on is as. e: as as ea .3 ea .3 assented
32.2 32.— .2 2:. <8 32.2 .2 2 .2 2.... 3 ceramic... 92
62+ 62+ 62+ 29.5215..—
2 2 2 33.7.2.2
ea 9e 28: ed 33 ed 22578.2
32.. 32.2 32.. 3 3 2335.2 «2
62+ as Ne 52+ «a .83.?an
@582 a a GM 2582 N a BM 9582 @582 @582 a a BM 292.392..
ea co ea .3 ea .3 as on as .3 ea .3 ea .3 ea .5 223.78..
32.. 2 .2 e32.— 232._ 32.. 32.. 32.. «.2 3 $32.32 an
Nc2+ 5+ 62+ 62+ a). «one...
9582 @582 2 3.52 a: «"2.—
281 ed 281 ed Ea an As: ed as «8.2
«32.. 2322 232.2 232.. m.n 3 ere—2.2.1.5... a:
5+ 62+ 52+ 6+ Ne2+ ea «on...
GM 9582 2852 BM .9582 a: nuza
ea as as ea as on as we as a: s. race
2 2 2 .2 2 3. 23.7.9.2 a:
m a e n e m N _ a. 62....
68 v
3332 e5 season: .82 3.2 ace. etc: cast... :3.

 

2.83 320238392 53.9.8 Bu 2.8883 838.39

Tm 03am.

134.

 

 

 

Table 3-2
Depositionparametersforion beam sputterion beamassisteddepositedfilmsfabricated
for direct TEM observation.
Figure I Assisted Ion-to-Atom Substrate Sputter Time Wn'on
Beam energy Arrival ratio Temperature (minute) Rate
(M (K) (A/sec)
4~7(a) 0 -- 373 22 0.55
4-7(b) 100 0.326 388 23 0.55
4-7(c) 200 0.596 423 23 0.55
4-7(d) 300 0.813 403 22 0.55
48 500 1.084 403 22 0.55
4.9 100 0.326 473 23 0.55
Table 3—3

Deposition parameters for ion beam sputter ion beam assisted deposited films.

 

Tubstrate and Assist- Current Deposition lonIAtom Assist- Sputter
Run Beam Densi Arrival Beam Time

ty Rate
(eV) (mAng) (Almin) Ratio Angle (minuteL
o

 

.5

 

 

 

 

 

 

 

 

 

 

 

 

 

. a
Control 1 0 42.4 0 ”In 307
$102 0 0 51.5 0 0/3
0 0 57.3 0 “/8
0 0 1 7.2 0 Ilia
Control 2 0 0 21. 0 n/a 180
8102 0 0 25.6 0 IV!
0 0 29.2 0 M
f 0 211.5 0 117a
Control 3 0 0 25.7 0 N8 125
8102 0 0 33.0 0 n/a
0 0 42.6 0 his
IBAD 1 50 6.5 40.1 0.085 7.8 178
8102 50 5.0 48.6 0.054 15.2
50 3.6 54.8 0.035 22.2
106 2 7 .1 1 1.1 0.51 0
IBAD2 100 24.1 15.1 0.42 7.8 308
“(Cl-+8102 100 18.7 17.0 0.27 15.2
100 13.1 17.4 0.17 22.2
566 62.3 0 1.13 0
IBAD 3 500 49.7 0 0.72 7.8 317
“(Cl-+5102 500 33.9 5.8 0.38 15.2
500 21.6 14.4 0.19 22.2

 

135

 

 

 

Comparison' of the d-spacing values of :13th shown in Figlne 4-5 and the TiN
phase [129]. ~
Precipitates Intensiy d-mq Plane Intensity
2.50 . s 2.499 (111) 72
2.1 1 vs 2.120 (200) 100
1.50 3 1.499 (220) 45
1.29 m 1.279 (311) 19
1.23 m 1.224 (222) 12
1.05 w 1.060 (400) 5
1.03 w 0.973 (331) 6
0.95 m 0.948 (420) 14
0.87 w 0.865 (422) 12
0.82 w 0.816 (333)(511) 7
Table 4-2

Compositions of target alloy and magnetron sputter-deposited films.

 

 

 

 

Materials Composition (at%)
ZAF Method Cliff-Lorimer Method
'11 Ni Cu '11 Ni Cu
TargetAlloy“ 50.2 45.7 4.1
SL Film 47.4 46.5 6.1 47.5 47.5 5.0
PML-l6 Film 49.2 44.8 6.0 49.2 45.5 5.4
PML-9 Film 51.0 44.4 4.6

 

* The composition of target alloy supplied by the donor is 50.05 at% Ti, 44.99 at% Ni and

4.96 at% Cu.

 

136

Table 4-3
Crystallization data for magnetron sputter-deposited Ti(NiCu) films.

 

 

 

 

WW Gystallizationinthalpy
Mmrial (K) (Id/111016)

2" 10" 30" 50* 2* 10" 30" 50*
SLFilm 749” 764“ 776 371807! 1.23 1.68 1.92 2.16
PML-9Film 733w 749 761 771”” 1.48 2.28 1.72 2.13
PML-9 Filmon Si 727 753 767 773.. 0.91 1.61 1.84 2.32
PML-lBFilmonCu 729 747 758 766 1.29 1.29 1.82 2.11
* : Heat rate in Kelvens per minute.

Table4-4

Transformation data for PML-16 films.

 

 

Heat'rteatment Method Mg Mf Ag Af AT Hysteresis 0,32!”

923 K. one hour ER 287 228 238 297 59 10 -
(2xlO'5 tort)

923 K, one hour DSC 275 253 266 284 20 9 685
(21:105 torr) .

923 K, six hours DSC 275 254 266 284 19 9 427
(2x105 torr) '

923 K, one hour DSC 258 251 259 267 8 9 1016
(2x105 torr)

823 K, one hour DSC 222 202 243 273 35 31 854
(2x105 torr)

 

Temperatrne in Kelvins

Table 4-5
Crystallographic data for Type-I twinning

 

 

 

 

Solutionl 551mm" 2
(1 -x)=0.32m6 x=0.67994 (1-x)=0.32m6 x=0.67994
0.58112 0.43012
n1: -0.81382 llz= 0.27789
-0.(XX)89 0.85894
' M11 M21
1.02518 0.09869 - 0.13399 1.02608 0.15603 - 0.08709
0.02036 1.02179 - 0.05033 -0. 03877 1.01381 - 0.1 1627
0.05428 - 0.02355 0.94746 -0. (X1223 0.04698 0.94713
M12 M22
1.02518 -0.09832 0.06301 1.02608 -0.04866 0.11760
0.02036 0.94782 0.02365 -0. 03877 0.95667 - 0.05913
0.05428 - 0.10070 1.02461 -0.(X1223 - 0.02423 1.01835
E1 Es
1.02518 - 0.03527 - 0.0004 1.02608 0.01685 0.05209
0. 02036 0.97150 - 0W3 -0.03877 0.97496 - 0.077418
0. 05428 - 0.07601 0.99992 -0. (X1223 - 0. (11144 0.99556
S=O. 1088 S=0.1088
0.39866 0.55758
1111: 0. 32205 11128 -0. 82876
0.85869 -0. 04760
0.41720 0.57077
m'1-- 0.29652 ln'2= -0. 82085
0. 85908 -0. 0207 2

 

 

 

n1, ng=Normals to the habit plane in the cubic basis.
gin. MusThe distortions to which the major and minor twins are subjected, in the cubic

asis.

(1-x), x=Vollnne fractions of major and minor twins respectively.

138

 

 

Tab1e4-6
Transformation data for PML-9 films.
Heat Treatment Method Ms Mf As Af AT Hysteresis (I file)
923 K. one hour ER 305 225 247 322 78 17 --
(2x10'5 1011')
923K.on6hour DSC 297 285 300 312 12 15 482
(2x105 1011')
923 K, one hour DSC 293 286 299 307 8 14 643
(2x10'6 torr)
823 K,onehour DSC 292 286 297 306 8 14 949
(21110" torr)

 

Temperature in Kelvins

139

Table 4-7
Summry oftlansformation data for PML films

 

 

 

 

 

 

 

 

 

 

PML- 16 Films
Heat Treatment Method Mg Mf As Af AH AT
“973's 0W9
, 1 hr
(2.6x103 Pa) ER 287 228 238 297 59
DSC 275 253 266 284 685 20
W6 hr
2mm. Pa) DSC 275 254 266 284 427 19
, 1 hr
(25x10. Pa) DSC 258 251 259 267 1016 9
XRD >251 ”201 >50
1231.1 hr
(“was Pa) DSC 222 202 243 273 854 25
XRD >221 ~198 >23
PML-9 Films
m, 1 hr
(2.6x10'3 Pa) ER 305 225 247 322 78
DSC 297 285 300 312 482 12
XRD 143< Q22
12316.1 hr
(25x10. Pa) DSC 293 286 299 307 643 8
XRD ~22]
m 1 hr
(2.6x10'4 Pa) DSC 292 286 297 306 949 8
XRD 229< Q70

 

Temperature in Kelvins

140

 

 

 

 

88.. 88...- 88...
88... 88..... 88... . .
88... 88...- 88.. .88... .28.... .88.... .8: .8. .2... e
88.. 88... 88...- ,
.2.. .. 88... 88.. 88...- . .
88... 88... 88..... .88... .88... .38.: E... w 8 .8: n
88... 88... 88...
88...- 88.. 88... . .
88...- 88... 88.. .88.... .88... .88.... 8. : P... . .8. e
88.. 88... 88...-
..8.....- 88..... 88... . .
88...- 88...- 88. .88... .28... 88..... 8.. .8: .2... m
. . 88.. 88...- 88...
w .. . 88...- 88. 88...- . .
88...- 88... 88..... .88... .88.... .88.... c... E... .8: N
88... 88...- 88...-
88... 88.. 88... . . .
88... 88... 88.. .88... .88... 88.... 8.: ch. ..8. .
2:8. 2.8.... 2.8:
.8... 4.52 SEE 258...... assign... 88>

 

8.8.. 88858888983356.2388;

 

«two—nah.

141

 

 

 

 

 

 

wuv 035—.

88.. 88... 88...
88...- 88...... 88...- .
88... 88... 88.. .88... .38... .88.... ...... 8.. ....... a.
88.. 88...- 88...-

..w .. >. 88...- 88.. 88... . .
88... 88...- 88... .88... .88.... .88.... E. ...... .8: ..
88..... 88...- 88...
88... 88.. 88...- . .
88...- 88...- 88.. .38... .88... .88.... ...w ...... ..8. ...
88.. 88...- 88...-
..8..... 88...... 88...- . .
88...- 88... 88.. .88... .28.... 88.3 8.. ...... .2... ..
88.. 88... 88...-

...... ... 88... 88.. 88...- .. .
88... 88... 88..... .88... .88... .28.... ...... u... .8: a
88..... 88... 88...-
88...- 88.. 88...- . .
88... 88...- 88.. 38...- .88... .88.... ....- .... : ..8. .-

....8. 2.8.... 3.8..
numb has. .5ng 35.3. singers-r. 83 .3.;
63580.

 

 

 

 

\\\\\\\\\\\\\\T

 

 

; 8s 8 8%
E8 EQQ:

 

 

 

 

 

 

 

 

 

//
V/

 

F‘ l-.1 Schemati cdrawingo fthe(si)iape_memorye effect: (a)austenitesingle ecrystal;
martensitew consistingo of two variants, variant tcoalesoenoeon loading; and(d) martensite
revertingtoaustem niteon heating.

143

 

....

- . 9‘00
~8w.»-

 

a»...

Martensite

 

 

 

 

 

m
m
m w
am
...... m.
m m
e t
.- m
mm m
u .m
- .
m

 

(0)

Figure 2-1: Two dimensional schematic drawing of the martensitic transfonmtion: (a)

crystal structures of austenite and martensite respectively; (b) lattice defamation

(C)

mnafmmingmstafimhuicemmnsimhtfice;(c)hfioeinmfimtsiwmainmidngm

undistorted habit plane; and (d) rigid body rotation maintaining a continuous intelface.

144

 

 

 

 

 

Figure 2-2: Section through a martensite plate, showing the banded structure of relative
amountsxoftwinZand(l-x)oftwin l. 'I'heplaneofpaperisperpendiculartothetwin

planes [25].

Weoght Percent Nickel

 

 

 

 

 

 

 

 

 

 

 

o o 20 30 40 so so 70 so so lOO
neoo ’ .. . . T . - ..
* :
i '--._...1 t
V :
i455°C'
. . i ao°c :-
3» .
'to°c °
2 I
: (iii) :
a a
5;: °c t
E 711“ a 3
cu "" 2
5- 25 ~
I" .'
t
cafe
600 Y . T 1 ' r W ' '1 l
o :o 20 so 40 so so 70 so so 100
T: Atonnc Percent Nickel K.

Figure 2-3: Equilibrium phase diagram for Ti-Ni alloys [29].

145

 

 

 

NilTil

Ni

 

Figure 2-4: Isothermal cross section through the 'l'i-Ni—Ophase diagram at 1200 1([42].

146

 

 

 

 

 

 

 

 

Figure 2-5: Isothermal cross sections through the Tr-Ni-Cu phase diagram (a) at 1073 K
and (b) at 1143 K [44].

147

I = Parent Phase
II =Premartensitic Phase
LO . I [II = Martensitic Phase

HI+II

II

 

 

Electrical Resistance (arbitrary units)

 

 

T (°C)

Figure 2-6: Electrical resistance as a function of temperature showing a two-step
transformation in a TiNiFe alloy [50].

 

     

I/’ ”I\
ll \\
“‘s A-o'l’ \A-e' "/
\\~B-C 19' C’O”’
‘n‘ C ‘:3
9°"\ ,I’Ioo
I \\ 1’ \ - -
o I \\ III \ '0'
8|?i'5 I" A \\ .50
a' 0"0 TI TIO ‘-
: ’8 | A‘D l‘
I
rfF’ ’ >1?!”
‘~--~ imo """"" ' \‘
""‘I’Kl’c'
:8-0
I
I
I
. i .
”0' I 0'?
BC C10

Figure 2-7: (Tl l)BZ stereographic projection showing R-phase variants A, B, C and D
and twinning relations between them [53].

’20 I: \Q
I} O
O \

 

Figure 2-8: Self-accommodation morphology of R-phase (above) and corresponding
schermtic variant-combination (bottom) [53]-

149

 

 

Figure 2-9: Self-accommodation morphology of monoclinic martensite (above). and a sub-
micro model depicting crystallographic relationships between variants in the triangular
morphology (bottom) [23].

150

 

 

 

 

200 7 I l l l
e
150 L- . ‘
I
100 — ‘ “
L . r ‘
t o, t .
50— 9 “
I I n‘.‘ l '
A0;
AA

0 — "1a fl

4! o
_50 _ 0 Wang et al. a e
AiHarrIson et al. '
olHanlon et al. i '
400 __ uzPurdy et al. A _i
I
o i
—150 I 1 1 1 1 "
53

47 48 49 50 51 52
Nickel atomic (%)

Figure 2-10: 'Ihedependenceofthetransformatio ° '
in Tim alloys [4]. n temperature (M5) on ntaruum content

151

...»... ...: o. 88 a so... .2....- s. an... .3.-.8 8.. 2.. .... 838.8 .8882:- HNE as»...

80.. ..E ...-3.... .2.-.- 92.85
8.. .388 3.83.8 o... ...... 83809.8. Sauna—Saba o... .«e 8.89.2.3 comm-8:980 .— TN 95mm

 

 

 

 

 

 

 

 

F ...-.. .2 G88. 23:8 ..r
0. m0 mo «.0 No 0 m... a... 8... m... to m...
. . a q . q . 1 . CON « c — a com
- e 0 e e e e 0.... - ” cm W m m .
n u ...-lot. m. com-.10 o o m m m ...
- . - W. “.-.----“- ........ . ......... .u ........ ”- ....... 18.. 3
.. ...... n m ... m m U m
, n . w ...... s...- m is” m m . . ....
o I _ . . . a m
.98: 3 m n ......... .- -------- .I-JE-.. ...... ....... took a
... m 8...- . m . m .... . m.
Lame—tor m I.\ n O" m m “u H U
l v. . o u n _ . I.
X 9 fi- _ o _ H O. . a
m ..r. ...... o -- stoma w
.I n n n .0 fl _ t .W
.W E . . ... o u n . r
.5.. .....- .m .....m. m m 0qu m H m
NO_I .l\ W. .- .......... .. ........... u ........... L. ----- .---#00,--Loom N
i n m m m . 4 )
007 M. com... ” u " AWE. H mm
mm- -..m... . m m m o m .
m n ........ n .......... ........ n ....... 18..
w .- ..omu . n u n .
......roc . u n u u .
vm- . m m m m .
o8.- . . . . ......

 

 

 

 

 

 

152

 

 

  

 

 

 

 

 

 

I 'UIVTII I 1 I'F'IUUI I I I VIl'l' I I'LL...—
‘Ho’II, N69, u’o, It“, r"- rsrzpmosuen a. 2m ,-—"'”' _ —
'0 _~_ ~..,M.O.KI. 0.10" weuscurno oi ol u. 260,327) ,d:,-.—-—-°"" K? :
. m'o,~'o,u.'., u’emc‘ o Mummies N m /-’2{a’ (c 0' -
- ‘w‘o , m‘ 6.10’6 noseuaencmmsn u. an I,’,- f) ’ .4.—4" - ~
5 ~ weave usemoywunu 2m :1’ ’,-—;". ‘3:
. u' our afa.‘ u moms" u oI M 62) ’x” --.... g ‘
.. u- a north“ 336) /),-_.—-—-—-' ~~-\\\ -
0'03““, Av‘o,m°o,xo°o am or at u. 337) . 1/‘0 ‘\
2- we POAlEfloH‘DSl / N" -‘
M0 Incas“ am .0 .
/ ’ 0
0
IO : ‘:
. . O "
S ~ ~
2 » -. .. —-—-- —~— ~-\\ -
‘1
-I
IO : ‘2
5 ~ ;
Y ‘ ~a-\\ -
_ ‘..\ d
\.
2 . “‘ -
-.\ \
.2 ..g
'0 :- ‘ .-
- \o I
5: N'DJig'v .- stunnwcmcuu 3w ‘
are rut «a to. Jul "
_ , we scmmwnz neoI -
.- u.‘ o routerr,ameutt I.”
2. /, . w‘o "scuttocutum -
' ,/ Co'o ,Cu' 6 ,cc‘ - :Smm u ‘9!
,3 x “9‘9 «menu. Jon
lO / new wemmmszmneu m, -
/ Hg" ASKEIIOV,S£NAN nor :
7 now ISMAIt.5£PrIcnII-Josr -
s / . -
. / NICKEL
9' e
2 11111 i r Irrrrrl r r rrrrul r r irrrul
3 5 0.1 2 5 1.0 2 5 ‘IO 2 5 100

Figure 2-13: Sputtering yield of nickel as a function of ion energy and ion mass [84].

ENERGY lkeV)

153

 

 

l
Art. AgJaJ’iAl :
7'. Al \
"I '\
l/cos 9 ’i /E\ g
2.0 ~— {/0 '0 \.
I X
0:
5 /-’ A
A- 'A
5 I, / To ‘A
IL
0 D/lA/ \
-J l.5 - A -—.
LE ' //A O’O.O\
: i / I, / Ag 0 D
5 2/9 g)’ 0
Z i 485’ o/ \ A
, l /
d 5.18%“ \
E l.0 ’0 o -i

- o
0.5 - \.
a
l l
0 so so so

ANGLE OF INCIDENCE (degrees)

 

 

 

Figure 2-14: Variation of sputtering yield with angle of incidence for 1 keV Ar+ incident
on Ag, Ta, Ti and Al [85].

154

 

     

. 0- 5 SUBSTRATE

ARGON TEMPERATURE (T/Tm)

I‘RESSURE
(mlorr)

Figure 2-15: Structure diagram for thick films produced by sputtering [87].

155

.3...» 2.82.. o... ... :9... 82.08 388.830 2.. mg 5.... .85 ...m 2.5."—

 

 

 

 

, , - .
r 8:23.... 8258 182...... I335... .5. 1%:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a
.8: a
E cooaolm 3.8.889... T cflfluhwau
7962...: E... 2.3-fl.- , 33...... r 7 . .9.
- -. 5E5. - , . .8238
72.5238: 253.8: 3.61m T 85.85.: _t 338.935.»
_ 8.538888... 382...... _L 8... on .3: _-
fig: Sufism 8.8.285 _ .558.» no... _ I— 8...- finance
. r .— 533m »
7888...... .8... .- 288; .
_ $838.2 Sufism 9.58m -
_ . >2- 2 >92.
— ..g x 2.. a... o m:_-.
_ 8868.2: ...... a x .8: 810
.. , ....u. .6
8»? E 82%.... - 589.. 33.8.. 58m 8. ”gain
sumac... 8.38% ‘ 3.5% 58 8. ,
- magma-.282 >2... 028095 3.26 - 35.82 - 8.23...
.883 5.88.8 . 82.2223
R8923: 52803 0.8.5.5... e£=8noo§3e0nm . ._ Soon .3. ,
3338332 __n _ 50m c... .9.

 

 

 

156

 

 

  
 

 

  

V
' Aperture for
Thickness Monitor |
Substrate Holders

Rotary Feedthrough

 

  

Rotary Table

Substrate Holders ‘ ' I I ]

TargetShutters I I I [ J

 

 

 

 

 

 

 

Rotary Feedthrough
(a)

Figure 3-1 Schematic drawing of (a) the triode magnetron sputtering apparatus, and (b)
the substrate holder assembly.

157

CmrBlock

 

 

.2.
‘

 

\\

 

 

 

 

 

 

 

 

 

(b)

Figure 3-2 (oont'd)

158

 

 

 

 

 

 

   

Substrate 7' Apertures '

5-cm
Ion Gun

l

I

|

|

I

| /
I

| x’ .
I
I I Rotary I
| I Feedthrough :
I I
Faraday Probe \- | ,I .
Shutter " :
Substrate Aperture :
I
I
I

Ionfiun

 

 

Quartz Lamp

Rotary Feedthrough

Figure 3- 3: (a) Arrangement of the ton sources, the substrates and related devices for ton
beam assisted deposition. The devices shown were contained' tn a 14" metal chamber. (b)
Schematic drawing of the substrate holder.

159

 

Substrate Shutter

 

Substrate Holder

 

--Bearing

 

 

 

‘ Rotary Frrdthrough

Figure 3-3 (cont‘d)

160

...oumuonoc c832. .50.. co.
5.. «33...... o... ...... mono...— »Eu 0... .89... o... .8959. ..o. o... ..o Scanned—... v2.5.0: .1 onE

 

so; on... _ 5.. 8. 32> .8...

Evin

«9.3.0.? .... _ _
III v.36 vogue:
H - I _

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

an"
.....m .8: .858 _
3.89.032. .50 co.
89m

 

 

 

 

 

 

 

161

 

 

L Clean Substrate I
1

 

L Installation I

 

 

I Evacuate :Chamber I

LCool Substrates to 200 K I
[Adjust Gate Volve toI yield a P . O. 3Pa I
Hum on Triode Source I<

I Pre-Sputtejr Target I

I Open Target Shutter I

 

 

 

 

 

 

 

 

 

 

 

 

 

I Position Thickness Monitor Aperture I

 

 

IMeasure Deposition Rate I
uilose Targlet Shutter I
Janut Sample Information I

 

 

 

 

 

 

 

IOpen Substrate Shutter I

 

 

LOpen Target Shutter I
L Deposition j
LCIose Target Shutter I
IClose Substrate Shutter I

 

 

 

 

 

 

 
 

 

 
 

Substrate Temperature >350 K

 

 

 

I Turn off Triode Source I

(a)

maggiwposition procedures for (a) triode magnetron sputter—deposition and (b) ton

162

 

I Clean Substrate I

 

 

I lnstallaItion I

 

I Evacuate Chamber I

 

fium on Sputter-Beam Gun I
, I ,

 

I Turn on Quartz Lamp I

1
I Pre-Sputter Target I
7 l

I Turn on Assist-Beam Gun I
I

—.IAdjust Assist-Bealm Parameters

 

 

 

[Turn off Spultter-Beam I

 

I Measure loin Current I

 

I Monitor Substrate Temperature

 

ITum on Sputter-Beam I

rOpen Target Shutter J

 

 

 

 

I Deposition I

 

 

 

, I

I Close Target Shutter I
r a

I Chagge Substrate J

 

 

 

 

(b)

Figure 3-5 (cont'd)

163

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

80.0
--0- Control1(85mA)
A 70.0 --I— Control2(50mA)
.5 --e- CanolaIesInA)
3 60.0
9. 50.0
at
m
c 40.0
.2
"2 30.0
a
o
0 20.0
10.0
Distance (cm)
1.20II‘IIUT'1'UUrIl'U—VVV TIUT IIYT ifjl.
- r i A ’ .l
. +IBADI(500V) / .
1.00 _ —-I— IBAD 2 (100 eV)
0 _ —-e—Iera(soooVI / .
:3 r / t
as 0.800
a: I / 1
75 ’ / ..
.2 0.600. / j ‘
t _ r
< b / // -I
< 0.400. /‘ .
2 I / ‘
I. ///./ .
0.2007“ /‘ .
: —————-~r—"‘ :
0.001'11 11.11 1111 1111 Llll All] 11in
1 2 3 4 5 6 7 8

Distance (cm)

b

100

b O 00
O O O

2
(aura/VIII) Mgsuea Iueung uoI

Figure3-6: (a)Depositionratesandioncurrentdensitiesand(b)I/Aratiosplottedasa
funcdonofdwhwrddisplwememofdnspecmnfiommespuuaingtargacentalhw.

164

$8839. 2.089589... b.3338 E038... 2.. ..o ":35 0.882% ...-m BEE

 

 

 

 

 

 

 

03:30.59... .m

...... 03...; ...

...... amazo> .m

.30 .2250 .N .38:
c. acotsu .— /
baasm .36..
on <98 ....

G o00m

 

 

 

.020: 0....an *

mac canon—.2
Sou

 

.______.
>~—

    

 

 

 

 

.0933. C.
(wVON a...

 

 

 

          

)\

 

 

 

8.533... 53
...o_<.m.s .525

 

 

 

 

 

.ousnEoo
...<-u._ 2m_

 

 

‘165

 

 

 

 

/Liquid Nitrogen Stopper

 

   
   

—— Liquid Nitrogen Dewar

Cold Finger

Sample Holder
(with Build-in Heater
and Thermocouple)

 

 

 

 

 

 

—> Vacuum Pump
X-Ray Wlndow —

 

 

 

 

Rotary Stage

 

 

 

Fi 3-8: Schematic drawing of the cooling stage attached to a Rigaku X-ray
' tar.

166

 

Figure 4-1: Secondary electron micrograph of its-sputtered SL film showing the fracture
surface and the top surface.

 

167

 

 

     

 

    
 

 

 

 

 

. . . . . _ m
1 u u o u n
r u n u u
. o a
v n u . n
. o a
. m m m s
1 ........ m. ...... .......n.. ..w..- l". 5
. u m 4 . ... m
l . c o . l
n u M n M n
1 . o o . A
. m m m m .
. o c c
I ........ .r ...... .n ...... a». ..... 4. w \l
v u n u u a
. o o c
. m m m m m
. u u u 9
. m m m n W
.. ........ . ....... .H ....... . ..... + a a
... . . u t
I . o n u
. o o o m
u . u u u
. c . u
. . u u u m
l ....... r ....... . ....... p ...... ..r T
u u u w
I . -
u o
. o u
I . . .
c c
1 _ m m
T m m . s
1 ..... .1 ..... -n ....... . ....... ..3
1 u u u .
r n u u u .
u n n u
. n u n u .
w . u u u .
b b P n h b h o
m w m m m m3
3 2 2 1 1

A38 5.9.3:.

sputtered SL, PML-9, PML- 12, and PML-16

films showing a broad first-order peak belonging to the amorphous phase.

Figure 44: X—ray diffraction patterns of as-

 

Figure 4-3: TEM results for Ti-Ni thin films: (a) ion-sputtered at Tsub=373 K; (b)th IBAD
with UA=0. 33, Tsub=383 K; (c) IBAD with I/A=0. 60, Tsub=423 K; (d) IBAD with
UA=0.81, Tsub=4o3 K.

 

Figure 4-4: IBAD film irradiated with a 500 cV assist-beam and 1.08 I/A ratio, Show;
extensive argon gas incorporation.

 

Figure 4-5: IBAD with I/A=0. 33, Tsub=473 K: (a) bright field trmge; (b) diffraction
pattern; (c) dark field' image formed with innermost diffraction ring and (d) dark field image
formed with third diffraction ring.

170

 

 

 

 

5
q I I I I I I I I I I I I 10
M. a a a o
I m I ”m m m I
. m . ...W M m .
n . II n u n
. m.” . u m m .
m II. m m a
u ........ i. ....... . ........ .. .......... m ......... - ....
u u u .
1 m II m m l o
I m m m M a
u u u n
. . . .
. m m u u .
u n u m 5
.. ........ .n .......... m + ........ . w
n u " nu.
" n u ._
fl m m m
m m 3 m
1 u u u a n . 5
'IIIIIIIIM'CIIIIIOOUW IIIIIIIIII m'llIfl'fiflr lllml IIIIIIII I y
m m m a...» m 0.
. m m m ......L .o
m m m ..W.
. m m m .
n n “ ...».
V - g u I. A
m m m "..u
h P b — P p b P r& b n b n b — vi 0
0 0 0 0 0 O
6 4 2 O 8 6
5 5 5 5. A A
O O 0 O 0 0

3.33 28:00 5353:.

 

tn / (tn +tauoy)

Figme 4-6: Comparison between thecalculated composition and the measured composition

of PML films.

171

 

 

         

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

s m.
canned-1:: o m IqqmflqqmuJid «um qquqi..*
9 m 3 v m m . w m
u w c ,llrll " u m
..... u - mmm s ..m. .. vv
I..- out iiiiiiiiii 10. d V O... ._ .. .—
o e o o o o
m E m _ mm? m
p m JI. ... ....
1m, ........ .n ................ 44.0w Win-o.” ........... .n ...... .u ...... .H .....
a u n on a o u u u u
m m m % .... m Ill." m m m m
mu ..... M ........ M ................ ....m m .. ..... ..... .......... ...... ......
m .M. .m .. m. m M w
o m M .m. o m m m m .
.l ttttttt im- tttttt am rrrrrrrrrrrrrrrr Izmm 2' iiiii w. iiiii .M oooooooooo .M tttttt am. totem
2 u u 0 n . u . u . u
m. a. m .m b m m m m. m . m
m M o m m m
m m m .. .. m. m .m.-- -. .......... .u ...... .u ...... .u ..... -
m .. ....... ... ........ .. ....... . ... mm m m . m m m
.. mm . .
IbbW-pPWTb n" OM. rm PPpmpththm-hL-p-
1 3 6. . 2 00 mm 0 Z 6 8 1 2 40
. o . o .7... m W m m .. ... m
3a 335ng SEE mm see E.
F.»

0.6
Ti-Ni films as a

0.5

processed

pectedcanpositionofthefilmintheabaenceofan

0.3

0.2

Ion/Atom Arrival Ratio

Figure 4-8: The change of titanium concentration of the IBAD

0.1
function of VA ratio, relative to the ex

assist-beam. Data are shown for 50. 100 and 500 eV beam.

172

 

 

 

\ B 0 GWOV

Calculated (2.1.75)
- - - - Calculated (2.9)

 

0 l 1
:‘ O 500V
3 D 100 ov

 

 

 

 

IILIJJJIIIL

 

 

\

-o.1 ’ = A

\ 0

’3 -o.os ‘

9} : ‘. \:
i: -0.08 ‘

4

ll] 441

A
V

 

O/O“
/1
>0

 

 

 

 

 

 

 

 

-0.12 \:
- \
P \ I
-o 14 _ ‘
‘ o 0.2 0.4 0.6 0.8 1

Fraction Resputtered (Xr)

Figure 4-9: The change in titanium content for IBAD processed Ti-Ni films as a function of
dretotalfractionofthefilmresputteredb theassist—beam.relativetothee

composition ofthefilmintheabsence anassist-beam. Dataareshownfor50,1mand
SOOeVbeams.

172

 

 

50 IV

100 IV

500 IN

Calcuatcd (2.1.75)
- - - - Calaulatcd (Z-O)

1? _ .
% 0.06 : t‘ \:
i: -o.oa ’ ‘ x .
Q - ‘ \ <> :
-0.1 " 8 V N
n \ 0 j
I . o .
-o.12 _ .‘ \‘ .
h- ‘ \
o 0.2 0.4 0.6 0.8 1
Fraction Resputtered (Xr)

 

 

 

 

 

e.
. . o
O
N

17)-
OUO"
lllTllL

 

 

llLlll

 

 

0

 

 

 

 

 

 

 

 

-0.14

Figure 4—9: The change in titanium content for IBAD processed 'I’i-Ni films as a function of
tlretotalfractionoftlrefilmresputteredbytlreassist-bemrelative totheexpected
msition of the film in the absence of an assist-beam. Data are shown for 50, 100 and

e beams.

173

 

       

-- ----—-—-q

Tc-749 K

  

 

 

  

    

T-753K

 

 

b- PML-9
r-----;;-I::r-------------r

b------------‘b------------
II
I
I
I

 

 

 

 

Tc-747K

-------‘ - ---------d

    

 

/ mole

SOC

55 J/

 

P---J

623 723 823
Temperature (K)

523

423

 

 

.8

 

 

25.“. so:

323

and (d) PML-18 film on Cu

try data: (a) free-standing SI. film; (b) free-

on Si (100) substrate

crime
standing PML-9 film. (c) PML9
substrateataheatingrateof lOK/min.

533'

scannin

Figure 4-10: Differential

174

 

 

 

 

 

 

 

u
IIIIHIIId‘IdIIq‘I‘J-IIII‘II‘I‘II‘I 10
c u - o - .
t u u - . . . n
- u u u u _
. . . . . . a
I . — u . J
) u u u .
. mm H u u . . .
I m M ---4“ """ "fl---' “ -' -.1"-l “
. ... Mu m m m u m .1..
amum "w" m"
I . . . . 4
u n. J u
v u u n 5 n ._
. - u n - 1
. m m u. m m .
. . .
i u u . W . u . i
. u . u n .
n u . 3 . n Z
l..'. -' ' 'h--m-r" * .... r-"l 3
U c u o
. u u n n u 1
. a n m n n .
c c u c - .
1 . . . w . . a
c . . . .
a . . . . 6 . .
m n m «m m .
I - +I---L.- --¢ .... T'---b- ..... T-'-J m
. . . - .
r. . . . . .
. . . . . . . 4
- . . u - .
. o - . - _
. c . u c . a
o . - c . .
. g . c . .
_ - . - c . A
— - . c c - 8
DPhhlbhb—PPPbbhhbn—bbb bPPPPD-I -

 

 

450
400
350
300
2.50
200L
150
1092

33:.

1000/Tc (1/K)

 

 

 

 

 

 

 

A33 33:35

°37

m...
mm
.flw
m m a .u .1 ... n.
m m m A
n u u m. '
m M w .m w m he 7
- ...... m ....... m ...... -. a, ..... ...... . 4
.mm m m .... ...: .fi
. u u d. v n A

m. M M a a 2 ..

.... 1 ...... m ....... m ....... .u ..... a, ...... “if“.
.mo m m m a. m A.

w M m t M .
mm 8:5. . 3
mm - ...... ....... M ....... M- ...i- -.. ..... ..
men 36:. . . H
mm M m m M
mm . ...... m ....... m ....... .n ...... . ...... .-..
Mm
......Mv - ...... W ....... w ....... .m ..... . ..... -.....
mm. M w H m
.. m m m A m
mm M m a r M
m... m. m m
an

Two Theta (degrees)

magnetron sputter-deposited films

Figure 4—12: X-raydifi‘ractionpatternsofanumberof
afteranisotropicannealintheDSCcell.

175

 

V

O T1x(NiCu)1-x thin film

 

 

IqI‘IIIdi

fill-I'll.

 

 

---..}.--.--------.----|.-------.------

III‘IIIMIIIwflIImIII

-------P--------------

-II

1%!
l J
07

0.8

 

m
r"
m...
d t. \m
.. ... U. m .
. mm x. n .
. 1 u u .
. \ . .
1. T T ............ fl ...... m ...... ”.---i 3
. n u 0
. DI Dm “ u .
I . . b m m m n
c . o u o g
. m m ...o. m m m -
. . \ . \. . .

1:.-.” ...... met!“- ..... 1.8-. ...... m ...... ”.----1 5.
. m m m. cm m m .0
m m .0. m m m

Um. " m m m m m
T u a u u u u u .
a m m m m m m m .
1:--.” ........... m. ..... .m. ----- .H ...... m ...... ”.----.. A
. . . n n u u 0
.. m m D m u u u m .
. m m Um m m m u .
g c o g - . c
.. m m m m m m m .
. m n m m m m m .
PPP-bbb PFbB-nhp h-n bun nbh 3

 

O: 838888. c3§=39¢u

800 r-

780

Ti content (at%)

tier-deposited films, and melt-quenched ribbons. (Both ion sputter-

deposited films and ribbons are binary Ti-Ni alloys.)

Figure 4-13: Comparisonofthecrystallization temperatm'esofion sputter-depositedand

magnetron spu

 

 

 

 

 

 

 

IIII IIII II: qIII III :II IIII
- M M M . . -
u n u u no
. n u u u o u .
. m u u u u u .
u u u u u
. m m m m m m o.
I .II yr--. tttttt . ttttttt . ttttttt . aaaaaa .I t]
rim-o "- .u m "- .m.
w warm " u u u n a
1 n u n u u
. )x u u u n u .
u . . . . . .
mm m m m m ... .
. Ax m m m .n.\ m .
It “first. llllll . ttttttt .1It .r tttttt . ttttt I.
m- ...) .-
v . . o. . u a
. co m "\ m m m
. m \w u m m
- m ..o m u m m
u xx" ” u u n L
"x u u "e- “ H
II 11111 p 1111111 r 111111 L 1111111 L. 1111111 r 111111 or 11111 1
u u u \o n u
o o .\ c u
. . \ . . . a
m mxxxm m m m .
m o m m m m m
9. o. m m m m .
. mo m m m m m .
1-»--. ..... ... ...... . ....... m ....... ”- ...... + ..... -
. u we m m m m
. u m m m m a
m N no u n a .
m m m m m m .
u m m m m m
DIPWDPD.-PPI-IDPP_IDDIP_-bpr-IDIP

 

 

600 _
550
500

450
400
350

O
3

Engage 3.2m 5562

0.6 0.7 0.8
Ti content (at%)

0.5

0.4

3.
O 00
S
2

Figure 4-14: Comparison of the activation energies of crystallization of magnetron sputte-

deposited films and melt-quenched ribbons.

176

 

 

 

 

 

Figure 4—15: Cross-section secondary electron micrograph for the SL film annealed at 923
K for one how showing grain size and grain morphology.

177

 

Figure 4-16: In-plane TEM micrograph for the SL film annealed at 923 K for one hour
showing grain size and microstructure.

 

Figure 4-17: Electron diffraction patterns acquired fiom the matrix phase ofthe 923 K
annealed SL films at room temperature in (a) <111>32 and (b) <110>32 zone axes showing
streaks in <110> and <112> reciprocal directions respectively-

 

 

178

 

 

Figure 4-18: A room temperature bright field image of the SL film annealed at 923 K for
one hour showing small precipitates with distorted Moire fringes roughly perpendicular to
the [100] BZ direction.

179

'2=[oo1] 2=[o101

 

m

Figure 4-19: Convergent beam diffraction patterns taken from small precipitates in a 923
K annealed SL film. A tetragonal cell with a = 0.31 nm and c = 0.80 nm is evident fiom
the diffraction patterns and the rotation angles between them.

 

(

 

180

 

 

 

 

 

 

 

 

  
 

 
  
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

----Jm
- ----- ..----
- a ..... ..- m m
m ----T--- . n u .....
111.11 . n . 1r111
m ---“.-- . u . -..-1--"
1111411 . u . 11r111 " . .
IIIII “I u n — . 0%--- u — u . 8
11 1.” u . u u .n u n .
...... u u u . ----"11..--“ n u .m 1 .
1 llll*l|lllml “ " IIL. IIIII "Pl " n ._ ”I'll-.4101- . a
. --.----..- u u .. ..... ..-- m u " mun ... ..... u u .- . --
..... . . u . 1" u . . " -1.C.- . . u 1... 1..
m m m ”-1.-1--. m m -1--..--.. mini - ..... T--.-

. . . 11111 .11 . . . 111-.1111 u - “ 11P1111“ .

-. m m .u -.n ..... ..- m . m m - m m m ..... e
. . . . . 0 11A . . . . 11111111 . . . . 111111

m m " Willllmllc M nu“ m Illvlllllml m m " 11m lllll W1

" " -.. ..... . u . u . u u . u

. --.--- . n . . ..... w . . u . ..- n n .

U- - . n '19 - — . II c

.141 u . i u 1111 u n u " "11114.11 n u A... 111 .

. W . 1111 . . 11111... . . 11111

"NJ-"H” --.---.m. u u in ..... u. u m _ -T----m- - u

“ .11111411 “ . “1111“.111 . u . -._T111. . n u 1 4
. lllll . . . n 1.41 . n . -. . IL . . 111511
11.9 _ . -II . _ c .+ c - LII .

u u . u u . u n _ -1.--..”. _ n u u

. u u .1111111 u . . 111:.11 . |.r . 11L1111 " . .

. . . 11111 .- u . . 1111—11 p 1 11r11 “ . . n

n “-1111." u n #11141 H u -..111 u . u n . 11

. 1111 . 1 1 . . 11 . . . tall
1141 . n . . . . . . Ilbll . . n 11.1111 .

. . . . . 1 . . 111—111 . . . 1+111 . .

m m . --.. ..... m. - .1. n1... . -.. ..... m m ..----..--1. m m

n .T--._--- u 1. n ..--1.7-- " u u “ ..-----” u u n " --- 2
1 ----- . m N n u 1“- u u u n 1..---- ._ I u n “ L11-..-

. c P - . III¢ - u u . I'lrll " a . . IIVIII . .

u . P1111171 . u n . 111w11 " _ a " 11J.111 . u “
I'll-4”.“ 11111 . n u . .1111411 . . u . .111 . u . u

., l--. . . . ---. . _ . . .1 W _ . . . --
... ... n u u n u . n u u e" u u --.-1.-
T" a m u u wit-......- u u m u ----“.1--.“ u t" m L ----- ..- N.1

. m . . . 11111 . . . n . 111111 " . . .m. . 111.111 . . _ T

. . . 1116 . . _ . 1111.1 . . . . v.11 . . .

. . 11.11 . . . . 1111 . . _ . . .I 1 . . . O
. 111 . . . .111 . _ . 1.3 . . . h

- e v"- . - - . '4 . _ . - . u u - c P O
1... t _ . . . 11 . . . . . 1. .. . . .

u a u . u .n I114-..- - u . . 1|“llll 1“ — n w P o m

u .m. u " T111411 " m .m " "111-W111 " u w P m 0 0

. .1 . 1. 11111 . . . . t .1111J1 . . . - o ..l
“...-m... u m m -... m... m u . m m m ....- 2
.1.. . u u . -4... u 2 . . . _ m 4

. + u . . 1.11.. . . B . . _ w 5

- . ...II Iql - - . .—l O 6

— x . .IIIIH - . . P O O o

. .n n 1111.1 . u . — o O O «I

“L t 111 u u . . O O o 2
r1...— m . - . - — O 0 O 3

u u . . — O O 4.

. 2 u . p m 0 5 C300

u 8 _ . O m 0 6 my

.Io o m m 1

o o m 3

m w

6

Energy (keV)

. ora
3.1111138

. 111-am

=3“ 014mm.Ix

yam"

. x-ra

. VC

1 mm

4.20=

rim

hon!“
at 923 K for {our
W
sL film

181

 

Figure 4-21: A bright field micrograph of the (NiCuh'li precipitates taken almost parallel
to <100>32 direction showing the morphology and habit plane of the precipitates: The
misfit dislocation network (see enclosed rectangular area) lies on the (001) precipitates
interface.

182

 

Figure 4-22: (a) Bright field micrograph of the annealed SL film showing three precipitate
variants in matrix; (b) corresponding diffraction pattern showing [1(1)] and [010]

(N 1010211 zone axes parallel to <100>32, and (c-d) dark field images taken from spot c
and d m (b) respectively.

183

 

Figure 4-23: A bright field image of the annealed SI. film showing grain boundary ,
precipitates. The associated diffraction pattern is in an <112> zone axis of TizNi phase.

 

“‘1‘...-

------C--...‘--------.-

184

.--. --‘-------.---

E
P
P
i
00
F3
05
V

58.25 ....-

--OCCOOOOu-‘-------------b-------c --‘n—o--.--o.---b—cooooo-u---‘--.--.-----

 

...-........~............. 0

-J-----------

  

Ildmdd‘wdddm‘

1

‘

E 0.25 111
l .l L L

 

m
. ..... -. ...... 4-
. Ewe...

 

V -----.-----.‘------..-.-

   

.1

l

 

   

-‘1---.-..---.

 

.--. --4.-.---.----

--.---.----{-----------

p
p
p

---.“¢‘

      

   

 

-..-..-.--‘|--.--------.- --.---

 

 

 

 

5.88...

  

41 43 4s 47 49
Two theta (degrees)

39,

 

37

bounand(c)4hours.

24: X-raydiffractionpatmsoftheSLfilmannealedat923Kfor(a)0.25

Figure4-
hour- (b) l

185

    
 

~11 ‘ ‘ ‘1‘

.' ’P,//"'
mi

Figure 4-25: Electron micrograph showing the grain size and microstructure of a PML-16
film annealed at 923 K for one hour .

 

Figure 4-26: Electron micrograph of the grain boundary precipitates for a PML-16 film
annealed at 923 K for one hour. The associated diffraction pattern is in an <110> zone axis
of TizNi phase.

186

 

Figure 4-27: Electron micrograph of a PML-l6 film annealed at 923 K for one hour
showing TizNi precipitates observed in the grain interior. The associated diffraction pattern
corresponds to two (111) twin-related <123> zones of TizNi phase.

 

Figure 4-28: Electron micrograph of a PML.16 film annealed at 923 K for one hour
showing blade-like precipitates. The corresponding diffraction patterns are in an <110>32
zone axis and a [011] precipitate zone axis, indexed using an orthorhombic unit cell with a
= 0.441 nm, b = 0.882 nm and c = 1.35 nm-

 

 

Figure 4-29: (a) Bright field micrograph of a PML-16 film annealed at 923 K for one hour
showing misfit dislocations lying on the interface of NigTiz type precipitates. (b) 'lhe
corresponding diffraction pattern, taken from a rod-like precipitate located between two
Ni3T12 particles, corresponds to an <110> zone axis of TizNi phase

Figure 4-30: X-ray diffraction spectra for the PML- 16 films annealed at 923 K for (a) 5
minutes; (b) 1 hour and (c) 6 hours.

188

 

 

 

 

 

 

 

 

 

 

 

 

   

  

 

 

 

 

q I u w 1 I w u q a I “-0
m M m __ . w.
. m “ mm mm: m .
. m 6 m1 5 .
l ----------- .- ------------ 1.----..-- -.1-L -- 7
. m . . . w 4
1 ........... M LES: ._ L .
. M31253: - w fl 9
. . . .
. . . . m
. . m . m.
. . _ .w . d
.1 ----------- .4 ------------ 1. ---------------- 1 3 (\
. . 4 m
. e
. . m
.5.-28133 . O
I ----------- w-goos- - - - M ..w-
r- ----------- m. ----------- $6va - - -.-~ “we
r p p M p n p M p n p p 7
3
3.88...

(b) 1 hour and (c) 6 hours.

Figure4-30: X-raydifi’ractionspecu'afordrerr-l6filmsannealedat923xfor(a)5

minutes;

189

 

Figure 4-31: Electron micrograph of a PML-9 film annealed at 923 K for one hour
showing the grain size and microstructure.

 

Figure 4-32: Electron micrograph of a PML-9 film annealed at 823 K for one hour
showing the grain size and microstructure.

 

Figure 4-33: Bright field micrograph of a PML-9 film annealed at 923 K for one hour
showing the size and morphology of second phase particles in the grain interiors.

 

Figure 4—34: Selected area diffraction patterns in (a) [110]32 and (b) [111132 zone axes
showing that the precipitates have an fcc unit cell with a = 1.132 nm with (100)precipitate //
(100)32 and [0101precipitatc // [010132-

191

 

Figure 4-35: 3 g dark field electron micrograph showing strain contrast around the
precipitates.

Intensity

 

37 39 41 43 45 47 49
Two Theta (degrees)

Figure 4-36: X-ray diffraction spectra for the PML-9 films annealed at 923 K for (a) 5
minutes and (b) one hour.

192

 

 

 

’ Target Alloy L

 

ooooooooooooooooooooooooooooo

 

’-.-..-.------.---r----------------” ...............

 

a. 1 ----1------

 

 

 

 

 

 

 

 

 

 

 

 

 

Resistivity (Arbitrary Unit)

 

 

 

 

 

 

 

 

 

llllilllllLlllillllLllll (c)

73 123 173 223 273 323
Temperature (K)

 

Figure 4-37: Electrical resistivity as a function of temperature for (a) the target .03)
greSLfilmannealedat923Kfor0.25hourand(c)theSLfilmannealedat923K orone
our

1!)?!

(a)

 

.......

.......

     
  

      
  

   

 

  

---------1-

IIIIIII

 

 

 

[Target alloy I

 

 

15.21.1711111/111111

 
 

. . q . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . .) .
.
. . . .x .
. . . .0 .
. . . .3 .
. . .3 .
a ................ J.(---.1'
. u .f n
. .\l .A .
. .K .
. . . .
. .7 .
\I. .11 .
K" "3 u .
1. .I\ .
1. . . .
.
a. . . .
m . _
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
.
.

1
1
1
1
l

 

.-------------‘-------------J.-- ...-....--L

b------—------+--------——---J.---- ----

h-------------‘

 

P-4 p.

 

 

All 05 , 85 1|!
3o... .8:

13Ei3

iZiVEB

EZEEB

 

 

 

 

    

 

   

 

   

 

 

 

)
.hu
(

. . . m .

g o u o o

. . . . .

c c o c c

g n o o .

a . o c -

. . . . .

. . . . .

. . m . o.

m m m m u m
...... M ..--.iizzti
. . . . 1. .

m m m 1 m
m m m m
o a o o
. . . .

. . .
p p u
.............. m-.----m-.nnn- - --.-.--m- -
. . an .
u u 9 u
. n a o
. m ( u
u o
. hm .
g o
. .
--------------------- .------ .-------.---..-..
c u
m m
. m m
r . n m m m
.nu . .ux...n..1 . .
... . ”a m . m .e
o ( u .m
’ o c p
.4 K J 11111111 u 111111111111 . 11111111 “1 1 1.1
.4. . . . m “a
2 m m _ .m
an. n n . u 14
n . . . o .
... . . . . . .,_
.\. . . . . . 11
.I. m . . III.
. o .

 

 

AI 90 85.. IV
:0... .8:

33113

£953

2!

227'23

225533

25323

2!113

Temperature (K)

Figure 438: Differential scanning calorimetry data for (a) the target alloy and (b) the SL

fihnntarunrnalerliatSJJESll(:fior'cnne:hururz

194

 

 

 

 

 

 

 

 

 

3
§
1: V E. 3000cps
fl -: ‘
a g i
a : 5 A
9 e
A - VA] 297K
$5 _
L)
V
>1 _Jr\ .. JAKJ ‘99K
:1 ..
U)
C
or
H
E
K _- M 193“

 

 

ii
X

 

 

 

1111111111111rrrrLrlrrrrirrrr

35 4O 45 50
Two Theta (degrees) Cu Ka

 

 

Figure 439: {Gray diffraction spectra for the 81.. film annealed at 923 K for 0.25 hour
acquired at various temperature.

195

 

 

 

 

 

 

 

 

 

g ..
§
" 1200cps
FN
1:“ g
a "
g S
2 I V
A 5 I
(D
Q M
3 Z97K ’
>5
u
'17)
C
0
g 22K
11111111111111:1111111111111,
35 4O 45 50

Two Theta (degrees) Cu K a

Figure4-40: X-ray diffraction specu'afortheSLfilmannealedat923Kforonebour
aoquiredatvarioustemperature.

 

1 200 nm

Figure 441: TEM micrograph of an one-hour annealed SL film taken at 255 K. The
associated [012]32 diffraction pattern shows well defined 1/3 spots in the <321>32
direction.

 

Figure 4-42: Electron micrograph of one-hour annealed SL film taken at 255 K. The
associated diffraction patterned which can be identified as an [11 20]}: zone axis using a
hexagonal unit cell with a = 0.738 nm and c = 0.532 nm-

lli

197

 

Figure 443: Bright field' image of an one-hour annealed SL film taken at 160 K. The
corresponding diffraction pattern showing [100] zone of martensite.

198

 

Figure 444: Electron micrograph of an one-hour annealed SL film taken at 160 K
showing small martensite variants. The associated [3 21] diffraction patterns and dark filed

micrograph showing (115) twin relation between neighboring variants.

199

 

Figure 445: (a) Bright field micrograph showing plate-like R-phase variants; (b)

associated diffraction pattern revealing the R-phase variants oriented perpendicular or
parallel to [110]B2 reciprocal direction and (c-d) dark field' rrnages taken from spots c and
d in (b) respectively showing labyrinth-like R-phase domains ns-

 

 

 

   

 

  

   

 

 

A """"""""" f """""""""" 5'"""""'X§'iéf3’é’i£)' """""" 5' """"""""""
11:: . : y : :
C t """""""""" r """"""""" r """""""""" r """""""""""" g- ---------------
.5 3" """"""""" j """"""" """" W """ 5 """" Xirz'jifi'is """"""
fig """""""" i """"""""" i """"""""" i” - """"""""
S ---------------- ----------------- 5- ----------------- ------- 11-121-711) ----------- -
l l l l i l l l 1 Li I l l l l i l l l l l i 1 1 j l 1
73 133 193 253 313 373
' Temperature (K)

Figure4-46: ElecuicalresistivityasafuncfionoftemperanneofaPmrléfilmannealed
at923 K for one hourin avacuum of 2.6x10'3 Pa.

 

-i

......‘C.......‘

    
  

 

; M3 (275 K)

...................................................

  
 

  
  

Heat Flow

 

 

 

 

T 3 6 J/sec/mole .
o . 1
'35 i
g i
i E kamn i
j i '1 i
213 233 253 273 293 313
Temperature (K)

Figure4-47. Differential scanningcalorimen'ydataforaPWrIGfilmannealedat923K
foronebourinavacuumof2.6x10'3Pa. .

201

 

  
  
 
 
   

   

 

 

 

 

 

 

 

 

 

"""""""" “ 111(202 K)
3 t .
g ..............................................
LL
.4
a ..............................................................
'53, .................. T124210
- r
WP? . . . : 4 i :
""i"? """" ’ """" ' . """" 1 """" f """"
0.84 J/sec/g: E E ' ? :
l l 1 J r I 1 1 1
173 193 213 233 253 273
g i 3 -
z -------------------- As(244 K) -----------------------------
H 3 i 5 ‘
W 1 - 7
Q) ——: :
I ........ i, ....... i, ............................................................
0.21 J/sec/g§
-... - 1 . . - . Af(273'1<)
1 m 1 I 1 1 L 1 m
193 213 233 253 273 293

Temperature (K)

Figure4-48: Differential scarmingcalorimeu-ydataforaPme-mfilmannealedat823 K
forone hour in a vacuum of 2.6x10'4 Pa.

202

 

 

 

 

 

 

 

fi
W K‘. K K K
3 Q; m fl %
T 1| " .l V 2 2 2
z 2:»
2m” . aim—mi
288
38.: . l
55 fig
IANOOV 2:... AVNNV

3:2: 28.:

 

£83 -..lrdnullnHWI. ,. w, i...

50

45
Two Theta (degrees) Cu K a

40

 

A88 $28...

35

Figure 4-49: Cold-stage X-ray diffraction patterns of a PML-16 film annealed at 923 K in

a vacuum of 26le Pa.

203

 

 

 

 

 

 

 

 

 

823K

 

0 m
A J S a m
s m K
m 2 m
K K v K2 K U
m 8 ..l A 9 A 5 C h
m m... n... . H2 mu. “m
— — 4 A ) l
_ _ 1 .w 5 w W
4
saw—me F»: a. _ w. m
d f
28.8 ( m
38.: III... 1 m m
...E 5 “z”: 88 ..m
p 388 m o
w . .
288 2633 T . W
25:38.: W
:23 11.1.1 1-1.1”, a m
x mm
mm
Amauv 53:35 mm
m m
L.
Mm
a
m... .m

 

Figure 4—51: Electron micrograph and corresponding [1 34313 diffraction pattern of 823 K
annealed PML-16 film taken at 215 K showing R-phase variants with plate-like variants.

 

Figure 4~52z Electron micrograph and corresponding [023]}32 diffraction pattern of 823 K
annealed PML-16 film taken at 215 K showing plate-like R-phase variants. The [023]32

diffraction patterns is superimposed with two (100) twin-related [3145513 diffraction
patterns.

205

   

(3134)

 

Figure 4-53: Electron micrograph and corresponding [052mg diffraction pattern of 823 K
annealed PML-16 film taken at 215 K showing labyrinth-like variant-combination of the R-
phase

206

 

Figure 4-54: Electron micrograph of 823 K annealed film taken at 215 K showing
martensite plates in R-phase matrix. The corresponding diffraction pattern shows a [010]”
pattern and a [011]];2 pattern with 1/3 spots in <111> reciprocal directions.

207

 

 

Figure 4-55: Microsu'ucture of monoclinic martensite in annealed PML-16 films taken at
170 K: (a) band-like variants and (b) zigzagged variants.

208

 

Figure 4.56: TEM micrographs of self-accommodated monoclinic martensite taken at 170
K: (a) bright field image; (b) [010] and [21 1] diffraction patterns; (c) dark field image
taken from (lOOHOl 1) spots labeled in (b); and (d) [110] and [101] diffraction patterns.

 

A.

0.5 pm

 

Figure 4-57: An electron micrograph of the 823 K annealed film illuminated by converged
electron beam showing zigzagged martensite plates retained in the 82 matrix. The
associated diffraction pattern are in [0011M and [011]32 zone axes.

 

 

Figure 4-58: An electron micrograph of the 823 K annealed film illuminated by converged
electron beam showing zigzagged martensite plates retained in 32 matrix. The associated

diffraction pattern shows two (1 l l) twin-related [3 12]” zone, corresponding to two sets
of martensite plates.

210

[01 1132
( [1134]; )
[100g2
[011132 ( [85341; )

 
 
  
   

. . |(100)32

- Variantl El Variant 2

Figure 4-59: Schematic drawing showing the labyrinth-like morphology of Rophase.

211

171 (0.46)

 

 

 

100 101
(0.24) (0.37)

habit plane - (0.58112, -0.81382, -0.00089)
shear direction - [0.41720, 0.29652, 0.85908]

Figure4—60: Orientationdependenoeofmc minf _ - - -
solution 1 listed in Table 4-5. - shape or Type Itwrnnrng dam from

 

212

 

 

                   

      
   

           

 

   
  
   

 

        

 

 

 

 

 

.w
3 m
o c g 7 o o c o o
u n u ..3 m m m m m m
m m mm .. 9 W W W M
u u 2 u 1 W m m m
W M mm or... p ......... m. ............. m. ............... m. ...... .
. ooooooooooooooo f 151 u n . _
M .u Ar". m a m m m m ...
m m m % n m n m m u.
M W . . , ........ ........ m. ...... ”aim... ...... 1
m n u 1 M M ... M
.om ........ .m ..... inn m m m m u
7. . 12( fh m m m
4" u u e O. . . .
Q." .) u .. .u. mm m m m
”.1 x n a . 1 ..... ... ....... . .............. .. ...... .. ...... 1
n a u n r ma ” n u
m a m . m ... m _ . m
T ...... T- Arm ..... u ..... 1 a m acm .TV m m
n u n 2n u m m
n u “ .Wm ........ .. ........ T ........ . ...... .c ........ n. ...... 1
n u u . .mm m _ mo 2 m
n u " ... . V . m5 m m
u u n ma 2 “u u u
u u n .. .m Tm . "( n n
. . . 3 u n . n n
. ...... u ..... .. ...... .. ..... in . m u. Hm ..... a. ........ u. ...... 1
m m . .n , ........ $.57 . m m
. . .. e . n n u
m m m w my m . m m
m m m cm mm m m m m
n u u 3. “x J.|_ m m m
1 . - 2 3
9.5 55.3 .. mm. 1| so 85 Iv
Em

$583. so: so:

333

313

Temperature (K)
3 Pa.

293

 

273

Figure 4—62: Differential scanning calorimetry data for the PML-9 film annealed at 923 K

for one hour in a vacuum of 2.6x10

213

 

 

 

 

 

 

 

 

 

      

 

 

 

a
m 2 K
c K K . K W K; We
1| e \.I
T|L_ 2 ..w
1.28 258$ 4 m
2:: e
288 w
a
32:: lulllllllH '1‘. and
..E: 2338 m
o
4 w
188 2338 T
a 22712:
Cfiwoov .‘Illqllllr J
3

 

A33 €235

Figure4-63: Cold-stageX-raydifl‘r'actionpatternsofther’9filmannealedat923 Kin

Pa.

3

a vacuum of 2.6x10'

214

 

 

 

 

 

 

 

 

 

 

 

 

g .-
§
1900cps
E
a?
g ..
’8‘- .
l a:
U r A
V 1 8 1 6?
>5 ! a Is- N
3: , '- O x x
w + .2 s “so
5 3 8 "5
o :v
E r: 9
V 2215
2791‘-
z. 3“ 2.. T
F g; F
A m A
E 8 i
L ... Ni V 323K
JJJIlJJJIJJIllllllJIJlelJJjJ
35 40 45 50

Two Theta (degrees) Cu Ka

Figure 4—64: Cold-stage X-ray diffraction patterns of the PML-9 film annealed at 923 K in
a vacuum of 26le Pa. ‘

215

 

 

 

 

 

 

 

 

 

 

 

 

 

N ‘l—
at
E?
3 2300cps
12
A A
3 8‘
U 8 a:
V C:
I
s? E .25 a
g + x '5 8,5,3
or ,3 S c; ”:5!
E 9 5. 58
V 2ggt<
270K
2
2 N 2
if“ a; r?
K. .9". §
E ... :1 Ezlxmrk
rrrjrrrrlrrrrrrrrrw
35 40 45 50

Two Theta (degrees) Cu Ka

Figure 4—65: Cold-stage X-ray diffraction patterns of the PML-9 film annealed at 823 K in
a vacuum of 2.6x10-4 Pa.. .

 

Figure 4-66: Retained austenite observed at 160 K: (a) bright field micrograph; (b)
corresponding [110] diffraction pattern and (c) dark field image from (200)32 spot labeled
in (b).

217

 

Figure 4-67: Onhorhombic martensite at ambient ternperatrn'e: (a) dark field image; (b)
corresponding diffraction pattern in the [010].” and [011132 zone axes; (c)drffractr
pattern in a [101].“l zone axis.

218

 

Figure 467 (cont'd)

219

 

Figure «#68: TEM micrographs of self accommodated orthorhombic martensite taken at
room temperature: (a) bright field micrograph; (b) [110] diffraction pattern taken from area
B; (c) [110] diffraction patterns taken from area C; (d) [101] diffraction pattern taken from
area D; (e) dark field image taken from spot a in (b) showing four martensite variants a, b,
c and d; (1) dark field image taken from spot g in (d) at higher magnification showing three
martensite variants e, g and f; and (g) schematic drawing showing the variant distribution.

220

 

Figure 4-68 (cont'd)

221

 

Figure 4-68 (cont'd)

222

 

Figure 4-69: Self accommodated orthorhombic martensite taken at room temperature: (a)
bright field micrograph; (b) corresponding diffraction pattern showing three (111) twin-
related [101] patterns and (0) dark field image taken from spot a in (b) showing three
martensite variants a, b and c.

223

{111}

 

 

 

 

 

Figure 4-70: Crystallographic nature of the martensite variants shown in Figure 4-69.

224

 

Figure 4-71: Electron micrograph and corresponding [121] diffraction pattern showing
(111) twins in a thin-plate form of orthorhombic martensite taken at room temperature.

225

 

Figure 4-72: TEM micrographs of self accommodated orthorhombic martensite taken at
room temperature: (a) bright field micrograph; (b) [100] diffraction pattern taken from area

B showing one set of (011) twins and (c) [100] diffraction pattern taken from area C
showing another set of (011) twins.

226

 

Figure 4-73: Electron micrograph taken at 123 K showing the monoclinic martensite
inherits (011) twin-related orthorhombic variants. The associated diffraction pattern
showing the electron beam is almost parallel to [10mm

 

Figure 474: Self accommodated monoclinic martensite at 140 K: (a) bright field image;
(b) diffraction patterns of martensite variants a, b and c in [101] zone axes; (c) dark field
image shows variant a, b and c. Fine striations were observed inside the martensite
variants.

 

 

Figure 4—75: Self-accommodated monoclinic martensite at 123 K: (a) bright field
micrograph; (b) corresponding [110] diffraction patterns and (c) dark field image taken
from spot c in (b). The streaks of diffraction spots in [001] reciprocal directions result
from (001) twins.

229

 

Figure 4-76: Electron micrograph and corresponding [110]” diffraction pattern taken at
123 K showing (001) twins with an average spacing of4 nm.

230

 

 

 

 

 

 

Figure 447: (a) Untransfcrmed 82 structure in which a fct cell is delineated. (b)
orthorhombic cell transformed from the fct cell. Principle axes (1', j',k') are obtained by
45° rotation about the k axis.

231

 

 

 

 

010
3-9
5-11
10-4
6-12
2-8
1-7

 

 

Figure 448 Stereographic projection in [001]]32 direction showing the twinning
relationships between orthorhombic martensite variants.

232

111 (0.26)

 

 

 

100 1 10
(0.24) (0.49)

habit plane - (0.5506, 0.5506, 0.6274)
shear direction - [0.4437, 0.4437, -0.7787]

Figure 4.79: Orientation dependence of the shape strain for orthorhombic martensite.

10.

ll.
12.
13.

14.

15.

16.
17.

18.
19.

233

BIBLIOGRAPHY

W. J. Buehler and F. B. Wang, Ocean Eng, 1 (1968) 105.
C. M. Jackson, H. J. Wagner and RJ. Wasilewski, NASA-SP 5110 (1972).

M. H. Wu in ”Engineering Aspects of Shape Alloys", eds. T. W. Duerig,
K. N. Melton. D. Stackel and C. M. Wayman,69 (1 ), Butterwarth-Heinemann,

K. N. Melton in "Engineering Aspects of Shape Me Alloys", eds. T. W.
Duerig, K. N. Melton, D. Stockel and C. M. Wayman, l (1990), Butterworth-
Heinemann, London.

D. S. Grumman, S. NamandL. Changin"ShapeMemoryMaterialsand
Phenomena", eds.C.T. Liu,l-I. Kunsmann, K. OtsukaandM. Wuttig,259(l991),
Materials Research Society Symposium Proceedings, Vol. 246.

D. Johnson, J. Micromech. Micraeng., l (1991) 34.

W. J. Moberly and K. N. Melton in "Engineering Aspects of Shape Memory
Alloys", eds. T. W. Duerig, K. N. Melton, D. Stockel and C. M. Wayman, 46
(1990), Butterworth-Heinemann, London.

0. Mercier and K. N. Melton, Metall. Trans, 10A (1979) 387.

R. E. Reed-Hill, Physical Metallurgy Principles, 611 (1973), Van Nostrand Reinhold
Company, New York.

Zog‘ishiyama, ”Martensitic Transformation”, 276 (1978), Academic Press, New
Y .

M. W. Burkart and T. A. Read, Trans. AIME, 197 (1953) 1516.
D. S. Lieberman, M. S. Wechsler and T. A. Read, J. Appl. Phys, 26 (1955) 473.

CM. Waymanin"Engineer-rn gAspectsafShape Memory Alloys",eds.T. W.
Duerig, K. N. Melton, D. Stackel and C. M. Wayman, 3 (1990), Butterworth-
Heinemann, London.

11 c Tang and C. M. Wayman, Acta Metall., 22 (1974) 887.

K. ShimizuandK.Otsukain ”ShapeMemaryEffectsinAllays”,ed.J.Perkins,59
(1975), Plenum. New York.

K. N. Melton and O. Mercier, Metall. Trans, 9A (1978) 1487.

3.615.. Buehler and R. C. Wiley: Nickel-Base Alloy, U. 8. Patent 3174851, Mar. 23,

R. G. deLangeandJ. A. Zijderveld,J. Appl. Phys., 39 (1968) 2195.

figsukaT. Sawamura,K.ShimizuandC. M.Wayman, Metall. Trans, 5(1971)

21.

135-‘3

32.
33.
34.
35.

37.
38.
39.

41.
42.
43.

45.

47.

49.

234
K. Otsuka and K. Shimizu. Scfipta Metall., 4 (1970) 469.

R. A. Swalin, "Thermodynamics of Solid", 33 (1972), John Wiley & Sons, Inc,
New York.

It Warlimant, L. Delaey, R. v. Krishnan and H. Tas, J Mater. Sci., 9 (1974) 1545.
s. Miyizaki, K. Otsuka and C. M. Wayman, Acra Metall., 37 (1989) 1873.

J. s. Bowles and J. K. Mackenzie. Acta Metall., 2 (1954) 129.

M. s. Wechsler, a. s. Leiberman and 'r. A. Read. Trans. AIME, 197 (1953) 1503.
a. s. Lieberman, M. s. Wechsler and r. A. Read, J. Appl. Phys, 26 (1955) 473.

gfishiyama, "Martensitic Transformation", 355 (1978), Academic Press, New

E. C. Bain, Trans AIME, 70 (1924) 25.

Binary Alloy Phase Diagrams, vol. 2, ed. T. B. Massalski, 1768 (1987), American
Metals Society, Metals Park.

P. Duwez and J. L.Tay10r, Trans, AIME, 188 (1950), 173 .

D. M. Pool and W. Hume-Rathery, J. Inst. Met, 83 (1954-1955), 473.

D. Koskimaki, M. J. Marcinkawski and A. Sastri, Trans AIME. 245 (1969), 1880.
S. P. Gupta. K. Mukherjee and A. A. Johnson, Mater. Sci. Eng, 11 (1973). 283.
H. Margolin, E. Race and J. P. Nielsen, Trans. AIME, 197 (1953), 243.

G. R. Purdy and J. G. Parr, Trans. AIME, 221 (1961), 636.

R. J. Wasilewski, S. R. Butler, J. E. Hanlon and D. Warden, Metall. Trans, 2
(1971), 229.

G. F. Bastin and G. D. Rieck, Metall. Trans, 5 (1974), 1817.
M. Nishida, C. M. Wayman and T. Honma, Metall. Trans, 17A (1986) 1505.
T. Saburi, S. Nenno and T. Fukuda, J. Less-com. Met, 125 (1986) 157.

J. Beyer, R. A. V. D. Brakel and J. R. T. Lloyd, Prac. Int. Conf. on Martensitic
Transformation (1986) 703.

M. V. Nevitt, Trans. AIME, 218 (1960), 327.
G. Chattapadhyay and H. Kleykamp, Z Metallkd. 74 (1983), 182.

R.Kainuma,M.MatsumataandT. HonmsProc. Int. Conf.onMartensitic
Transformadon (1986) 717.

1171 J. J. van Lao, G. F. Bastin and A. J. H. Leenen, J. Less-com. Met, 57 (1978)

R. H. Brickell, K. N. Melton and 0. Mercier, Metall. Trans, 10A (1979) 693.

A. R. Pelton, P. H. Adler, W. Yu, R. Holt and R. McCanville, Proc. XII th
IntCong. Electron Microscopy (1990).

L. Chang and D. S. Grumman, Scripta Metall. Mater., 25 (1991) 2079.
G. D. Sandrock. A J. Perkins and R. F. Hehemann. Metall. Trans, 2 (1971) 2769.
H. C. Ling and R. Kaplow, Metall. Trans, 11A (1980) 77.

57.

58.
59.

61.

63.

$.33

67.

68.
69.
70
71.
72.

73.

235

891845)};wang, M. Meichle, M. B. Salaman and C. M. Wayman, Plil. Mag. 47

E. G00 and R. Sinclair, Acta Metall., 33 (1985) 1717.

S. K. Wu ans C. M. Wayman, Acta Metall., 37 (1989) 2805.

S. Miyzaki and C. M. Wayman, Acta Metall., 36 (1988) 181.

K. Otsuka, T. Sawamura and K. Shimizu, Phys. Stat. Sol.(a), 5 (1971) 457.
G. M. Michal and R. Sinclair, Acta Cryst., B37 (1981) 1803.

W. Buhrer, R. Gatthardt. A. Kulik. O. Mercier and F. Staub, J. Phys. F: Met.
Phys, 13 (1983) L77.

:0490d0h, M. Tokanami,pS. Miyazaki and K. Otsuka, Acta Metall., 33 (1985)

C. M. Hwang, M. B. Salaman and C. M. Wayman, Phil. Mag. A, 47 (1983) 177.
S. K. Wu and C. M. Wayman, Mater. Sci. Eng, 96 (1987) 295.

K. M. Knowles and D. A. Smith, Acra Metall., 29 (1981) 101.

C. Y. Xie, L. C. Zhao and T. C. Lei, Scripta Metall. Mates, 24 (1990) 1753.

R. Sinclair, AIP canf. Prac., 53 (1979) 269.

91. 3Matsumoto, S. Miyazaki, K. Otsuka and H. Tamura, Acra Metall., 35 (1987)
7.

31.931511; 1;, S. Hanada and T. Honma, Bull. Res. Inst. Min. Met., Tohoku Univ., 41

S. Miyazaki and K Otsuka. Metall. Trans, 17A (1986) 53.

{.198 Proft, K. N. Melton and T. W. Duerig, Proc. of the Inter. MRS Conf., Tokyo
7).

K. Otsuka in Engineering Aspects of Shape Memory Alloys, eds. T. W. Duerig, K.
N. Melton, D. Stockel and C. M. Wayman, 36 (1990), Butterworth-Heinemann,
Landon.

M. Nishida and T. Honma, Scripta Metall., 18 (1984) 1299.

R. H. Bricknell, K. N. Melton and O. Mercier, Metall. Trans, 10A (1979) 693.
R. H. Bricknell and K. N. Melton, Metall. Trans, 11A (1980) 1541.

T. Tadaki and Wayman, Metallagraphy, 15 (1982) 233.

Y. Shugo, F. Hasegawa and T. Honma, Bull. Res. Inst. Min. Met., Tohoku Univ.,
37 (1981) 79.

1Y1.7Shugo and T. Honma, Bull. Res. Inst. Min. Met., Tohoku Univ., 43 (1987)

73a. K. Tsuji and K. Namura, Scripra Metall. Maren, 24 (1990) 2037.

74.
75.
76.
77.

10 H. Buschow, J. Phys. F: Met. Phys, 13 (1983) 563.

K. H. Buschow, J. Appl. Phys, 56 (1984) 304.

K. ntutc. H. Fujita, M. Ikeda and s. Yamashita, IEEE CH2832-4 (1990) 38.

1. Battezzati, G. Cocoa, L. Schiffini and S. Enza, Mater. Sci. Eng., 97 (1988)121.

236
78. R. B. Schwarz. R. R. Petrich and C. K. San. Non-Cryst. Solids. 76 (1985) 281.

79. J. D. Busch, A. D. Johnson, C. H. Lee and D. A. Stevenson, J. Appl. Phys, 68
(1990) 6224.

80. G. F. Bastin and G. D. Rieck, Metall. Trans 5 (1974) 1827.
81. W. L. Johnson, Mater. Sci. Eng., 97 (1988) 1.

82. E. D. McClanadhan and N. Laegreid in "Sputtering by Particle Bombardment III",
eds R. Behrisch and K. Wittmaack, 339 (1991), Spnnger-Verlag, Berlin.

83. P. Sigmund in "Sputtering by Particle Bombardment I”, eds. R. Behrisch, 9 (1981),
Springer-Veriag, ' ‘

84. 11.11. Anderson and H. L. Bay in "S uttering by Particle Bombardment I", eds. R.
Behrisch, 145 (1981), Springer-Vet g, Berlin.

85. R. Kell in ”Ion Bombardment Modification of Surfaces", eds O. Auciello and R.
Kelly, 01 (1984), Elsevier Science, Amsterdam.

86. J. Bohdansky, J. Roth and H. L. Bay, J. Appl. Phys, 51 (1980) 2861.

87. J. A. Thronton in "Deposition Technologies for Films and Coatings", ed. R. F.
Bunshah, 170 (1982), Noyes Publications, Park Ridge.

88. B. M. Clemens, J. Appl. Phys, 61 (1987) 4525.

89. 16341Hendersan, M. H. Brakdsky and P. Chaudhari, Appl. Phys. Lett, 25 91974)

90. G. Betz and G. K. Wehner in ”Sputtering by Particle Bombardment II", eds. R.
Behrisch, 11 (1983), Springer-Verlag, Berhn.

91. P. Ziemann and E. Kay, J. Vac. Sci. Technol., A1 (1983) 512.

92. J. M. E. Harper in ”Thin Film Processes", ed. J. L. Vessen and W. Kern 175
(1978), Academic Press, Inc., New York.

93. H. F. Winters and D. Horne, Phys Rev. B, 10 (1974) 55.

94 F. A. Smidt, Intern. Mat. Rev., 35 (1990) 61.

95 J. K. Hirvonen, Mat. Sci. Reports, 6 (1991) 215.

96. J. M. E. Harper and R. J. Gambina, J. Vac. Sci. Technol., 16 (1979) 1901.
97 J. J. Cuama and R. J. Gambina, J. Vac. Sci. Technol., 14 (1977) 152.

98 M. Marinav, Thin Solid Films, 46 (1977) 267.

99. L. Pranevicius. Thin Solid Films, 63 (1979) 77.

100. G. E. Lane and J. C. Anderson, Thin Solid Films, 26 (1975) 5.

101. 12112.1 Huang, G. Lim, F. Parmigiani and E. Kay, J. Vac. Sci. Technol., A3 (1985)

102. B. H. Hirsch and I. K. Varga, Thin Solid Films, 69 (1980) 99.

103. R. W. Knoll and E. D. McClanahan, J. Vac. Sci. Technol., A1 (1983) 271.

104. P. J. Martin, R. P. Netterfield and W. G. Sainty, J. Appl. Phys, 55 (1983) 235.
105. M. L. Tarng and G. K. Wehner, J. Appl. Phys, 42 (1971) 2449.

106. E. V. Kornelsen, Can. J. Phys, 42 (1964) 364.

237
107. C. M. Gilma'e, A. Haeri and J. A. Sprague, Thin Solid Films, 165 (1988) 359.
108. S. M. Rossnagel, R. S. Robinson and H. R. Kaufman, Surf. Sci., 123 (1982) 89.
109. B. M. Clemens, Phys Rew. B, 33 (1986) 7615.
110. R. J. Gaboriaud and J. Delage, Thin Solid Films, 200 (1991) 275.
111. x. C. Zheng and R. A. Dadd. Nuclear Instr. Meth.(in press)
112. J. J. Kim, P. Maine and D. A. Stevenson, Scn'pta Metall., 20 (1986) 243.
113. A. K. Kai and R. S. Bhattacharya, Mat. Sci. Bngr., 85 (1987) 139.

114. J. D. Busch, A. D. Johnson, D. E. Hadgson. C. H. Lee and D. A. Stevenson
presented in ImMAT (1989). '

115. J. a. Busch, A. a. Johnson, C. H. Lee and D. A. Stevenson, J. Appl. Phys, 68
(1990) 6224.

116. J. D. Busch,M.H.BerksanandA.D.Johnson,presentedinMat. Res. Soc.
Meetingat Anaheim(1991). ,

117. 1c Kurihayashi, M. Yasaaki and S. 08am. 0128324/90 (1990) 217.

118. J. A. Walker, K. J. Gabriel and M. Mehregany, The 5th Int. Conf. on Solid State
Sensors and Actuators, Switzerland, June 8, 1989, 123.

119. J. A. Walker, K. J. Gabriel and M. Mehregany, Sensors and Actuators, A21-A23
(1990) 243.

120. J. G. Miller and D. I. Bolef, J. Appl. Phys, 39 (1968) 5815.

121. J. C. M. Li in "Ultrarwid Quenching of Liquid Alloys", ed. H. Herman, 326 (1981),
Academic Press, New York.

122. H. E. Kissinger, Analysical Chemistry, 29 (1957) 1702.

123. a. w. Henderson, J. Non-Crystalline Solids, 30 (1979) 301.

124. K. Matusita and s. Sakka, Physics and Chemistry of Glasses, 20 (1979) 81.
125. J. A. 'lhomton, J. Vac. Sci. Technol., 11 (1974) 666.

126. J. L.Vossen and J. J. Cuama in 'Thin film processes”, eds. J. L. Vossen and w.
Kern, 59 (1978), Academic Press New York.

127. H. S. Chen and K. A. Jackson in "Ultrarapid Quenching of Liquid Alloys", ed. H.
Herman, 215 (1981), Accaemic Press, New York.

128.1}2Murri, F. Gozza, N. Pinto and L. Schiavulli, J. Nan-Crys. Solids, 127 (1991)

129. JCPDS file 38-1420. .

130. n S. Chen, Rep. Prog. Phys, 43 (1980) 353.

131. E. Balanzat and J. Hillairet, J. Phys F: Met. Phys, 12 (1982) 2907.

132. H. S. Chen, Appl. Phys. Lett., 29 (1976) 12.

133. H. A. Davies, J. Aucote and J. B. Hull, Scripta Metall., s (1974) 1179.
134.11. a. Schwarz, Mater. Sci. Eng., 97 (1988)

135. I. 0. Popoola. J. P. Villain and P. Maine, Mater. Sci. Eng, A115 (1989) 133.

238

136. R. A. Kant, S. A. billion, a. D. Sartwell and J. A. Sprague, Mater. Res. Soc.
Symp. Prac., 128 (1989) 79.

137. a. A. Kant and a. a. Smell. J. Vac. Sci. Technol., A6 (1988) 2945.

138. V3181? Technology: Its Foundation, Formulae and Tables, Leybald-Heraeus Inc.

139. R. Behrisch in ”Sputtering by Particle Bombardment 1", ed. 1, n. Behrisch, 1 (1981),
Springer-Vedag, Berlin.

140. H. L. Bay and J. Bohdansky, Appl. Phys, 19 (179) 421.

141. Y. Yamamura and J. Bohdansky, Vacuum, 35 (185) 561.

142. J. L. Vomit, J. Vac. Sci. Technol., 8 (1971) s12.

143. a. Bruning, 2. Altaunian J. 0. Strom-Olsen, Mater. Sci. Eng., 97 (1988) 317.

144. L. Chang, unpablished data.

145. u. Koster, z. Metallkd. 75 (1984) 691.

146. M. Nishida and C. M. Wayman, Metall. Trans, 18A (1987) 785.

147. D. A. Porter and K. E. Easterling, "Phase Transformations in Metals and Alloys",
159 (1981), Van Nostrand Reinhold Company, New York.

148. O Mercier, E. Torak and B. Tirbanod, Prac. Inter. Conf. on Martensitic
Transformation, (1979) 702.

149. J. Raa, Y. He and R. Ma, Metall. Trans, 21A (1990) 1322.

150. C. M. Hwang and C. M. Wayman, Scripta Metall., 17 (1983) 1449.
151. C. Kim and C. M. Hwang, Scripta Metall., 21 (1987) 959.

152. S. K. Wu and H. C. Lin, Scripta Metall. Maren, 25 (1991) 1529.

153. Giggglfiang, M. Meichle, M. B. Salamon and C. M. Wayman, Plil. Mag. 47

154. T Saburi, M. Yoshida and S. Nenno, Scripta Metall., 18 (1984) 363.
155. P. L. Ferraglio and 1c Mukherjee, Acta Metall., 22 (1974) 835.

156. A1 £51123“, E. F. Heintzelman, W. J. Buehler and J. V. Gilfrich, Trans. ASM, 58

157. A. Hedayat, J. Rechien and K. Mukerjee, J. Mater. Sci., 27 (1992) 5306.

158.‘£.1§aburi, T. Komatsu, S. Nenno and Y. Watanabe, J. Less-Cam. Met., 118 (1986)

159. B. J. Walles, Master Thises, Michigan State University, (1993).
160. N. Afify, J. Non-crystalline Solid 126 (1990), 130.

"I11111111111111“