 

 

This is to certify that the
thesis entitled

NEW METALLURGICAL METHODS FOR JOINING NITINOL

presented by

a)

at!
>- {‘3 z;
E c 2 JOHN W. FOLTZ IV
a: w 0;
5&3E
.J .9 D

E has been accepted towards fulﬁllment

of the requirements for the

 

 

 

Masters degree in Materials Science

  

 

 

Major Professor’s Signature

8124/07

Date

 

MSU is an afﬁnnative-action, equal-opportunity employer '

 

 

. ”wink”... _,, .4..-

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

DATE DUE DATE DUE DATE DUE

 

NEW METALLURGICAL METHODS FOR JOINING NITI
By

John W. Foltz IV

A THESIS
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
MASTER OF SCIENCE

Department of Chemical Engineering and Materials Science

2007

ABSTRACT
NEW METALLURGICAL METHODS FOR JOINING NITI
By

John W. Foltz IV

Nickel titanium is an intermetallic alloy used in many applications because it
demonstrates shape memory and pseudoelasticity at near room temperature. Historically,

methods of joining nickel titanium have had limited success.

This thesis reports on a variety of preliminary experiments on creating metallurgical
bonds with NiTi that take advantage of a pseudo-binary equilibrium system between NiTi
and Nb. This system allows pure Nb to act as a melting point depressant for NiTi. Bonds
between NiTi sections have exhibited high strength and ductility. The technique has been
used to create low density honeycomb topologies capable of geometrically amplifying

pseudoelastic and shape memory effects.

The liquid formed by this pseudo-binary is highly reactive, and aggressively wets many
metallic and non-metallic surfaces. DC magnetron sputtered multilayers of NiTi and Nb
can be used to control the reactivity of the liquid. These multilayers themselves exhibit
shape memory and high transformation temperatures, which demonstrates achievement of
good composition control. When combined with niobium cladding, multilayers also
allow NiTi to be bonded with many other materials without the formation of undesired

intermetallic compounds.

Copyn'ght
JOHN w. FOLTZ Iv

2007

Dedication

To mom and dad. Thanks for the support.

iv

ACKNOWLEDGEMENTS

I would like to acknowledge the many people whom I learned a great deal from while
working on this thesis. Thank you Dr. Grummon, Dr. Shaw, Dr. Crimp, Dr. Case and Dr.
Bieler. I would like to also thank the undergraduate students that have helped with this
work, including Jason Carter, Neil Murphy, Corey O’Connell, Scott Gillard, Arthur
Matteson and Jason Johnson. Finally, I would like to thank my family and Stacey

Schroeder for supporting me through this process.

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES VIII
LIST OF FIGURES IX
1 INTRODUCTION 1
1.1 Shape Memory Processes 2

1.2 Bonding N iTi 6

1.3 Organization 16

2 JOINING SIMILAR NITINOL 17
Discovery of 17

2.1 N iTi-Nb reactive brazing with N iTi tubes 17

2.2 Construction of pseudoelastic tubes in a low-density topology ........... 28

2.3 Construction of smooth corrugation honeycomb structures ............... 31

2.4 Construction of hexagonal corrugation structures 37

3 REACTION CONTROL USING MULTILAYERS - 47
3.1 Multilayers 49

3.2 Application of multilayers to spot welding 56

4 BONDING DISSIMILAR NITI SECTIONS 82
4.1 Bonds between NiTi thin ﬁlms and wrought Nitinol 83

4.2 Thin ﬁlm to Porous N itinol 88

4.3 Thin film to Chemical Nitinol 93

4.4 Porous NiTi to Wrought Nitinol 95

4.5 Porous Nitinol to Chemical N itinol 97

4.6 Shape memory to pseudoelastic N itinol 101

5 JOINING NITI TO OTHER MATERIALS - 107
5.1 Alumina 108

5.2 Molybdenum 112

5.3 Cobalt Chrome 117

5.4 FezAlV -124

5.5 Boron Carbide 132

5.6 Magnesium Oxide 137

5.7 Tantalum 140

5.8 Ti-6AI-4V 144

5.9 Tungsten 154

5.10 304 Stainless Steel 161

6 CONCLUSIONS 165

 

7 FUTURE WORK

APPENDICES

 

A.1 Brazing using Centorr Annealing Furnace
A.2 Brazing using MILA ULVAC rapid thermal heating system ...........

A.3 Polyimide backed thin films

 

 

A.4 Freestanding thin ﬁlm creation

A.5 Multilayer creation

 

 

A.6 Oxide Removal Etch

 

A.7 Metallurgical Etch

 

A.8 Tacking NiTi to Nb using AC spot welding

A.9 Post Braze Heat Treatment

BIBLIOGRAPHY

 

 

vii

168

170
170
171
172
183
185
186
188
188
189

190

LIST OF TABLES

Table 1: DSC results from post brazed pseudoelastic strip NiTi. ..................................... 43
Table 2: DSC analysis of phase change temperatures in sample 111306A ....................... 53
Table 3: Experimental parameters covered in experiment series 121906A. ..................... 57
Table 4: Experimental procedure for AC spot weld samples 022107. .............................. 58
Table 5: Experimental parameters for series 030107. ....................................................... 58
Table 6: Experimental parameters for series 040107. ....................................................... 59
Table 7: Experimental Procedure of experiment series 04122007. ................................... 63
Table 8: Experimental procedure and results of experiment series 03272007. ................. 77
Table 9: Spot welding parameters for sample 03272007-5. .............................................. 77

Table 10: Transformation temperatures from DSC results corresponding to Figure 46. ..84

Table 11: EDS analysis of spots shown in Figure 48. ....................................................... 90
Table 12: EDS results of chemical NiTi to thin ﬁlm bond, shown in Figure 49 ............... 94
Table 13: EDS analysis ofX-ray scans shown in Figure 57. ......................................... 110
Table 14: Sputtering conditions for sample O62306D ..................................................... 118
Table 15: EDS analysis, in atomic percent, of X-ray spectra in Figure 70. .................... 130
Table 16: EDS analysis of the areas shown in Figure 78. ............................................... 143
Table 17: EDS analysis, in atomic percent, of spots indicated in .Figure 83. ................. 151
Table 18: EDS Data collected from spots shown in Figure 89. ...................................... 159
Table 19: EDS data collected from spots in Figure 90. ................................................... 160
Table 20: EDS semi-quantitative analysis corresponding to spots in Figure 92. ............ 164
Table 21: Transformation temperatures for thin ﬁlm 0523 07A. Obtained ﬁom DSC

results displayed in Figure 99. ......................................................................................... 180
Table 22: Sputtering conditions for ﬁ'eestanding NiTi thin ﬁlms. .................................. 184

viii

LIST OF FIGURES

Figure 1: Schematic stress strain curve of a pseudoelastic Nitinol. The upper curve is the
loading curve, while the bottom is the behavior of the material upon unloading. Plastic
deformation can begin at the completion of the pseudoelastic plateau on the stress strain
curve. ................................................................................................................................... 3

Figure 2: Ti-Ni phase diagram in atomic percent Nickel. ................................................. 13

Figure 3: Ternary phase diagram between Ni, Ti and Nb displaying solidus surfaces and
terminal phases. ................................................................................................................. 14

Figure 4: Stress-strain response of a NiTi butt-joint bonded using the Ni-Ti-Nb ternary
system. Fracture surface shows ductile cup and cone type fracture, likely due to the Nb
phase. ................................................................................................................................. 15

Figure 5: SEM image of NiTi cross-section that has been wetted by NiTi-Nb liquid in a
vacuum furnace. Upon freezing, faceted precipitates form, which are morphologically
identiﬁed as TizNi .............................................................................................................. 16

Figure 6: Macroscopic bond of NiTi-NiTi using Nb ﬁller metal. Note that the eutectic
liquid wets the NiTi surface without ﬂuxes or other wetting agents. ................................ 20

Figure 7: Compressive stress strain curve for a pair of joined NiTi tubes. Loading is
shown as solid lines, unloading as dashed lines. The pseudoelastic effect of nickel
titanium is ampliﬁed from approximately 7% reversible strain. Little permanent
deformation was observed up to 30% compression. ......................................................... 21

Figure 8: (Top left) Secondary electron image of bond between NiTi tubes using Nb ﬁller
metal. (Top right) Nb L-a X-ray ﬂuorescence showing no signiﬁcant solid state diffusion
of Nb into NiTi. (Center right) Ni K-a X-ray ﬂuorescence shows a lower concentration Of
Ni inside the eutectic than inside the NiTi. (Lower right) Lower concentration of Ti in the
proeutectic phase than in the base NiTi, which is shown using the Titanium K-a map.
(Center Left) Low magniﬁcation SEI of the bond showing the presence of a eutectic
structure and proeutectic phase. (Bottom Left) High magniﬁcation image of eutectic
structure and proeutectic phase .......................................................................................... 24

Figure 9: Pseudo-binary eutectic between NiTi and Nb. The eutectic point lies at
approximately 26 atomic percent Nb and a temperature of 1170°C. The maximum
reported solubility of Nb in NiTi is near 6 atomic percent. ............................................... 25

Figure 10: SEI of NiTi —NiTi bonded tube fracture surface. Dimpling present throughout
the entire fracture surface implies a ductile tearing mode fracture. .................................. 26

Figure 11: High magniﬁcation SEI of fracture surface inside the bond area. Dimpling
implies ductile fracture inside the bond ............................................................................. 27

ix

Figure 12: Macro photo Of close packed array of NiTi pseudoelastic tubes bonded with
127 um Nb wires. (A) Etching effect on Nb shown in the cross-section. ........................ 29

Figure 13: Etching effect of 127 um Nb wires cutting through the NiTi tubes, which is
likely driven by capillary forces. Wires seem to have migrated and etched against the
force of gravity, suggesting that gravity does not play a major role at the length scale of
the wires ............................................................................................................................. 30

Figure 14: Schematic of the corrugation design for the NiTi strip material. This
formation was termed the sine corrugation due to its resemblance to a sine wave, but it is
deﬁned by alternating circular radii ................................................................................... 33

Figure 15: Schematic of corrugation construction using (a) 127 um Nb wires (b) 54 um

Nb foils. While the foils are thinner, they are considerably wider than the wires, thereby
supplying more Nb to the liquid. The foil does not cut through the NiTi, unlike wires. ..34

Figure 16: (Top) Stress strain curves of cyclic loading OfNiTi sandwich. The sample
was strained up to 50% with little permanent deformation, demonstrating how the
pseudoelastic effect can be geometrically ampliﬁed. (Bottom) Time versus strain plot,
which shows the loading cycles for the sample ................................................................. 35

Figure 17: Typical heating schedule for initial sine wave corrugations. Furnace cooling
continued until the sample was removed at room temperature. ........................................ 36

Figure 18: (a) Schematic of shape-set tool honeycomb corrugation made of stainless
steel. All sides of the corrugation are 3 m. At the intersection of the 3 mm lengths is a
0.18 mm kerf. (b) Stacking sequence showing 0.051 mm thick Nb pieces. (c) Final
sandwich assembled and awaiting brazing. ....................................................................... 38

Figure 19: Sample 0308053, honeycomb assembled using a small thermocouple spot
welding unit to tack the pieces of NiTi to Nb foil. The array was placed in a
molybdenum ﬁxture as shown, with a small weight applying a compressive load to
ensure the NiTi strips were well aligned. .......................................................................... 39

Figure 20: This image shows the ﬁxtures and set-up for sample 03 0805B, for honeycomb
production. The honeycomb was encased in a molybdenum ﬁxture with a molybdenum
weight applying a compressive load to align the NiTi strips. ........................................... 40

Figure 21: Heating schedule for 030905D. As shown, the sample was heated to 1185°C
for a dwell time of 5 minutes in the Centorr vacuum furnace. During this process, the
pressure began at 4.4 uTorr, and reached a maximum pressure of 24 uTorr around 100°C
before decreasing. .............................................................................................................. 41

Figure 22: If the NiTi strips are misaligned, a meniscus forms as shown. The meniscus,
being rich in Nb, has a shinier and smoother surface. ....................................................... 43

Figure 23: SEI collage of 032805A. Bond taken from a honeycomb sandwich, which
used a 54 um Nb foil to bond pieces of NiTi together. ..................................................... 44

X

Figure 24: DSC results ﬁom pseudoelastic NiTi. From top to bottom, the austenite start
and ﬁnish temperatures shift as thermal processing continues .......................................... 45

Figure 25: (a) Strain vs. time plot of sample 030905D showing the loading times and
strains per cycle. Sample was strained to a maximum of 60% strain. Low permanent
deformation from each loading cycle can be seen by the unloading area of the curve.
After 60% strain, the sample had 8% permanent strain. (b) Stress-strain curve for each
increasing cycle number. Note the sample pseudoelastically deforms even when
compressed to 60% strain, which can be seen by little permanent deformation upon
unloading. (c) ( Top to bottom) Images of the sample before any compressive loading,
deformed on cycle 12 to 60% strain, and unloaded ﬁ'om the 60% strain of cycle 12. ...... 46

Figure 26: BE] of a cross-sectioned piece of multilayer 111306A using a 25keV ﬁeld
emission gun. The bright phase inside the multilayer ﬁlm is niobium and the darker grey
phase is nickel titanium. In total, the ﬁlm is made of 127 discrete layers and is 34 pm
thick. .................................................................................................................................. 53

Figure 27: One exothermic peak upon heating (bottom half of curve) showing the
austenite start and ﬁnish temperatures. Two endothermic transformations occur during
cooling, which can only be interpreted as formation of the R-phase upon cooling between
60°C and 40°C, followed by the B19’ martensite phase between 25°C and 5°C. ............ 54

Figure 28: X-ray diffraction results from sample 111306A after annealing for 3 hours at
550°C. Theoretical Nb and NiTi (BZ) X-ray peaks are shown at the top, while actual
results are on bottom. There are no signiﬁcant amorphous peaks, and based on thermal
shape recovery, the sample appears fully crystalline. ....................................................... 52

Figure 29: Comparison between eutectic structures produced by 25 um Nb foil (top) and
a 34 um multilayer ﬁlm (bottom). The eutectic region is larger in the Nb foil due to the
base metal NiTi melting during the brazing process. ........................................................ 55

Figure 30: Schematic drawing of spot welding design. (A) Spot weld nugget (light grey)
formed by local melting at the interface between the two pieces of the sample (B).
Melting is limited to the interface by rapid heat conduction (lines) through the electrodes
(C). The electrodes apply a compressive stress to the sample .......................................... 64

Figure 31: Grain boundary niobium phases and small precipitates of Nb inside the NiTi
base metal of 121906A. Temperatures exceeded 1300°C locally, melting the base metal.
Black spots are etch pits. ................................................................................................... 65

Figure 32: Sample 121906A2 (A) SE1 of multilayer transitioning from unmelted to fully
melted, demonstrating the processing temperature gradient. (B) The multilayer did not
adhere well to the base metal, which could be a result of increased heat conduction or less
joule heating in the base metal. ......................................................................................... 66

Figure 33: SEI of post-ﬁactured 022107D. Specimen broke at the edge of the spot weld.
Eutectic solidiﬁed multilayer can be seen at the fracture surface. This implies that the
spot weld heated outside of the electrodes (overheated). .................................................. 67

xi

Figure 34: Ni4T i3 precipitates in needle shaped morphology can be seen in the center of
this SEI of sample 022107D. Niobium precipitates and residual porosity decorate the
surrounding microstructure ................................................................................................ 68

Figure 35: Sample 010307A. Higher spot welding voltage causes increased residual
porosity, grain boundary Nb precipitates and irregular shaped melting paths. ................. 69

Figure 36: AC spot weld from experiment 032307. Sample was tested in shear and failed
throughout the weld area. The small lighter grey areas on the multilayer, visible on the
left half, show the area that the electrodes melted. This area is smaller than the diameter
of the electrode. ................................................................................................................. 70

Figure 37: Low magniﬁcation BEI of spot weld 010407D. Niobium mixed through the
base metal towards the outer surface. There is evidence of convection currents mixing
the multilayer into the base metal. ..................................................................................... 71

Figure 38: High magniﬁcation BEI of Nb phase separation. Bright streaks inside the dark
NiTi phase suggest solutionized Nb. ................................................................................. 72

Figure 39: SEI of 053107A AC spot welding sample with niobium. Mixing of Nb with
base metal suggests the multilayer foil is not critical for spot welding. ............................ 73

Figure 40: Specimen 053107D, AC spot weld without Nb. Melting occurred only near
the interface between the NiTi pieces. The spot weld is approximately as wide as
electrode contacts. The weld has melted more material to the right, suggesting the
waveform of the power supply was non-symmetric. ......................................................... 74

Figure 41: Specimen 03272007-5. Fracture surface at the edge of the spot weld. There is
no oxide discoloration in the spot weld. Material ejected from the sample as a result of
the electrodes pushing liquid from between the NiTi strips. ............................................. 79

Figure 42: Stress-strain curve for the strongest spot weld ................................................. 80

Figure 43:SEI of 040207. Niobium appears to be solutionized or pushed out of the spot
weld. The niobium that can be seen has phase separated, as shown in the insert. ........... 81

Figure 44: Thin ﬁlm bonded to wrought pseudoelastic NiTi. Film, originally the same
width as the wrought NiTi, was tacked to top faces of corrugation. During brazing, the
ﬁlm sagged into contact with the wrought material. ......................................................... 85

Figure 45: Thin ﬁlm 012907A bonded to the inner surface of a NiTi pseudoelastic tube.
The thin ﬁlm was oriented with the 750 nm Nb coating facing towards the tube. The ﬁlm
was wrapped approximately 2.5 times around the inside of the tube, as shown at the top
center of the image. TizNi is present at the bond line ...................................................... 86

Figure 46: DSC data from freestanding thin ﬁlm 012907A. The ﬁhn transforms from B2
austenite to R-phase to B19’ martensite upon cooling, based on the two large upward
peaks. ................................................................................................................................. 87

xii

Figure 47: NiTi low density closed cell foam brazed to SMA NiTi thin ﬁlm. ................. 91

Figure 48: Thin ﬁlm bond to NiTi foam material. (1') Thin ﬁlm, (ii) bond line between
thin ﬁlm and (iii) NiTi foam. (iv) TizNi phase is present, not the Nb phase. ................... 92

Figure 49: SE1 of sample 060107A, a thin ﬁlm (top) bonded with chemical nitinol
(bottom). The bond line between the ﬁlm and bulk was probed using EDS at spots A and
B. The thin ﬁlm and bulk compositions were explored at points C and D respectively, as
shown in Table 12 .............................................................................................................. 95

Figure 50: Sample KB052806. NiTi foam (top) bonded to wrought SMA plate Nitinol
(bottom) using 54 um Nb foil. ........................................................................................... 96

Figure 51: Etched rnicrostructure of NiTi foam (top) bonded with chemical NiTi
(bottom). The eutectic rnicrostructure can be clearly seen forming a bond between the

materials and has ﬁlled some of the holes in the porous structure that are near the
interface. ............................................................................................................................ 99

Figure 52: Microstructure of chemical NiTi to chemical NiTi bond. Unmelted pure Nb
foil remains in the center of the bond; eutectic NiTi and Nb have solidiﬁed between the
Nb foil and NiTi base metal. ........................................................................................... 100

Figure 53: Heating schedule for birnetal. The heating proﬁle consists of a ten minute
dwell time at 1190°C, then furnace cooled until below 100°C. The maximum pressure of
10.5 uTorr was reached at 900°C, ................................................................................... 102

Figure 54: 39.2% cold rolling of sample 040605A. Eutectic has been compressed in the
vertical direction 40%. NO cracking is visible ................................................................ 105

Figure 55: Eutectic shows no major change in the rnicrostructure. 40% compression in
vertical direction. Niobium deforms easily without cracking. ....................................... 106

Figure 56: Metallurgical bond between pseudoelastic NiTi ribbon (top) and alumina
(bottom) using multilayer 111306A. A eutectic microstructure of NiTi and Nb has
formed between the A1203 and NiTi, with small precipitates of TizNi. ......................... 110

Figure 57: X-ray scans corresponding to points shown in Figure 56. ............................. 111

Figure 58: NiTi bonded to alumina (center), formed by a 0.504 mm thick Nb wire.
Etching ﬁ'om the Nb wire can be seen on the left side of the image. Sample broke
through alumina, as evident by the alumina fracture surface on the right side of the
image. Scale divisions on the right are millimeters. ....................................................... 112

Figure 59: Specimen 031607, showing the bond between molybdenum (top) and NiTi
(bottom). A mixed two-phase rnicrostructure between NiTi and Mo can be seen, which is
a result of liquid formed by Nb lowering the melting point of the NiTi. Bright phases
have high concentrations of Mo and almost undetectable levels of Nb. The linescan

xiii

across the bond begins on the left of the graph, with high counts moving to the right of
the graph. ......................................................................................................................... 114

Figure 60: Specimen 031607. Eutectic liquid is wetting the NiTi and forming a new
morphology of two-phase rnicrostructure. The Nb-rich phase contains most of the Mo,
and is the bright phase in the two-phase region. The dark line between the NiTi and two-
phase region is a polishing effect. ................................................................................... 115

Figure 61: Microhardness of sample 032107B, a bond between NiTi and Mo. A 50 gram
load was used for the measurements. The hardness was the highest inside the bond.
None of the indents chipped or showed any signs of microcracking. ............................. 116

Figure 62: Heating proﬁle of furnace. One minute dwell time at 1185°C. Sample
062806A, created from 062306D and 062306E. ............................................................. 119

Figure 63: Overview of CoCrMo ﬁ'acture surface. Upper left corner was not brazed most
likely due to capillary forces drawing the eutectic liquid away. Failure occurred inside
the porous NiTi. The rough appearance of the CoCrMo (top left of CoCrMo) could be a
result of signiﬁcant Co sublimation during the brazing process. .................................... 121

Figure 64: Composite image of pore and fracture surface. Note the different texture of
the pore versus the fracture surface, implying fracture did not propagate through a pore.
NiTi half of bond. ............................................................................................................ 121

Figure 65: Fracture surface in area that was brazed between NiTi pores and left material
on CoCrMo side ............................................................................................................... 122

Figure 66: Hardness data ﬁ'om CoCrMo material before and after brazing. The overall
hardness increased by 13.7%, while the standard deviation in hardness decreased by 9.8
VHN. Many irregularly shaped indents were observed in both as-received and post
brazed conditions. ............................................................................................................ 123

Figure 67: Global view of sample 031607A. The apparent wetting angle of the liquid by
the NiTi has been changed compared with NiTi — NiTi bonds. Fe2AlV is adequately
wetted by the liquid, and has ﬁlled cracks in the material. ............................................. 127

Figure 68: Low magniﬁcation SE1 of 031607A. The solidiﬁcation products of the bond
appear different than the standard NiTi-Nb solidiﬁcation products. The wetting angle of
the FezAlV is similar to other materials, however, the NiTi wetting angle is greater than
90°, suggesting a non-wetting condition. F ezAlV appears susceptible to microcracking,
as suggested by a crack inﬁltrated with solidiﬁcation product. This crack must have been
present during the brazing process. ................................................................................. 127

Figure 69: Sample 031607A. The solidiﬁcation product does not seem to be a cellular or
dendritic structure. A third possible phase could be at the top of the image, appearing to
have solidiﬁed ﬁom the FezAlV ...................................................................................... 128

xiv

Figure 70: SEI and corresponding X-ray maps of NiTi-FezAlV bond. Niobium content
appears to be the largest chemical difference between the two solidiﬁcation products.
Titanium, aluminum and vanadium also appear different between the phases, while iron
and nickel content are sirrrilar. ......................................................................................... 129

Figure 71: SEI of rnicrostructure. X-ray information was collected from points A-D and
were analyzed in Table 15. .............................................................................................. 130

Figure 72: Hardness data for specimen 032107-A. The NiTi phase has a lower hardness

than the FezAlV. The hardness of the bond was higher than the hardness of the parent
phases; however no cracking was visible inside the bond ............................................... 131

Figure 73: Optical image of NiTi side of NiTi-B4C fracture surface. Units on the scale to
the left is millimeters. (a) Area remained unbonded during brazing. (b) B4C remains
bonded with the NiT i. (c) Solidiﬁcation product of the liquid. ...................................... 134

Figure 74: Cross-section of ﬁ'acture surface. While the base metal appears to be wetted
by the eutectic liquid with a low contact angle, the brittle bond with the boron carbide
fractured from thermal contraction. ................................................................................. 135

Figure 75: High magniﬁcation SE1 of solidiﬁcation product formed by NiTi-Nb liquid
reacting with RC ............................................................................................................. 136

Figure 76: Mg rich region ﬁlled with ﬁne Nb-based and Mg-based precipitates. The
phases were too small to analyze using EDS. Based on EDS data, only the Mg-rich
region appears to contain Mg. ......................................................................................... 139

Figure 77: Stress-strain plot of the sample shown in the photo insert. Sample necked
inside the tantalum, demonstrating the joint is stronger than the tantalum. (Low 2007) 142

Figure 78: Proeutectic NiTi eutectic Nb/T a and NiTi ﬁll the bond between NiTi and Ta.
This SEI shows the simple eutectic between NiTi and Ta that results ﬁom brazing with
an interlayer of 127 pm thick Nb foil. X-ray information collected at points A, B, C and
D shown in Table 15. ....................................................................................................... 143

Figure 79: LOM of Nb coating, the bright phase on the right, on Ti-6Al-4V. The Nb ﬁlm
is approximately 10 um thick on sample 053007B, which was taken from the thin ﬁlm
deposition 051507B. ........................................................................................................ 148

Figure 80: LOM of (a) Ti-6Al-4V base alloy with (c) 30 um Nb thin ﬁhn deposited on it
to prevent extra titanium ﬁom reaching (b) the eutectic between niobium and nickel
titanium. Niobium successfully prevented Ni diffusion into Ti-6Al-4V. Four pieces of
multilayer M032007A were placed between the niobium coating and the NiTi before
brazing, serving as a liquid source ................................................................................... 149

Figure 81: LOM of 060207C, (a) NiTi bonded to (b) Ti-6Al-4V with a dwell time of 1.5
minutes at a temperature of 1185°C. Niobium did not prevent the liquid from melting the

XV

Ti-6A1-4V, which enriched the liquid in titanium, producing (c) TizNi upon cooling. The
TizNi phase fractured during hot mounting. .................................................................... 150

Figure 82: Sample 040607A. This BEI shows the transition in microstructure from BCC
Nb (left, bright phase) to the tweed microstructure of Ti-6Al-4V. A two-phase region is
seen between the thin ﬁlm and base metal; differences in etching rate have produced a
ledge that causes darkness at the interface. ..................................................................... 150

Figure 83: SE1 of bond between NiTi (left) and Ti-6Al-4V (right). The coating of Nb
appears to be well adhered to the Ti-6Al-4V as evident by Z-contrast in the B region. X-
ray information from spot A, B and C were analyzed using an Oxford Instruments EDS
software package. ............................................................................................................ 151

Figure 84: 042407A. NiTi to Ti-6Al-4V bonded without multilayers, only an Nb coating.
This process was completed using a thick coating of Nb on Ti along with fast brazing
times, especially above 1150°C. ...................................................................................... 152

Figure 85: 040607A NiTi - Ti64: The microstructure of the solidiﬁed eutectic contains
Nb and NiTi phases, along with a small phase fraction of TizNi, which is expected due to
oxygen contamination on the multilayer foil ................................................................... 153

Figure 86: 040607A NiTi - Ti64. The Nb coating serves as a barrier for nickel diffusion
from NiTi to Ti-6Al-4V. The soft Nb coating has proeutectic Nb solidiﬁed from the
bond with NiTi using multilayer foils. The sputtered coating mixes with the Ti-6Al-4V
in a two-phase region a+B region. ................................................................................... 154

Figure 87: Microstructure of NiTi-W bond. (a) Unmelted NiTi base metal (b) Proeutectic
NiTi phase coherent with unmelted NiTi (c) Proeutectic NiTi lamellae (d) Voids in the
microstructure (e) Tungsten wire .................................................................................... 157

Figure 88: LOM of sample 053007A. (a) Pseudoelastic NiTi bonded with (b) 254 um
diameter tungsten wire and mounted in (c) Konductomet ® phenolic mount material. A
small void is present in the bond between the NiTi and tungsten, as well as a eutectic
rnicrostructure and a proeutectic phase consisting of Nb and tungsten. Also, note the low
wetting angle between the wire and eutectic liquid ......................................................... 158

Figure 89: NiTi side of 053007A. (a) Unmelted NiTi, (b) proeutectic coherent NiTi (c)
eutectic solidiﬁed Nb phase. ((1) Faceted precipitate. (e) Bright eutectic phase. .......... 159

Figure 90: Microstructure of 053007A near tungsten wire. (a) ‘Matrix’ phase that is dark
grey, (b) tungsten wire approximately 15 um inside the wire, (0) dark precipitate phase,
which could be based on TizNi, and (d) bright eutectic phase. ....................................... 160

Figure 91: Low magniﬁcation SEI of post-ﬁacture bond between NiTi and stainless steel.
Bond was formed using Nb foil ....................................................................................... 163

Figure 92: Three distinct solidiﬁed microstructures in the bond between NiTi and 304
stainless steel. .................................................................................................................. 164

xvi

Figure 93: (a) DC Magnetron sputtering gun, (b) 75 mm heater for post deposition heat
treatment and (c) the substrate holder showing thermocouple placement ....................... 177

Figure 94: As-rolled thin ﬁlm of 10 um NiTi deposited on 7.6 pm thick polyimide.
Puckering is present around the edges of the ﬁlm from the deposition plastically
deforming the polyimide. The ﬁlm was cold-rolled to approximately 5% strain. ......... 178

Figure 95: Thin ﬁlm (left) and aluminum plates after cold rolling (right). ..................... 178

Figure 96: Collective results of cold rolling experiments. As rolling strain increases, the
thermally recoverable strain also increases. 0.8% of the rolling strain is permanent.
Cracking becomes present in thin ﬁlms rolled beyond 6 to 7% strain, shown in Figure 97.
......................................................................................................................................... 1 80

Figure 97: Microcrack developed after the sample was cold rolled beyond 6% strain.
This SE1 from a Phillips Sirion SEM equipped with a ﬁeld emission gun operating at 25
keV. Bright area is charging on the underlying polyimide ............................................. 181

Figure 98: BEI of NiTi thin ﬁlm on polyirrride (bright stripe along bottom of ﬁlm). The
grain boundaries of the ﬁlm appear to terminate roughly one um away from the
polyirrride ﬁlm, which could be an etching effect or transition to amorphous NiTi. ...... 182

Figure 99: DSC results from sample 052307A, a 10 um thin ﬁlm on a 7.6 pm thick
polyimide substrate. ......................................................................................................... 182

xvii

KEY TO SYMBOLS AND ABBREVIATIONS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Symbol Unit of measure

°C Degrees Celsius

mm Millirneters

urn Micrometers

mTorr x 10'T Torr

uTorr x IOFTorr

MPa x 106 Pascal

EDS Energy dispersive spectroscopy
XRD X-ray difﬁ'action

SEM Scanning electron microscope
SE1 Secondary electron image

BEI Backseattered electron image
LOM Light Optical micrograph

PID Programmable interface device
NDC Nitinol Devices Corporation, Inc.
Nitinol Trade name for nickel titanium

 

 

xviii

 

1 Introduction

The interesting properties of the interrnetallic alloy nickel titanium were ﬁrst discovered
between 1962 and 1963 at the United States Naval Ordinance Laboratories. After the
alloy demonstrated shape memory, it was named Nitinol, which is an acronym for NIckel
Tltanium Naval Ordinance Labs. (Otsuka and Ren 2005; Memry 2006) Since its
creation, nickel titanium has been used in many applications, due to its possession of
several unique features: biocompatibility, pseudoelasticity, and shape memory
transformation temperatures near human body temperature. (Memry 2006) Some alloys
OfNiTi can exhibit wide hysterisis, or a wide temperature difference between the
austenite start (As) and martensite start (Ms) temperatures, which is useful for aerospace
applications. (He and Kong 2004) In addition, NiTi is corrosion resistant at low
temperatures, so it can be used in harsh environments. (Otsuka and Ren 2005) Finally,
the unique physical properties of NiTi make it especially useful in electronic applications
where large amounts of mechanical work must be developed using a small mass. (Gill,
Ho et al. 2002) Sensors, motors and pseudoelastic antennas are just a subset of the

potential electronic applications of MT i alloys. (Memry 2006)

Nitinol is a near equiatorrric formulation of nickel and titanium. Near room temperature,
the material can undergo a phase change ﬁ'om a B2 (CsCl) austenite crystal structure to a
B19’ (monoclinic) martensite structure. This martensitic phase transformation is
reversible and gives rise to shape memory and pseudoelastic effects. (Otsuka and Ren

2005)

The common oxide formed on NiTi alloys is Ti4Ni20, which shares the same spacegroup
and morphology as TizNi. For this reason, Ti4Ni20 and TizNi will be assumed
interchangeable, because without TEM analysis, the two phases cannot be distinguished.

(Otsuka and Ren 2005)

1.1 Shape Memory Processes

The single way shape memory effect (SWSME) encompasses the following
transformation. First, the martensite is deformed, then, upon thermal reversion of the
austenite the material will restore its previous shape. The crystallography of the phase
change will be discussed later, but the effect is dependent on a rearrangement of
reversible martensite variants under an applied external stress ﬁeld. When the material is
heated to above the austenite start temperature (As), the original austenite crystal structure
is elastically restored due to the higher stiffness of the austenite. This effect is known to
restore up to 7% strain in a polycrystalline material, and up to 10% in a single crystal.

(Otsuka and Ren 2005)

Pseudoelasticity occurs when a material undergoes no permanent deformation when
subjected to a strain substantially beyond the elastic limit of other metals. For monolithic
polycrystalline NiTi, up to 7% strain can be elastically recovered by reversion of stress-
induced martensite. During this martensite formation, the strain of the material increases
dramatically, while the stress does not increase substantially. After the austenite phase is

transformed to martensite, it can slip induce plastic deformation. (Hou 1998)

Pseudoelasticity can be observed during a strain controlled tensile test. If the material
exhibits pseudoelasticity, only the endpoints of the loading and unloading curves are

coincident, shown schematically in Figure 1.

 

 

 

 

 

8

Figure 1: Schematic stress strain curve of a pseudoelastic Nitinol. The upper curve is the loading curve,
while the bottom is the behavior of the material upon unloading. Plastic deformation can begin at the
completion of the pseudoelastic plateau on the stress strain curve.

1.1.1 Two-way shape memory effect

Two-way shape memory (TWSME) is another process that shape memory alloys, such as

NiTi, can exhibit. In this process, an internal stress ﬁeld drives material to a taught shape

in the martensite phase. When the Nitinol is heated into the austenite phase, it still shows
the single-way shape memory effect. With this process, the material can remember two

distinct shapes: one shape at low temperature and another at high temperature.

Cyclic theme-mechanical treatments are used to induce a stress ﬁeld into the nickel
titanium. Exact procedures vary, but during this process the material is either under
constant load and is cycled in temperature or is cycled both in load and temperature
between the martensite and austenite phases. The TWSME is not as robust as the single-
way shape memory effect, and shows less strain amplitude and less cyclic stability. Over
time, the stress ﬁeld can relax, which diminishes the effect. Typically, during the ﬁrst
few heating cycles, it is possible to get above 2% strain change, but after approximately
10 to 100 cycles, the effect diminishes to a asymptotic 1% strain change between cold

and hot cycles. (Otsuka and Ren 2005) (Zhang 2006)

The TWSME effect can also be developed using external biasing of a SWSME. Here the
material is redeformed upon cooling of the material by an external force. This is a less

elegant form of the TWSME, but it can support higher maximum strain amplitude.

(F oltz, Grumman et a1. 2006)

1.1.2 Deformation systems

The functionality of Nitinol is dependent on a diffusionless, thermo-elastic, ﬁrst order
martensitic phase transformation. The B19’ structure of Nitinol is a thermo-elastic

martensite, which forms by inter and intra-martensite twinning. (Otsuka and Ren 2005)

4

The phase transformation must have a coherent phase front to create the martensite by
shearing the austenite. Due to differences in lattice size between B19’ and B2, this
coherent phase front is accomplished through a Type II twinning process, which results in
a twinned martensite. The strain associated with this shearing of the austenite is
neutralized by other twin variants nearby. Due to this, martensite in Nitinol is referred to

as forming by ‘self accommodation’. (Otsuka and Ren 2005)

The B19’ martensite has a monoclinic crystal structure, which is a low symmetry
structure. The austenite phase is a high symmetry, CsCl , or BZ crystal structure. The
austenite phase supports dislocation motion, making nickel titanium unusually ductile for
an interrnetallic. If the onset of slip plasticity is delayed using strengthening
mechanisms, Nitinol can stress induce martensite, which is the basis for the pseudoelastic

effect. (Otsuka and Ren 2005)

Stress inducing martensite is explained by the Claudius-Clapperyon equation. This
relationship dictates that martensite transition temperatures rise as the shear stress in a
material increases. Therefore, as the austenite is loaded, the start temperature of the
martensite rises until the testing temperature is below the, now elevated, martensite start
temperature. At this point, martensite stress induces and can deform by detwinning.
Upon unloading of the sample, the detwinning and martensite nucleation reverses and

leaves the Nitinol as austenite once more. (Otsuka and Ren 2005)

There is also a rhombohedral martensite (r-phase) which can form in Nitinol; it behaves
differently than the monoclinic martensite, but still forms by self accommodated
twinning. The strain-temperature hysteresis loop between austenite and the r-phase is
smaller than the loop between austenite and B19’ martensite. This has two important
consequences: the r-phase completes the phase transformation at higher temperatures

than 319’, but has smaller recoverable strain. (Otsuka and Ren 2005)

1.2 Bonding NiTi

Several methods of joining nickel titanium have been developed in an effort to increase
the complexity of devices. Thus far, some joining methods are acceptable in limited
applications; while other applications, such as biomedical ones, do not have robust
joining techniques due to biocompatibility problems. The following is a brief summary

of some joining methods for nickel titanium.

1.2.1 Diffusion bonding

Joining nickel titanium has been historically difﬁcult due to the slow diffusivity of both
titanium and nickel in nickel titanium. According to Li and Rong, the bulk diffusivity of
nickel in Nitinol is 1.5 x 10'12 at 1223 K, while titanium in Nitinol is 2.0 x 104° at 1223
K (Li, Rong et a1. 1998) A ﬁrst order approximation of the time required to diffusion
bond a NiTi surface with an average roughness of 10 pm is given by (W ert and Thomson

1970):

«513—:X
X2 = 21):
X2
27) =
0.0012

2 -1.5 . 10‘12
Equation 1: Approximation of average diffusion distance (W ert and Thomson 1970)

t

 

= t = 333,333 sec = 3.85 days

Where D is the average diffusivity, t is time and X is distance. With a lengthy high
temperature process such as this, diffusion bonding leads to grain growth, and can cause
precipitates to enlarge and lose their strengthening ability. This loss of strengthening can

lead to easier slip plasticity in the austenite phase, poor pseudoelastic effect and poor

strength. (Otsuka and Ren 2005)

1.2.2 Silver solders

There is some research on silver soldering of Nitinol. All published techniques require
aggressive ﬂuxes to remove the Ti4Ni20 complex Oxide, which forms readily at elevated
temperatures. These ﬂuxes are generally not biocompatible and therefore cannot be used
for biomedical devices. The strengths of these solders are low, usually between 150 and

250 MPa. (Memry 2006)

Three distinct soldering techniques have been explored by the Edison Welding Institute
(EWI). Halogen based ﬂuxes, used commonly for aluminum alloys, have limited success
removing surface oxides in NiTi alloys, but must be mechanically removed after

soldering. Nickel and other metals have been plated onto NiTi, which indirectly allows

NiTi to be soldered to other materials. This technique, however, is difﬁcult and time
consuming due to the multiple stage process. The third method involves ultra-sonic
soldering NiTi using tin. This technique is not biocompatible, and achieves low shear
strength. The highest shear strength obtained among all three soldering techniques was

50 MPa. (Hall 1997)

1.2.3 Welding techniques

Welding and spot welding of Nitinol have provided reasonable success; but bond
strengths of 550 MPa are the highest reported, which is approximately 50% that of the
base metal. In metal inert gas (MIG) and tungsten inert gas (TIG) welding, nickel
titanium bonds are typically brittle, and possess T12N1 precipitates in the fusion zone,
which is where the solidiﬁcation takes place. (Falvo, Furdiuele et al. 2005; Li, Sun et a1.
2006) Devices using these techniques are limited by the low bond strength, and the small
cross sectional area of the weld, which reduces the functionality of the material near the

weld.

Laser welding of NiTi sheet material using a NngAG laser has provided positive results.
Both shape memory and pseudoelasticity of NiTi plates can be preserved when joining
them by laser welding with argon shielding gas. This method, however, can only be used

for line-of-sight joining; not within a structure. (Schlossmacher, Hass et a1. 1997)

1.2.4 Transient Liquid Processing

Transient liquid processing (T LP) is a method of joining materials that has existed for
thousands of years. (MacDonald and Eagar 1992) Only in recent years has it begun to
receive attention again, due to advancements in metallurgy. This joining method
involves a liquid forming by a low melting temperature eutectic or peritectic reaction
between two materials and an interleaved foil of reactive material. By way of solid state
diffusion, the liquid slowly freezes as the atomic species of the interleaved foil diffuses
into the bulk. The microstructure of the bond, upon solidiﬁcation, is nominally the same

as the parent materials, making the bond difﬁcult to ﬁnd later.

TLP requires high difﬁrsivity of the minor species in the solid state of the parent phase,
and a low melting eutectic or peritectic reaction between the two materials. (MacDonald
and Eagar 1992) Without high diffusivity in the solid state, the processing time is
prohibitive in terms of high cost and poor material properties. The process is not possible
without the use of a low melting point peritectic or eutectic reaction; without this liquid

formation, the process would be the same as diffusion bonding.

This process has been used in recent years to join NiAl and NiTi. (Gale and Guan 1997)
TLP was mildly successful in joining these materials, however, the process greatly
disturbed the transformation temperatures and microstructure of the NiTi material. The
addition of a copper interlayer between the NiTi and NiAl allowed formation of a liquid,
which later solidiﬁed from both the NiTi and MA] sides. The bond contained excess Ti

as a result of faster diffusion of Ti versus Ni in NiTi. As a result, the microstructure of

the NiTi was inhomogeneous as a ﬁrnction of distance from the bond. In addition to
forming secondary phases, the process required over an hour Of processing at 1150°C,

which would cause excessive grain growth in the NiTi material.

To join NiTi to itself or other materials effectively and at a reasonable cost, the thermal
exposure must be limited. This is due to the reduction of the mechanical properties of
NiTi when exposed to temperatures above 500 °C, as well as the cost incurred by high

temperature processing.

1.2.5 Ni-Ti-Nb ternary system

The Ni-Ti binary phase diagram, shown in Figure 2 is a complex system containing
several intermetallics. The most important of these intermetallics is NiTi, but the stable
interrnetallic TizNi and the metastable Ni4Ti3 and Ni3Ti2 phases are important in
determining the strength of NiTi alloys. TizNi is brittle and weak, and can embrittle and
weaken NiTi alloys when it forms at grain boundaries. Ni4Ti3 and Ni3Ti2, however, can
strengthen NiTi alloys by precipitation hardening if the particles are small and well
dispersed in NiTi grains. Ni4Ti3, in particular, is commonly used as a strengthening
mechanism against dislocation motion in the BZ phase of NiTi, which increases the

pseudoelastic functionality of the alloy. (Otsuka and Ren 2005)

Niobium, a common alloying addition for NiTi, has low solubility in the B2 NiTi phase.
With alloying additions above 10 at.% Nb, phase separation produces BCC niobium and

BZ NiTi. (Piao, Miyazaki et a1. 1992) The ternary system between Ni, Ti and Nb is

10

shown in Figure 3. The nickel-rich half of the ternary system contains several ternary
intermetallics, while the titanium-rich half contains no ternary intermetallics. (Gupta

1991; Prima, Tret'yachenko et al. 1995)

Literature suggests a possible quasi-binary eutectic between NiTi and Nb, shown in
Figure 3, with a reported eutectic temperature of 1170°C with an atomic composition of
36Ni-3 8Ti-26Nb. (Piao, Miyazaki et a1. 1992) (Gupta 1991; Prima, Tret'yachenko et a1.
1995) The exact melting temperature of the quasi-binary eutectic point is only known to
lie between 1150°C and 1170°C, because of the steep solidus surfaces in the ternary
system. The quasi-binary eutectic temperature is roughly 140°C below the melting
temperature of pure NiTi, and therefore could be used to join NiTi together in a process
similar to TLP. Contact melting of niobium and NiTi will take place above 1150°C,
joining the materials together. Upon solidiﬁcation, based on the pseudo-binary eutectic,
phase separation will occur and the bond will be composed of B2 nickel titanium and
disordered BCC niobium. The key feature of these microstructures is that both phases in

the eutectic are ductile.

Based on Kc Bin Low’s work at Michigan State University, butt-joints made using the
NiTi-Nb quasi-binary eutectic system have high breaking strengths, as shown in Figure 4.
The fracture is comprised of a cup and cone type failure inside the solidiﬁed zone, which
is most likely due to the ductile Nb phase. Microalloying the Nb phase with other

elements could possibly lead to higher breaking strengths. (Low 2007)

ll

The quaternary system between Ni, Ti, Nb and O is also of great importance to this
joining method, which has not been studied in depth. If the eutectic liquid created by
heating NiTi and Nb above 1150°C interacts with oxygen, even in a vacuum furnace, it
stabilizes TizNi precipitates. TizNi precipitates that are solidiﬁed from the eutectic liquid
can contain niobium, based on the Ti-Nb binary phase diagram. Niobium is assumed to
substitute for Ti inside the TizNi precipitate. Without oxygen, a mixture of NiTi and Nb
freezes into NiTi and Nb upon cooling. In the presence of oxygen, it is theorized that the
liquid ﬁ'eezes into a 3 phase reaction plane consisting of NiTi, Nb and TizNi.

Microstructural evidence for this hypothesis can be seen in Figure 5.

12

2500

 

Ni—Ti BInary‘

 

2000‘

1500- .. .g ér: I455

 

Temperature, °C

1000 - Z984

 

 
  

    
 

 

h -r ' ' ' '
cp I _ NI41"'I3 T }Metastable
,‘_— .

 

 

 

 

500

 

 

 

T 20 'TIzNi 4'0 . so ' “is“ 8'0 ' Ni
Atom % NI

Figure 2: Ti-Ni phase diagram in atomic percent Nickel.

 

Ti

Figure 3: Ternary phase diagram between Ni, Ti and Nb displaying solidus surfaces and terminal phases.

 

 

 

 

 

soc
— 253m
— 9
70°“ (Nelle Response
6w. /
as”
2 r.
7400 ,' n
a ,x .'
2 :
5 300‘ ' x :
zoo-
100 :1 "'1
' ' 1'2 14 1'5 1'3 2'0 22

 

 

Figure 4: Stress-strain response of a NiTi butt-joint bonded using the Ni-Ti-Nb ternary system. Fracture
surface shows ductile cup and cone type fracture, likely due to the Nb phase.

(Low 2006)

 

Figure 5: SEM image of NiTi cross—section that has been wetted by NiTi-Nb liquid in a vacuum furnace.
Upon freezing, faceted precipitates form, which are morphologically identiﬁed as TizNi.

1.3 Organization

The topic of this thesis is the creation and basic understanding of bonds between nickel
titanium to itself and other materials. Numerous experiments were conducted for this
thesis; these will be described in detail, with individual introductions, lists of materials

and procedures. The results will be summarized after discussing all of the experiments.

16

2 Joining similar Nitinol

The ﬁrst NiTi metallurgical joints made using Nb-based reactive eutectic brazing were
made in 2004 by Shaw and Grummon, working at MSU, as an outgrowth of attempts to
create low-density topologies in Nitinol. The ﬁrst experimental joint was made along an
axial contact line between a parallel pair of 3 mm NiTi tubes. Subsequent metallurgical
analysis revealed the aggressive wetting capability of the liquid that was formed by
contact melting, and the simple 2-phase eutectic structure in the solidiﬁcation product.

Preliminary mechanical tests suggested that the joint was mechanically robust.

2.1 Discovery of NiTi-Nb reactive brazing with NiTi tubes

2.1.1 Introduction

Brazing and diffusion bonding are the only processes that can create bonds on the inside
of structures. By utilizing furnace brazing, nickel titanium requires none of the special
equipment necessary for welding. Furthermore, brazing can bond large surfaces
metallurgically, which is difﬁcult to impossible with traditional welding techniques. This
allows for tougher mechanical joints that can support higher loads and deformation

without the high stress concentrations that can occur from welding.

The discovery of the NiTi-Nb reactive brazing process was one of accident while
attempting to diffusion bond pseudoelastic NiTi tubes along an axial direction. The tubes
were held in contact using a Nb wire, which melted during the experiment. After this

discovery, the experiment was repeated, but a 0.508 mm wire of Nb was placed along the

17

contact line of the tubes. The following section describes this initial attempt at using the

Ni-Ti-Nb ternary system to join NiTi tubes.

2.1.2 Brazing

To make use of the reported quasi-binary eutectic between NiTi and Nb, two 3 mm
diameter tubes of nickel titanium were joined together using a small supply of high purity

niobium wire.

As—received NiTi tubes with an inner diameter of 3.2 millimeters were sectioned into 19
mm lengths. A single 19 mm length of 500 um diameter 99.995% pure Nb wire was
placed between the tubes. The tubes were held in contact with a wrap of tungsten wire
going around the circumferences of the tubes. The tubes were then suspended from the

top of a Centorr vacuum fumace using another tungsten wire.

The tubes were joined by vacuum furnace brazing with a Centorr vacuum annealing
ﬁrmace. The furnace was allowed to pump until the base pressure was below 10'5 Torr,
after which it was heated to a temperature of 1185°C at an approximate rate of
20°C/minute. It was held at 1185°C for 5 minutes and then allowed to cool by turning
off power to the heating elements. The temperature of the fumace was measured by an R
type thermocouple attached to a Honeywell Versa Pro UDC 3000 thermocouple
controller. The thermocouple registered below 500°C after approximately 15 minutes of
cooling. Once cooled, the sample was heat treated in an air-fumace for 1 hour at a

temperature of 400°C.

18

2.1.3 Results

The tubes were completely fused by formation of a low melting eutectic liquid as a result
of the niobium braze ﬁller. There was no discemable bond line in the cross section of the
tubes, as shown in Figure 6. It is simple to determine the distance the liquid ﬂowed by
observing the color and shade of the NiTi tubes after being brazed. The liquid initially
formed from contact melting between a single 500 um diameter Nb wire and the NiTi
tubes. As the liquid formed, it ﬁlled the local capillary space nearby. The liquid crept up
the sides of the tubes, to balance the surface energies by reducing the wetting angle

between the nickel titanium and the liquid, as shown in Figure 6.

A tubular shape is one geometric conﬁguration that can amplify the pseudoelastic effect
of nickel titanium. The initial experiment to join the tubes using diffusion bonding was
to create macroscopic transformation foams. To demonstrate the ampliﬁcation of
pseudoelastic effect, the sample was cyclically compressed with increasing compressive
strain until fracture was observed, as shown in Figure 7. This fracture occurred at nearly
30% strain, measured by the initial height of the tubes, and was outside of the joined area.
The pseudoelastic effect, which can be observed by the reversible strain upon unloading

of the tubes, was above 20% on the ﬁnal compressive loading curve.

19

 

Figure 6: Macroscopic bond of NiTi—NiTi using Nb ﬁller metal. Note that the eutectic liquid wets the NiTi
surface without ﬂuxes or other wetting agents.

20

Ac
(MPa)
40 _
30 -
20 _

10~

 

 

 

30
———I-- OIL (%)

Figure 7: Compressive stress stain curve for a pair of joined NiTi tubes. Loading is shown as solid lines,
unloading as dashed lines. The pseudoelastic effect of nickel titanium is ampliﬁed from approximately 7%
reversrble stain. Little permanent deformation was observed up to 30% compression.

The microstucture of the bond was especially useful in the attempt to verify the
existence of the pseudo-binary eutectic reported in literature. The tubes were sectioned
and polished to a mirror ﬁnish using silicon carbide grinding papers and alumina media
on a polishing wheel. The ﬁnal polish was conducted with 0.05 pm diameter diamond

paste, after which the specimen was observed using a scanning electon microscope.

One half of the bond between the NiTi tubes can be seen in Figure 8. A Hitachi S-2500C
scanning electon microscope operating at 25keV and equipped with an Oxford

Instuments EDS system was used to characterize the bond. The top left image in Figure

21

8 shows a low magniﬁcation secondary electon image of the bond. Below this image is
a higher magniﬁcation image of the bond, where the dendritic solidiﬁcation stucture of

the proeutectic NiTi phase is visible to the left and right of the cental brighter area.

The dendrites are also present in the bottom right image of Figure 8. These dendrites of
NiTi form as a result of the composition of the liquid; the liquid enriches in NiTi after the
exhaustion of the Nb source. The liquid composition reaches the NiTi liquidus
composition at the processing temperature, and upon cooling, the ﬁrst phase to solidify is
dictated by the two-phase region that the super-cooled liquid moves into, given by the
phase diagram in Figure 9. The solidiﬁcation of the melted NiTi of the tubes results in a

dendritic stucture, based on solidiﬁcation rate. (Porter and Easterling 1989)

The bottom left image in Figure 8 also clearly shows the eutectic stucture that forms
below the eutectic temperature of 1170°C, which is a combination of Nb and NiTi
phases. The morphology of the eutectic stucture is consistent with furnace cooling in

other alloy systems. (Sinha 2003)

To help understand the composition of the proeutectic NiTi and the eutectic mixture,
X-ray ﬂuorescence images were formed from the Nb L-a, Ni K-a and Ti K-u
characteristic X-rays. As shown in Figure 8, the titanium concentation in the proeutectic
NiTi is lower, and thus has a darker image, than the unmelted NiTi. Additionally, there is

some Nb present in the proeutectic NiTi phase, but none is visible in the unmelted NiTi.

22

This implies that diffusion of Nb in solid state NiTi is slow, and that the Nb solidiﬁed

into the proeutectic NiTi preferentially substitutes for Ti in the NiTi stucture.

To obtain a fracture surface, another section was taken and the tubes were manually tore
apart using pliers under moderate force. A low magniﬁcation secondary electon image
of this fracture surface is shown in Figure 10. As indicated in the ﬁgure, dimples cover
the ﬁacture surface, implying a ductile tearing fracture mode and a tough bond. As seen
in Figure 11, these dimples have coalesced into large voids, again implying a tough
metallurgical bond. While the ductile tearing ﬁacture by microvoid coalescence is
commonly observed in the eutectic zone, the commonly observed ﬁacture mode in nickel

titanium is ductile cleavage.

23

1‘32“." IU‘I'I ‘ L'C‘ ul'I
— _

1:"; II '1':
_

 

1 (I!) ll IT
_

Figure 8: (Top left) Secondary electon image of bond between NiTi tubes using Nb ﬁller metal. (Top
right) No L—a X-ray ﬂuorescence showing no signiﬁcant solid state diffusion of Nb into NiTi. (Center
right) Ni K-a X-ray ﬂuorescence shows a lower concentation of Ni inside the eutectic than inside the NiTi.
(Lower right) Lower concentation of Ti in the proeutectic phase than in the base NiTi, which is shown
using the Titanium K-u map. (Center Left) Low magniﬁcation SEI of the bond showing the presence of a
eutectic structure and proeutectic phase. (Bottom Left) High magniﬁcation image of eutectic stucture and
proeutectic phase.

24

 

 

’“ a5;
mewsﬁwrsem.ﬁ§ggessmesma
ermsmmamseam amass ass?”

« Ae=z§avt§a3§2§a= :5
.mnoﬁmrxexa xi
.1. ....... a ..... j =4

mngE-éagxmumgi ‘ -
Enzcsﬁigﬁﬁizz 535::

33%;
$5“ ﬁﬁWWﬁiﬁi-i’mmﬁﬁggii “u” ‘58

gammaeeeessaeeee
; s mmrmm
" are:

mg,

‘ "‘2'” --ei‘eggjﬂgg j , mg
§§jﬁ¥§ﬁifm ﬁgﬁfé‘ Agriﬁﬁﬁi‘ﬁ
'ﬁfﬁﬁﬁﬁéﬁﬁﬁ‘zuﬁﬁﬁﬁ __ L"--.
{message-we Esteem-re:

§

I.':;,a‘F1t:;. Ema
WWMQW' s_-.gm~m.m.wmm
ﬁmmﬁrﬁm%ﬁﬁaﬁﬁsﬁ%w . ..
gs: «awfeég? 15%? it"? .. ﬁﬁﬁ ....;§- we...“
mm m=m

Temperature, °C

ﬂﬁiqa‘g: 333552335.

3.. '12 Smile? ﬁmmmmagae: ﬁmﬁﬁ

 

sea @ waﬁ
as: awmatwswwmmmmmwﬂgmyge E
ewes: emmmwm-mwww

40Atom % wa

 

Figure 9: Pseudo-binary eutectic between NiTi and Nb. The eutectic point lies at approximame 26 atomic
percent Nb and a temperature of 1170°C. The maximum reported solubility of Nb in NiTi is near 6 atomic
percent

25

 

Figure 10: $131 of NiTi —NiTi bonded tube fracture surface. Dimpling present throughout the entire fracture
surface implies a ductile tearing mode ﬁ'acture.

26

 

Figure 11: High magniﬁcation SEI of fracture surface inside the bond area. Dimpling implies ductile
fracture inside the bond.

27

2.2 Construction of pseudoelastic tubes in a low-density topology

2.2.1 Introduction

To further amplify the pseudoelastic effect, larger arrays of pseudoelastic NiTi tubes

were bonded to one another using the NiTi-Nb reactive brazing technique.

2.2.2 Materials and Methods

Pseudoelastic NiTi tubes were purchased from NDC in an etched and polished condition
to ensure no excessive surface oxides existed. These tubes were cut into 2.5 cm lengths
and were stacked into a close packed array inside a molybdenum ﬁxture. Three mm
lengths of 125 um diameter Nb wires were placed atop the junctions between NiTi tubes.
The assembly was put in a Centorr vacuum ﬁrrnace and was allowed to reach a base
pressure of 10—5 Torr. The sample was then heated above 1185°C for 5 minutes and
allowed to furnace cool to room temperature. The array was removed tom the

molybdenum ﬁxture and was photographed and sectioned, as shown in Figure 12.

2.2.3 Results

As shown in Figure 12, the Nb wires used to form the eutectic liquid can chemically etch
the walls of the NiTi tubes due to the shape and melting dynamics of the wire. As the Nb
wires melt, the wires are pulled close to the NiTi tubes to minimize the surface area, and
thus the free surface energy, of the liquid. To maintain equilibrium with both NiTi and

Nb interfaces, the liquid continues to melt NiTi locally. As the Nb wires further melt the

28

NiTi tubes, the wires are drawn deeper into the newly formed crevice to minimize the
surface energy of the liquid. In this manner, the Nb wires cut through the NiTi tubes,
causing the etching effects visible in both Figure 12 and Figure 13. The effect does not
seem due to the differences in density between the liquid and the niobium wires because
many of the wires etched against the force of gravity. It is therefore suggested that on
this length scale, tens to hundred of microns, gravity does not play an important role in

determining melt dynamics.

 

Figure 12: Macro photo of close packed array of NiTi pseudoelastic tubes bonded with 127 um Nb wires.
(A) Etching effect on Nb shown in the cross-section.

29

 

Figure 13: Etching effect of 127 um Nb wires cutting through the NiTi tubes, which is likely driven by
capillary forces. Wires seem to have migrated and etched against the force of gravity, suggesting that
gravity does not play a major role at the length scale of the wires.

30

2.3 Construction of smooth corrugation honeycomb structures

2.3.1 Introduction

To improve the compressive properties of an array of pseudoelastic NiTi, thin strips of
Nitinol were corrugated and stacked in an alternating stucture. The corrugation was
designed with two circular arcs, each with a radius of 2.54 mm, intersecting at tangential
points deﬁned by a 7.65 mm lateral separation between cycles. This design is shown in

Figure 14.

2.3.2 Materials

The corrugations of pseudoelastic NiTi were shape set by clamping strips of material in a
die and annealing the strips at 500°C for 15 minutes. The die was then quenched in water
before the samples were removed. Both before and after the anneal, the nickel titanium
strips were acid etched as discussed in Appendix A.6. The etch removed the surface

oxides, which can embrittle the metal at high temperatures.

Following etching, the corrugations were tacked together using a small custom-built AC
thermocouple spot welder. For the initial experiment, specimen 071604A, 127 um
diameter, 3 mm, Nb wires were placed between the corrugations, as shown in Figure 15
(a). Once these corrugations were tacked together, as discussed in Appendix A.8, the
assembly, also known as a sandwich, could be manipulated without the corrugations
breaking apart. The sandwich was then placed in a Centorr vacuum annealing furnace,
and was allowed to reach <10'5 Torr before brazing. In initial experiments, the dwell

period was 10 minutes at 1200°C, which was reduced to 2 minutes at 1185°C once it was

31

discovered how quickly melting occurred. The heating schedule for the initial

experiments is shown in Figure 17.

The sandwiches were then post-braze annealed in an air furnace for 1.5 hours at 350°C to
restore the pseudoelastic functionality of the NiTi. This anneal has been reported to

reprecipitate metastable Ni-rich phases that are solutionized above 800°C. (Saburi)

2.3.3 Results

The sample was sent to Dr. John A. Shaw at the University of Michigan for compression
testing. The sample was cyclically compressed with increasingly compressive stains

tom 5% stain to 50% stain. The results of this test can be seen in Figure 16.

From the testing, it was shown that the compressive stain to failure of the sine wave
corrugated NiTi strips was superior to the stain to failure in the NiTi tubes. The
pseudoelastic effect of NiTi was further ampliﬁed by the corrugations compared with the

tubes. The density and stiffness of the array also decreased.

Niobium wires used in initial specimens etched holes in the NiTi, which decreased the
mechanical properties Of the array. To eliminate this problem, later samples were
constucted using Nb foils and multilayer ﬁlms that altered the contact geomety as

discussed in Section 2.2.3. The schematic design for later specimens is shown in Figure

17 (b).

32

 

 

t: 0.10 mm
t

 

w = 7.65 mm

/
S
- —d
d = 4.8 mm

 

 

Figure 14: Schematic of the corrugation design for the NiTi strip material. This formation was termed the
sine corrugation due to its resemblance to a sine wave, but it is deﬁned by alternating circular radii.

33

O

 

(b)

:0)
o w

O
O

 

Figure 15: Schematic of corrugation constuction using (a) 127 um Nb wires (b) 54 um Nb foils. While the

foils are thinner, they are considerably wider than the wires, thereby supplying more Nb to the liquid. The
foil does not cut through the NiTi, unlike wires.

34

 

 

p 0.5 7
A0
(MPa)
1 0.4 "
8'0 160
r- 6 I L (96)
is
95 e
(96)
l

 

 

 

-- t (min)

Figure 16: (Top) Stess stain curves of cyclic loading of NiTi sandwich. The sample was stained up to
50% with little permanent deformation, demonstating how the pseudoelastic effect can be geometically
ampliﬁed. (Bottom) Time versus stain plot, which shows the loading cycles for the sample.

35

 

Heating Schedule for Corrugated NiTi

 

1200 -*

 

1000 -~

°C

800 ‘-

Temperature,
Os
8

 

 

 

400 v
200 ~-
L
L
O a T t t ‘ t
O 10 20 30 40 50 60

Time, minutes

 

 

 

Figure 17: Typical heating schedule for initial sine wave corrugations. Furnace cooling continued until the
sample was removed at room temperature.

36

 

2.4 Construction of hexagonal corrugation structures

2.4.1 Introduction

Nickel titanium honeycombs with hexagonal corrugations were designed to further
amplify the pseudoelastic effect in Ni-rich NiTi strips at a lower density than sine wave

corrugations.

2.4.2 Materials and Methods

In a process very similar to that described in Section 2.3, strips of NiTi were acid etched,
shape set and assembled as shown in Figure 18. At the contact points between NiTi
strips, small 2 mm x 4 mm squares of 0.051 mm thick Nb foil were tacked into place as
described in Appendix A.8. A molybdenum ﬁxture was used to hold the sandwich inside
the vacuum furnace, as shown in Figure 19 and Figure 20. The entire honeycomb
assembly was then brazed using the method described in Appendix A.1. The heating

schedule of the process is shown in Figure 21.

The sample was then given a post braze anneal to restore the pseudoelasticity of the alloy.
The sample was heated in an air furnace for 1 hour at a temperature of 450°C. The
honeycombs were ﬁnally sent to Dr. John A. Shaw at the University of Michigan for

compression testing.

37

 

25.4mm

 

 

 

 

 

(b) (C)

 

 

 

 

 

Figure 18: (a) Schematic of shapeset tool honeycomb corrugation made of stainless steel. All sides of the
corrugation are 3 m. At the intersection of the 3 mm lengths is a 0.18 mm kerf. (b) Stacking sequence
showing 0.051 mm thick Nb pieces. (0) Final sandwich assembled and awaiting brazing.

38

 

Figure 19: Sample 030805B, honeycomb assembled using a small thermocouple spot welding unit to tack
the pieces ofNiTi to Nb foil. The array was placed in a molybdenum ﬁxture as shown, with a small weight
applying a compressive load to ensure the NiTi stips were well aligned.

39

 

Figure 20: This image shows the ﬁxtures and set-up for sample 0308053, for honeycomb production. The
honeycomb was encased in a molybdenum ﬁxture with a molybdenum weight applying a compressive load
to align the NiTi stips.

40

 

030905D Heating Schedule

 

 

 

 

400.0 -
200.0 s
0.0 I I 7 I T I
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0

Time, minutes

 

 

 

Figure 21: Heating schedule for 030905D. As shown, the sample was heated to 1185°C for a dwell time of
5 minutes in the Centorr vacuum furnace. During this process, the pressure began at 4.4 uTorr, and
reached a maximum pressure of 24 uTorr around 100°C before decreasing.

2.4.3 Results and Discussion

The geomety of the honeycombs allows Nb foils to be placed between the contact nodes
of the individual pieces of strip NiTi. This deviation in Nb geomety from wires
previously changed the melt dynamics; Nb no longer chemically etched through the NiTi.
As shown in Figure 22, the bond between NiTi and NiTi using Nb foil produces a low
wetting angle meniscus between the individual stips. This can also be seen in the SEM,
as shown in Figure 23. The liquid is also capable of spreading out and moving along

surfaces for several hundred microns.

41

Several honeycombs were made throughout this work at Michigan State University. Of
particular interest, 030805D was a large hexagonal array produced for compression
testing. As shown in Figure 25, sample 030805D was subjected to increasing stain every
cycle of testing. The maximum compression stain for sample was 60%, with only 8%
permanent stain after unloading. The density of the honeycomb was estimated by the
mass and volume measurements to be just under 0.5 g/cm3. With such a large
deformation and a low density, this honeycomb could also be described as an open celled,

tansforrnational NiTi foam.

The high temperature brazing process does change the thermo-mechanical properties of
the base alloy. Samples of the alloy before and after the brazing process were sent to the
Composite Center at Michigan State University to ﬁnd the tansformation temperatures
of the material using a DSC. A third sample was sent to the Composite Center after an
in-air heat teatment at 350°C for 1.5 hours in an attempt to restore pseudoelastic
functionality. The tansformation temperatures acquired tom the tests are shown in
Figure 24 and are summarized in Table 1. Based on these results, the austenite start (As)
and ﬁnish (Af) temperatures change signiﬁcantly, raising the potential for a signiﬁcant
phase fraction of martensite at room temperature. The martensite start (Ms) and ﬁnish
(Mf) temperatures also increase during and after brazing, compared with the as received

material.

42

 

Figure 22: If the NiTi stips are misaligned, a meniscus forms as shown. The meniscus, being rich in Nb,
has a shinier and smoother surface.

43

    

r

, , ‘ c '. . L I,
Figure 23: SE1 collage of 032805A. Bond taken from a honeycomb sandwich, which used a 54 um Nb foil
to bond pieces of NiTi together.

 

\

 

xv.

 

 

 

-60 -40 -20 0 20 4O 6O 80 100 120 140 160
°C

 

 

 

-— As Rec NiTi — Post Brazed NiTi —Aged NiTi

 

Figure 24: DSC results from pseudoelastic NiTi. From top to bottom, the austenite start and ﬁnish
temperatures shift as thermal processing continues.

45

a! r'lm

 

 

 

 

Figure 25: (a) Stain vs. time plot of sample 030905D showing the loading times and stains per cycle.
Sample was stained to a maximum of 60% stain. Low permanent deformation from each loading cycle
can be seen by the unloading area of the curve. After 60% strain, the sample had 8% permanent stain. (b)
Stess-stain curve for each increasing cycle number. Note the sample pseudoelastically deforms even
when compressed to 60% stain, which can be seen by little permanent deformation upon unloading. (c) (
Top to bottom) Images of the sample before any compressive loading, deformed on cycle 12 to 60% stain,
and unloaded from the 60% stain of cycle 12.

46

3 Reaction control using multilayers

Many ﬁlm deposition techniques exist, such as DC magneton sputtering, radio frequency
magneton sputtering, co-sputtering, plasma spray, and thermal evaporation. In
sputtering, atoms are electrically attacted towards the cathode by an electrical ﬁeld.
These atoms ballistically knock atoms from the cathode; these released atoms attach to
the substate. (Hou 1998) Sputtered nickel titanium ﬁlms, if grown at room temperature,
are amorphous and require annealing to crystallize. NiTi ﬁlms are commonly deposited
between 673-873 K in a crystalline form, which obviates the need for a post-deposition
crystallization heat teatment. (F emandes, Martins et al. 2002; Cho, Kim et al. 2005)
Amorphous ﬁlms have been reported to crystallize above 400°C when annealed for one
hour or more. (Femandes, Martins et al. 2002) Once deposited, NiTi thin ﬁlms can
perform a multitude of tasks including work generation and fatigue life improvement.

(Grummon, Nam et al. 1992)

Many researchers have attempted to make NiTi thin ﬁlms ﬁ'om multi-layers of pure
components. (Cho, Kim et al. 2005) These ﬁlms are later heat teated to synthesize NiTi
from the pure elemental nickel and titanium. Unfortunately, local ﬂuctuations in
chernisty create TizNi, NizTi, Ni3Ti and Ti3Ni4 precipitates that reduce the performance

of these ﬁlms.

Others have examined the performance of single-layer Ti-rich NiTi thin ﬁlms deposited
at above room temperature. (Zhang, Sato et al. 2005) These ﬁlms were examined using

tansmission electon microscopy to observe the evolution of Ti-rich precipitates in the

47

Nitinol ﬁlms. Small Gunier-Preston zones of TizNi were observed inside the NiTi gains
and at grain boundaries. These precipitates changed the twinning mechanism in the ﬁlm

and stengthened by precipitation hardening.

DC magreton sputtering is complicated by composition ﬂuctuation over the life of the
cathode. F ilrn composition can also vary based on the lateral distance away from the
center of the cathode due to different take-off angles of different elements. The
differences in take-off angles can be ampliﬁed by varying the distance between cathode
and substate. (Hou 1998; Zhang 2006) These ﬂuctuations can substantially change the

phase tansition temperatures of the ﬁlm and the mechanical properties.

Thin ﬁlm adhesion is critical for many applications; if a ﬁlm delarninated, it could
decrease fatigue life or lower functionality of the device. Surface oxides and oils must be
removed in order to develop high interfacial stength. (Zhang 2006) A ﬁnal sputter
cleaning of the substate inside the vacuum chamber prior to deposition can improve
adhesion. To sputter cleaning, the conductive sample is placed at a highly negative
electrical potential so that ionized argon atoms will be attacted to the substate. During
the argon - substate collision, the kinetic energy of the argon atoms is imparted to the

substate surface, which etches the surface.

Thin ﬁlm deposition techniques for the Creation of multilayers were taken from another
work at Michigan State University. (F oltz, Grummon et al. 2006) Ductile thin ﬁlms of

NiTi that dernonstated shape memory at room temperature were deposited on polyimide.

48

These ﬁlms were later cold co-rolled to develop uniaxial stains that could be thermally
recovered. These works, and the conditions necessary for creating ductile thin ﬁlms of

NiTi, are included in Appendix A.3.

3.1 Multilayers

3.1.1 Introduction

When brazing nickel titanium to itself and other materials it is often desirable to form a
liquid of NiTi and Nb from a single source, rather than multiple sources. As previously
described, capillary forces can draw liquid away from its origin to other areas, which can
create stuctural ﬂaws in a component. This can be eliminated if the eutectic liquid is

created by a ﬁller material, rather than a foil of pure niobium and NiTi.

Additionally, if the eutectic liquid source were rich in NiTi, it could deposit unmelted
NiTi to an area, potentially stengthening sections of a device. Conversely, if the liquid
were rich in Nb it would remove NiTi from the surrounding area, which could potentially
remove small amounts of surface oxides. There are several ways to manufacture such a
eutectic liquid source. With the equipment available at Michigan State University, DC
magneton sputtering of thin ﬁlms to create a multilayer of NiTi and Nb was the most
practical. The term multilayer ﬁlm (ML) will be used to describe alternating layers of the
NiTi compound and the Nb phase. This method of creation has the added beneﬁt of
drastically increasing the interfacial area between NiTi and Nb, on which melting is

dependent. With discrete thin layers of NiTi and Nb in intimate contact, the melting rate

49

of a multilayer should be orders of magritude faster than a single layer Nb foil (Low

2007)

3.1.2 Materials

Several multilayer ﬁlms of alternating NiTi and Nb layers were made using DC
magneton sputtering according to Appendix A5. The ﬁrst sets of multilayer ﬁlms,
specimen 111306A, were made from a sputtering cathode of Ti53,5-Ni46.5 at a working
distance of 51 mm. The ﬁlms were gown at ambient temperature with a base pressure of

3 pTorr and a working pressure of 4 mTorr of high purity argon gas.

The substates for these ﬁlms were glass microscope slides, which were cleaned with
acetone before deposition. The Nb was sputtered onto the substate for two minutes at
145 W, while the NiTi was sputtered for three rrrinutes at 228 W. The ﬁlms were
desigred to have 127 alternating layers of NiTi and Nb, with Nb on the top and bottom of

the ﬁlm.

After deposition, the ﬁlms were removed from the sputtering chamber and post
crystallized in a Centorr annealing furnace for 3 hours at 575°C. The ﬁlms were checked
for crystallinity using a Scintag 2000 XDS equipped with a Cu-K (1 X-ray source
operating at -3 SkeV. The results of a 2-0 angle scan can be seen in

Figure 26. Based on the two-theta scan, the ﬁlms showed a crystalline stucture. Using a
glass cutter, the glass slides were broken in half and the ﬁlms were manually

delarninated.

50

3.1.3 Results and discussion

A section of ﬁlm 111306A was imaged in cross-section using a Carnscan SEM, operating
at 17 keV, equipped with a backseattered electon detector. As shown in Figure 27, the
sample contains alternating, discrete layers of NiTi and Nb. The volumetric ratio of NiTi
to Nb was calculated using SEM imagery to be approximately 2.2:], which is very close
to the 2.18 volumetric ratio needed to produce the eutectic composition. Using the SEM
and a conventional micrometer, the thickness of the ﬁlm was found to be 34 pm. The
multilayer exhibited the shape memory effect in a ‘pinch test’, where the ﬁlm is pinched
and then placed on a hotplate. The foil was sent to the Composite Center at Michigan
State University for DSC analysis. The ﬁlm was found to exhibit both 819’ and R-phase
tansformations. The tansformation temperatures for the ﬁlm are shown in Figure 28

and Table 2.

A piece of specimen 111306A was placed between two pieces of pseudoelastic NiTi strip
material, and brazed in a Centorr vacuum furnace at the same time as an identical sample
with a 25 pm thick Nb foil. The results of the comparative rnicrostucture can be seen in
Figure 29. The overall rnicrostuctures are the same, but the thickness of the multilayer
bond is signiﬁcantly smaller because the 25 um Nb foil creates 80.7 pm of liquid, while
the multilayer creates only 34 pm. The bonds are thinner than the calculated thicknesses
because of capillary forces pulling material away, however the cross-section conﬁrms the

approximate volume ratio of NiTisz to be approximately 2.2:].

51

In addition to serving as a braze foil for NiTi — NiTi bonds, the multilayer foil could be

used for joining NiTi to dissimilar materials that have been coated in an inert, but wetable

material.

2-Theta X-ray, 1 11306A

 

 

 

 

Counts

 

 

 

r

40 45 50 55 60 65 70
Angle. degrees
I; 1 1 1306A — Theoretical Nb — Theoretical 132 NiTi]

r

30 35

 

Figure 26: X-ray diffraction results from sample 111306A after annealing for 3 hours at
550°C. Theoretical Nb and NiTi (BZ) X-ray peaks are shown at the top, while actual
results are on bottom. There are no signiﬁcant amorphous peaks, and based on thermal

shape recovery, the sample appears crystalline.

52

 

Figure 27: BEI of a cross-sectioned piece of specimen 111306A using a 25keV ﬁeld emission gun. The
bright phase inside the multilayer ﬁlm is niobium and the darker grey phase is nickel titanium. In total, the

ﬁlm is made of 127 discrete layers and is 34 pm thick

 

Table 2: DSC anal is of base chan e te tures in sam le 111306A.
l M I M | I R. A. 1 A I

f s Rf I
I 7°C I 22°C I 485°C 1 575°C [ 62°C [ 76°C I

 

DSC Results of 051407A

 

 

 

 

 

 

-20 -l 0 0 IO 20 30 40 50 60 70 80 90 100 l 10 l20

Figure 28: One exothermic peak upon heating (bottom half of curve) showing the austenite start and ﬁnish
temperatures. Two endothermic tansformations occur during cooling, which can only be interpreted as
formation of the R—phase upon cooling between 60°C and 40°C, followed by the 319’ martensite phase
between 25°C and 5°C.

54

25 In Nb Foil

34 m Multi . er Foil

 

Figure 29: Comparison between eutectic structures produced by 25 pm Nb foil (top) and a 34 um
multilayer ﬁlm (bottom). The eutectic region is larger in the Nb foil due to the base metal NiTi melting
during the brazing process.

55

3.2 Application of multilayers to spot welding

3.2.1 AC Spot welding

3.2.1.1 Introduction

Rapid thermal processes that limit thermal exposure to local areas are very attactive
torn a ﬁnancial and production standpoint. In addition to removing the need for
furnaces, spot welding units have high production rates and consume less energy.
Alternating current (AC) spot welding is a method of joining metals where pieces of
metal are locally joule heated using large currents and low voltages of electricity. The
electricity is applied to the sample through electodes that are usually made tom copper
based alloys having low electrical resistivity and high thermal conductivity. The
electodes are desigred to conduct the heat away tom the outside of the sample so that
melting is limited to the interface between the two halves of the sample. This is

schematically shown in Figure 30.

Spot welding relies on two types of resistance for heating: resistivity of the material and
resistance of the interface between the two halves. While the resistivity of the material
can be calculated and assumed consistent from sample to sample, the interface resistance
is dependent on many factors and must be considered a dynamic value. A proper spot
weld relies on both of these sources of resistance to heat the sample: the inherent
resistance of the alloy locally heats the sample near the melting point, and the interface
resistance raises the temperature at the interface to just above the melting point. Heavy

reliance on only one of these mechanisms can produce inconsistent welds. While the

56

inherent resistance of the material cannot be altered, the interface resistance can be
drastically altered by surface preparation and by adjusting the applied compressive load.
Contact resistance is an inverse function of compressive stress. Therefore, having a large
compressive force on the sample can reduce the interface resistance such that spot welds
become consistent; and relying mostly on the resistivity of the base metal. (Karagoulis

1991)

3.2.1.2 Materials and Methods

In this section, all samples were made from stips of 0.180 mm thick, pseudoelastic NiTi.
All multilayers that used were taken from sample 111306A. An SEWA AC spot welder

with a 5 kA power supply was used to create lap shear specimens for tensile testing.

A baseline for experimental parameters was established in sample 121906A. This sample
was made with round 1.588 mm diameter copper wire electodes. The experimental
parameters for this sample are shown in Table 3. A second sample, 010407D, was later
created to examine the rrricrostuctural effects of longer pulse times and higher pulse

power settings.

Table 3: Experimental parameters covered in experiment series 121906A.

 

Sample Applied Applied Pulse 1 Pulse 1 Pulse 1 Dwell Pulse 2 Pulse Pulse 2
Load Voltage ramp up time, sec Power, time, sec ramp up 2 time, power,
time, sec Amp time, sec sec Amp

 

121906A 0.1 kg 3.5V 0.0833 0.5 700 0.167 0.0833 0.5 300

 

 

 

 

 

 

 

 

 

 

 

010407D 0.1 kg 3.5 V 0.483 1.65 550 0.333 1.65 1.65 550

 

57

 

The ﬁrst series of experiments was desi grated 022107, and used dissimilar-shaped
copper electodes. The top electode was spade shaped, measuring 5 mm wide by 1 mm
long. The bottom electode was an 8 rrnn diameter electode. The experimental

parameters of experiment series 022107 can be seen in Table 4.

Table 4: for AC weld 022107.

olts up 1
' sec rme

         

.0833
.0833

.0833
.483
.483
.483

A second series of experiments, 030107, was conducted after consultation with Dr.
Michael Karagoulis. Ramp up and down times were eliminated to simplify the
waveform. New electodes were desigred and machined with a symmetric,
hemispherical shape with a 3.2 mm diameter ﬂat at the contacts. These electodes were
used in all subsequent AC spot welding. The electodes were made tour a chromium-
copper, Alloy 182, 15.88 mm diameter rod purchased from McMaster-Carr. The
electodes can be seen schematically in Figure 30. Experimental parameters for 030107

can be seen in Table 5.

Table 5: Experimental parameters for series 030107.

 

 

 

 

Applied Load Electrical Applied Pulse 1 Time, Pulse 2 Time, |

Current Voltage sec sec
IMin I41.SI(L I250A |2.5v |0.167 lo |
[Max | 41.5 It; | 450 A | 2.5 v | 0.666 | 0 |

 

Series 040107 was a third and ﬁnal experiment desigred to ﬁnd the optimum parameters.

The bounds for experiment 040107 are in Table 6.
58

Table 6: Experimental parameters for series 040107.

 

 

 

Applied Load [Electical Applied Pulse 1 Time, I Pulse 2 Time, I

Current Voltage sec sec
| Min [41.5 kg | 750 A | 2.5 v | 0.333 | o |
IMax [41.5 kg | 1200A |2.5v |0.333 [0 |

 

3.2.1.3 Results and Discussion

Sample 121907A was made from a small piece of multilayer 111306A, described in
Section 3.1, placed between NiTi strips. The results of the spot welds can be seen in
Figure 31 and Figure 32. Near the center of the spot weld, the multilayer mixed with the
molten base metal, as evident by taces of niobium near the surface. Near the edge of the
spot weld, in Figure 32, poor adhesion between the multilayer and the base metal was
observed. Joule heating in the multilayer was geater than in the base metal due to the
number of metallurgical interfaces. Heat conduction away from the weld zone in the
base metal was high; however, the conduction tom the multilayer to the base metal was
low. This resulted in poor adhesion between the multilayer and base metal. From
literature, this suggests that the contact pressure of the electodes should be increased to

improve the heat conduction tom the thin ﬁlm to the base metal. (Karagoulis 1991)

In sample 010407D, niobium appears to be solutionized in some areas, and phase
separated in other areas, shown in Figure 37 and Figure 38. The backscattered electon
images show that the base metal has sigriﬁcantly rrrixed with the multilayer, meaning the
sample was heated above the melting temperature. This stongly suggests that the
heating time should be reduced, the current increased and the electodes redesigned for

more efﬁcient thermal conduction. (Karagoulis 1991)

59

In an attempt to discover the microstuctural dependencies of each setting, the ﬁrst series
of experiments varied power, voltage and pressure according to Table 4. Compared with
other samples in this series, the multilayer adhered to the base metal better in sample

022107D and had less residual porosity. This is because the sample used a higher power

setting, which melted the multilayer and surrounding base metal more than other samples.

The resulting microstucture of 022107D can be observed in Figure 33. The sample was
imaged after breaking in a lap shear test. While the sample featured a eutectic solidiﬁed
area and higher stength than other samples, it was not an Optimal spot weld. Some large,
brittle precipitates of Ni4Ti3 gew during the process, as shown in Figure 34; these
precipitates are undesirable because they are too large to stengthen against dislocation

motion.

At 3.5 V and 0.1 kg pressure, sample 022107B showed the best stength. Specimens
022107E and 022107F, which were made with 4.5 V, did not join the multilayer to the
base metal as well as 0221078, and contained more microvoids than the other samples.
Sample 022107F contained gain boundary Nb that gew out perpendicular from the

multilayer because of increased voltage, shown in Figure 35.

Experiment 032307 adjusted the power and time length, to ty to improve the

microstucture of the spot welds. Throughout this experimental series, as power

60

increased, the breaking stength of the spot welds increased. All the specimens fractured

through the spot weld in a tensile fashion, as shown in Figure 36.

To better match the power settings of DC spot welding, experiment 04122007 used
shorter pulse times and larger applied currents. Twenty-e1 ght samples were made, each
with an identical ﬁve spot weld pattern. The experimental results from this series can be
seen in Table 7. Specimen 04122007-13 had the highest breaking stength at 500N, or

591 MPa, using an applied current of 950 amps.

A sample was made for metallurgical inspection using a 950 amp power setting, shown in
Figure 39. The multilayer ﬁlm has mixed with the base metal, suggesting that spot
welding can be completed without the multilayer. However, the power required for spot
weld with the multilayer foil is lower than that required without the multilayer, which is a
result of the increased joule heating inside the multilayer and at the contact interfaces.

By increasing the resistance of the sandwich, the spot weld using the multilayer foil

requires less power to heat the above the melting point.

During experiment 032307, it was speculated that the multilayer was not required to spot
weld NiTi pieces together. Based on this hypothesis, samples 04122007-27 and
04122007-29 were made without any multilayer foil. Specimen 04122007-29 had the

highest breaking stength, 510 N, with a power setting of 1200 amps.

61

A sample was made without a multilayer ﬁlm, using a 1200 amp current pulse for
metallurgical, shown in Figure 40. The spot weld has a non-symmetric nugget,
suggesting the waveform of the power supply was biased to one half of the AC sine
wave. The nugget does not penetate more than halfway through the NiTi base metal,
which is an improvement ﬁ'om initial results and suggests good adhesion in the bond.
However, to reduce residual void porosity inside the nugget, a higher pressure should be

applied through the electodes.

Based on the AC spot welding results from 04122007, it is unnecessary to use the
multilayer ﬁlm to join NiTi. The multilayer increases the resistance in the spot weld due
to the metallurgical interfaces, which lowers the required current to make a spot weld.
There is insufﬁcient temperature contol in spot welding to melt only the multilayer while
heating the base metal for good adhesion. Multilayer ﬁlms do not improve the stength

of AC spot welding, and would add sigriﬁcant manufacturing cost.

62

Table 7: Procedure of ' series 04122007.

 

Load
00

ML foil

ML foil

63

 

Figure 30: Schematic drawing of spot welding design. (A) Spot weld nugget (light grey) formed by local
melting at the interface between the two pieces of the sample (B). Melting is limited to the interface by
rapid heat conduction (lines) through the electodes (C). The electodes apply a compressive Stess to the
sample.

 

 

Figure 31: Grain boundary niobium phases and small precipitates of Nb inside the NiTi base metal of
121906A. Temperatures exceeded 1300°C locally, melting the base metal. Black spots are etch pits.

65

 

Figure 32: Sample 121906A2 (A) SE1 of multilayer tansitioning from unmelted to fully melted,
demonstating the processing temperature gradient. (B) The multilayer did not adhere well to the base
metal, which could be a result of increased heat conduction or less joule heating in the base metal.

66

l()( 1 11m

 

Figure 33: SE1 ofpost—ﬁ‘actured 022107D. Specimen broke at the edge of the spot weld. Eutectic solidiﬁed

multilayer can be seen at the fracture surface. This implies that the spot weld heated outside of the
electodes (overheated).

67

 

Figure 34: Ni4Ti3 precipitates in needle shaped morphology can be seen in the center of this SE1 of sample
022107D. Niobium precipitates and residual porosity decorate the surrounding microstucture.

68

 

 

Figure 35: Sample 010307A. Higher spot welding voltage causes increased residual porosity, grain
boundary Nb precipitates and irregular shaped melting paths.

69

"" T W'TQ~- . 3FW%§,se-w ..s.
. , _. -4 ..

 

Figure 36: AC spot weld from experiment 032307. Sample was tested in shear and failed throughout the
weld area. The small lighter grey areas on the multilayer, visible on the left half, show the area that the
electodes melted. This area is smaller than the diameter of the electode.

70

lIlIIlIII lllllym I____.I

 

Figure 37: Low magniﬁcation BEI of spot weld 010407D. Niobium mixed through the base metal towards
the outer surface. There is evidence of convection currents mixing the multilayer into the base metal.

71

 

Figure 38: High magniﬁcation BEI of Nb phase separation. Bright steaks inside the dark NiTi phase
suggest solutionized Nb.

72

 

Figure 39: 8131 of 053 107A AC spot welding sample with niobium. Mixing of Nb with base metal suggests
the multilayer foil is not critical for spot welding.

73

 

Figure 40: Specimen 053107D, AC spot weld without Nb. Melting occurred only near the interface
between the NiTi pieces. The spot weld is approximately as wide as electrode contacts. The weld has
melted more material to the right, suggesting the waveform of the power supply was non-symmetric.

74

3.2.2 DC Spot welding

3.2.2.1 Introduction

Direct current (DC) spot welding is also commonly used in industry. This form of spot
welding can be more reliable because the waveform of the power output more precisely
controlled. The DC spot welding supply used for this research could only adjust between
1 and 9 millisecond pulse times, compared to 0.0833 to 1.5 second pulses in AC spot

welding.

3.2.2.2 Materials and Methods

Initial attempts at DC spot welding used niobium foil to join NiTi to itself, which proved
unsuccessful. It was hypothesized that the niobium foil could not melt sufﬁciently,
because of the very short heating times involved, which resulted in poor adhesion. To
solve this problem, the multilayer sample 111306A was created, which was designed to
melt much faster. The interfacial area between NiTi and Nb was drastically increased by
the number of alternating layers of Nb and NiTi. The melting rate of the ﬁlm is
dependent on the interfacial area, so the foil should melt several orders of magnitude
faster than an Nb foil of similar thickness. The ﬁlm also supplies the necessary NiTi to
create the liquid. If the temperature of the spot weld remained below 1310°C, mixing of

the base metal and the liquid should not take place.

75

Spot welds utilizing the multilayer ﬁlm had much higher adhesion than spot welds using
Nb foils. A series of experiments was run to determine the best power settings for the
DC spot welds. Using a set of copper-chrome electrodes with hemispherical ends and an
applied pressure of 40.5 kg, the power settings were adjusted from 300 amps to 2700
amps. The electrical pulses delivered by the power supply started at 8 ms pulses, and
were increased to 9 ms pulses to strengthen the bonds. Experimental details can be seen

in Table 8.

76

Table 8: Experimental procedure and results of experiment series 03272007.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample P4 T3 Thickness Max load (N)
1 0.93 7 0.353 495
2 0.93 _ 9 0.358 500
4 0.95 7 0.369 480
4a 0.95 7 0.367 560
5 0.95 9 0.373 575
5a 0.95 9 0.367 485
7 0.97 7 0.356 275
7a 0.97 7 0.367 520
8 0.97 9 0.367 325
8a 0.97 9 0.368 440
10 0.99 7 0.361 410
11 0.99 9 0.363 460
13 1.01 7 0.372 325
14 1.01 9 0.367 460
16 1.03 7 0.334 215
17 1.03 9 0.357 510

 

 

 

3.2.2.3 Results and Discussion

The optimum DC weld conditions were determined for the 0.190 mm thick pseudoelastic

strip material, which can be seen in Table 9. A lap shear sample, 03272007-5, was made

with these conditions. The actual pulse current applied using the DC power supply was

1.7 kA, not the selected 0.95 kA. The resulting fracture surface ﬁom the tensile can be

seen in Figure 41; the associated stress-strain curve can be seen in Figure 42. The sample

broke with the force of 575 N inside the base metal. The calculated tensile stress in the

metal at failure was 644 MPa, based on the cross section of the strip NiTi.

Table 9: Spot weldinﬂiarameters for sample 03272007-5.

 

 

 

Applied Load, Pulse Time, msec Pulse Current, Thickness, mm Width, mm ML thickness,
ES 3&3 um
41.5 9 1700 0.190 4.7 34

 

 

 

 

 

 

77

 

Specimen 040207 was made for metallography using the same spot welding conditions as
previously described. The spot weld process appears to have pushed the liquid from the
multilayer away from the spot weld, as seen in Figure 43. The niobium that remained in

the spot weld has phase separated.

Optimum AC and DC spot welds melt signiﬁcant ﬁ'actions of the base metal. The
multilayers are not necessary because the base metal melts. Spot welding NiTi to NiTi
without Nb or a multilayer should be possible; using higher power settings would

increase the joule heating in the base metal.

78

 

Figure 41: Specimen 03272007-5. Fracture surface at the edge of the spot weld. There is no oxide
discoloration in the spot weld. Material ejected from the sample as a result of the electrodes pushing liquid
from between the NiTi strips.

79

DC Spot Weld Tensile Strengths

 

700

Tensile Strength, MPa
U)
8

200 --

100 *-

 

 

 

 

 

0 4 * 1 : . 1 ‘ . 7* 1 ' ‘ 1 : ' ' ‘ :
0% 1% 2% 3% 4%
Strain, mm/mm

 

[—03272007-13 ]

 

Figure 42: Stress-strain curve for the strongest spot weld.

80

5%

 

Figure 43:SE1 of 040207. Niobium appears to be solutionized or pushed out of the spot weld. The niobium
that can be seen has phase separated, as shown in the insert.

81

4 Bonding dissimilar NiTi sections

Establishing a reversible shape memory effect is critical in many designs where actuation
or a pre-programmed response is required. Reversible shape memory, or two-way shape
memory, can be created in a variety of ways. One approach is to train the nickel titanium
by inducing plastic deformation in a controlled manner, which will result in a self-biasing
effect in the material. When the material heats up, it will form one shape, and when it
cools down, the nickel titanium will remember a different shape. This can develop large
forces, but rather small displacements because this type of reversible shape memory is

roughly limited typically to 1% strain.

Shape memory alloys can also be biased using other materials like steel; at a low
temperature, Nitinol is more compliant than steel and visa versa at high temperature.
Larger strains can be developed using this method, up to the elastic limit of the biasing
spring, which can be mechanically ampliﬁed. Examples of this have been shown in
devices such as the Krulevitch microgripper and the Johnson valve. (Johnson, Gupta et

a1.) (Lee, Ciarlo et al. 1996)

The ultimate biasing spring would be one that elastically deforms much beyond the
typical 1% strain of most engineering metals: a pseudoelastic metal. It is therefore
important to examine the capability of the niobium brazing system in order to join
dissimilar Nitinol formulations. In addition to transformational shape memory and
pseudoelastic Nitinol, NiTi can also be created as a thin ﬁlm, a non-transforming porous
foam or a fully dense non-transforming ‘chemical’ Nitinol. Both porous and chemical

82

Nitinol contain impurities and usually do not transform into the B19’ phase upon cooling,
these formulations of nickel titanium are very important for biomedical applications.
Porous nickel titanium is a biocompatible, high compliance, low-density structure that
allows tissue growth inside the pore structure. Chemical Nitinol is less expensive than
other forms of NiTi, and could be used in biomedical devices that do not require
pseudoelasticticity or shape memory. The latter form of nickel titanium could be used as
a structural material to join Nitinol sections together without drastically raising the price
compared with using active Nitinol. (Grummon, Shaw et al. 2003; Chu, Chung et al.

2005; Greiner 2005)

4.1 Bonds between NiTi thin ﬁlms and wrought Nitinol

4.1.1 Introduction

Bonds between macroscopic sections of NiTi and thin ﬁlms are important in the
production of MEMS devices and other thin ﬁlm applications for NiTi. Without bonds
between thin ﬁlms and macroscopic materials, installation and assembly of thin ﬁlm
devices can be difﬁcult. These bonds must possess microstructural stability and ductility,

but do not require high strength.

4.1.2 Materials

Shown in Figure 44, thin ﬁlms of NiTi can be bonded to bulk NiTi alloys. This specimen
was made from 0.254 mm thick wrought pseudoelastic strip NiTi bonded to a SMA

ﬁeestanding thin ﬁlm approximately 10 pm thick. The ﬁlm was cut to the same width as

83

the strip NiTi and was tacked to the top faces of the strip NiTi using a small AC spot
welder. The sample was brazed using the MILA ULVAC furnace with a heating rate of
30°C per second and a dwell time of 30 seconds. After cooling below 300°C, the sample

was removed from the furnace and quenched in water.

Thin ﬁlms can also be rolled and bonded to the inner faces of pseudoelastic NiTi tubes.
A section of the same thin ﬁlm was bonded inside NiTi tubes using the heating process as
previously described above. The sample was then sectioned, mounted and polished for
metallurgical examination. An etchant, described in Appendix A.7, was used sparingly to

develop contrast between the phases, as shown in Figure 45.

 

 

 

Table 10: Transformation tern eratures ﬁ'om DSC results corresponding to Fi e 46.
Sample Mf Ms Rf R, A, Af
012907A 5°C 26°C 57°C 61°C 66°C 77°C

 

 

 

 

 

 

 

The bond is ﬁlled with TizNi, rather than eutectic NiTi and Nb. Based on the
transformation temperatures of the thin ﬁlm shown in Table 10, it is unlikely the ﬁlm is
rich in titanium. The NiTi tube is nickel rich, so the bond must contain TizNi as a result
of oxygen stabilization of the TizNi phase. EDS data from the area also supports this

theory; niobimn is found in the TizNi, which might further stabilize the phase.

84

 

 

 

Figure 44: Thin ﬁlm bonded to wrought pseudoelastic NiTi. Film, originally the same width as the
wrought NiTi, was tacked to top faces of corrugation. During brazing, the ﬁlm sagged into contact with the
wrought material.

85

 

 

Figure 45: Thin ﬁlm 012907A bonded to the inner surface of a NiTi pseudoelastic tube. The thin ﬁlm was
oriented with the 750 nm Nb coating facing towards the tube. The ﬁlm was wrapped approximately 2.5
times around the inside of the tube, as shown at the top center of the image. TizNi is present at the bond
line.

86

DSC Results of 012907A

 

-40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160

°Celsius

Figure 46: DSC data from freestanding thin ﬁlm 012907A. The ﬁlm transforms from BZ austenite to R-
phase to 819’ martensite upon cooling, based on the two large upward peaks.

4.1.3 Results and Discussion

The ﬁlms used in both Figure 44 and Figure 45 were made from DC magnetron
sputtering using Ti535Ni465 and 99.999Nb cathodes, as described in Appendix A.3. In
Figure 45, the microstructure of the bond shown is dominated by TizNi, but small phase
fractions of NiTi and Nb are also present. Without considering oxygen, these terminal
phases normally are only in equilibrium if the local alloy composition is rich in titanium
and not located on the quasi-binary eutectic. With the high transformation temperatures
shown in Table 10, it is possible that the thin ﬁlm could be rich in titanium. Another
indication the ﬁlm is rich in titanium is the presence of the R-phase, shown in Figure 46,

which Ti-rich thin ﬁlms typically form upon cooling from the 82 austenite crystal

87

structure before forming the 319’ martensite. (Otsuka and Ren 2005) TizNi can also be

stabilized by oxygen, which could also make it a more favorable solidiﬁcation product.

The bond between the thin ﬁlm and the bulk wrought material is similar to the bond
between two wrought pieces of Nitinol. The bond might be further improved if the phase
fraction of TizNi was reduced by either better vacuum control or the addition of extra
nickel to compensate for Nb substituting for Ti and forming TizNi. Higher vacuum is

difﬁcult, so the simple solution would be adding extra nickel to the multilayer source.

The mechanical performance of the thin ﬁlm was lowered by the brazing process. The
formation of TizNi dulled the surface ﬁnish of the ﬁhn, and increased the stiffness of the

ﬁlm.

4.2 Thin film to Porous Nitinol

4.2.1 Introduction

Complex biomedical implants may eventually employ thin ﬁlms that work as pumps,
valves or even artiﬁcial muscles. These devices may be physically attached to bone or
tissue by cellular material ﬁlling the pores. Bonds between these thin ﬁlms and foam
materials must be strong, tough, and able to resist damage from high cycle fatigue and
corrosion. The ﬁrst step in producing devices like this is demonstrating that NiTi foams

can be bonded metallurgically to thin ﬁlms.

88

4.2.2 Materials

Specimen 040607A was created by brazing a Nb-coated thin ﬁlm to a closed-cell type
NiTi foam received from NDC. A small section was taken ﬁom a 3.8 cm diameter round
stock of NiTi foam and was ultrasonically cleaned for 5 minutes in cyclohexane. The
thin ﬁlm was sectioned from sample 013007A, which was made by DC magnetron
sputtering. NiTi was deposited at room temperature from a Ti53.5Ni45.5 sputter target for 8
hours at 200W, followed by 10 minutes of Nb deposition at 200W. The base pressure of
the deposition was 3 uTorr, while the working pressure of the sputtering guns was below
4.5 mTorr throughout almost the entire deposition. The ﬁlms were deposited on glass
microscope slides, which were a distance of 50 rrrrn away from the cathode. To help
delarrrinate these ﬁlms from the slides, the working pressure of the sputtering guns was
maintained at 7 mTorr during the ﬁrst 5 minutes of the deposition, Following deposition,
the ﬁlms were post-crystallized in a Centorr vacuum annealing furnace at a temperature

of 575°C for 1.5 hours before being peeled away from the glass slides.

To create sample 040607A, the 3 mm square of thin ﬁlm was placed on top of the foam
inside the MILA ULVAC vacuum ﬁrrnace. The furnace was allowed to pump to low
vacuum, 0.1 mTorr, before heating. The temperature of the sample was determined with
a direct contact thermocouple probe placed beside the foam NiTi. The sample was
heated to 1185°C in 40 seconds, held at 1185°C for 10 seconds and then furnace
quenched by turning off the inﬁared heating elements. Once the sample was cooled
below 300°C, it was removed from the fumace and quenched in water. The sample can

be seen in Figure 47.

89

4.2.3 Results and discussion

The sample was cross-sectioned for metallurgical examination in the Hitachi S2500-C
SEM. The niobimn coating on the thin ﬁhn was in direct contact with the foam and
created a volume of liquid approximately 2.5 pm thick. This liquid ﬂuxed the surface
oxides and formed a mixture of TizNi, Nb and NiTi phases upon solidiﬁcation. The Nb
from the ﬁlm substituted for the titanium in the TizNi phase rather than forming a
niobium rich phase, as seen in Figure 48. Based on EDS information shown in Table l 1,
the bond and the ﬁlm appear to be rich in titanium. Traces of niobium can be found in
both the ﬁlm and the bond, but not in the porous material. These observations are
consistent with the ﬁlm being rich in titanium and fornring TizNi, with the Nb

substituting for Ti in TizNi.

Table 11: EDS analysis of spots shown in Figure 48.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Spot Ti Ni Nb

A 55.00 44.53 0.5

B 48.22 51.62 0.16
c 50.39 49.10 0.51
D 52.84 46.83 0.33
E 55.66 43.92 0.41
F 53.94 45.57 0.49
G 51.83 48.19 0.00
H 47.48 52.57 0.00
1 47.06 52.96 0.00

 

90

 

 

 

 

Figure 47: NiTi low density closed cell foam brazed to SMA NiTi thin ﬁlm.

91

r

50 um

 

Figure 48: Thin ﬁlm bond to NiTi foam material. (1’) Thin ﬁlm, (r'i) bond line between thin ﬁlm and (iii)
NiTi foam. (iv) TizNi phase is present, not the Nb phase.

92

4.3 Thin ﬁlm to Chemical Nitinol

4.3.1 Introduction

Bonds between NiTi thin ﬁlms and chemical, or non-transforming, NiTi could be
important in building advanced biomedical implants that make use of thin ﬁlm valves.
Scaffolding would ﬁx the valve in place; bonds between chemical NiTi and NiTi thin
ﬁhns could provide the mechanical and electrical connections necessary for a complex
device. If the mechanical requirements of the scaffolding were low, chemical NiTi
would offer a lower cost alternative than bonding the thin ﬁlm to pseudoelastic or shape-

memory NiTi.

4.3.2 Materials

Sample 060107A, a 40 um DC magnetron sputtered ﬁhn bonded to a piece of chemical
Nitinol, was made based on the procedure in Appendix A.2 with a dwell time of 30
seconds at a time of 1185°C. The ﬁlm was taken from sample 013007A, which was a
freestanding ﬁlm of approximately 40 pm of NiTi and 1 pm of Nb. Both NiTi and Nb
were deposited at 200 W; the NiTi was deposited for 8 hours using a Ti53,5-Ni45,5
sputtering cathode, and the Nb for 10 minutes. The sputtering followed the procedure
listed in Appendix A.4. The chemical Nitinol was cut ﬁom a used sputtering cathode
with a composition of Ti53,5-Ni46,5. Following brazing, the sample was sectioned and

metallurgically mounted and polished.

93

4.3.3 Results and Discussion

Sample 060107A can be seen in Figure 49. The microstructure of the joint between the
ﬁlm and the bulk material is ﬁlled with TizNi, which is stabilized by the presence of
oxygen. The EDS data supports the morphological evidence for TizNi, as the titanium
content inside the bond is signiﬁcantly higher than in the ﬁlm or the base metal. When
TizNi forms, it often changes the local chemistry of the metal, enriching the NiTi with Ni
atoms. If some level of surface oxides stabilized the TizNi phase, Nb could substitute for

Ti in TizNi, helping force the phase in the bond.

While the TizNi phase does limit the strength of the bond between bulk NiTi and a thin
ﬁlm, this could be overcome with the proper design considerations. Additionally,
metallic foams have limited strength, so low bond strength would be expected and

acceptable.

Table 12: EDS results of chemical NiTi to thin ﬁlm bond, shown in Figure 49.

 

 

 

 

 

 

Spot Ni at% Ti at% Nb at%
A 43.79 56.04 0.13

B 45.16 54.67 0.16

C 50.81 49.04 0.15

D 51.53 48.46 0.008

 

 

 

 

94

 

 

Figure 49: SE1 of sample 060107A, a thin ﬁlm (top) bonded with chemical nitinol (bottom). The bond line
between the ﬁlm and bulk was probed using EDS at spots A and B. The thin ﬁlm and bulk compositions
were explored at points C and D respectively, as shown in Table 12.

4.4 Porous NiTi to Wrought Nitinol

NiTi foams can be bonded to wrought NiTi using Nb to stimulate local melting of the
interfaces, as shown in Figure 50. This sample, KB052806, was formed using the
procedure given in Appendix A.1, with a dwell time of 2 minutes and a temperature of
1185°C. As shown, the 54 um Nb foil formed a liquid that inﬁltrated some of the lower

pores of the foam, which appear brighter in the ﬁgure. These types of bonds could
95

provide biomedical implants with strength and pseudoelastic or shape memory
performance in one area, while allowing tissue inﬁltration in another area. Such an
implant could be useful in the replacement of bones that require both tissue inﬁltration
and strength. These bonds do not signiﬁcantly differ from microstructures seen in NiTi

foams bonded with chemical NiTi.

 

Figure 50: Sample KB052806. NiTi foam (top) bonded to wrought SMA plate Nitinol (bottom) using 54
um Nb foil.

96

4.5 Porous Nitinol to Chemical Nitinol

4.5.1 Introduction

Bonding fully dense sections of NiTi to porous NiTi foams may be useful for biomedical
implants, because cellular growth can occur in the porous structure while the fully dense,
structurally active NiTi protects the body by dampening loads. Similarly, bonding porous
NiTi to chemical NiTi, which does not transform but is less expensive than traditional
NiTi due to processing standards, could help lower the overall cost while improving the
functionality of biomedical implants. Chemical NiTi traditionally is less expensive than
transformational NiTi because it contains higher concentrations of impurities. The
purpose of this experiment is to determine if chemical NiTi’s higher impurity level could
lower the strength of the NiTi brazing process by forming TizNi or other undesirable

intermetallics.

4.5.2 Materials and Methods

Chemical Nitinol can be bonded to NiTi foams using either niobium foils or multilayers.
Bonds between these materials are stronger than the porous material, and thus are
untestable using standard mechanical tests. Due to the inherent inability to test it, this
combination of materials was only brieﬂy explored. A piece of chemical NiTi was
bonded to porous NiTi, received from NDC, using a small piece of multilayer ﬁlm,
specimen 111306A. The sample was processed in the MILA furnace as according to

Appendix A.2, with a set temperature of 1185°C and a dwell time of 1 minute. The

97

sample was sectioned, polished and metallurgically etched using the procedure in

Appendix A.7, and was observed in a Hitachi S-2500C SEM operating at 10 keV.

4.5.3 Results and Discussion

As shown in Figure 51, a eutectic of NiTi and Nb phases ﬁlls the bond between materials.
This eutectic structure is similar to the microstructure of chemical NiTi to chemical NiTi
bonds, shown in Figure 52, and SMA NiTi to SMA NiTi bonds. The microstructure is
different, however, than the TizNi that is observed in thin ﬁhn to foam bonds, shown in
Figure 48. This difference in rnicrostructure is not due to by surface preparation
techniques, as all samples with visible black layers of surface oxides were etched clean.
The difference could, however, be due to the differences in oxygen content between the

multilayer and the 40 um ﬁlm.

98

 

Figure 51: Etched microstucture of NiTi foam (top) bonded with chemical NiTi (bottom). The eutectic
microstructure can be clearly seen forming a bond between the materials and has ﬁlled some of the holes in
the porous structure that are near the interface.

99

 

Figure 52: Microstructure of chemical NiTi to chemical NiTi bond. Unmelted pure Nb foil remains in the
center of the bond; eutectic NiTi and Nb have solidiﬁed between the Nb foil and NiTi base metal.

100

4.6 Shape memory to pseudoelastic Nitinol

4.6.1 Introduction

Bonding shape memory NiTi to pseudoelastic NiTi is an attractive method of producing a
thermally reversible birnetal actuator that employs a small temperature range, unlike

traditional birnetal actuators that require large temperature changes.

Shape memory NiTi can be fastened to other metals, so that a spring can re-deform the
Nitinol during cooling, creating a biased shape memory effect. The amount of
deformation capable in these devices is limited however, to the stiffness and local strains
inside the biasing material. Utilizing a thin plate of pseudoelastic Nitinol to bias the

shape memory NiTi could provide a larger reversible strain.

4.6.2 Materials and Methods

Two pieces of 1.588 mm thick plate SMA NiTi and two SE NiTi pieces were EDM cut
into 12.5 mm by 45 mm coupons. These coupons were then shape set ﬂat, in a die held at
500°C for 30 minutes. The samples were subsequently etched to remove any surface

oxides.

To create birnetal plates, the pseudoelastic plate was placed atop the shape memory plate,
with 127 um Nb foil between the contacting faces of the materials. The samples were
loaded into a Centorr vacuum annealing furnace and degassed at a temperature of 375°C

for 12 hours to obtain a base pressure of 2.6 uTorr. A brazing schedule of 10 minutes at

101

1190°C was desired; the actual heating proﬁle is shown in Figure 53. Following 10
minutes of brazing, the sample was allowed to furnace cool in a vacuum chamber to
below 100°C. In air, the sample was post annealed at a temperature of 350°C for 1.5

hours to restore pseudoelastic functionality in the nickel-rich half of the sample.

Heating Proﬁle

 

 

Temperature, degrees Celsius

 

 

 

 

200.0 I l T T I I
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
Time, minutes

Figure 53: Heating schedule for bimetal. The heating proﬁle consists of a ten minute dwell time at 1190°C,
then furnace cooled until below 100°C. The maximum pressure of 10.5 uTorr was reached at 900°C,

4.6.3 Cold rolling thin plates for actuators

The samples created in 4.6.1 were subsequently cold rolled to induce a two-way shape
memory effect in the material. As the sample is heated, the shape memory half of the

plate should transform to austenite, which is signiﬁcantly stiffer than the prior martensite.

102

As this happens, the pseudoelastic material, which is softer than the austenite in the SMA
half, should deform by twinning mechanisms to martensite. This apparent ‘ﬂip’ of the
microstructures will allow the bimetal to straighten towards the original ﬂat shape. Upon
cooling, the material should re-deform as the SMA material converts back to martensite
and the some stress is relieved ﬁom the pseudoelastic NiTi, causing it to convert back to
austenite. Through this entire process the bond between the halves of the bimetal must be
able to transfer the stress from one half to the other without yielding. If the bond is too
thick it could deform signiﬁcantly, and allow one half of the bimetal to slip. The bond
must also accommodate any bending strains that might arise from the unbalanced elastic

modulus in the material.

Specimen 041406A was cold rolled to 3% strain by a Penn cold rolling mill. At this
point the specimen curled towards the martensite, which could have been a result of
unbalanced roller speed in the mill. The sample was then cross-rolled perpendicular to
the prior rolling direction, which did not change the overall curl of the sample. Upon
heating to 100°C, the sample did not change shape. This could have many causes,
including improper straining of the material, plastic versus elastic deformation of the
pseudoelastic material during cold rolling, or improper transference of the stress through
the bond. The sample was then rolled to 7.7% strain; at which point it curled towards the
SMA half slightly less than before. The sample was also hot, approximately 45°C, upon
exiting the mill. Upon heating to 100°C the sample did not recover the stored curl, but
did reduce in length by 0.52 mm, or 1% strain. This implies that the curvature of the

specimen was imposed by improper cold rolling; unbalanced strain in the sample could

103

develop if the rollers of the mill moved at different rates then an unbalanced strain could

develop throughout the specimen.

The sample was subsequently rolled to 12.3% strain and still curled towards the SMA
half of the sample. Finally, the sample was taken to failure at 39.2% strain, where it
failed violently in shear with a rapid release of energy. As a note to readers, cold rolling

should be done with extreme care following all available safety precautions.

The sample was sectioned along transverse and longitudinal directions for metallurgical
inspection. The sample was observed using a Hitachi S-2500C SEM; the unfractured
bond can be seen in Figure 54 and Figure 55. Additionally, the microstructure is
compressed in the vertical direction, which demonstrates that the bonding process is
compatible with subsequent forming techniques. Forming techniques, including cold

rolling, could be used to design bimetals that demonstrate reversible shape memory.

104

 

Figure 54: 39.2% cold rolling of sample 040605A. Eutectic has been compressed in the vertical direction
40%. No cracking is visible.

105

 

Figure 55: Eutectic shows no major change in the microstructure. 40% compression in vertical direction.
Niobium deforms easily without cracking.

106

5 Joining NiTi to other materials

Bonding nickel titanium to dissimilar metals and nonmetals, such as titanium or alumina,
is exciting from a technological standpoint as it allows complex devices with either
transformational pseudoelasticity or shape memory to make use of the strengths of other
materials. An example would be combining the high wear resistance of alumina with the

low density and high stiffness of titanium and the pseudoelasticity of NiTi.

NiTi was bonded to numerous materials using niobium to form a liquid. The resulting
bonds were examined, and compared to NiTi;NiTi bonds in terms of the presence of
second phases and wetability of the materials. Through this investigation, two materials
were commonly used: 0.180 mm thick pseudoelastic NiTi strip material and 1.588 mm

thick pseudoelastic plate nickel titanium.

When bonding with some materials, like alumina, the eutectic liquid created by Nb and
NiTi melting above 1170°C wetted the surface of the dissimilar material without
substantially melting it. With other materials, secondary phases precipitated quickly due
to melting of the dissimilar metal. In the latter case, creative solutions are needed to bond
the dissimilar metal to MT i while keeping the eutectic liquid from reaching the dissimilar
metal. The solution is a multilayer ﬁlm composed of alternating layers of NiTi and Nb,

made as described in Appendix A5.

107

5.1 Alumina

5.1.1 Introduction

Alumina (A1203) has frequently been used for ﬁxturing experiments, and has been
accidently bonded to NiTi several times. NiTi and A1203 can be bonded with either Nb or
a multilayer foil placed between them. The bond is formed by only NiTi and Nb melting;
not A1203. One accidental bonding was during an attempt to make a butt joint in
pseudoelastic strip NiTi material, where A1203 was used for ﬁxturing. The butt joint was
successful, as shown in , but A1203 was also bonded with the specimen as shown in

Figure 56.

5.1.2 Materials and Methods

The sample was made in the MILA ULVAC furnace using a dwell time of 1 minute and a
set temperature of 1185°C, following the procedure described in Appendix A.2. Two
pieces of 0. 1 80 mm thick NiTi strip material were placed adjacent to one another with
one 35 um multilayer foil and one 10 um NiTi thin ﬁlm tacked to each side of the butt
joint. The sample was loaded into a MILA ULVAC furnace with a small piece of A1203
placed underneath as a ﬁxture. Following the brazing process, the A1203 was joined with
the NiTi. The sample was mounted and metallurgically polished to reveal the butt joint

and the bond between A120; and NiTi.

108

5.1.3 Results and Discussion

The sample was observed with a Hitachi S-2500C SEM using an accelerating voltage of
10 keV. EDS information from the bond between NiTi and A1203 was taken using an
Oxford Instruments X-ray detector that was equipped with a beryllium window capable

of detecting elements as light as boron.

In Figure 56, the dark line between the eutectic microstructure and the A1203 in the bond,
at the bottom of the image, is a ledge caused by differences in hardness during polishing.
X-ray spectrums, shown in Figure 57, were taken ﬁom the spots shown in Figure 56.

EDS semi-quantitative analyses of these X-ray spectrums are listed in .

The TizNi in the bond between NiTi and A1203 is similar to TizNi found in the butt joint
between the NiTi pseudoelastic materials shown in . The TizNi in the NiTizA1203 bond is
most likely not related to the alumina; therefore it is probably stabilized by oxygen

content in the multilayer ﬁlm.

Mechanical tests of NiTi bonds with A1203 have not yet been designed, in part due to the
inherent low tensile and shear strength of alumina. Based on the microstructure however,
the bond should be strong. This has been conﬁrmed by fracture surfaces produced by
manually breaking bonds between NiTi and A1203, as shown in Figure 58. While some
of the A1203 remains bonded to the NiTi, fractured occurred with little strength. This
tends to imply that the alumina is somehow weakened by the process, presumably by a

form of microcracking.

109

 

 

Alumina

Figure 56: Metallurgical bond between pseudoelastic NiTi ribbon (top) and alumina (bottom) using
multilayer 111306A. A eutectic microstructure of NiTi and Nb has formed between the A1203 and NiTi,
with small precipitates of TizNi.

 

31:43.4:- ‘ ’

 

110

 

Figure 57: X-ray scans corresponding to points shown in Figure 56.

111

 

Figure 58: NITl bonded to alumina (center), formed by a 0.504 mm thldr Nb wire. Etching from the Nb wire can be seen

on the left side of the Image. Sample broke through alumina, as evident by the alumlna fracture surface on the right side
of the Image. Scale divisions on the right are millimeters.

5.2 Molybdenum

5.2.1 Introduction

Molybdenum (M0) is a common refractory metal used for ﬁxturing materials in vacuum
furnaces. On several occasions, NiTi-Nb eutectic liquid has encountered Mo inside a
fumace and bonds have accidentally been created upon cooling. This MoszzNiTi bond
is difﬁcult to break, and as a result, it could be an important bond in joining NiTi to
ceramics, that might require a diffusion barrier to prevent formation of undesired

intermetallics.

112

5.2.2 Materials and Methods

To examine this bond, a diffusion couple was made between NiTi and Mo with a 54 pm
thick Nb foil as the liquid source. For this diffusion couple, a dwell time of two minutes
at a temperature of 1185°C was used in a MILA ULVAC vacuum furnace at a pressure
less than 0.2 mTorr. The procedure followed is listed in Appendix A.2. The sample was
sectioned, metallurgically mounted and polished to examine the microstructure and

hardness of the bond.

5.2.3 Results and Discussion

The sample was observed using secondary electrons with a Hitachi S-2500C SEM. The
bond between NiTi (bottom) and Mo (top) can be seen in Figure 59, with a linescan
overlaid upon the image. From the linescan, the two-phase mixture appears to be made
of NiTi and Mo phases. Niobium is almost undetectable in the linescan and occupies the
same location as Mo. This is expected because Nb and Mo have similar atomic radii and
both have BCC crystal structures at room temperature. (William D. Callister 2003)
Additionally, Mo is miscible in Nb according to the binary phase diagram, so it is
unlikely to have any non-terminal phases if bonded to nickel titanium using a niobium
braze. (Massalski 1986) The M0 appears to change the overall morphology of the
solidiﬁed eutectic, as shown in Figure 60. The solidiﬁed structure appears to have higher

phase ﬁactions of proeutectic NiTi.

Vicker’s microhardness testing was used to compare the hardness of the bond with the

base metals using a load of 50 grams. As shown in Figure 61, the hardness of the bond is

113

higher than either base metal, and no cracking or chipping was observed following the

indentation. Based on these observations, this robust metallurgical bond with NiTi

warrants further investigation.

 

Figure 59: Specimen 031607, showing the bond between molybdenum (top) and NiTi (bottom). A mixed
two-phase rnicrostructure between NiTi and Mo can be seen, which is a result of liquid formed by Nb
lowering the melting point of the NiTi. Bright phases have high concentrations of Mo and almost
undetectable levels of Nb. The linescan across the bond begins on the left of the graph, with high counts
moving to the right of the graph.

 

114

 

Figure 60: Specimen 031607. Eutectic liquid is wetting the NiTi and forming a new morphology of two-
phase rnicrostructure. The Nb-rich phase contains most of the Mo, and is the bright phase in the two-phase
region. The dark line between the NiTi and two-phase region is a polishing effect.

115

450

 

400 — x
350 — x X X
x
x x X x
300 - X
no
G

 

 

 

150 -
100 -
50 -
O l l l l I
O 2 4 6 8 10
Measurement

 

<> NiTi D Mo X Inside bond

Figure 61: Microhardness of sample 032107B, a bond between NiTi and Mo. A 50 gram load was used for
the measurements. The hardness was the highest inside the bond. None of the indents chipped or showed
any signs of microcracking.

 

 

 

116

5.3 Cobalt Chrome

5.3.1 Introduction

Cobalt chrome (CoCrMo) is an important material for biomedical implants, as well as
tribological applications. Several experiments were conducted to ﬁnd the optimum
method of brazing NiTi to CoCrMo. In a vacuum chamber at elevated temperature, the
cobalt in CoCrMo has a signiﬁcant vapor pressure; often times coating the ﬁxtures in a
thin ﬁlm of Co. This proved to be a challenge for bonding NiTi to CoCrMo; thin ﬁlm
coatings put on CoCrMo had poor adhesion. In addition to poor adhesion of thin ﬁlms at
ambient temperatures, thin ﬁlms depositions were attempted at 400°C, which completely

failed to adhere to the CoCrMo, most likely due to outgassing of the CoCrMo.

5.3.2 Materials and Methods

A bond between porous NiTi and CoCrMo was made to examine the resultant
microstructure. This sample was made before NiTi-Nb multilayers, and was an initial
attempt in preventing other elemental species ﬁom melting into the eutectic liquid.
CoCrMo samples were provided by Nitinol Devices & Components, and were polished to
colloidal silica before deposition. Sputtering conditions are summarized in Table 14. 20
um of NiTi were initially deposited onto the CoCrMo. Ten pm of niobium were then

deposited on the CoCrMo and porous NiTi.

Samples were sectioned with a diamond saw, cleaned and placed in a Centorr vacuum

fumace. A section of porous NiTi was set directly above a section of the CoCrMo, with

117

the two niobium coated faces toward one another. The furnace was degassed for ﬁfteen
hours at 150°C to reach a base pressure of 5 uTorr. The heating proﬁle is shown in

Figure 62, and follows the procedure given in Appendix A. l.

The bond did not fully wet both materials. Following brazing, the sample was fractured
in 3-point bend to observe the fracture surface without collecting data on the stresses

involved.

Table 14: Sputtering conditions for sample 062306D

 

 

 

 

 

 

 

 

 

 

Base Pressure < 3 uTorr
Power 200 Watts
Working Pressure <4.5 mTorr

Nb Sputtering Rate 7.5 urn/hr

NiTi Sputtering Rate 4.4 rim/hr
Working Distance 50 mm
Deposition Temperature Ambient, ~18°C

 

118

Heating Proﬁle

1200
l

1000 ~

800 ~

600~

Temperature,

400«

200-

 

 

0 T l l l I
0 5 10 15 20 25

Time, Minutes

Figure 62: Heating proﬁle of furnace. One minute dwell time at 1185°C. Sample 062806A, created ﬁ'om
062306D and 062306E.

119

5.3.3 Results and Discussion

As shown in Figure 63, the interface between NiTi and CoCrMo was only partially
wetted by the liquid. It appears that most of the liquid was pulled into capillary space in
the porous NiTi rather than spreading across the bond interface. Additional multilayers

of NiTi and Nb might solve this problem by forming more liquid.

The ﬁacture surfaces were examined using a Carnscan SEM. Fracture surface features
ﬁ'om the CoCrMo side can be seen in Figure 63 and Figure 65, while a portion of the
NiTi half of the fracture surface can be seen in Figure 64. As shown in Figure 63, most
of the CoCrMo material was not bonded with the NiTi. Figure 65, however, was taken in
the area that bonded with the NiTi and shows that the fracture occurred in the NiTi, not at

the bond.

The hardness of the CoCrMo was tested before and after brazing. The materials hardness
raised by the brazing process, but the standard deviation in hardness decreased. Hardness

data is summarized in Figure 66.

The difﬁculty of bonding CoCrMo with other materials is the high vapor pressure of Co
inside a vacuum chamber. If an adherent thin ﬁlm of niobium were able to coat the
CoCrMo and remain adherent at 1170°C or higher, multilayers could provide liquid for

metallurgical bonding with CoCrMo.

120

 

Figure 63: Overview of CoCrMo fracture surface. Upper left comer was not brazed most likely due to
capillary forces drawing the eutectic liquid away. Failure occurred inside the porous NiTi. The rough
appearance of the CoCrMo (top left of CoCrMo) could be a result of signiﬁcant Co sublimation during the
brazing process.

 

Figure 64: Composite image of pore and fracture surface. Note the different texture of the pore versus the
fracture surface, implying fracture did not propagate through a pore. NiTi half of bond.

121

 

Figure 65: Fracture surface in area that was brazed between NiTi pores and left material on CoCrMo side.

122

500

 

mm
4w~
Mm
420- C] D DD El DD CI

. U
4m DD

Vickers Hardness

mm A

%m AAAA

340 -

320 -

 

 

300

 

Indent #

 

A As received CoCrMo C1 Brazed CoCrMo

 

 

 

Figure 66: Hardness data from CoCrMo material before and aﬁer brazing. The overall hardness increased
by 13.7%, while the standard deviation in hardness decreased by 9.8 VHN. Many irregularly shaped
indents were observed in both as-received and post brazed conditions.

123

5.4 FezAlV

5.4.1 Introduction

Thermoelectrics are currently researched materials designed to reclaim wasted heat in
engines and other devices. These materials, however, usually have low toughness and are
susceptible to failure in environments that have large vibrations and applied loads.
Pseudoelastic materials, such as nickel titanium, could be used as a support structure

capable of dampening these forces, thereby protecting the thermoelectrics.

5.4.2 Materials and Methods

Bonds between FezAlV were made using a MILA ULVAC vacuum furnace following the
procedure in Appendix A.2. Multilayer ﬁlms of NiTi and Nb, taken ﬁom sample
111306A, served as the liquid source. The samples were brazed for 30 seconds at a
temperature of 1185°C, followed by a rapid ﬁrmace quench. The bonds were

subsequently sectioned and metallurgically polished and etched for SEM microscopy.

5.4.3 Results and Discussion

The resultant microstructure can be seen in Figure 67, Figure 68, and Figure 69. The
microstructure inside the bond consists of two phases, both intermetallics. These phases
were characterized using EDS, the results of which are shown in Table 15. Both phases
appear to be rich in nickel and titanium, but without crystallographic data it is difﬁcult to

further identify the phases.

124

As seen in Figure 68, the NiTi-Nb eutectic liquid readily wets the thermoelectric, which
is the material on top. It is interesting to note is that the low wetting angle observed in

NiTi;NiTi bonds was not observed in this sample.

The thermoelectric material appears microcracked in Figure 67 and Figure 68. One of
the cracks contains solidiﬁcation product ﬁom the bonding process, suggesting the crack
was created before the brazing process. Other cracks do not contain solidiﬁcation
products, which suggests they were formed by the rapid heating and cooling of the braze.
It should be noted that the rapid heating and cooling of the sample in this experiment was

done for convenience, and not as a preplanned heat treatment.

X-ray mapping shows the variation in composition between the two solidiﬁcation
products in Figure 70. Nb and Ti have the most notable differences in composition. The
phase that appears light in Figure 69 appears to have signiﬁcantly more Nb than the dark
phase; conversely, the dark phase appears to have signiﬁcantly more titanium than the

light phase.

The hardness of the NiTi, Mo and the bond were all examined using Vickers
nricrohardness indentation; the results are shown in Figure 72. The hardness of the bond
was higher than the hardness of the parent materials, but the bond did not show any signs
of nricrocracking or chipping. The F ezAlV microcracked from the thermal processing,

while the solidiﬁcation product between the FezAlV and NiTi did not crack.

125

The results of this investigation demonstrate that the thermoelectric is more susceptible to
damage from thermal shock than the bond or NiTi. Further identiﬁcation of the phases
present in the solidiﬁed rnicrostructure is needed, but the bond appears stronger than the
thermoelectric material. As identiﬁed in Figure 71, thermal shock of the material can
result in fracture of the thermoelectric in the vicinity of the bond, which could be a result
of a mismatch in coefﬁcients of thermal expansion. In future studies, a similar sample

should be created and allowed to heat and cool at a more moderate rate.

126

 

Figure 67: Global view of sample 031607A. The apparent wetting angle of the liquid by the NiTi has been
changed compared with NiTi — NiTi bonds. Fe2AlV is adequately wetted by the liquid, and has ﬁlled
cracks in the material.

 

Figure 68: Low magniﬁcation SEI of 031607A. The solidiﬁcation products of the bond appear different
than the standard NiTi-Nb solidiﬁcation products. The wetting angle of the FezAlV is similar to other
materials, however, the NiTi wetting angle is greater than 90°, suggesting a non-wetting condition. FqulV
appears susceptible to microcracking, as suggested by a crack inﬁltrated with solidiﬁcation product. This
crack must have been present during the brazing process.

127

 

Figure 69: Sample 031607A. The solidiﬁcation product does not seem to be a cellular or dendritic
structure. A third possible phase could be at the top of the image, appearing to have solidiﬁed from the
FquIV.

128

Secondary Electrons V

   

Nb-L a Ti-K u Ni-K a

   

Al-K a V-L a. F e-K a

Figure 70: SEI and corresponding X-ray maps of NiTi-FezAlV bond. Niobium content appears to be the
largest chemical difference between the two solidiﬁcation products. Titanium, aluminum and vanadium
also appear diﬂ'erent between the phases, while iron and nickel content are similar.

129

l
l
r.
l
i

. .
.4....‘_. ~
. \ \ .‘~ 4 § .

- V .. ‘ v "
-MW—e‘ :_ “3:51. “in-.. -\-—. <
J. f . ._ , . _. .- g

 

Figure 71: SE1 of microstructure. X—ray information was collected from points A-D and were analyzed in
Table 15.

 

130

032107-A Microhardness

 

 

 

 

600
x
5004 X
X x
x
x

.00. I I I I I I l I I §

x x
5
E300- X
5 o
g 9 O O O Q 9 Q Q 0
$200-
>
100-
0

Indent#

 

O NiTi I Fe2AlV X Inside bond

 

 

 

Figure 72: Hardness data for specimen 032107-A. The NiTi phase has a lower hardness than the Fe2AlV.
The hardness of the bond was higher than the hardness of the parent phases; however no cracking was
visible inside the bond.

131

5.5 Boron Carbide

5.5.1 Introduction

For specimen ﬁxturing in research and industrial production, it is important to ﬁnd
materials that do not react with the eutectic liquid. Boron carbide, BC, was bonded to
NiTi to determine if materials with strong covalent bonds would react with the eutectic

liquid.

5.5.2 Materials and Methods

A small, fully dense piece of boron carbide was bonded to NiTi inside the Centorr
vacuum furnace using a 34 pm thick multilayer foil, specimen 111306A. The heating
schedule followed the procedure listed in Appendix A.1. The dwell time of the sample
was two minutes at a temperature of 1185°C. Upon opening of the chamber to retrieve
the sample, the B4C had broken off ﬁ'om the NiTi bond. It is likely the sample fractured
due to thermal stresses, which occur because of different thermal expansion coefﬁcients.
The NiTi half of the sample was mounted in cross section and polished and examined

with a Hitachi S-2500C SEM.

5.5.3 Results and Discussion

The nickel titanium half of the fracture surface is shown in Figure 73. The boron carbide
was partially bonded to the NiTi, and the fracture occurred through the bond. A mixture

of the two parent materials appears to have adhered to the base NiTi metal around the

132

ﬁacture surface, which can be seen in cross section in Figure 74. The bond between the
NiTi and the solidiﬁcation product shows a low angle wetting with the NiTi, but a high
wetting angle with the solidiﬁcation product. Furthermore, the microstructure of the
solidiﬁcation product, shown in Figure 75, shows a complex morphology, from which no
meaningful X-ray data could be obtained due to interaction volume and difﬁculties
quantifying boron. From the microstructure, it is speculated that the solidiﬁcation
product is most likely a complex non-terminal solidiﬁcation product that has nearly no
toughness or strength. This bond is the least adherent and interacts the least of all the

materials bonded to NiTi thus far.

133

 

 

Fractured bond

 

 

 

Figure 73: Optical image of NiTi side of NiTi-B4C fracture surface. Units on the scale to the left is
millimeters. (a) Area remained unbonded during brazing. (b) B4C remains bonded with the NiTi. (c)
Solidiﬁcation product of the liquid.

134

 

 

Figure 74: Cross-section of fracture surface. While the base metal appears to be wetted by the eutectic
liquid with a low contact angle, the brittle bond with the boron carbide fractured from thermal contraction.

135

 

Figure 75: High magniﬁcation SE1 of solidiﬁcation product formed by NiTi-Nb liquid reacting with B4C.

136

5.6 Magnesium Oxide

5.6.1 Introduction

The bond between NiTi and magnesimn oxide, MgO, was examined in the attempt to ﬁnd
a material that does not bond to the eutectic liquid. Similar to boron carbide, magnesium
oxide has strong covalent bonds, and a high melting point. Magnesium oxide was chosen
based on magnesium having a higher afﬁnity for oxygen than titanium, which was
assumed to be a limiting factor in producing new solidiﬁcation products with oxygen.

(Gaskell 1995)

5.6.2 Materials and Methods

Reagent A.C.S. magnesium oxide powders, from Matheson Coleman & Bell, were spread
across a multilayer foil, specimen 111306A, that was on top of a piece of 3.127 mm thick
pseudoelastic NiTi. The sample was carefully placed in the MILA ULVAC furnace,
which was gently evacuated to avoid disturbing the powder. The sample was brazed for
two minutes at 1185°C in the furnace, followed by a quench according to the procedure
listed in Appendix A.2. Upon opening the furnace, much of the powder remained
unbonded to the multilayer. These loose particles subsequently blew away during sample
removal. Some particles did attach to the NiTi, as evident by the rough texture of the free

surface. The sample was mounted in cross section and polished.

137

5.6.3 Results and Discussion

The microstructure of the bond, shown in Figure 76, appears similar to that of the
solidiﬁcation products in a NiTi — NiTi bond. A eutectic structure of NiTi and Nb is still
observed in the material, as well as TizNi, ﬁ'om the liquid forming on a ﬁee surface and
oxidizing. Magnesium oxide agglomerates have been inﬁltrated by the eutectic liquid.
The solidiﬁcation products inside the magnesium oxide agglomerates are too small to
quantify using the SEM, and thus are assumed MgO, Nb and NiTi based on phase
contrast. If MgO powders have been inﬁltrated by the eutectic liquid, it is an example of
a metal matrix composite. Based on the microstructure in Figure 76, this bond should be
recreated using fully dense magnesium oxide to better understand the solidiﬁcation

products.

138

 

 

Figure 76: Mg rich region ﬁlled with ﬁne Nb-based and Mg—based precipitates. The phases were too small
to analyze using EDS. Based on EDS data, only the Mg-rich region appears to contain Mg.

139

5.7 Tantalum

5.7.1 Introduction

Cheng, Cai and Zheng used tantalum as a surface modiﬁcation for NiTi implants to
improve radiopacity for viewing of medical X-rays. (Cheng, Cai et al. 2006) Coating the
NiTi surface with approximately 3 pm of tantalum resulted in higher implant visibility
within the body when imaging using X-rays. Additionally, this reduced the nickel release

rate of the NiTi by over an order of magnitude.

5.7.2 Materials

Ke Bin Low made a butt joint between pseudoelastic NiTi and tantalum using the Centorr
vacuum annealing furnace at Michigan State University, following the procedure in
Appendix A.1 (Low 2007). A foil of 127 um Nb was placed between the NiTi and Ta,
which subsequently melted and bonded the materials when brazed at 1185°C for two
minutes. After standard furnace cooling, the sample was sectioned using a diamond saw

for mechanical testing.

5.7.3 Results and Discussion

Results of the tensile test, conducted at a strain rate of 0.1 mm/sec, can be seen in Figure
77 (Low 2007). Necking occurred in the Ta, showing that the butt joint between the
tantalum and nickel titanium is stronger than Ta alone. The maximum engineering stress
experienced by the sample was roughly 210 MPa, which was the ultimate tensile strength

of the Ta. This implies that if Ta coatings on NiTi had poor adhesion, the Ta could be

140

brazed to the nickel titanium using an intermediate layer of Nb. An intermediate layer on
Nb would melt when heated above 1170°C and improve the interface between the thin

ﬁlms and the NiTi by removing any porosity and oxides present.

A portion of the bond between Ta and NiTi was mounted and polished for inspection.
The sample was imaged using secondary electrons on a Hitachi S-2500C SEM. As
shown in Figure 78, the resultant braze rnicrostructure is a simple eutectic with
proeutectic NiTi and Ta phases. As suggested by the completely miscible binary phase
diagram between Nb and Ta, the image morphologically demonstrates that niobirnn and
tantalum have phase separated from the NiTi, producing a eutectic rnicrostructure.
(Massalski 1986) EDS analyses of the phases, Table 16, show that the Nb and Ta have
phase separated from the NiTi. Due to the X-ray interaction volume, EDS could not

accurately quantify the eutectic phases.

141

‘IAA
5w

 

220'

Sims (MPa)
'9'

 

 

 

 

too-
so-
so-
40-
20-
o l l I l f
o 5 to 15 20 25 30
Strain (96)

Figure 77: Stress-strain plot of the sample shown in the photo insert. Sample necked inside the tantalum,
demonstrating the joint is stronger than the tantalum. (Low 2007)

142

 

 

Figure 78: Proeutectic NiTi eutectic Nb/T a and NiTi ﬁll the bond between NiTi and Ta. This SE1 shows
the simple eutectic between NiTi and Ta that results from brazing with an interlayer of 127 pm thick Nb
foil. X-ray information collected at points A, B, C and D shown in Table 15.

 

143

5.8 Ti-6Al-4V

5.8.1 Introduction

Ti-6Al-4V is an important biocompatible alloy used extensively due to its low density,
high strength and high corrosion resistance. Complex devices combining these features
with the pseudoelastic or shape memory effects of NiTi could usher in a new generation
of implants. From the Ni-Ti phase diagram, nickel titanium and titanium contact melt
above 942°C and can produce equilibrium mixtures of titanium, TizNi and NiTi upon

cooling.

5.8.2 Materials and Methods

To prevent the formation of TizNi, coupons of Ti-6A1-4V were coated with 30 um of Nb
by DC magnetron sputtering at a temperature of 400°C. (Liitjering and Williams 2007)
Heating the titanium to 400°C improves adhesion by allowing the titanium alloy to

reduce the surface oxide, thus allowing a metallurgical bond with the Nb ﬁlm. The

titanium alloy was sectioned with a diamond saw after the ﬁlm deposition.

Samples were created using four squares of 34 um thick multilayer foils to bond a pieces
of pseudoelastic NiTi strip material to a piece of 1.588 mm thick Ti-6A1-4V plate
material. The pieces of multilayer, specimen 111306A, were tacked to the NiTi, and
placed against the Nb coated face of the titanitun alloy. The assemblies were heated in
the MILA ULVAC vacuum furnace to 1185°C in one minute and were held at 1185°C

for 10 seconds before furnace quenching. Sample 042407A was held for 1 minute at

144

1185°C. During the dwell, the multilayer melted and the Nb coating on the titanium
came in direct contact with the liquid. Upon cooling, the liquid solidiﬁed, forming a
metallurgical bond between the NiTi and Nb, while the Nb and Ti-6Al-4V metallurgical
bond remained. The samples were later sectioned and metallurgically mounted and

polished.

5.8.3 Results and Discussion

The process formed a metallurgical bond between NiTi and Nb, and a sputtered
metallurgical bond between Nb and Ti-6Al-4V. The bond was polished and etched; it
was imaged in a light microscope shown in Figure 79. The sample was coated with only
10 um of Nb, which appears intact after the brazing process. This means that the liquid
did not signiﬁcantly melt the Nb coating on the titanium. A lower magniﬁcation image
of the joint between NiTi and an Nb coated section of Ti-6Al-4V can be seen in Figure
80, which shows the low wetting angle of the liquid on the NiTi and Nb coating.
Furthermore, the low magniﬁcation image shows that capillary action drew the NiTi
close to the Nb coating near the bottom of the image without signiﬁcantly affecting the
coating itself. Figure 81, shows a sample that was brazed too long, allowing Ni atoms to
penetrate the Nb coating. When this happened, the Ti-6Al-4V and NiTi formed TizNi in

the bond, which resulted in cracking and uncontrolled melting of the sample.

DC magnetron sputtered thin ﬁlms can have porosity and pinholes from irregularities
during sputtering. It was important to examine the bond between the Nb and Ti-6Al-4V

to determine if any nickel was able to penetrate the Nb layer through holes. In a higher

145

magniﬁcation SE1, Figure 82, a transition ﬁom the BCC Nb to BCC and HCP Ti-6Al-4V
can be seen, however no TizNi appears to be present. This transition is further
characterized by EDS along the Nb to Ti-6Al-4V bond line, which did not detect nickel,
as shown in Figure 83 and Table 17. In Figure 84, the X—ray linescan performed across
the image shows a simple transition from NiTi to eutectic with NiTi and Nb, to the Nb
coating and then the Ti-6Al-4V. No nickel is detected beyond the Nb coating. A small
two phase region between the Nb and Ti-6Al-4V, shown again by the linescan,
demonstrates the metallurgical nature of the bond between the Nb coating and Ti-6A1-
4V. The bond between Nb and Ti in also appears to be well mixed based on the Z-
contrast in the titanium side of the interface. The bright features in the interface between
Nb and Ti are differences in secondary electron emission at a ledge from different etching
rates between Ti and Nb. The dark interface shown in the BEI of Figure 82 demonstrates
that beta-stabilizing elements in the titanium, such as vanadium, diffuse faster into the
niobium than the niobium diffuses into the titanium. This results in a higher volume

fraction of the a-titanium phase at the bond than in the bulk of the sample.

In Figure 85, the solidiﬁed microstructure of the braze and Nb coating can be observed.
Proeutectic Nb, identiﬁed by Z-contrast in the larnellar structure, appears to have
solidiﬁed from the Nb coating. This demonstrates that melting of the Nb coating can be
controlled by properly designing the thickness of the ﬁlm for the intended thermal

exposure above 1150°C.

146

By using rapid cooling, samples that do not use multilayers produce similar solidiﬁcation
microstructures as samples with multilayers. For comparison, a specimen was made
using the same brazing technique mentioned earlier, but did not contain multilayer foils.
The microstructure of this bond can be seen in Figure 86. The braze interface appears
identical near the niobium, but this is a result of the rapid cooling. With conventional
fumace heating and cooling rates, NiTi and Nb would continue to melt until the Nb
interlayer was exhausted. The multilayer establishes a large diffusion distance between
the pure NiTi and pure Nb layers, over which diffusion is slow. This prevents severe

dissolution of the Nb coating, which protects the Ti-6Al-4V from nickel atoms.

147

'l‘i—(rAl—JV

 

Figure 79: LOM of Nb coating, the bright phase on the right, on Ti-6Al-4V. The Nb film is approximately
10 pm thick on sample 0530078, which was taken from the thin ﬁlm deposition 0515073.

148

l\l) coating '

Eutectic .

9
33
a.

a;
.5-

' ' ‘3‘..." f

u

. a. .. . g».
‘s’aé’em bdcﬁiﬂ 53.5; 154%

 

..£ 1‘

Figure 80: LOM of (a) Ti-6Al-4V base alloy with (c) 30 um Nb thin ﬁlm deposited on it to prevent extra
titanium from reaching (b) the eutectic between niobium and nickel titanium. Niobium successfully
prevented Ni diffusion into Ti-6Al—4V. Four pieces of multilayer M032007A were placed between the
niobium coating and the NiTi before brazing, serving as a liquid source.

149

Ti-(1.r\l--l\'

 

Figure 81: LOM of 060207C, (a) NiTi bonded to (b) Ti-6Al-4V with a dwell time of 1.5 minutes at a
temperature of 1185°C. Niobium did not prevent the liquid from melting the Ti-6Al-4V, which enriched
the liquid in titanium. producing (c) TizNi upon cooling. The TizNi phase fractured during hot mounting.

Nb coating

Ti-(rAl-«lV

 

Figure 82: Sample 040607A. This BEI shows the transition in microstucture from BCC Nb (left, bright
phase) to the tweed microstructure of Ti-6Al-4V. A two-phase region is seen between the thin ﬁlm and
base metal; differences in etching rate have produced a ledge that causes darkness at the interface.

150

 

 

Figure 83: SE1 of bond between NiTi (left) and Ti-6Al-4V (right). The coating of Nb appears to be well
adhered to the Ti-6Al-4V as evident by Z-contrast in the B region. X-ray information from spot A, B and
C were analyzed using an Oxford Instruments EDS software package.

Table 17: EDS analysis, in atomic percent, of spots indicated in .Figure 83.

 

151

 

Ti-(rAl-4V'

 

Figure 84: 042407A. NiTi to Ti-6A1—4V bonded without multilayers, only an Nb coating. This process was
completed using a thick coating of Nb on Ti along with fast brazing times, especially above 1150°C.

152

 

Figure 85: 040607A NiTi - Ti64: The microstucture of the solidiﬁed eutectic contains Nb and NiTi phases,
along with a small phase ﬁaction of TizNi, which is expected due to oxygen contamination on the
multilayer foil.

153

N1) coating

'I‘i-6Al-4V

 

Figure 86: 040607A NiTi - Ti64. The Nb coating serves as a barrier for nickel diffusion from NiTi to Ti-
6A1-4V. The soft Nb coating has proeutectic Nb solidiﬁed ﬁ'om the bond with NiTi using multilayer foils.
The sputtered coating mixes with the Ti-6A1-4V in a two—phase region n+0 region.

5.9 Tungsten

5.9.1 Introduction

The niobium brazing process also wets tungsten, a commonly used refractory metal.
Tungsten is attractive because it could possibly act as a diffusion barrier to prevent other

materials from being attacked by the reactive liquid.

154

5.9.2 Materials and Methods

To explore the metallurgical interaction between NiTi and tungsten, a small piece of
0.180 mm thick NiTi strip material was bonded to 254 um diameter tungsten wire using a
3 mm square of 34 pm thick multilayer foil taken from sample 111306A. The sample
was created using the MILA brazing furnace, with a dwell time of 2 minutes at a

temperature of 1185°C. See Appendix A.2 for more information on the brazing process.

The sample was cross sectioned, metallurgically polished and then etched following the

procedure described in Appendix A.7.

5.9.3 Results and Discussion

The sample was photographed using a Nikon D70 digital camera attached to a Nikon
reﬂection microscope. Figure 88 is a collage of light optical micrographs showing the
microstructure of the bond. Some microvoids were present inside the bond, which could
ultimately lower the strength of the bond. These voids were not interconnected however,
so the hermetic qualities of the bond are probably not compromised. This image also
shows the liquid has a low wetting angle with the 250 um diameter wire. The wire is

very close to its initial diameter, suggesting that very little melted.

The sample was imaged in a F El Quanta operating at 20 keV using secondary electrons;
EDS data was collected from the various microconstituents inside the bond. The results
of these can be seen in Figure 89, Table 18, Figure 90 and Table 19. The rnicrostructure

of the bond has a signiﬁcant phase ﬁaction of a dark, faceted precipitate phase, which

155

EDS data suggests could be an interrnetallic based on TizNi. The phase fraction of TizNi
could be lowered by etching the trmgsten before bonding it to NiTi, thus lowering oxygen

contamination.

Niobirun and tungsten are known to be miscible from the binary phase diagram.
(Massalski 1986) In the regions very close to the tungsten wire, the tungsten has changed
the morphology of the Nb rich phase in the eutectic into a more pronounced rod shape.
The Nb phase near the tungsten wire has higher concentrations of tungsten than the NiT i
phase nearby. This suggests that most of the melted tungsten has solidiﬁed in the Nb

phase.

The NiTi phase appears unchanged; EDS data conﬁrms very low concentrations of
solutionized tungsten. The concentration of tungsten in NiTi is consistent near the
unmelted NiTi and near the tungsten wire, which suggests sufﬁcient mixing of the
eutectic liquid. The titanium concentration inside the proeutectic and eutectic NiTi is
lower than the base metal by approximately 6%, which is roughly equal to the
concentration of Nb in the phase; again suggesting that Nb is substituting for Ti. The
addition of a small concentration of tungsten into the material does not change this

substitution.

156

la)

Ni'l‘i ._
HV Spot WD Det Mag , ' 100 Oilm
20.0 W 4.0 10.9 mm ETD613x

~

 

Figure 87: Microstructure of NiTi-W bond. (3) Unmelted NiTi base metal (b) Proeutectic NiTi phase
coherent with unmelted NiTi (c) Proeutectic NiTi lamellae (d) Voids in the microstructure (e) Tungsten
wire

157

 

Figure 88: LOM of sample 053007A. (a) Pseudoelastic NiTi bonded with (b) 254 um diameter tungsten
wire and mounted in (c) Konductomet ® phenolic mount material. A small void is present in the bond
between the NiTi and tungsten, as well as a eutectic microstructure and a proeutectic phase consisting of
Nb and tungsten. Also, note the low wetting angle between the wire and eutectic liquid.

158

(I1)

HV Spot WD Det Mag , , ~ 1.1 ~
20.0 kV 4.0 10.9 mm ETD 1 138x Microstructure on NiTi side

 

Figure 89: NiTi side of 053007A. (a) Unmelted NiTi, (b) proeutectic coherent NiTi (c) eutectic solidiﬁed
Nb phase. ((1) Faceted precipitate. (6) Bright eutectic phase.

 

159

I

, WD Det Mag .
20.0 kV 40 10.9 mm ETD 2500x Microstructure around tungsten wire

 

Figure 90: Microstructure of 053007A near tungsten wire. (a) ‘Matrix’ phase that is dark grey, (b) tungsten
wire approximately 15 um inside the wire, (c) dark precipitate phase, which could be based on TizNi, and
(d) bright eutectic phase.

 

160

5.10 304 Stainless Steel

5.10.1 Introduction

Stainless steel is an important engineering material due to high corrosion resistance in
many environments. The corrosion resistance is partly due to the protective surface oxide
formed by alloying additions of chromium and nickel. This nickel can cause problems
when bonding NiTi to stainless steel. Methods of joining, such as pulsed laser welding,

have met with limited success. (Wang 1997) (Schlossmacher, Hass et al. 1997)

5.10.2 Materials and Methods

An attempt was made to bond NiTi to 304 stainless steel received from ESPI Metals
using 127 pm thick Nb foil and 1.588 mm thick pseudoelastic NiTi plate material. The
bond was made using a Centorr vacuum annealing furnace with a dwell time of two

minutes and a temperature of 1185°C according to Appendix A.1.

5.10.3 Results and Discussion

The bond was extremely brittle and fractured easily. The sample was cross-sectioned and
observed using a Hitachi S-2500C SEM operating at 10 keV and an Oxford Instruments
X-ray detector. A micrograph of the bond can be seen in Figure 91, which appears to be
comprised of at least three different microstructures. EDS semi-quantitative analyses

were performed on X-ray scans taken at the labeled spots in Figure 92; results are listed

161

in Table 20. Contrary to literature reports of NiTi to stainless steel welds containing

T iFe, no precipitates contain large amounts of iron. (Wang 1997)

The signiﬁcant microstructural difference between the bond shown in Figure 92 and a
NiTi to NiTi bond is the formation of a third phase, labeled B in the ﬁgure. EDS semi-
quantitative analysis suggests the phase is rich in nickel and titanium, with small amounts
of Nb, Fe and O. The phase could be a result of nickel provided by the stainless steel
during the brazing process. If the phase were based on the Ni-Ti-Nb ternary system, it
could a non-terminal phase rich in nickel and niobium. Literature reports these non-
terminal phases as the XA and X5 phase, which are not currently reported on. It is likely
that this phase could be avoided if the stainless steel were coated with a material like
niobium, tantalum or tungsten, which have been shown to not signiﬁcantly affect the

microstructure of the bond.

162

 

Figure 91: Low magniﬁcation SE1 of post-fracture bond between NiTi and stainless steel. Bond was
formed using Nb foil.

163

ﬁx?" 5 i

4" . . B
$0.13"- k“!
, .

i:
\‘M ‘ .311!

 

Figure 92: Three distinct solidiﬁed microstructures in the bond between NiTi and 304 stainless steel.

 

6 Conclusions

The Ni-Ti-Nb ternary is a unique system that combines biocompatibility with high
strength and ductility. Brazing NiTi with Nb produces a eutectic structure upon
solidiﬁcation, and is compatible with low and medium vacuum processing. Complex
structures can be produced using this brazing process: honeycombs, corrugations and
tube arrays. These structures have demonstrated pseudoelastic ampliﬁcation, low

density and high compliance.

Dissimilar formulations of NiTi can also be bonded together using this brazing
technique; producing a eutectic structure upon cooling. In some cases, TizNi forms
upon cooling instead of the eutectic because of oxygen contamination. Bonds
between dissimilar NiTi sections could be used in applications such as complex
biomedical devices that incorporate porous NiTi, wrought NiTi and chemical NiTi.
Other devices, such as MEMS, could be built using lower cost chemical NiTi and thin

ﬁlm NiTi for pumps, motors, actuators and valves.

Thin ﬁlms made on glass slides can be delarninated by carefully peeling them from
glass slides after crystallization. These ﬁhns, despite being amorphous during
deposition, show good shape recovery and could be used for low hysteresis
applications by taking advantage of the R-phase that is formed while cooling.
Transformation temperatures for these ﬁlms are above room temperature, which

implies the ﬁlms are very close to the ideal NizTi ratio for shape memory.

165

Thin ﬁlms with alternating layers of Nb and NiTi can also be produced on glass
slides. These ﬁlms also show shape memory with high transformation temperatures.
These ﬁlms drastically increase the contact surface area between NiTi and Nb for
brazing, which substantially increases the melting rate. Due to oxidation, higher
phase ﬁ'actions of TizNi are found in brazes with multilayers rather than with niobium

foil.

The NiTi — Nb multilayer ﬁlms can also be adjusted to have a speciﬁc ratio of nickel
titanium to niobimn. With this adjustment, the ﬁlm can be altered to either

selectively etch or deposit NiTi to the surfaces of braze joints.

Multilayers of NiTi and Nb thin ﬁlms can also be used to join dissimilar materials
with NiTi. Eutectic NiTi and Nb form in bonds with A1203, tantalum, niobium,
tungsten and magnesium oxide. Combining NiTi multilayers and Nb coatings on

dissimilar materials can eliminate undesired intermetallics from forming in Ti-6Al-

4V bonds with NiTi.

Other materials bonded to NiTi that were investigated produced complex
solidiﬁcation products upon cooling. Bonds with FezAlV, cobalt chrome, boron
carbide and 304 stainless steel produce complex solidiﬁcation products. Bonds with
boron carbide and 304 stainless steel are weak. NiTi bonds with stainless steel are

brittle because of a third, non-terminal phase containing a high concentration of Ni.

166

 

Coating these materials with a thin ﬁlm of niobimn could reduce the phase fractions

of these undesirable intermetallics.

The wetting angle between NiTi and the liquid changes when bonding FezAlV with
NiTi, and further understanding of this change could lead to discovery of a material
that does not bond with the liquid. The thermal expansion mismatch between NiTi

and boron carbide could help discover a material for ﬁxturing NiTi in brazing

furnaces.

DC spot welding has higher tensile strength than AC spot welding, in part due to
shorter processing times and greater control of the waveform. Using multilayers to

spot weld NiTi has proven unnecessary. Properly designed spot welds can be made

without ﬁller metals.

The niobium — nickel - titanium ternary system thus allows for a novel, robust

brazing system capable of joining NiTi to many materials with strong bonds.

167

 

7 Future Work

Now that the NiTi-Nb brazing process has been shown to be robust and applicable to
other materials, future research should be focused on industrial applications.
Additionally, a material must be found that is not wetted by the eutectic liquid to help
create manufactured devices. Bonds between Fe2A1V and NiTi should be reexamined,

focusing on the change in wetting angle between Fe2A1V and NiTi.

While in viva corrosion is not expected, a study should be conducted to determine if the

galvanic potential between NiTi and Nb is large enough to corrode NiTi-Nb bonds.

A fatigue study of brazed NiTi structures should be conducted to determine the fatigue
resistance in the bond. In particular, low strain tests should be conducted on the

honeycombs, to test the long term behavior of these designs.

Cold rolling of thin ﬁlms should be developed to produce working actuators. It should be
possible to make a large scale model of the Johnson valve using a cold rolled thin ﬁlm to

open and close a valve should be possible to make and will serve as a proof of concept.

This device could also be used for a fatigue study of NiTi thin ﬁlms.

High resolution TEM (HRTEM) work should be conducted to determine if the quasi-
binary truly exists. Additionally, HRTEM should be capable of determining if TizNi
forms inside bonds from oxygen stabilization or due to other reasons. Foils should be

produced from arc-melted buttons of material, and the overall chemistry of the buttons

168

 

should be veriﬁed by a method such as plasma ﬂame spectroscopy. Some researchers
have reported terminal phases in the Ni-rich half of the Ni-Ti-Nb ternary system that

would slightly intersect the quasi-binary, and these results should be veriﬁed.

Bonds with monolithic, fully dense sections of MgO, hydroxyapatite and other important
engineering ceramics should be investigated. If bonding NiTi to these materials using Nb
foil proves to form undesirable intermetallics, diffusion barrier coatings on the ceramics

should be investigated.

169

Appendices

A.1 Brazing using Centorr Annealing Furnace

A Centorr vacuum annealing furnace was used for many experiments in this thesis. It
accommodates large samples and uses simple ﬁxtures for positioning samples. The
furnace is heated by helical tungsten ﬁlaments inside the vacuum chamber, which are

controlled manually by a Honeywell Versa Pro UDC 3000 temperature controller.

The ﬁxtures used for many experiments had considerable thermal mass and therefore
limited the actual heating rate of the sample. By trial and error, it was determined that
the sample should be heated to 1200°C in no less than 7 minutes. Heating rates around
100°C per minute were the standard throughout this thesis, following the heating curve

shown in Figure 17, located in Section 2.3.3.

This Centorr vacuum furnace operated with base pressures between 1 uTorr and 10
uTorr, and working pressures between 5 uTorr and 30 uTorr, based on the exact
condition of the heating chamber. Typically, the furnace was heated to a moderate
temperature of approximately 200°C and allowed to reach a base pressure of < 5 uTorr.
Once this base pressure was reached, the sample was heated inside the chamber at a
heating rate of approximately 50°C — 100°C per minute until the hold temperature, or
dwell temperature, was established. The bold temperature was maintained for the desired

time, or dwell time, then the power was shut off to allow the furnace to cool.

170

 

Once the temperature was below 100°C, the chamber was opened and the sample was
retrieved. It should be noted that the samples could not be taken out at a higher
temperature due to the brittle nature of the heating elements. Once the samples were

retrieved, they were quenched in water for handling purposes.

A.2 Brazing using MILA ULVAC rapid thermal heating system :

The MILA ULVAC rapid thermal processing fumace was used for small samples. While i

 

it is not capable of high vacuum, it heats and cools much faster than other, more
conventional furnaces. The ULVAC furnace heats using reﬂective infrared heaters that
are outside the chamber. The ﬁrrnace also features a PID controller, which allows

heating schedules to be very accurate and repeatable.

A CTI Cryogenics pump provided the medium vacuum source for the MILA ULVAC
heater. While running standard tests, the samples were allowed to pump to a soft vacuum
of 0.1 mTorr. When this vacuum was achieved, the sample was heated to the dwell
temperature, typically 1185°C, in 40 seconds. The PID maintained this dwell
temperature within 1 — 2°C for the prescribed dwell time, which typically varied between
5 seconds and 2 minutes. The maximum use temperature of the MILA ULVAC heater is
1200°C, so long brazing times cannot be run on this machine, as it will overheat. The
sample was allowed to furnace cool, which was rapid because of the low thermal mass of
the chamber. Heating elements do not add to the thermal mass because the heating is
infrared ﬁ'om outside the chamber. Samples typically cool to below 800°C within the

ﬁrst 30 seconds.

171

A.3 Polyimide backed thin ﬁlms

Polyimide ﬁhns were included in this thesis because the previous work at Michigan State
University on thin ﬁlms has led to a greater understanding of the intricacies of DC

magnetron sputtering thin ﬁlms.

A.1.1 Introduction

Nitinol thin ﬁlms have high work density compared with piezoelectrics and
thermoelectrics. (Gill, Ho et al. 2002) Microelectro-mechanical systems can be made
from Nitinol thin ﬁlms and produce large amounts of work using very small ﬁlms. The
Johnson valve is capable of pumping one liter per minute of ﬂuid through a small oriﬁce
using NiTi thin ﬁlms. (Johnson, Gupta et al.) The device consisted of several Nitinol
thin ﬁlm ribbons released from single crystal silicon by etching. A silicon poppet was
connected to one of the ligaments, which closed a small oriﬁce below the silicon and that
allowed ﬂuid ﬂow through the device. The entire device was placed into a pressure
vessel and was connected to an electrical source. Pressure ﬁom a liquid deformed the
thin ﬁlm ligaments out of the plane, closing the valve. When electricity was applied to
the valve, the ligaments heated and shortened, which brought them back into plane and

opened the valve.

Other devices have been built using NiTi thin ﬁlms on silicon. Krulevitch et al. built a

micro-gripper ﬁ'om NiTi thin ﬁlms deposited on silicon. This gripper developed 13 mN

172

of clamping force between the jaws, and was capable of clamping onto features as large
as 110 um. (Lee, Ciarlo et a1. 1996) Carmen et al. created a microwrapper capable of
gripping an object as large as 320 pm in diameter. The wrapper is a single NiTi thin ﬁlm

on polyimide, both of which have been chemically milled. (Gill, D.T. et al. 2001)

The strain developed in these devices is based on the difference in thermal expansion
coefﬁcients between the substrate and the ﬁlm. Only 1-2% strain can be developed in
NiTi thin ﬁlms with large thermal expansion mismatches between the substrate and ﬁlm,
which does not take full advantage of the complete thermal hysteresis of NiTi. To take
full advantage of the thermal hysteresis available with NiTi, thin ﬁlms should be strained
to as much as 7% strain. This level of strain is not possible using thermal expansion
mismatches. To mitigate this problem, cold rolling of thin ﬁlms could introduce
unidirectional strains in the substrate and thin ﬁlm. While these unidirectional strains
could be somewhat elastically recovered in some substrates such as polyimide, other

substrates like silicon are not compatible with the process.

Uniaxial deformation is a method for developing larger strains in NiTi thin ﬁlms.
Grummon et al. reported large reversible strains in upwards of 7% using thin ﬁlms
deposited on polyimide. However, these large strains were developed using uniaxial
dead load experiments, which can be difﬁcult using cut-out sections of a thin ﬁlm. If a
small defect, such as a crack or tear, is present at the edge of a thin ﬁlm, the overall stress
before failure of the ﬁlm can be drastically reduced due to stress concentrations.

(Grummon 1995)

173

 

Cold rolling NiTi ﬁlms for actuators has been done by Miyazaki et al., with mild success.
Thin plates of NiTi were cold rolled to thicknesses of approximately 100 um, and were
further chemically milled as low as 25 um. (Miyazaki 2000) These thin plates showed
shape memory and were suggested to work well in actuator applications. However,
chemical milling is an expensive process that wastes large amounts of NiTi. Chemical
milling was used by Miyazaki because cold rolling is limited by the maximum force the
cold rolling mill can supply. As reported in literature, the minimum rolling thickness is

dependent on materials constants as shown: (Hosford and Caddell 1993)

h —C"R

min B —(0 o_0t)

Equation 2: Minimum height for cold rolling thin ﬁlms.

Where E’ is an effective Young’s modulus, R is roller height, ,u is the friction between
the rollers and the ﬁhn, a, is the ﬂow stress of the ﬁlm, a, is the unidirectional stress on
the ﬁlm if it was being pulled or pushed into the rolling mill and C is a geometric
constant assumed to be 7, based on the literature. For typical values for NiTi thin ﬁlms
and cold rolling mills, the minimum thickness is near 20 um, which is 25 times larger

than the ﬁlm used for the microwrapper.

To cold roll thinner ﬁlms, literature reports suggest sandwiching samples between thin
plates of a material that is softer than the thin ﬁlm. Assuming no slip between the layers,

the strain observed in the soft material also transferred into the thin ﬁlm. Therefore,

174

much thinner ﬁlms can be deformed uniaxially or even biaxially rolled using this process

without relying on elaborate rolling mills or expensive chemical milling.

A.1.2 Materials and Methods

Based on work by predecessors at Michigan State University, depositions on 7 .6 um
thick Dupont Kapton® 30 HN polyimide were conducted at elevated temperatures using
DC magnetron sputtering (Zhang 2006). The conﬁguration of the deposition and

annealing chamber can be seen in Figure 93.

The sputtering cathode composition used for all polyimide work was a T153_5-Ni45_5
composition, which produced thin ﬁlms with austenite start temperatures around 57°C
and ﬁnish temperatures around 93°C. Sputtering power during deposition was 200 W,
maintained for 2 hours per ﬁlm. After initial investigations to ﬁnd the best temperature,
depositions were conducted with a substrate topside temperature of 420°C using a 100
mm heater placed below the substrate. The substrates were at a distance of 51 mm from
the surface of the sputtering target to yield thicker ﬁlms per unit time than previous thin
ﬁlm work at MSU. It was discovered that if the substrate distance was lowered below 51
mm, the ﬁlms grown would be brittle in nature. This resulted in an approximate

sputtering rate of 4.5 pm per hour.

The polyimide substrates were held taunt in ﬁxtures that made 75 mm diameter samples,
the same diameter as the sputtering cathodes used. The base pressure of the high vacuum

chamber before sputtering was always below 1 uTorr. The working pressure during

175

deposition was between 5.5 and 6 mTorr of 99.9995% purity argon gas, which was

further puriﬁed by a Restek Thermal Gas Puriﬁer, model 21662.

A post-deposition heat treatment was also performed before breaking high vacuum to
ensure crystallization of the NiTi ﬁlm and to reduce internal stresses. During annealing,
the topside substrate temperature was held at 575°C for two hours, with a backside
temperature of less than 450°C to protect the polyimide, which decomposes above 480°C
in vacuum. This annealing occurred at the same time as the deposition of other ﬁlms, to

decrease processing time.

Following the post-deposition anneal, the thin ﬁlms on polyimide were removed from the
sample holders and were trimmed to remove the uncoated polyimide. This left the circles

of polyimide coated with approximately 9 pm of NiTi, shown in Figure 94.

The thin ﬁlms were sandwiched between thin metal plates of aluminmn, stainless steel or
brass and cooled to 0°C to fully martensitize the ﬁlms before being passed through a
Fenn 152.4 mm cold rolling mill. The material used did not affect the cold rolling of the
ﬁlms. Successive passes increased strain in the ﬁlm, in addition to curling the ﬁlm, as
shown in Figure 95. This curling effect was most likely due to differences in springback
between the polyirrride and the NiTi. To measure the strain, ﬁlms were ﬂattened and
digitally scanned with a ﬂatbed scanner at 1200 dots per inch before and after the cold

rolling procedure.

176

Several thin ﬁlms were rolled between 0% and 10% strain, by manual adjustments of the
rolling mill. Films were limited to 6% strain after several ﬁlms formed microcracks at
7% strain. To measure the recovered strain after heating, the ﬁlms were scanned after a
heat treatment in an air furnace at 150°C for 15 minutes to fully austenitize the ﬁlm and

compared with previous scans. An example of the as-rolled condition of a thin ﬁlm is

shown in Figure 94.

 

Figure 93: (a) DC Magnetron sputtering gun, (b) 75 mm heater for post deposition heat treatment and (c)
the substrate holder showing thermocouple placement.

177

 

Figure 94: As-rolled thin ﬁlm of 10 um NiTi deposited on 7.6 pm thick polyimide. Puckering is present
around the edges of the ﬁlm from the deposition plastically deforming the polyimide. The ﬁlm was cold-
rolled to approximately 5% strain.

 

Figure 95: Thin ﬁlm (left) and aluminum plates after cold rolling (right).

178

A.1.3 Results and discussion

Thin ﬁlms of NiTi are capable of strains up to approximately 6%, before developing
cracks or other damage. It was observed that the ﬁlms took on the surface features of the
material they were sandwiched between, so adhesives were not used to adhere the ﬁlm to
the sandwich. Microcracking was observed in ﬁlms that were inherently brittle from
irregularities in the sputtering process, or were strained beyond 6%. An example of a

microcrack from over-straining can be seen in Figure 97.

Thermally recovered strain was observed to increase linearly with rolling strain. A
rolling strain of approximately 0.8% was observed to be thermally irrecoverable, as
described in Figure 96. At the maximum limit of 6% rolling strain, it is possible to
induce up to 3.5% reversible strain in a thin ﬁlm. This represents approximately a 350%

increase in the maximum strain compared with silicon based NiTi ﬁlms.

The microstructure of the thin ﬁlms was inspected using a Carnscan SEM equipped with
a backseattered electron detector. A thin ﬁlm produced was cross-sectioned and imaged
with backseattered electrons. The sample was etched to reveal gain boundaries, shown
in Figure 98. The ﬁhn contained equiaxed gains ranging ﬁ'om approximately 1 to 10 um
in diameter. As previously reported, good mechanical properties are obtained from a thin
ﬁlm containing equiaxed gains rather than columnar gains. (Hou 1998) It is also
interesting that the gain boundaries fade away near the polyimide layer, which could be

a result of anisotropic etching or a transition to an amorphous NiTi layer.

179

DSC results, Table 21, obtained from the Composite Center at Michigan State University,
suggest that the thin ﬁlms created by this method are Ti-rich based on the transformation

from B2 austenite to R-phase to B19’ martensite. (Otsuka and Ren 2005)

Table 21: Transformation temperatures for thin ﬁlm 052307A. Obtained from DSC results displayed in

 

 

 

 

 

 

 

 

 

 

iigure 99.
Mr Ms Rt Rs A: At
7°C 22°C 48°C 53°C 62°C 76°C

4 _

N 0)
r I

Thermally Recovered Strain (%)

 

 

 

O

l I I l

2 4 6 8
Cold Rolling Strain (%)

Figure 96: Collective results of cold rolling experiments. As rolling strain increases, the thermally

recoverable strain also increases. 0.8% of the rolling strain is permanent. Cracking becomes present in
thin ﬁlms rolled beyond 6 to 7% strain, shown in Figure 97.

O

180

 

Figure 97: Microcrack developed alter the sample was cold rolled beyond 6% strain. This SE1 from a
Phillips Sirion SEM equipped with a ﬁeld emission gun operating at 25 keV. Bright area is charging on the
underlying polyimide.

181

 

Figure 98: BEI of NrTr thin ﬁlm on polyimide (bright stripe along bottom of ﬁlm). The grain boundaries of
the ﬁlm appear to terminate roughly one urn away from the polyirrride ﬁlm, which could be an etching
effect or transition to amorphous NiTi.

 

 

Figure 99: DSC results from sample 052307A, a 10 um thin ﬁlm on a 7.6 urn thick polyimide substrate.

182

A.4 Freestanding thin ﬁlm creation

Freestanding thin ﬁlms of NiTi are of geat technical importance. Not only can
freestanding thin ﬁlms show shape memory, but they can also be used as a basis for

nickel titanium multilayers.

Freestanding thin ﬁlms are grown on glass microscope slides at ambient temperatures,
then post-crystallized and later manually peeled from the glass slide surface. The
adhesion between the glass slides and NiTi thin ﬁlms must be engineered properly to
allow the ﬁlms to be peeled off. If the ﬁhns are not sputtered correctly, adhesion will be
either too weak or too strong. If the adhesion is too weak the ﬁlm can delaminate during

the deposition process, which distorts the surface ﬁnish and shape of the ﬁlm.

Conventional glass microscope slides were purchased precleaned ﬁom MSU stores.
Before putting the slides into the sputtering chamber, the slides were given a ﬁnal
cleaning in acetone to remove any ﬁngerprints or other contaminants. Acetone was
chosen because it leaves a faint residue on the surface of the glass slides, which helps

ﬁlm removal.

To deposit the thin ﬁlm, the chamber was allowed to reach a base pressure of < 3 uTorr.

The sputtering conditions for creation of freestanding thin ﬁlms will be discussed in

detail in Appendix A.5, but are summarized in Table 22.

183

Table 22: Sputterin conditions for freestanding NiTi thin ﬁlms.

 

 

 

 

 

 

 

Base Pressure Working distance Working pressure Sputter Sputter Rate
Power
< 3 x 106 Torr 50 mm Initial: 7.5 x 10'3 Torr, 0.1 hr 200 Watts 4.5 um/hr
Remaining: < 5 x 10'3 Torr
Ambient Post sputter anneal Post sputter anneal time Sputtering
Temperature temperature Cathode
20°C 575°C 1.5 hr Tisu-NigL

 

 

 

To ensure correct adhesion between the ﬁlm and slide, the working pressure during the
ﬁrst ﬁve nrinutes of the deposition was maintained at 7.5 mTorr, which is sigriﬁcantly
higher than throughout the rest of the deposition. After ﬁve minutes, the pressure was
lowered from 7.5 mTorr to below 5 mTorr. The pressure is further lowered to 4 mTorr if

the plasma can be sustained.

After the completion of deposition, the thin ﬁlms are crystallized using a Centorr vacuum
annealing furnace. The ﬁhns are heated to a temperature of 575°C for 1.5 hours and are
then furnace cooled. Once they have been annealed, one of the glass slides is placed in a
Scintag 2000 XRD to check for full crystallinity of the ﬁlm. If the ﬁlm is fully
crystalline and shows X-ray peaks corresponding to a martensite crystal structure, the
ﬁlms are detached from the glass slides. If the ﬁlm does not show a martensite structure,
however, the ﬁlms are heated in the annealing furnace for another hour. The variation in
annealing time stems from changes in ﬁxturing and thermocouple placement in the

vacuum annealing furnace.

To remove the thin ﬁlms, a glass cutter is used to score the glass and break it in half. The

thin ﬁlm is then peeled slowly away from the surface of the glass at a 45° angle to the

184

 

glass. The ﬁlm sometimes catches on the edges of the glass slide and begins to tear,

forceps are used to scrape the thin ﬁlm to release it from the glass slide.

Currently, this process creates cellular-type gain gowth in the thin ﬁlms. It is believed
that by raising the temperature of the deposition, the gain shape could be altered.
Depositing fully crystalline ﬁlms is also a possibility using high temperature depositions,
around 400°C, but has not been explored yet. If this technique works on glass slides, the
important factor in allowing ﬁlm removal will be a high working pressure for the ﬁrst

few minutes of the deposition.

A.5 Multilayer creation

Multilayers of nickel titanium and niobium were made using DC magretron sputtering in
an ultra high vacuum chamber at Michigan State University. The ﬁhns were created with
two sputtering guns, shown in Figure 93, which were equipped with an off-stoichiometric
nickel titanium cathode and a 99.9995% pure niobium cathode. The working distance
between each of the cathodes and the glass slide substrates was 51 mm during deposition.
The sputtering took place simultaneously on both sputtering guns with two sets of
substrates. The substrates were rotated from one sputtering gun to the other after each

deposited layer.

Before sputtering, the pressure in the vacuum chamber was less than 3 uTorr. To assist
pumping, halogen bake-out lights were used to heat the chamber to an elevated

temperature, which helped drive off water ﬁ'om surfaces inside the chamber. An inert

185

atmosphere ultra-high purity argon was maintained at a pressure of less than 5 mTorr
during deposition. The argon supply was 99.9995% pure, and was further puriﬁed with a

Restek thermal puriﬁer.

Niobitun was chosen as the top layer of the multilayer to avoid readily forming NiTi
oxides. The sputtering power of each sputtering gun was adjusted so that the sputtering
rates matched the desired foil composition of 2.2 NiTi : 1 Nb. The power of the NiTi
sputtering gun was limited to less than 250 W due, to concerns that compositional

changes in the NiTi ﬁhns could occur at high cathode powers.

The deposition was conducted at room temperature so that adhesion between the thin ﬁlm
and the substrate would be low. As adhesion is expected to decrease with increasing
working pressure, the ﬁrst layer of the multilayers were deposited at a working pressure
of 7 mTorr. Following deposition, the samples were annealed inside a vacuum furnace
for 1.5 hours at 575°C to crystallize the NiTi phase and reduce any residual stresses in the

thin ﬁlm. Multilayers are removed the same way as freestanding thin ﬁlms.

A.6 Oxide Removal Etch

Throughout this work, steps have been taken to minimize substantial oxide formation
from thermal exposure. Oxides were removed from samples before brazing. Low base
pressures were observed during brazing, and heating and cooling times were minimized
to reduce oxide formation. Brazing NiTi, as well as crystallizing thin ﬁlms, both take

place in hard vacuum.

186

Some of the material received from NDC had a thick black oxide coating on it, which had
to be removed before brazing. If the oxide was not removed before thermal exposure, the
metal became ernbrittled, most likely due to the diffusion of interstitials into the B2

crystal structure.

One etchant recipe, a combination of nitric and hydroﬂuoric acids, was used for all oxide
removal. These acids were mixed with liberal amounts of water to control the rate of
etching. After some initial experimentation, it was determined that a recipe of 10 HNO3 :
10 H20 : 1 HF removed the oxide layer from the nickel titanium quickly without
sigriﬁcantly attacking the base metal. This is also safer than other recipes that use larger

concentrations of hydroﬂuoric acid. (Chiu, Cheng et al. 2005) (Gill, Ho et a1. 2002)

The oxide removal etch was used primarily on the NiTi strip material, which was
received with an oxide layer approximately 0.025 mm thick. The material was etched
before and after shape setting the material. It was observed that the oxide layer was more

difﬁcult to completely remove if the material was not etched before shape setting.

Typically, samples were left to etch for no more than 2 minutes, with some material
needing as little as 10 to 20 seconds before the oxide was completely removed. Once
removed, the pieces were washed in tap water. The etching rate and variation in the
oxide thickness account for the wide range in etching time required. The etching rate is
dependent on the temperature of the chemical bath, which increases with the total amount

of metal that has already been etched and the amount currently being etched.

187

A.7 Metallurgical Etch

The standard metallurgical etchant used is 20 HNO3 : 100 H20 : 10 HF, which yielded

the best etching effect.

Etching time, as discussed in Appendix A.6, is dependent on many factors. Metallurgical
samples were etched for 10 seconds to 120 seconds with good results. To ensure proper
etching, the samples were etched, rinsed in tap water, and then viewed under an optical
microscope. The etching was completed when the gains of the nickel titanium were

distinguishable with a set of polarizing ﬁlters on an optical reﬂection microscope.

A.8 Tacking NiTi to Nb using AC spot welding

Using a low intensity pulse of AC power, Nb can be tacked to NiTi with enough strength
to allow manual assembly of large NiTi arrays. An AC spot welder was homemade from
a hand-wound transformer. The nominal output of the spot welder was 5 volts, and could
supply up to 100 amps, based on the resistance of the pieces that were to be joined. The
power output was limited to 0.10 seconds; with a square wave output centered around 0

volts.

The tacking process could only be used for thin gages of NiTi, such as strip NiTi, NiTi

thin ﬁlms, Nb wires and Nb foils. Thicker materials, like the plate NiTi, had a higher

resistance and thermal mass, which limited the current in the joint enough to prevent

188

 

tacking. Using a higher power would have solved this, but this machine was only

desigred to tack thin gages of NiTi and Nb.

The joint was made by placing two or three pieces of MT i and Nb together between two
14 gage copper wires. The wires were stiff enough to exert a small pressure onto the
pieces and push them together. The electrical pulse was carried through the wires and
into the sandwich of materials, where it made very small spot welds between the pieces.
These spot welds were the basis for the research into true spot welding technology, which

uses higher powers and much larger contact forces.

A.9 Post Braze Heat Treatment

To restore the pseudoelastic functionality of Ni-rich NiTi after brazing, a low temperature
heat treatment was performed. During this heat treatment, the Ni4Ti3 metastable phase
precipitates in ﬁnely dispersed particles, which raised the stresses required for dislocation
motion in the NiTi B2 phase. Samples were heated in an air fmnace to a temperature of
350°C for 1.5 hours and then quenched in water. This heat treatment was chosen by Ke
Bin Low, working under Dr. Grmnmon at Michigan State University, and was borrowed

for the production of NiTi arrays.

189

Bibliography

Cheng, Y., W. Cai, et al. (2006). "Surface modiﬁcation of NiTi alloy with tantalum to
improve its biocompatibility and radiopacity." J ourngl of Materials Science 41:
4961-4964.

Chiu, K. Y., F. T. Cheng, et al. (2005). "A preliminary study of cladding steel with NiTi
by microwave-assisted brazing." Materials Science & Engigeering A 407: 273-
281.

Cho, H., H. Y. Kim, et a1. (2005). "Fabrication and characterization of Ti-Ni shape
memory thin ﬁlm using Ti/N i multilayer technique." Science and Technology of
Advanced Materials 6: 678-683.

 

1!-..

Chu, C. L., C. Y. Chung, et a1. (2005). "Phase transformation behaviors in porous Ni-rich
NiTi shape memory alloy fabricated by combustion synthesis." Materials Science
& EngineeringA 392: 106-111.

Falvo, A., F. M. Furdiuele, et al. (2005). "Laser welding of a NiTi alloy: Mechanical and
shape memory behaviour." Materials Science & Engineering A 412: 235-240.

Femandes, F. M. B., R. Martins, et a1. (2002). "Structural characterisation of NiTi thin
ﬁlm shape memory alloys." Sensors gnd Actuators A Physical 99: 55-58.

Foltz, J. W., D. S. Grummon, et al. (2006). "Plastic Deformation of Sputtered NiTi Thin
Films on Polyimide Substrates by Co-Rolling." Shane Memory aﬁSuperelthig

Technologies Proceedings: 1-7.

Gale, W. F. and Y. Guan (1997). "Microstructural development in copper-interlayer
transient liquid phase bonds between martensitic MA] and NiTi." Journal of
Materials Science 32: 357-364.

Gaskell, D. R. (1995). Introduction to the Thermodynamics of Materials Taylor &
Francis.

 

Gill, J. J ., C. D.T., et al. (2001). "Manufactming issues of thin ﬁlm NiTi microwrapper."
Sensors and Actuators A: Physical 93(2): 148-156.

190

Gill, J. J ., K. Ho, et al. (2002). "Three-Dirnensional Thin-Film Shape Memory Alloy
Microactuator with Two-Way Effect." Journal of Microelectromechanical
Sﬁtems 11(1): 68-77.

Greiner, O., Dunand (2005). "High Strength, low stiffness, porous NiTi with superelastic
properties." Acta Biomaterialia 1(1): 705-716.

Grummon, D. S. (1995). "Progess on Sputter-Deposited Therrnotractive Titaniurn-
Nickel Films." J ourn_al de Physigi_1e IV (C8).

Grummon, D. S., S. Nam, et al. (1992). "Effect of superelastically deforming NiTi
surface microalloys on fatigue crack nucleation in copper." Materials Research
Society Symposium Proceedingg 246: 259-264.

Grummon, D. S., J. A. Shaw, et al. (2003). "Low-density open-cell foams in the NiTi
system." Applied Physics Letters 82(16): 2727-2729.

Gupta, K. P. (1991). "Phase Diagarns of Ternary Nickel Alloys, Part 2, Section D2, The
Nb-Ni-Ti (Niobium-Nickel-Titanium) System." 98-112.

Hall, P. C. (1997). Methods of Promoting Solder Wetting on Nitinol. SMST, Asilomar,
Paciﬁc Grove, CA, ASM International.

He, X. M. and L. J. Rong (2004). "DSC analysis of reverse martensitic transformation in
deformed Ti-Ni-Nb shape memory alloy." Scripta Materialia 51: 7-11.

Hosford, W. F. and R. M. Caddell (1993). Metal Forming: Mechgnics grid Metallurgy.
New Jersey, Prentice-Hall.

Hou, L. (1998). Processing, Microstructure and Thermomechanical Behavior of
Superelastic Nickel-Titanium thin ﬁlms sputter-deposited at elevated
temperatures. Michigan State University, Michigan State University.

Johnson, Gupta, et al. "Silicon Oxide Diaphragn Valves and Pumps with TiNi Thin Film
Actuation." TiNi Alloy Company.

191

Karagoulis, M. (1991). Control of Materials Processing Variables in Production
Resistance Spot Welding. Metallurgy. Meganics. ml Mategrls Science. East
Lansing, MI, Michigan State University. Doctor of Philosophy: 10-15.

Karagoulis, M. (1991). Control of Materials Processing Variables in Production

Resistance Spot Welding. Metallurgy, Mechanics, and Materials Science. East
Lansing, MI, Michigan State University. Doctor of Philosophy: 15-25.

Lee, Ciarlo, et al. (1996). "A practical microgipper by ﬁne alignment, eutectic bonding
and SMA actuation." Sensors and Actugtors A Physical 54.

Li, B. Y., L. J. Rong, et a1. (1998). "Porous NiTi alloy prepared from elemental powder
sintering." Journal of Materials Research 13(10): 2847-2851.

Li, M. G., D. Q. Sun, et al. (2006). "Effects of laser brazing parameters on microstructure
and properties of TiNi shape memory alloy and stainless steel joint." Materials
Science & Engineeringﬁ 424(1): 17-22.

Low, K. B. (2006). Current Research, Michigan State University.

Low, K. B. (2007). Current Research, Michigan State University.

Liitjering, G. and J. C. Williams (2007). Titanium, Springer.

MacDonald, W. D. and T. W. Eagar (1992). "Transient Liquid Phase Bonding
Processes." The Metal Science of Joining.

Massalski, T. B. (1986). Bing Alloy PhJase DIM. Metals Park, American Society
for Metals.

Memry, C. (2006). "Nitinol FAQ." Retrieved 6/20, 2007, from
http://www.memry.com/nitinolfaq/nitinolfirq.html#solderedWeb.

Miyazaki, S. (2000). "Texture of Ti-Ni rolled thin plates and sputter-deposited thin
ﬁlms." Joum_al of Plgsticity 16: 1135-1154.

Otsuka, K. and X. Ren (2005). "Physical metallurgy of Ti-Ni-based shape memory
alloys." Progress in Materials Science 50: 521.

192

Piao, M., S. Miyazaki, et al. (1992). "Effects of Nb Addition on the Microstructure of Ti-
Ni Alloys." Materials Transactions 33(4): 337 - 345.

Porter, D. A. and K. E. Easterling (1989). Phase Trgnsformations in Metals and Alloys.
London, Van Nostrand Reinhold.

Prima, S. B., L. A. Tret'yachenko, et al. (1995). "Phase Relationships In The Ti-TiNi-
NbNi-Nb Region of the Ternary System Ti-Nb-Ni." Powder Metgllurgy and
Metal Ceramics 34(3,4): 155-160.

Saburi, T. "Ti48Ni52 Quench from 10000C." Proc. Mgt. Res. Soc. 9: 78.

Schlossmacher, P., T. Hass, et a1. (1997). Laser-Welding of a Ni-ricl_< TiNi Shape
Memory Alloy: Pseudoelgstic Properties. SMST, Asilomar, Paciﬁc Grove, CA,
ASM International.

Sinha, A. K. (2003). Physical Metallurgy Haidbook, McGraw-Hill Handbooks.

Wang, G. (1997). Welding of Nitinol to Stainless Steel. SMST, Asilomar, Paciﬁc Grove,
CA, ASM International.

Wert, C. and R. Thomson (1970). Physics of Solids. McGraw-Hill Book Company.

William D. Callister, J. (2003). Materijals Science gndﬂrgineering: An Introduction. John
Wiley & Sons, Inc.

Zhang, J. X., M. Sato, et a1. (2005). "Deformation Mechanism of martensite in T-rich Ti-
Ni shape memory alloy thin ﬁlms." Acta Materialia 54: 1185-1198.

Zhang, Y. (2006). NOVEL TRIBOLOGICAL SYSTEMS USING SHAPE-MEMORY
ALLOYS AND THIN FILMS.

193

       

IIlilllllﬂljﬂlllmiﬂljllWilt]?I