Citation
Advanced high-temperature shape-memory alloy development and thermomechanical characterization of platinum and palladium modified NiTi based SMAs

Material Information

Title:
Advanced high-temperature shape-memory alloy development and thermomechanical characterization of platinum and palladium modified NiTi based SMAs
Creator:
Rios, Orlando ( Dissertant )
Dempere, Luisa A. ( Thesis advisor )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Copyright Date:
2006
Language:
English

Subjects

Subjects / Keywords:
Alloys ( jstor )
Austenite ( jstor )
Compressive stress ( jstor )
Deformation ( jstor )
Electrical resistivity ( jstor )
Heat stress disorders ( jstor )
Martensite ( jstor )
Materials recovery ( jstor )
Shape memory alloys ( jstor )
Stress tests ( jstor )
Materials Science and Engineering thesis, M.S
Dissertations, Academic -- UF -- Materials Science and Engineering
City of Orlando ( local )

Notes

Abstract:
A series of Ti-Ni-Pt and Ti-Ni-Pd high temperature shape memory alloys (HTSMAs) have been examined in an effort to find alloys with a suitable balance of mechanical and physical properties for applications involving elevated temperature actuation. Initially, more than 20 Ti-Ni-Pt alloys were prepared by arc melting high purity materials followed by a homogenization heat treatment under vacuum. Each alloy was then characterized using optical and scanning electron microscopy and differential scanning calorimetry. A strong link to stoichiometry and transformation temperatures was not evident which indicates that a very limited solubility for off stoichiometry compositions exist with in the B2 and B19 structures. The results from this study combined with the results of an advanced thermomechanical processing study conducted by colleagues at the NASA Glenn Research Center were used to select the Ni₂₅Ti₃₀Pt₂₅ alloy for more extensive investigations of their structure, property and processing relationships. For comparison, a Ti-Ni-Pd alloy, namely Ti₅₀Ni₃₀Pd₂₀, was selected from literature because it was found to have a maximum in the unconstrained recovery behavior. These materials were extruded characterized by advanced thermomechanical testing by measurement of the baseline mechanical properties and shape memory specific behaviors. In both alloys the work output reached a maximum as a function of applied stress (biasing load) as did the transformation strain. Through thermomechanical testing it was evident that slip mechanisms were detrimental to the performance of these alloy's performance as actuator materials. In both alloys the resistance to slip under a biasing load in the temperature regions near the transformation temperatures prevented complete recovery thus limiting the work performance of these alloys. A link between the difference in yield strength between the austenite and martensite and the performance under a biasing load was confirmed which is a good indicator in further alloy selection.
Subject:
actuation, actuator, aeronautical, aerospace, alloy, B19, B2, extrusion, high, mechanical, memory, Ni, Pd, Pt, resistance, resistivity, shape, temperature, thermomechancial, Ti
General Note:
Title from title page of source document.
General Note:
Document formatted into pages; contains 145 pages.
General Note:
Includes vita.
Thesis:
Thesis (M.S.)--University of Florida, 2006.
Bibliography:
Includes bibliographical references.
General Note:
Text (Electronic thesis) in PDF format.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Rios, Orlando. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
3/1/2007
Resource Identifier:
003757747 ( aleph )

Downloads

This item has the following downloads:


Full Text











ADVANCED HIGH-TEMPERATURE SHAPE-MEMORY ALLOY DEVELOPMENT
AND THERMOMECHANICAL CHARACTERIZATION OF PLATINUM AND
PALLADIUM MODIFIED NiTi BASED SMAs
















By

ORLANDO RIOS


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2006

































Copyright 2006

by

Orlando Rios


































To my father and family
















ACKNOWLEDGMENTS

I thank my mother and father. I thank my family for not letting distance separate

us. I thank the great friends I have made here. I would like to thank Dr. Nathal and Dr.

Ronald Noebe of the NASA Glenn Research Center' s Advanced Structures Division for

their support, guidance, materials and processing and unlimited use of the division' s

characterization and mechanical testing facilities. Without their support I would not have

an acknowledgments section to write nor would I have as interesting a study. I would

like to thank the kind effort of my committee members. I would like to thank Dr.

Donnelly for her support and guidance at all times, and all of my colleagues here and

afar.





















TABLE OF CONTENTS


page

ACKNOWLEDGMENT S .............. .................... iv


LI ST OF T ABLE S ................. ................. viii............


LIST OF FIGURES .............. .................... ix


AB STRAC T ......__................ ........_._ ........xi


CHAPTER


1 INTRODUCTION AND BACKGROUND .............. ...............1.....


Significance .............. ...............1.....
Background .................. ........... ...... ....... .............
General Shape Memory Alloy Behavior ................. ...............................3
S MA Structural Characteristics ................. ...............4............ ....
S MA Mechanical Behavior ................. ...............10........... ....
Thermoelastic Shear Transformations............... ............1


2 MATERIALS PROCESSING AND PROCEDURES .............. ....................2


Melting Procedures............... ...............2
Arc M elting .............. ...............22....
Arc Melt Machining .............. ...............22....
Induction M elting ................. ...............23.......... .....
Homogenization .............. ...............23....
Extrusion.................. .. ................2
Stress Relief Heat Treatment ................. ...............25................
Characterization Procedures .............. ...............25....

Dynamic Modulus .............. ...............25....
Compositional Analysis................... ...... ...............2
Nitrogen, Oxygen, Carbon and Sulfur Analysis............... ...............27
Therm al Analysis................ .......... ... .. .... .........2
Microstructural and Semi-Quantitative Compositional Analysis........................30
Dilatometry Measurements .............. ...............3 1....
Resi stivity Measurements ................. ...............3.. 1..............
Sample instrumentation ......___ ..... ... ._ .....___ ...........31
Resistivity apparatus .............. ...............32....












Thermomechanical Testing ............... ...............37....
Thermomechanical instrumentation ........_ ....... ..___ ........._ ....37
Uniaxial isothermal mechanical tests ....._____ ..... ... .__ ............._..39
Load free strain recovery tests .............. ...............40....
Load bias test............... ...............40..

3 ALLOY DEVELOPMENT .............. ...............50....


Characterization of the NiTiPt SMA system ................ .............. ......... .....50
Bulk Compositional Analysis............... ...............50
Transformation Temperature ................. ...............51.................
M icrostructure .............. ...............53....
Alloy Selection : NiTiPt .............. ...............55....
Alloy Selection : NiTiPd .............. ...............57....
All oy S el section Summary ................. ...............60................

4 RE SULT S AND DI SCU SSION ............... ...............6


Heat Treatment Optimization .............. ...............68....
Characterization ................... ...............70.......... ......
Material s Characterization............... ...............7
Properties and Transformation Temperatures .............. ...............70....
Thermomechanical Testing .............. .. .. ... ..............7
Isothermal Stress-Strain Behavior in Tension and Compression ................... .....74
Isothermal stress-strain behavior in tension and compression NiTiPd ........75
Dynamic elastic modulus determination NiTiPd .............. ..... ..................8
Isothermal stress-strain behavior in tension and compression NiTiPt.........83
Unconstrained Recovery Tests .................. ...............86........... ....
Unconstrained recovery tests NiTiPd............... ...............86.
Unconstrained recovery tests NiTiPt.................. ....... ............8
Constant-Load, Strain-Temperature Te sts and Work Output ...........................90
Constant-load, strain-temperature tests and work output : NiTiPd .............91
Constant-load, strain-temperature tests and work output:NiTiPt .................96

5 SUMMARY AND CONCLUSIONS ................ ...............116...............


Alloy Development. .............. .. .......... .. ...............116.....
Characterization and Thermomechanical Testing ................. .........................117
Conclusions Relevant to Alloy Design ................. ...............120........... ...
Future Studies ................. ...............12. 1..............


APPENDIX


A NiTiPd HTSMA MATERIAL DATA SHEET .............. ...............123....


Physical Properties............... ..............12
Electrical Resi stivity ................. ............ ...............123 .....
Thermal Coefficient of Resistivity ................. ...............123...............












Coefficient of Thermal Expansion .............. .....................123
Shape Memory Properties ................. ...............123................
Transformation Temperatures .............. ...............123....
Com position............... ..............12

B NiTiPt HTSMA MATERIAL DATA SHEET .............. ...............124....


Physical Properties............... ..............12
Electrical Resi stivity .........._.... .......__ ...............124...
Thermal Coefficient of Resistivity ........._.__........_. ...._._. ..........12
Coefficient of Thermal Expansion .............. .....................124
Shape Memory Properties. ........._.___..... .__. ...............124...
Transformation Temperatures .............. ...............124....
Com position............... ..............12

C CHEMICAL ANALYSIS OF EXTRUDED MATERIALS ................. ...............125


LIST OF REFERENCES ........._.___..... .___ ...............126....


BIOGRAPHICAL SKETCH ........._.___..... .__. ...............130....

















LIST OF TABLES


Table pg

1-1 Several characteristics common to metallic thermal SMA. ............. ...................13

3-2 Aim and measured compositions of all alloys investigated. ............. ...................61

3-3 Transformation temperatures of alloy set............... ...............62..

3-4 Semi-quantitative EDS analysis of the various phases observed..............._._..........64


















LIST OF FIGURES


Figure pg

1-1 Power-to-weight ratio versus weight diagram for common actuator types
currently used in aeronautics............... ..............1

1-2 Idealized plot of a property change vs. temperature .............. .....................17

1-3 Structure of the parent phase austenitee) and shear phases (Bl19 and Bl9'
m artensite) ........... .......__ ...............18..

1-4 Thermoelastic transformation and twin accommodated transformation strain........19

1-5 Two-dimensional lattice schematic of monoclinic structures ................ ...............19

1-6 TEM micrographs of twinned and untwined monoclinic martensite ................... ....19

1-7 Effects of thermal cycling through the hysteresis on the transformation
temperatures of several NiTi based shape memory alloys ................. ................. .20

1-8 Deformation and shape recovery by detwinning (twins marked with arrows) ........20

1-9 Isothermal stress strain behavior of a typical SMA strained in the fully
martensitic state ................. ...............21.................

1-10 Stress strain behavior showing the three main deformation regimes active in
SM As ................. ...............21.................

2-1 Induction melted Nil9.5Pd30Tiso.5 cast ingot with attached hot top on a quarter
inch grid................ ...............41.

2-2 Heat treatment and processing temperature schedule. ............. .....................4

2-3 Hot extrusion press schematic............... ...............4

2-4 Uniaxial sample (A) 5 X 10 mm compression sample (B) Threaded 17.4 mm
long by 3.81 mm diameter gauge sample ................. ...............43.............

2-5 Processing flow diagram of DSC, compression, and tensile samples. ................... ..43

2-6 The ICP using an Echelle type polychrometer ................. ................. ..........44










2-7 Example of a DTA scan showing the exothermic and endothermic peaks
characteristic of thermoelastic shape memory alloys ................. ............ .........44

2-8 Four-point probe resistivity configuration .............. ...............45....

2-9 Raw (blue) and conditioned (yellow) voltage signals for resistivity
measurements during inductive heating ................. ...............45................

2-10 Resistivity vs. temperature profile with regression analysis ................. ........._.....46

2-11 High conductivity Pt wire resistivity vs. temperature relationship demonstrating
the repeatability of the during heating and cooling ................. ................ ...._.46

2-12 NIST (resistivity standard) resistivity vs. temperature profile comparison of
NIST measurements and the measurements by the resistivity apparatus. ................47

2-13 Resistivity apparatus data flow diagram. ............. ...............48.....

2-14 Materials Testing Systems (MTS) tensile frame fitted with high-temperature hot
grips and induction heating configured for compressive testing. ................... ..........49

3-1 Ternary plot of the Ti-Ni-Pt compositions studied. The composition of all alloys
was confirmed by spectrographic analysis............... ...............61

3-2 A. Effect of Pt on the Ms transformation temperatures for Ni50-xPtxTi50 alloys,
including data from previous researchers B. Effect of Pt on all transformation
temperatures for Ni50-xPtxTi50 alloys............... ...............62.

3-3 SEM B SE micrographs of the non-stoichiometric alloys. ................ .................. 63

3-4 Phase diagrams (A) NiTi binary phase diagram from reference (B) TiPt binary
phase diagram from reference ................ ...............64................

3-5 Effect of ternary alloying additions on the Ms (or Mp) temperature for NiTi-
based high-emperature shape memory alloy systems .............. ....................6

3-6 Comparison of the specific work output for several conventional NiTi alloys,
SM495 NiTi, and the (Ni,Pt)Ti HITSMA ................. ...............65...............

3-7 Phase diagram of TiPd+TiNi alloys. ............. ...............66.....

3-8 Shape memory properties NiTiPd (A) Ms temperature resulting from ternary
alloy additions. (B) Average shape recovery in Tiso (NiSO-x) Pdx. ..........................66

3-9 Plots of martensitic transformation temperatures vs. composition for Tiso-
xP d ;, N i ,n 47. .............. ...............67











4-1 Stress Relief Heat Treatment Optimization by Analysis of Resistivity
Temperature Profiles (note the resistivity curves are offset for convenience on
the same resistivity scale)............... ...............98.

4-2 SEM BSE image of extruded Nil9.5Tiso.5Pd30 ........ ............... 99

4-3 NiTiPt Resistivity and Dilatometry Test Results ................... ...............9

4-4 NiTiPd Resistivity and Dilatometry Test Results .............. ....... .............10

4-5 NiTiPd Force Strain Curve at 3650C............... ...............100.

4-6 NiTiPd Alloy Uniaxial Isothermal Tensile Tests at RT, 2000C, 3000C and 4000C
(a) Engineering Stress Strain Curve (b) True Stress Strain Curve including
correction for non-uniform deformation of the 400C sample .........._... ..............101

4-7 NiTiPd Alloy Uniaxial Isothermal Compression Tests at RT, 2000C, 3500C,
3650C, 4000C and 5000C (a) Engineering Stress Strain Curve (b) True Stress
Strain Curve............... ...............101.

4-8 NiTiPd Alloy Uniaxial Isothermal Tensile Tests at 2250C, 2450C, 2550C and
2720C............... ...............102.

4-9 NiTiPd Alloy Uniaxial Isothermal Compression Test at 2550C, 2720C and
3000C............... ...............102.

4-10 Isothermal Uniaxial Stress Strain Curve with Resistivity Exhibiting a Stress
Induced Transformations............... ............10

4-11 NiTiPd Yield Stress vs. Temperature in tension and compression ................... .....103

4-12 Temperature Dependent Dynamic Elastic Modulus measured on heating ............ 103

4-13 NiTiPt Alloy Uniaxial Isothermal Tensile Tests at 4400C, 4700C, 5500C and
6000C............... ...............104.

4-14 NiTiPt Alloy Uniaxial Isothermal Compression Tests at 4400C, 4700C, 4900C,
5500C, and 6000C............... ...............105.

4-15 NiTiPt Alloy Uniaxial Isothermal Tensile Tests at 2000C, 3800C, 4000C and
4400C............... ...............105.

4-16 NiTiPt Alloy Uniaxial Isothermal Compression Tests at 2000C, 3800C, 4000C
and 4400C ........... .......__ ...............106..

4-17 Stress Strain Curve at 500 Celsius at Low and High Strain Rates ......................106

4-18 Yield Stress vs. temperature for NiTiPt .............. ...............107....










4-19 Fracture Stress and Strain vs. temperature for NiTiPt .............. .....................0

4-20 NiTiPd Unconstrained Recovery Test at 4 and 2 Percent Initial Strains. ..............108

4-21 Total Recovery Rate vs. Total Strain for NiTiPd ................. ................ ...._.108

4-22 Load Free Recovery Individual Components of Total Recovery for NiTiPd........109

4-23 Temperature Dependent Load Free Recovery Curve for Complete NiTiPt Test...109

4-24 Load Free Recovery Individual Components of Total Recovery for NiTiPt.........1 10

4-25 Load Bias in Tension (Specific Work Output) for NiTiPd ................. ................110O

4-26 Load Bias in Compression (Specific Work Output) for NiTiPd ................... .........111

4-27 Load Bias in Tension (Specific Work Output) Complete Thermomechanical
Path for NiTiPd ................. ...............111...............

4-28 Specific Work vs. Biasing Load for NiTiPd ................. ................ ......... .112

4-29 Transformation Strain (2nd cycle martensite to austenite) vs. Biasing Load for
N iTiPd ................. ...............112......... ......

4-30 Open Loop Strain vs. Biasing Stress for NiTiPd ................. ........................113

4-31 Load Bias in Compression (Specific Work Output) for NiTiPd ................... .........113

4-32 Transformation Strain (2nd cycle martensite to austenite) vs. Biasing Load for
N iTiPt ................ ...............114......... ......

4-33 Specific Work vs. Biasing Load for NiTiPt ................. .............................114

4-34 Open Loop Strain vs. Biasing Stress for NiTiPt ................. .........................115
















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

ADVANCED HIGH-TEMPERATURE SHAPE-MEMORY ALLOY DEVELOPMENT
AND THERMOMECHANICAL CHARACTERIZATION OF PLATINUM AND
PALLADIUM MODIFIED NiTi BASED SMAs


By

Orlando Rios

December 2006

Chair: Luisa Dempere
Major Department: Materials Science and Engineering

A series of Ti-Ni-Pt and Ti-Ni-Pd high temperature shape memory alloys

(HTSMAs) have been examined in an effort to find alloys with a suitable balance of

mechanical and physical properties for applications involving elevated temperature

actuation. Initially, more than 20 Ti-Ni-Pt alloys were prepared by arc melting high

purity materials followed by a homogenization heat treatment under vacuum. Each alloy

was then characterized using optical and scanning electron microscopy and differential

scanning calorimetry.

A strong link to stoichiometry and transformation temperatures was not evident

which indicates that a very limited solubility for off stoichiometry compositions exist

with in the B2 and Bl9 structures. The results from this study combined with the results

of an advanced thermomechanical processing study conducted by colleagues at the

NASA Glenn Research Center were used to select the Ni25Ti30Pt25 allOy for more










extensive investigations of their structure, property and processing relationships. For

comparison, a Ti-Ni-Pd alloy, namely TisoNi30Pd20, WAS selected from literature because

it was found to have a maximum in the unconstrained recovery behavior. These

materials were extruded characterized by advanced thermomechanical testing by

measurement of the baseline mechanical properties and shape memory specific

behaviors.

In both alloys the work output reached a maximum as a function of applied stress

(biasing load) as did the transformation strain. Through thermomechanical testing it was

evident that slip mechanisms were detrimental to the performance of these alloy's

performance as actuator materials. In both alloys the resistance to slip under a biasing

load in the temperature regions near the transformation temperatures prevented complete

recovery thus limiting the work performance of these alloys. A link between the

difference in yield strength between the austenite and martensite and the performance

under a biasing load was confirmed which is a good indicator in further alloy selection.















CHAPTER 1
INTRODUCTION AND BACKGROUND

Significance

A number of metallic alloys have been shown to exhibit the shape memory effect.

Wayman gives a broad definition of shape memory alloy (SMA) which encompasses the

bulk of all thermal shape memory alloys. He defines a thermal shape memory alloy as an

article that when deformed at a lower temperature will regain its original shape when

heated to a higher temperature.

Shape memory alloys are used in multiple engineering applications. The most

common commercial system is NiTi based SMAs. Applications thus far for NiTi alloys

include electrical switches, eyeglass frames, brassiere underwires, cell phone antennas,

appliance controllers, temperature sensitive valves, microactuators, and countless medical

and dental devices.3 4 In addition, the first large-scale commercial applications for shape

memory alloys were made using NiFeTi and NiNbTi alloys with sub-room temperature

transformation temperatures, for use as couplings for pipes, tubes, and electrical

interconnects.' These applications make use of NiTi alloys near room temperature. The

main reason that commercial applications have been limited to near room temperature is

that commercial NiTi SMAs have a maximum transformation temperature of about

100oC.

In addition, there are many control and actuation-type applications for materials

exhibiting the shape memory effect at higher temperatures. High-temperature shape

memory alloys (HTSMA) could be used in the aeronautic, automotive, power generation,









and chemical processing industries. While specific applications have been identified

based on some form of a HTSMA, no suitable materials have been developed. As is

common in the materials field the development of applications for advanced materials is

slightly ahead of the materials development itself, and such is the case for the

development of high-temperature shape memory alloys.

Integration of SMA actuators into aeronautic turbomachinery would result in

several inherent benefits. Aeronautics clearly emphasizes weight reduction in all stages

of engineering. Reduction in the net weight results in sizable gains in fuel efficiency.

Additionally, SMA actuators decrease the number of subsystems as compared to standard

pneumatic or more common hydraulic and motor-driven actuators, providing further

reductions in weight and cost. Figure 1-1 shows the typical weight to power ratios of the

more common commercial actuators currently used by the aerospace industry.6

Minimizing weight and maximizing power results in a performance index in which SMA

actuators are clearly superior.

The design and development of actively controlled SMA devices requires in-depth

characterization of the mechanical and shape memory specific properties. Past studies

have accounted for compositional effects of transformation temperatures and, in some

cases, load-free recovery, yet there is a complete lack of data required for the application

of shape memory alloys particularly in actuator-related applications. This study attempts

to characterize these properties and correlate them to material composition and

microstructure, which in turn can be used to identify possible areas for further alloy and

process development.










Background

General Shape Memory Alloy Behavior

SMAs are characterized by a set of temperatures at which a crystallographic

structural change begins and ceases. The high-temperature austenite or parent phase is a

high symmetry phase usually ordered while the lower temperature martensite phase is a

lower symmetry structure which forms from the high symmetry parent phase by a

diffusionless shear transformation. The various temperatures at which this

transformation begins and ends on heating and cooling are defined as As, Af, Ms and Mf.

The austenite start temperature, As, is the temperature at which the transformation of the

martensite to austenite phase begins on heating. Af is the temperature at which the

transformation is completed and the material is 100% austenite. The martensite start, Ms,

and martensite finish, Mf, temperatures are the temperatures at which the transformation

occurs on coo mng.2

Figure 1-2 is a classical schematic presented by Wayman which shows a material

property dependent change as a shape memory alloy is cycled through a thermal

hysteresis.l A discontinuity in the material properties arises at the onset of the

transformation on heating or cooling, which is characteristic of all SMA materials.

Throughout the transformation the material exhibits a reversible structural change that

results in a measurable change in material properties. This property may be for example

specific volume, electrical resistivity, modulus, or other structurally dependent property.

The most common test methods for the determination of the transformation temperatures

are thermal methods (DSC and DTA) dilatometric methods, and resistive methods. The

latter two result in similar hysteresis plots as exemplified in Figure 1-2. While this figure









is idealized it is representative of many of the relevant characteristics of thermal shape

memory alloys.

SMA Structural Characteristics

Shape memory alloys exhibit both a thermodynamically and crystallographically

reversible transformation. A crystallographically reversible transformation is most likely

when the interface between the martensite and austenite is essentially coherent and the

parent austenite phase is an ordered compound. The high-temperature austenite phase is

a higher-symmetry structure, usually ordered cubic (B2 structure) as in the case of NiTi

and NiTi modified alloys. The austenite phase transforms without appreciable long range

diffusion into a lower symmetry martensite structure at some lower temperature. Simple

cells schematically illustrating the structures common in NiTi above and below the

transformation temperatures are shown in Figure 1-3.2

The cubic B2 parent phase in NiTi based SMAs transforms to a number of different

martensitic structures. The final structure depends on alloying additions, impurities, and

processing history s. The following transformation reactions have been identified.

B2 tBl9
B2 t Bl9'
B2 t Bl9 t Bl9'
B2 -R -Bl9'

Each reaction is crystallography reversible with the exception of the B2 -R
transformation. The R phase is attained by {100} elongation of~ the B2,~, strutureresutin


in a rhombohedral structure.9 The Bl9 phase is an orthorhombic structure which is

formed from the B2 parent crystal in several steps, which consist of elongation about the

a, b and c axes and shearing of the basal plane in the c direction which is normal to the b

direction. The Bl19' phase is a monoclinic structure formed by additional shearing of the









Bl9 non-basal plane, which is normal to the a direction in the c direction. 10 The Bl9'

phase is the primary shear structure which appears in binary NiTi leading to the shape

memory effect.

The cubic parent phase in NiTi SMA is stable at high temperature (above the Af).

Multiple lattice variants of martensite can form from each parent austenite grain. Each

variant will follow a perfectly reversible path back to the parent austenite phase due to

the ordered nature of the alloy. If this did not occur, the material would undergo a

diffusional transformation in the chemical ordering of the original parent phase' s lattice.

In the Pt and Pd modified NiTi SMAs the Bl9 phase is of primary importance since

alloying additions greater than about 10% results in the primary martensitic structure

switching from the monoclinic Bl9' to the orthorhombic Bl9. 1 The Bl9 and Bl9' are

shear structures of the B2 and therefore exhibit a lower symmetry, with the Bl9' having

the lowest symmetry. 12 The symmetry of the structure is of importance as it is the

underlying factor in the determination of the number of equivalent martensitic variants

which may form from a parent B2 cubic structure. The Bl9 has 12 equivalent variants

that may form from a parent crystal. This translates to 12 different ways to shear the B2

structure in the formation of the Bl9. Each equivalent Bl9 structure may then be sheared

along the non-basal plane (001) in the positive and negative c directions to form a Bl9'

martensite which results in 24 equivalent variants. 13 In general when analyzing shear

structures the number of equivalent variants increases as the symmetry of the shear

structure decreases. This relationship will be examined in further detail as it relates to

the deformation behavior and mechanisms in the shape memory alloys encompassed in

thus study.









The fundamental theory of martensite formation, which applies to both

thermoelastic and thermoplastic transformations, relates a high symmetry parent structure

to a variant of the shear structure. Although this theory is applicable to many shear phase

transformations for descriptive purposes we will focus on the structures shown in

Figure 1-3. Three maj or deformations steps are required to relate to structures by a

purely shear transformation. 14 (1) Primarily there is a Bain distortion which attains its

name historically from the distortion observed in the thermoplastic transformation which

forms the metastable tetragonal martensite. The Bain distortion is simply the elongation

of the parent phase which is shown in Figure 1-3 as the elongations of the a, b and c axis.

(2) Secondly a shear deformation must occur in order to preserve the lattice symmetry

which combined with the Bain distortion forms the undistorted plane or habit plane. (3)

Finally there is a rotation which brings the undistorted plane into the same orientation in

both the parent and shear phase.

The accommodation of arbitrary shearing of the lattice (noted as step 2) is a

decisive factor defining whether a shear transformation is thermoelastic or thermoplastic.

Thermoplastic transformations such as those common in steels are non-reversible and the

maj ority of the transformation shear associated with step 2 is accommodated plastically.

That is, there is the formation of non-reversible defects, such as dislocation motion and

generation. Thermoelastic transformations on the other hand accommodate the maj ority

of the transformation shear elastically in combination with recoverable mechanisms.

There are three main ways the transformation shear may be accommodated, two of

which are reversible. The active shear mechanism depends in part on the mechanical

properties of the austenite and martensite as well as the magnitude of the transformation









shear. In both thermoelastic and thermoplastic transformations a significant amount of

shear is accommodated elastically which is dependent on the yield strength of the

martensite and austenite adjacent to the interface. If the transformation strain results in

an interface stress state which exceeds its local yield strength transformation strains are

accommodated irreversibly by plastic deformation or slip. Finally, if the shear stress at

the interface required to initiate deformation twins is sufficiently low and higher than the

yield strength the transformation strain is accommodated by twin formation.

Twin accommodated strain, which is the keystone of shape memory alloys, forms

during the transformation along the twin planes in the shear martensitee) phase. It is

important to note that the twinning plane is usually a low index plane that is not parallel

to the habit plane. Figure 1-3 is a schematic of the nature of a thermoelastic

transformation interface between parent and shear phases in which the transformation

strain is accommodated by twin formation. The macroscopic shear plane, which separates

the cubic and shear structures is dependent on the structural relationships between the

martensite and austenite which include Bain shear and rotation. The twinning plane

however is based on the symmetry of the shear structure as the deformation twin must

only reorient the structure by a consorted movement of atoms uniformly distributed over

the volume separated by the twinning plane. The transformation shear along the

macroscopic shear plane is partially accommodated by the shear associated with twinning

along the twin planes. In thermoelastic transformations the net transformation shear in

the martensite is accommodated in part by the elastic deformation of the twinned and

untwinned regions and partially by the formation of the deformation twins. If the

deformation twins did not occur the stress due to the transformation strain would exceed










the yield strength of the material and thus be accommodated plastically. Twins that form

during thermoelastic transformations, in order to accommodate the shear along the

macroscopic shear plane, have a twin related crystallographic relationship to one another.

As exemplified in Figure 1-4 the sense of the shear associated with a twin must alternate

between twinned regions. This criterion for the accommodation of the transformation

strain results in the formation of coupled pairs of twins referred to in the

phenomenological theory of martensite formation as correspondent variant pairs.15-20

Although many variant pairs may form from an austenite crystal each variant pair is

equivalent, thus it is possible for the austenite to transform to a single correspondent

variant pair which accommodates the transformation strain by twin formation. This

however does not occur in an un-biased (no external stress) sample. What is observed is

that austenite transforms into a more or less random orientation distribution of variant

pairs. An addition mechanical constraint must be considered to examine the driving force

for the observed distribution of variant pairs. The underlying mechanism driving such a

distribution stems from the minimization of the macroscopic shape of the bulk material. 1

The transformation product of the austenite is in different regions of the crystal

transforms in such a manner that there is no macroscopic shape change. This behavior is

referred to as self accommodation. Self accommodation is an arrangement of martensitic

variants such that the sum of their displacements within the boundary suffers no net

displacement. It is possible to place a self accommodating arrangement within an

austenitic matrix and not induce any macroscopic strains. Self accommodation is a

fundamental characteristic of all thermoelastic transformations as it minimizes interface

stresses assuring interface coherence and elastic accommodation of strains. In other










words, if self accommodation did not occur as the interface progresses the stress in the

interface would continue to rise and quickly surpass its yield strength resulting in plastic

deformation and a loss of coherency. Such is the case in twinned ferritic martensites,

where although the twinning process is reversible, the arrangement of twins is such that

the interface stresses result in plastic deformation and thus a non-recoverable

transformation.

In shape memory alloys the thermoelastic transformation results in a twinned

structure. Figure 1-5 shows a simplified representation of two equivalent monoclinic

variants separated by a twin boundary. Although more strain could be accommodated by

the additional translation associated with the formation of an incoherent twin the

interfacial energy of a coherent twin boundary is on average an order of magnitude lower

than the energy of an incoherent twin thus additional energy is required to form

incoherent twins. 16 As a result the formation of coherent twin interfaces between

martensite variants is thermodynamically favorable and additional energy is required to

form incoherent twins.

Figure 1-6 are TEM micrographs of a binary NiTi shape memory alloy.2 This

figure shows the differences between a twinless martensitic structure (6.a.) and a finely

twinned structure (6.b.). Both structures are monoclinic differing only by the presence of

a fine distribution of deformation twins that form during the phase transformation. The

detwinned structure shown in micrograph 5.a. resulted from deformation by detwinning

of the martensite. Another fundamental aspect of shape memory alloys is deformation of

the martensitic structure through detwinning.









A completely reversible structural transformation requires that the parent-

martensite interface be glissile in the forward and reverse directions. In addition the

thermal hysteresis must be small. Ling and Owen have shown that sessile dislocation

loops and other defects in the matrix facilitate the movement of the interface. 1

Furthermore these sessile defects increase the plastic flow stress of the matrix hence

making the accommodation of strain by slip more difficult. Mechanically the matrix is

effectively strengthened and the energy required to move the parent martensite interface

is lowered. This has been correlated to NiTi bases SMAs as well as other SMA systems.

The density of these defects increases with thermal cycling up to a limit resulting in a

decrease in transformation temperatures with increased thermal cycles. Decreasing

transformation temperatures with thermal cycling has been observed experimentally as

shown in Figure 1-7.i

SMA Mechanical Behavior

Metals that exhibit a thermal shape memory effect deform through twin boundary

motion. Recoverable deformation of the martensite by twinning reactions must occur at

stresses lower than those for non-crystallographically reversible reactions. Non-

crystallographically reversible reactions include dislocation generation and motion.

Structurally, twins are formed in the martensite during the forward reaction separating

equivalent variants. Multiple martensite variant formation is driven by the minimization

of the net transformational stresses. Hence, twins are present in the microstructure after

transformation; therefore, nucleation by an applied shear is not necessary in contrast to

standard deformation twinning. Deformation occurs by the growth of variants most

favorably aligned with the largest principle shear component or Schmidt factor at the

expense of those with the lowest component. This mechanism is commonly referred to









as detwinning. Deformation by this mechanism decreases the number of twins in the

alloy. This is schematically shown in Figure 1-8. The arrow indicates twin planes in this

schematic. An aligned shear results in the twin boundary motion in the direction normal

to the shear and the growth of a corresponding variant. Upon fully detwinning the alloy,

the material theoretically exists as a single variant although the extent of detwinning

depends on crystal structure and the associated number of equivalent variants as well as

the existing variant distribution prior to deformation.

The typical macroscopic mechanical behavior of a shape memory alloy is

represented in Figure 1-9. This figure is a schematic of the general stress strain curves

exhibited in these systems below the Mf temperature. It should be noted that the alloy's

composition, and mechanical and thermal history may change this curve. It is also

possible to have multiple active deformation mechanisms, which will affect the work

hardening rate during the detwinning region of this curve. This figure represents the

ideal case for the shape memory effect.

The initial portion of the stress strain curve is attributed to elastic deformation of

the undeformed martensite. Upon reaching a critical stress, detwinning of the martensite

begins. The detwinning stress is independent of twin density and therefore a region in the

stress strain curve exists in which the stress required to deform the material is

independent of strain. A critical level is reached at the point where favorably oriented

variants are most prevalent in the microstructure and thus reactants of the detwinning

reactions that supply the growth of the favorably oriented twins are consumed. At this

point twins with similar Schmidt factors may impede on each other. The result is an

increase in stress-strain relationship that is attributed to elastically deforming the









detwinned martensite. A second yield point is evident at which the critical stress for slip

is reached. Non-reversible deformation mechanisms are active in this region thus the

strains are not recoverable by shape memory processes.

Figure 1-10 is a series of hypothetical stress-strain curves that graphically represent

the three distinct deformation behaviors exemplified by SMAs and the temperatures at

which they may be active. 19 At temperatures above the At the shape memory alloy is

austenitic and deformation occurs by elastic loading of the austenite followed by slip.

Below the Mf temperature the shape memory alloy is fully martensitic and deformation

occurs by the detwinning mechanisms described above. SMAs demonstrate an

extraordinary superelastic effect which occurs when the material is deformed above the

Ms temperature and below the Md temperature (Figure 1-10). The Md temperature is

defined as the temperature at which mechanical stresses can induce a martensitic

transformation. Subsequent removal of the external stress results in a non diffusional

reversion to the thermodynamically stable parent phase. Elastic strains attainable in NiTi

shape memory alloys are 20X those of carbon spring steels.

Thermoelastic Shear Transformations

The thermodynamics of shape memory alloys and the relevant shear

transformations is a complex subject, which involves a competition between the chemical

and non-chemical driving forces. We have stated that martensite forms from the parent

phase by a purely diffusionless shear transformation. The transformation front progresses

by shear atomic motions and the interface between the martensite and parent phase is

coherent. Structurally the martensite results in a net shape change of each equivalent

variant. As in the case ofNiTi addressed previously a cubic structure transforms to a

monoclinic or orthorhombic. This net shape change results in an accommodation strain.









The local strain around each variant can be accommodated plastically, elastically or as a

mixture of both. This phenomena has been reviewed and the key thermodynamic

parameters identified by Reed and Abbaschian.20

Table I was compiled from publications [1,16,28]. The listed characteristics are

common to SMA systems. The structural and mechanical characteristics already have

been briefly addressed. In addition to and as a result of these structural characteristics a

specific set of thermodynamic properties arise.

Table 1-1 Several characteristics common to metallic thermal SMAs.

Structural characteristics of SMAsl
Ordered parent -> ordered martensite
Martensitic transformation is thermoelastic
Martensite is crystallographically reversible
Thermodynamic characteristics defined by Dunne and Wayman21
Small chemical driving force at Ms
Small transformational volume and shape change
High flow stress parent matrix
Additional mechanical characteristics defined by Ling and Owen 16
Parent-martensite interface must be glissile in both transformation directions
Premartensite elastic softening


A fundamental condition for the shape memory effect is that the transformation

must occur reversibly. Accommodation of the transformational strains adjacent to the

interface could be plastic, elastic, or a mixture of both. In the case where the majority of

the strain is accommodated elastically the interface is able to move in both directions

referred to as a thermoelastic transformation. As a result the chemical driving force

required to drive the reaction is small. This is the case in alloys exhibiting the shape

memory effect, which are more precisely defined as thermoelastic transformations.

In thermoplastic transformations the transformation strain is accommodated

plastically due to a low flow stress in the parent phase and a large transformational strain.









In this case, the transformations on the forward and reverse direction occur at much

higher chemical driving forces through the nucleation and rapid growth of the martensite.

Typically individual shear plates are nucleated at defects and grow irreversibly.

Subsequent thermal cycles result in new plates nucleating rather than reversible interface

movement. A classic instance of such a case is the martensitic transformation of carbon

steels where the thermal hysteresis is large and most of the transformational strains are

accommodated plastically.

Ortin and Planes elegantly treated the thermodynamics of thermoelastic effects and

systematically defined conditions for a thermoelastic energy balance.22 Thermoelastic

transformations are driven by the chemical free energy. At equilibrium it would be

expected that the transformation occurs when the chemical free energy of the parent

phase is a small amount larger than that of the shear phase. This however has been

shown not to be the case. In actuality, the chemical driving force in thermoelastic

transformations are opposed by non-chemical forces, thus the equilibrium transformation

occurs when these forces are nearly equal. Following the notation and approach

presented by Ortin and planes, thermodynamic equilibrium is represented by the

following equation.


AGp..m = -AGch + AGnch = 0


G,..m is the molar free energy of transformation, Gch is the molar chemical free energy

and Gnch is the molar non-chemical energy. At equilibrium the molar free energy of

transformation is equal to zero thus the chemical contributions are equal to the non-

chemical contributions.









The non-chemical contributions consist of several factors, the most prominent are

the elastic free energy and work done against frictional forces.


A~ch = AGel+ Eriction
AGch = ch
AGch = AGel+ Eriction


The elastic and frictional terms consist of several components. It was stated that a

condition for thermoelastic transformations is that the maj ority of the transformational

strains are accommodated for elastically. Therefore adj acent to each interface we have a

Einite amount of stored elastic energy. In addition there exists an interfacial energy

associated with the parent martensite interface as well as the interfaces between variants

(twin boundaries). Both contributions are reversible therefore they have been grouped

into the elastic term even though the interfacial energy is not truly an elastic contribution.

This is in line with the convention set forth by Ortin and Planes.

Frictional energy losses are non-reversible losses primarily due to interface

movement. This term may be treated as irreversible work done on the system. Three

significant parts have been identified by Olsen and Cohen.23 24 These include 1)

frictional stresses required to move interfaces 2) irreversible free energy related to defects

induced during the transformation 3) frictional stresses required to move interfaces.

It is important to note that if all or most of the accommodation occurs plastically

the elastic term will be very small and the frictional energy loss term will be the main

opposing energy to the chemical driving force. As a result a large non-reversible

hysteresis will be evident as is the case in carbon steels as mentioned earlier.

The underlying mechanisms and thermodynamics which control general behavior

of shape memory alloys have been review. Principally the structural relationships









between the parent in shear phases result in thermomechanical properties unique to shape

memory alloys. Additionally, driving forces and the relating thermodynamic for

thermoelastic transformations have been briefly discussed and compared to thermoplastic

transformations. A persistent problem that will be discussed in subsequent chapters

points out that plastic accommodations by non-recoverable processes hinder the sought

after characteristics of shape memory alloys, principally the materials ability to do work.

Essentially under certain conditions these shape memory alloys start to behave more like

thermoplastic materials where a significant portion of the transformation strain is

accommodated plastically. The combination of structure, mechanical properties and

thermodynamics are used as tools to understand deformation and propose possible

deformation mechanisms while identify possible target area for future alloy

improvement.

10000 .

Shape
Memory
looo Metals





10 -



10


0.01 0.1 1 10 100

Weight (Kg)

Figure 1-1 Power-to-weight ratio versus weight diagram for common actuator types
currently used in aeronautics.6




















































tempe~fratr --


M.l magtyder~ earl ana
Me IPcmairine Enbl Oempehrhe
a. rtori ofA reeetransfonion
o ~f mortite
Agi fniah of 0Wi alb gno
of mordenaik


Figure 1 -2 Idealized plot of a property change vs. temperature.l


Eletrical







Vorlumea Change
e~te

















(a) Parent B2


8 -rr
O Ni


(b Martensite B19


b ems| {I Oj,,


b m* (|[1 1 0],


Figure 1-3 Structure of the parent phase austenitee) and shear phases (Bl19 and Bl19'
martensite).2






















Figure 1-4 Thermoelastic transformation and twin accommodated transformation strain.14


Iwla5w


Figure 1-5 Two-dimensional lattice schematic of monoclinic structures


(a) (b)


Figure 1-6 TEM micrographs of twinned and untwined monoclinic martensite.3





HTT360
* HTT420
I H~TT460
+ HTT520
*HTT580


WI


I _


__


Illrlllllllll.lll


r


I


I


0 10 20 30 40 50 60 70 1100


400


300


200


No. of cycles



Figure 1-7 Effects of thermal cycling through the hysteresis on the transformation
temperatures of several NiTi based shape memory alloys.l


Deform


Figure 1-8 Deformation and shape recovery by detwinning (twins marked with
arrTows).1





































CCldTg ,#, Ck.


u a~bhs#Cd dreformation of
derwinned manreasit

onset o

detwinning asser o* D(sl









martensitic state.3


Figure 1-10 Stress strain behavior showing the three main deformation regimes active in
SMAs.31


Ti: Amtmit


Tg Paumebatio


Tn: C1rtemits


~ln --














CHAPTER 2
MATERIALS PROCESSING AND PROCEDURES

Melting Procedures

Arc Melting

The experimental portion of the alloy development phase of this study was initiated

with a cursory examination of a series of over twenty alloys along and on either side of a

line of constant stoichiometry (constant Ti atomic fraction of 50%) between TiNi and

TiPt in order to survey the basic properties of the Ni-Ti-Pt system in a region where the

potential for shape memory behavior is likely. The experimental alloys were produced

by non-consumable-arc melting of high purity starting components (99.95% purity Ti,

99.995% purity Pt, 99.98% purity Ni) using a water-cooled copper crucible in a high-

purity argon atmosphere. Due to the substantial density differences and melting points

between the starting materials, it was difficult to melt the platinum completely in one

step. Consequently, the buttons were turned over and remelted 4-6 times in an attempt to

insure homogeneity.

Arc Melt Machining

Sectioning for thermal, microstructural, and hardness measurements was performed

by wire EDM (electrical discharge machining). For example, 10mm length by 5 mm

diameter cylinders were EDM' d for thermal analysis from the center of the arc melted

buttons. Planar specimens for microstructural and hardness were fabricated by

transversely sectioning the arc-melted button followed by pressure mounting into

phenolic bound thermosetting polymers mounts. The exposed planar section was then









ground using SiC papers and polished using diamond suspensions using standard

metallographic preparation techniques, yielding a highly polished, low residual stress

surface.

Induction Melting

The experimental alloy of a target composition of Nil9.5Pd30Tiso.5 was produced by

vacuum, induction melting of high purity starting components (99.95 Ti, 99.995Pd,

99.98Ni) in a graphite crucible, which is tilt poured into a 25.4 mm diameter by 102 mm

in length copper mold. The mass of each ingot is approximately 450 grams including the

hot top, which feeds the casting during the shrinking associated with solidification

(Figure 2-1). Induction melting was chosen over arc melting in order to circumvent

problems inherent to arc melting materials with large density difference as in the case of

Ti 4.506 g/cm3 and Pd 12.023 g/cm3 Or Ni.8.908 g/cm3. Induction melting also induces a

mixing action in the melt which assures a homogenous melt. Graphite does introduce a

limited amount (approximately 0.5 at.%) of carbon during the melting process resulting

in the formation of carbides with the excess off-stoichiometry Ti.

Homogenization

Each induction melted ingot or arc melted button was simultaneously sealed in a

vacuum furnace and homogenized at 1050 oC for 72 h. This was followed by a furnace

cool as shown graphically in step 1 of

Figure 2-2. This figure summarized the thermal history of the extruded material.

The arc melted material is homogenized (step 1) and as it is not extruded or mechanically

machined a stress relief heat treatment is not required and stabilization of the

transformation temperatures is accomplished by multiple thermal cycles during hysteresis

measurements. The homogenization heat treatment is employed in order to remove fine









atomic segregation and insure full reaction to form the ordered structures with a

minimum amount of ordering defect and resulting point defects. Uniformity in the hot

zone is assured though multiple thermocouples and independent zone control though

several calibrated temperature controllers and gradual temperature ramp rates.

Extrusion

After homogenization the ingots were individually sealed in steel extrusion cans.

The ingot was placed into the extrusion can and a vacuum cap was sealed by tungsten

inert welding. The vacuum cap specifies a cylindrical cap with a fitted mild steel vacuum

tube. Through this tube the sealed cavity was evacuated following by a crimping and

spot welding operation. At this point the canned ingot is ready for extrusion. Following

the canning operation the sealed ingots were extruded with a 7:1 reduction ratio at 9000C

in a hydraulic press. A schematic of the extrusion operation is given in Figure 2-3. Prior

work at the NASA Glenn Research Center has demonstrated the feasibility of this

technique and determined these are optimal conditions for the thermo-mechanical

processing of comparable HTSMAs.

The extrusions were X-rayed in order to have an accurate determination of the

location of SMA core's start and finish position in the steel covered rod and the excess

ends of the extrusion removed by abrasive cutting. The extrusion rods were then cut into

various lengths.

Compressive samples were fabricated by wire EDM methods and centerless

grinding, yielding 5 mm diameter by 10 long samples (Figure 2-4). Additionally 1 X 4

mm rectangular samples, x mm in length were sectioned by wire EDM from the core of

the extruded bars for resistivity measurements. Finally, a section of the extrusion is










removed for ICPS (spell out) by which the bulk alloy composition was measured and

documented.

Tensile samples were fabricated by turning, using a computer controlled lathe

(C&C). The canned extrusion was sectioned by EDM into 50.80 mm sections, which

were individually center punched. The cut sections were loaded with the extrusion can

still lining the sample into the C&C lathe. The entire machining procedure was

performed in a cutting mode which completely excludes costly grinding operations.

Successive cutting passes were made to yield threaded 17.4 mm long by 3.81 mm

diameter gauge sections (Figure 2-4). A summary of all the sample preparation steps and

methods are graphically represented in the flow diagram shown in Figure 2-5.

Stress Relief Heat Treatment

To complete the sample preparation phase, all samples are given a stress relief heat

treatment at the At plus 200 oC for 1 hour followed by a furnace cool. The function of

this heat treatment is to relieve any residual stresses on the surface of the samples

resulting from the extensive machining operations during the fabrication stage.

Structurally this heat treatment is performed in the austenite phase thus alloying for

recovery in the high symmetry phase. This heat treatment was optimized based

temperature resistivity measurements.

Characterization Procedures

Dynamic Modulus

A dynamic modulus testing apparatus facilitated the determination of the modulus

of elasticity as a function of temperature for each phase. A 34 mm bar was machined and

fixed with an electrodynamic vibrator at one end a piezoelectric transducer at the other

end. By locating the resonance frequency as a function of temperature it was possible to









calculate the dynamic modulus as a function of temperature during heating. The sample

was heated at 10 OC/min in a convection furnace to a maximum temperature of 800 oC.

A detailed description of the melting, heat treatments, extrusion, and Einal

machining of thermo-mechanical Nil9.5Pd30Tiso.5 samples is outlined here. As these

alloys behave in a similar manner and in order to allow for direct comparison between the

Pt and Pd modified alloys, the processing scheme was intentionally followed as closely

for both materials. The Pt modified alloy Ni24.5Pt25Tiso.5 selected for this study was

processed in parallel following a comparable processing scheme as the Pd modified alloy

described above and as outlined in Figure 2-5. The only significant differences is that

since the austenite has a lower flow stress in the Pd modified alloy the extrusion pressure

is slightly lower. All melting and processing steps are comparable.

Compositional Analysis

The bulk alloy compositions were determined by inductively coupled plasma

spectroscopy and the interstitial impurity concentrations were determined using standard

LECO O/N and C/S determinators. For this analysis, samples were prepared by first

cutting buttons into smaller sections of approximately 100mg followed by a petroleum

ether rinse in order to remove any surface contamination. These sections were then air

dried, reweighed and placed in a Teflon digestion vessel. 3 mL HC1, 1 mL HNO3, and 1

mL HF (all concentrated and trace metal grade) were added to the vessel which was then

placed in a block digester at 100-130 OC until dissolution was complete (generally 1-2

hours). Finally, the dissolved sample was transferred to a 100 mL volumetric flask for

ICP analysis.









The ICP solutions were analyzed for alloying elements and impurities using the

Varian Vista-Pro Inductively Coupled Plasma (ICP) Emission Spectrometer. The

composition of the ingots was measured by inductively coupled plasma spectroscopy post

heat treatment and thermo-mechanical processing to insure measurement of any

additional contamination that may have been picked up during these procedures. A

schematic of the instrument is shown in

Figure 2-6.32 In this process, the solution is inj ected by capillary action into an

argon plasma. The sample is ionized and excited by the plasma, which results in each

element emitting a characteristic wavelength. The emitted photons are allowed to pass

into the system though the entrance slit and are diffracted by a fixed grating Echelle

polychrometer, and finally the intensity of each wavelength is measured using a CCD

(charge coupled device) detector. The detector is capable of simultaneously detecting up

to 73 different elements in the range of 167 785 nm.

Several emission lines are selected for quantitative analysis of each element by

extrapolation of measurements using calibrated NIST-traceable solution standards. The

emission intensity at each wavelength is proportional to the concentration of the element

present in the solution, which is determined from extrapolation of the calibration curves.

The mean intensity over several peaks is used in calculating the reported concentration of

each element.

Nitrogen, Oxygen, Carbon and Sulfur Analysis

Again, samples of approximately 100mg were analyzed for oxygen and nitrogen

using the Leco TC-436 Nitrogen/Oxygen Determinator. In this technique, a graphite

crucible is baked out by a high current in order to expel any gas trapped within the

crucible. The sample is then placed into the crucible and the chamber is evacuated.









Resistive heating raises the crucible and sample temperature thus causing the sample to

expel any nitrogen and oxygen. A gas solid reaction occurs between the oxygen and the

carbon crucible which yields CO. Subsequently the CO is passed through a rare earth

copper oxide reactant producing CO2 which is measured by an IR cell. The CO2 is

separated from the remaining gas by ascarite absorption which is a powerful CO2

absorber. Finally, nitrogen is measured by a thermal conductivity cell which measures

the temperature difference between a heat source and a thermocouple which is protected

from radiative heating.

Samples parallel to those used for O/N analysis were analyzed for carbon and

sulfur using the Leco CS-444LS Carbon/Sulfur Determinator by the combustion

instrumental method. In this method, the sample is combusted in the presence of oxygen

yielding carbon dioxide and sulfur dioxide. The amounts of CO2 and SO2 are determined

by measuring the absorption of specific IR wavelengths, which are proportional to the

partial pressures of these gasses.

Thermal Analysis

Transformation temperatures were determined by differential scanning calorimetry

(DSC) for low to intermediate temperature analyses and differential thermal analysis

(DTA) was used for intermediate to high temperature transformations. DSC Data

analysis was performed using the TA Universal Analysis 2000 Software package. Each

alloy was cycled through two full thermal hysteresis cycles (approximately Af +100 oC to

Ms-100 oC) assuring the reproducibility of the thermoelastic transformations. It was

experimentally determined that in these arc melted and homogenized samples, the

transformation temperatures stabilize after two cycles and therefore remain constant with









subsequent cycles. The acquired data is formatted as the temperature dependent heat

flow as exemplified in Figure 2-7. A combination of techniques was employed in order

to circumvent temperature limitations of each instrument by maintaining the

measurements within the designed temperature intervals thus assuring high precision

measurements. For both techniques, a heating and cooling rate of 10 oC per min was

used.

DSC and DTA are common techniques used in determining the thermal properties

of materials with a high degree of precision. Although both are differential methods, the

DSC in contrast to the DTA calculates heat flow directly during heating, cooling or

isothermal holds by measuring the amount of energy required to maintain the specified

temperature, while the DTA measures differences in temperature between a sample and a

reference while heating both at a constant rate. Therefore, the main differences between

the data acquired from a DTA and that of the DSC is that the determination of heat flow

requires calculations from standards of known heat capacities similar to the unknown

sample and the sample heating rate is not as precisely controlled.

As a consequence of these differences, it is generally considered difficult to directly

compare the thermodynamic properties measured without stringent calibration. The

application of the DSC and DTA in the current SMA study is solely for the determination

of the transformation temperatures without focusing on the heat capacities or the

magnitude of the transformation enthalpy.

Characterization of the transformation temperatures through DSC and DTA

measurements were made by the extrapolated onset method by which the transformation

temperatures are recorded as the intersection of the base line and the best fit of the linear










portions of the increasing and decreasing regions of the exothermic or endothermic peaks

in the lambda type curve. 33 A characteristic of the extrapolated onset method is that the

relative amplitudes of the heats of transformation as a function of temperature determines

the transformation start and stop temperatures independent of the absolute or actual

transformation enthalpy, facilitating the comparison of transformation temperature

measurements made on a DSC and DTA.

Thermoelastic transformations, which are characteristic of metallic shape memory

alloys, involve displacive shear transformations. This type of transformation exhibits a

strong temperature and stress dependence and relatively fast transformation kinetics. Fast

transformation kinetics result in the reaction reaching its temperature and stress-

dependent equilibrium rapidly in contrast to diffusional transformations, which require

extended times. As a result, thermoelastic transformation temperatures are not highly

time or more specifically heating rate dependent unlike diffusional transformations.

Consequently, although the heating or cooling rate of a thermoelastic sample might vary

slightly in a DTA where only the hot zone's temperature is precisely controlled, the

kinetics of the reaction are so fast that the dependence of the transformation temperature

on these slight variations are insignificant. Therefore it is feasible to compare the

transformation temperature measurements of thermoelastic transformations in a DSC and

DTA.

Microstructural and Semi-Quantitative Compositional Analysis

The sectioned and polished alloys were examined using a JEOL 6400 scanning

electron microscope (SEM) using backscattered electron (BSE) mode. An annular

detector was used in order to maximize the signal. The sample surfaces were kept normal

to the 15 KeV beam and parallel to the detector' s exposed face. This configuration










maximizes phase contrast and minimizes any topographical effects by measuring the sum

of the back scattered electrons around a 3600 ring perpendicular to the electron beam and

normal to the sample's surface. All comparable imaging was done at the same

magnification in order to allow for simple comparisons. Semi-quantitative ZAF

corrected EDS analysis of the various phases was performed using the NiKa, TiKa, and

PtMa lines.

Dilatometry Measurements

The transformation temperatures were also determined by measuring the strain-

temperature response of the material under essentially zero load (stress free condition)

using a differential thermal dilatometer. This instrument compares the measured change

in length of a test specimen to that of a standard as a function of temperature thus

allowing for correction for any thermal expansion in the apparatus itself. In this

measurement, the furnace heating rate is controlled at 10 OC per minute and the sample

temperature is measured directly by thermocouple contact. A differential thermal

dilatometer is used to measure the transformation strain in the longitudinal direction.

Several cycles through the transformation hysteresis are performed until the

transformation strain vs. temperature relationship stabilizes. This technique is used to

determine the transformation strains associated with the SMA reaction as well as the

thermal expansion coefficient of the parent and austenite phases.

Resistivity Measurements

Sample instrumentation

1 X 4 X 20 mm rectangular samples were sectioned by wire EDM from the

extruded bars described in a previous section and prepared for electrical resistivity

measurements. The surface was cleaned by light polishing prior to the stress relief heat









treatment. The sample was instrumented with a four point probe and a K-type

thermocouple as shown schematically in Figure 2-8. Two pairs of Pt or Ni drawn wires

were spot welded to the sample functioning as the voltage sensing and current excitation

leads. The current excitation leads were spot welded near the ends of the sample while

the voltage sensing leads were spot welded further in from the ends between the two

current excitation leads. The sample dimensions (cross-sectional area) and distance

between the voltage sensing leads was measured using a vernier caliper.

Resistivity apparatus

Test specimens were heated through a thermal hysteresis in an ATS 3200 series

split tube furnace at 10 OC/min followed by a subsequent furnace cool in an ambient

atmosphere. Furnace control was achieved via a Eurotherm programmable PID

(proportional integrating differentiating) temperature controller. Thermal cycling though

a thermal hysteresis was achieved by running a 3 leg program consisting of a ramp of

100C/min to the final temperature followed by a step command which tumns off the power

to the furnace until the minimum specified temperature was attained and finally an end

command which signals the end of a thermal cycle. The independent programmable

control loop in the resistivity apparatus is the call for power from the furnace controller.

Alternatively the resistivity apparatus is capable of sending control commands to the

furnace controller allowing for external furnace control and integration with moving hot

zone cyclic furnaces.

The stress-dependent transformation temperatures and resulting electrical

properties of a material were determined by high-resolution resistivity measurements. A

virtual digital instrument that acquires high-resolution, real time, temperature versus










resistance profies of materials was developed for this purpose. Additional integration of

this instrument with an MTS uniaxial testing apparatus facilitated the direct measurement

of stress-induced transformations and the effects of external stresses on transformation

temperatures. Two operational modes, level crossing acquisition or timed acquisition,

depending on the function of the measurement (cyclic thermal measurements or stress,

strain, resistivity profiles) were employed. This instrument was designed to be automated

such that, once configured, it would allow for the unattended measurement of many

cycles.

The components of the measurement apparatus include a PC, National Instrument

DAQ (data acquisition card) and a NI SCXI (signal conditioning) chassis fitted with an

analog signal amplification and filtering card. Current excitation was supplied by an

Agilent power supply with internal shunt resistors while the calibrated National

Instruments DAQ system and analog signal conditioner were used to measure voltage

signals and linearize the thermocouple. The digitized signals were used to calculate real

time resistance values for each specimen, which in turn allows for the calculation of the

samples resistivity, based on sample dimensions. In addition, external calibrated devices

have been integrated facilitating the calibration of the apparatus and the documentation of

calibration prior to each measurement. In principle, it was required by design criteria that

the system measure accurate absolute values of resistivity as well as resolve

transformation temperatures thus necessitating external calibrations. Finally, external

digital and analog channels have been programmed into the instrument so that it could

function as the controller for a cyclic furnace or comparable device.










Due to the high conductivity of the SMA alloys, the resulting voltages measured

across the sample were on the order of fractions of a mV. Signals of this magnitude are

very susceptible to noise thus necessitating several signal conditioning processes.

Hardware and software low pass filters have been employed in this system in order to

allow for measurements to take place while the sample was under an inductive load as

well as filtering out cyclic noises above 4 Hz which may enter the signal via the extensive

connecting wires or at the sample itself. Figure 2-9 shows the in-situ filter response for a

sample that was heated by an inductive field. The signal in this type of instrument may

be separated into two main components. The primary component is the DC signal of

interest which results from the interaction of the material with the excitation source and a

superimposed AC component which is caused by induction heating of other external

sources. The lower blue graph shows the unfiltered signal as measured by a documenting

oscilloscope while the upper curve shows the filtered response which demonstrated the

efficiency of the filter system at isolating the DC component of the signal..

A secondary benefit of using a powerful precession external power supply is the

capability of resistively heating samples by driving high current densities along their

length. Such an experiment may be used to monitor the power requirements of SMAs

and SMA actuators. Routines for monitoring these parameters have been written into the

instrument. The processed data is sample temperature, power, current, voltage across the

sample and resistance. Again, these are all monitored real time so that this data can be

used to characterize the power requirements of SMAs in several geometries. It is

possible to use this routine to characterize model and real SMA actuators.









As mentioned previously, the instrument was designed to be automated and acquire

data in two modes (level crossing acquisition or timed acquisition). The level acquisition

mode is used for the generation of resistivity vs. temperature profies. Level acquisition

refers to the monitoring of a trigger signal and acquiring a new set of data points or data

obj ect when the trigger is crossed. In the case of the resistivity apparatus, the

temperature signal is defined as the trigger which implies that when the absolute value of

the difference between the transient or trigger temperature and the previous trigger value,

is greater than a specified amount a new data point (temperature and resistance) is

recorded. Additionally, detection of a complete thermal hysteresis is required in order to

allow the separation of data into groups forming a complete hysteresis defined as a heat

and cool cycle. This again was accomplished by level crossing triggers where the

algorithm finds a maximum followed by a minimum temperature before triggering a new

hysteresis command, which records the data and clears the dynamic memory. Timed

acquisition implemented in the integration of stress, strain, and resistivity measurements

records data obj ects at predefined clock intervals.

Post processing measurements of the transformation start and finish temperatures

for the forward and reverse shape memory reactions was also automated, thus facilitating

the measurement of many cycles in a typical high cycle test. The temperature

dependence of the resistivity of the material determined under zero load conditions is

shown in Figure 2-10 as a representation of the method for determining transformation

temperatures via resistive measurements. The transformation temperatures are

determined from this data by the construction of linear, polynomial curve best fits

through the low temperature, intermediate temperature, and high temperature portions of









the heating and cooling curves, respectively. In the intermediate temperature range

where a mixture of the parent and martensite phases exist, the best linear fit is found by

scanning a window of a specified temperature width through the entire intermediate

temperature range. An array of linear best fits and associated R values is constructed.

The series of linear regressions are then ranked by R values and the best fitting regression

is used. Through the utilization of the equations for these best fits, start and Einish

temperatures were determined by interrogating the intersection points. Figure 2-10

exemplifies why resistance measurements are a prominent characterization technique in

the study of shape memory alloys as there is a substantial difference in the electrical

resistivity between the parent B2 and martensite phases. A summary of the complete

algorithm is summarized in the data flow diagram shown in Figure 2-13.

The timed acquisition mode was employed when a continuous acquisition of data

was required. There are two main functions for this mode. Timed acquisition may be

used to optimize heat treatment times and temperatures as a result of the effects of

internal stresses, dislocation structure, grain size and precipitate-matrix interactions and

their respective kinetics. In the current study, the timed acquisition mode was developed

for integration of thermo-mechanical testing methods with in-situ resistivity

measurements. For this purpose, an external communication channel was developed,

which continuously sends a signal to the MTS servo hydraulic controller proportional to

the measured resistivity values thus allowing for a resistivity record on the same system

that is acquiring thermo-mechanical data. This allows for a point by point correlation

between the measurements ultimately leading to the simultaneous documentation of

resistivity, stress, strain, and temperature.










Experiments were conducted on a Pt wire (Figure 2-11) and a NIST steel resistivity

(Figure 2-12) standard34 in Order to determine the accuracy of the instrument and

calibration. The resistivity vs. temperature relationships for both materials are shown in

the figures below. For the platinum wire, the resistivity vs. temperature relationship

should be linear in the temperature range of RT to 900 OC. The measurement shown in

Figure 2-1 1 which confirms the linearity of this relationship indicating the accuracy of

the resistivity apparatus in both heating and cooling modes. The NIST steel resistivity

standard was measured in order to verify the calibration of the absolute values of

resistivity as a function of temperature. These results are shown in Figure 2-12 and

confirm a high degree of accuracy between these resistivity measurements and the

measurements made by NIST. Figure 2-12 compares NIST measurements with our

results via level crossing acquisition in a sample configured with a 4 point probe and spot

welded thermocouple. A low and high range exist in analog amplifiers which allow for

higher precision measurements in lower resistivity samples. Figure 2-12 includes

measurements for the low and high range both of which are within the margin of error

specified by NIST for the resistivity standard.

Thermomechanical Testing

Thermomechanical instrumentation

Mechanical testing was performed on an MTS servo-hydraulic test frame equipped

with an MTS 484 controller and MTS software. MTS 646.10B hydraulic collet grips

with a modified 680 LCF grip set were used to grip the threaded specimens. A 20 kip

load cell was used and strain measurements in tension were taken with an MTS Model

632.51B-04 extensometer using a 12.7 mm gage length. This extensometer is equipped

with 85mm long quartz probes with a v-chisel edge having a maximum range of









+20/-10% strain. Strain measurements during compression testing were acquired using a

laser extensometer. Specimens were induction heated using an Ameritherm Novastar 7.5

power supply.

Resistivity measurements were integrated with the uniaxial mechanical tests,

facilitating measurement of transformation temperature, determination of the matrix

phase as well as determine if a phase change occurred during the test while

simultaneously determining mechanical properties. Additionally this test technique was

used to verify the materials phase fractions. The tensile or compressive sample was

instrumented with a four point probe and a K-type thermocouple. A four point probe

configuration was again chosen in order to eliminate the effects of contact resistance.

Nickel wires were spot welded to the sample functioning as the voltage sensing leads.

Current excitation was supplied through the hot grips. This instrument has been

developed at the NASA Glenn Research Center advanced metallics branch. Figure 2-14

with key components labeled shows this working configuration on a Materials Testing

Systems (MTS) tensile frame fitted with high temperature hot grips and induction

heating.

K thermocouples were spot welded to the sample. Temperature gradients across

the gauge were to within +/-0.5% of the test temperature by calibration on a control

sample fitted with three thermocouples while actual test samples contain a single

centrally located thermocouple. A problem inherent to spot welding is the formation of

stress raisers during the rapid melting and solidiaication of a narrow region adj acent to the

spot weld base and the wire (weld nugget). The radius of the nugget is a critical factor in

determining the stress concentration at the weld and impacts the fracture stress and strain









of a sample. During a spot weld a predefined amount of energy is stored in a capacitive

system and subsequently discharged through a welding probe following the path of least

resistance. As the discharge passes from the wire to the sample or base metal a small

region melts first. This event leads to the contact resistance in this very narrow region

dropping significantly which results in the remaining discharge to focus through this

path. The result is a very sharp defect on the sample's surface. This problem was

minimized by using a multistep spot-welding procedure. Initially multiple low energy

spot welds are made at closely spaced distinct locations, which assures numerous wide

wire to sample contact points. Although these wide contact points are not mechanically

strong enough to withstand the stress during sample handling, loading, and deformation

they provide multiple wide low resistance interfaces. Step two consists of a second pulse

that is approximately 10X greater in discharge energy than those employed in step one.

The secondary pulse forms a strong weld nugget with a wide radius thus reducing the

stress concentration at the weld.

Uniaxial isothermal mechanical tests

Tensile specimens were strained to failure in strain control at a rate of 1 x 10-4 SOC-1

If a specimen reached the 20% limit of the extensometer, the sample was unloaded under

strain rate control until reaching a ON trigger after which the control mode was switched

to load control. The specimen was then allowed to cool and then unloaded. Compression

tests were run in displacement rate control at approximately the same strain rate. Use of

displacement control was necessary due to limitations in the maximum scan rate of the

extensometer and the acquisition rate required by the MTS controller for proper and safe

PID (proportional integrating differentiating) control. Analysis of the stress strain curves









and determination of the proportional limit was facilitated by proprietary NASA Glenn

Research Center Advanced Structures Division software.

Load free strain recovery tests

In the unconstrained or load free strain recovery tests, tensile specimens were

deformed in strain rate control. The maximum strain level trigger was set, which upon

crossing reverses the strain rate to a compressive 1 x 10-4 SOC-1 until the sample is fully

unloaded. When the load reached O N the controller was switched to load control. Load

was held at 0 N while the specimens were thermally cycled to a temperature in the range

of 400 oC. Heating rates were maintained at 10 oC/min after which the samples were

allowed to air cool to well below the Mf before further loading. The recovery rate was

determined by monitoring the sample's strain during the thermal cycling.

Load bias test

Load bias testing measures an SMA' s ability to perform work. This is

accomplished by measuring strain under a constant load. Three modes of load bias test

were employed in this study. The primary tensile load bias tests were run in a series of

progressive loads on the same sample. Specimens were deformed in strain rate control

(of 1 x 10-4 SOC-1) near room temperature to the predefined holding load. At this point the

controller was switched to load control holding a constant load. Specimens were finally

thermally cycled twice from room temperature to about 100 oC above the austenite finish

temperature. Heating rates were maintained at 10 oC/min followed by air cooling. An

auxiliary fan was turned on after the sample temperature dropped below the Mf,

providing additional cooling. In this test specimens were unloaded at near room

temperature and then strained again to the next higher load level. This procedure was

repeated for each load. The work output was calculated by measuring the resultant










change in strain during the martensite-to-austenite transformation during the second

heating cycle and multiplying by the applied stress. Additionally a series of tests were

run on samples with an identical procedure except that the samples were unloaded hot

while austenitic or above the Af. Compression load bias tests were run in a parallel

procedure in displacement rate control rather than strain rate control with near room

temperature unloading.

In summary several test modes are utilized each targeting the measurement of

specific material and shape memory properties. Resistivity measurements targeted

determination and stability of the no load transformation temperatures as well as quantify

the temperature dependence of the resistivity of the austenite and martensite phases.

Isothermal uniaxial tests combined with data from dynamic modulus tests were used to

determine the baseline mechanical properties of each phase as well as the mechanical

behavior of the alloys near the transformation temperatures. Furthermore, load free

recovery experiments measured the effectiveness of the alloys in recovering elastic and

plastic strains while the load bias testing was used to determine the alloys specific work

output. Finally the cool at load test measured the stress dependence of the transformation

temperatures as well as the transformational strains under load.


Figure 2-1 Induction melted Nil19.5Pd30Tiso.5 cast ingot with attached hot top on a quarter
inch grid.











72 hours

-- --- xtrusion

1l hour




----------------------- L ~--- ------JL



time


1050oC
900 oC






RT


Figure 2-2 Heat treatment and processing temperature schedule.


Shape memory alloy |
SMA start position


Figure 2-3 Hot extrusion press schematic





















Figure 2-4 Uniaxial sample (A) 5 X 10 mm compression sample (B) Threaded 17.4 mm
long by 3.81 mm diameter gauge sample.



Chemical cleaning
(etching) Ti and Ni






Precision
measurement of
starting materials


Vacuum induction
melting of starting
materials





homogenized in
vacuum at 1050 oC
for 72 h






Ingot sealed in a
vacuumed mild
steel extrusion can


Canned extrusion
at 90C b grahite Inductiv ym coupled Extrumsins X- a ed
ram at a reduction spcrsoylength~
ratio of 7:1


Figure 2-5 Processing flow diagram ofDSC, compression, and tensile samples.































Figure 2-6 The ICP using an Echelle type polychrometer32


Figure 2-7 Example of a DTA scan showing the exothermic and endothermic peaks
characteristic of thermoelastic shape memory alloys











Cunrent Excitation Solure


Figure 2-8 Four-point probe resistivity configuration


Figure 2-9 Raw (blue) and conditioned (yellow) voltage signals for resistivity
measurements during inductive heating


*/~I1/Y\~1~E~*rlhSu~~


















y = -0.025x + 176.88







y = 0.1245x + 115.26


100


200 300 400
Te mpe ratu re (C)


500


600


Figure 2-10 Resistivity vs. temperature profile with regression analysis












,0.01


700 800 900 1000


Figure 2-11 High conductivity Pt wire resistivity vs. temperature relationship
demonstrating the repeatability of the during heating and cooling


0 100 200 300 400 500 600
Temperature (C)









































--NIST measurement

-*- Standard test low range

Standard test high range


0 100 200 300 400 500
Temperature (C)


600 700 800 900


Figure 2-12 NIST (resistivity standard) resistivity vs. temperature profile comparison of
NIST measurements and the measurements by the resistivity apparatus.

















Voltrage
across
sample
measured


Excitation
current
measured


Temperature
measured


Signals amplified and filtered


Filtered
voltage
current
measured


Filtered
excitation
current


Resistivity calculated


Figure 2-13 Resistivity apparatus data flow diagram.





















Hydraulic Co~ll't Grip








Sample Cooling Fan


Hot~ Grip Ram


Induction Coill

Samlple




La~ser Miromneter


Current excitation lead~


Voltage:


Figure 2-14 Materials Testing Systems (MTS) tensile frame fitted with high-temperature
hot grips and induction heating configured for compressive testing.















CHAPTER 3
ALLOY DEVELOPMENT

The alloy development section of this study consists of two parts addressing the

selection of the Pt and Pd modified NiTi alloys studied in detail in Chapter 4. The NiTiPt

alloy was selected based on the results from the study of over 20 NiTiPt alloys, which

focused on elevated transformation temperatures, effects of stoichiometry on

microstructure, and the results of prior baseline thermo-mechanical studies showing good

work output. The NiTiPd alloy was selected based on studies others have performed,

which have focused on transformation temperatures and no-load recovery tests.

In Chapter 4, the thermomechanical properties of the chosen baseline

Nil9.5Pd30Tiso.5 alloy will be presented. The majority of the thermo-mechanical tests

performed on the Pd modified alloy were also performed in a parallel effort on the chosen

Pt-modified Ni24.5Pt25Tiso.5 alloy. Data from investigations on the Pt alloy will be drawn

upon for comparison to the Pd alloy, which exhibits similar structural transformations yet

has quite different mechanical properties and shape memory performance. The

underlying motivation of such a study is to compare the mechanical properties and

materials performance of the two HTSMA systems, thus allowing for the identification of

areas where further alloy development may improve the materials performance.

Characterization of the NiTiPt SMA system

Bulk Compositional Analysis

Selection of a baseline Pt-modified alloy was accomplished by a cursory study of

the effect of near stoichiometric Pt alloying additions on transformation temperatures and









microstructure of NiTi alloys. Stoichiometric alloys refers to alloys which lie

approximately along the NiSO-xPtxTiSO iso-stoichiometric line between the binary NiTi and

NiPt intermetallics. This term will be used throughout for the NiTiPt and NiTiPd (NiSo-

xPdxTiSO) alloy. Prior studies have shown that ternary Pt additions produce elevated

transformation temperatures well above the transformation temperatures ofNiTi. Thus

one would expect to find potential high-temperature shape memory behavior in this

sy stem. As a matter of fact, Pt additions are superior to all other current alloying

modifications to NiTi at increasing transformation temperatures, even so, this system has

received little attention in the form of characterization studies or alloy development.

Three groups of alloys were chosen for characterization: stoichiometric alloys, Ti rich

alloys in which the fraction of Ti in the alloy is greater than 50 at.% and Ti deficient

alloys which have atomic fractions of Ti less than 50 at.%.

A ternary compositional plot of the experimental compositions selected for this

alloy development study is shown in Figure 3-1. The stoichiometric alloys are denoted

with a E### designation and Ti rich and deficient alloys are designated with an F###

designation. Subsequent to this study two stoichiometric alloys of 20 and 30 at% Pt

denoted as baseline alloys as designated in Figure 3-1 were selected for detailed

thermomechanical testing and characterization.35

Transformation Temperature

Transformation temperatures as determined by DSC or DTA are given in an

inclusive list tabulated in Table 3-3. It was determined that the transformation

temperatures in NiSO-xPtxTiSO alloy system are strongly dependent on Pt content along the

NiSO-xPtxTiSO iso-stoichiometric line between the binary NiTi and TiPt intermetallics. The

transformation temperatures in the alloys of stoichiometric compositions designated with









an E prefix are in good agreement with several other reported values.36-3 Figure 3-2 (A)

compares several reported values of the Ms temperature as a function of Pt for Ni

stoichiometric alloy additions while the complete set of transformation temperatures are

documented in this study are presented in Figure 3-2 (B). Minor alloying additions of Pt

resulted in a depression in transformation temperatures up to approximately 10 at% Pt

after which further alloy additions resulted in a potent increase in transformation

temperature with additional Pt content.

A slight discrepancy exists between the transformation temperature measurements

taken by Lindquist and Wayman and the results of the current study. This discrepancy is

evident primarily in the compositional region where the transformation temperatures are

depressed by alloying additions. The current study and the measurements found in

(37,38) determine the transformation temperatures by calorimetric methods (DSC or

DTA) while Lindquist and Wayman employed electrical resistivity measurements to

determine the transformation temperatures. As described previously, transformation

temperature analyses by thermal methods are based on the extrapolated onset method

where the baseline is forced to have a nearly flat slope. Resistivity measurements, on the

other hand, rely on deviations in the slope of the p vs. T curves determination of

transformation temperatures the base line is formed by the temperature dependence of the

samples resistivity. Therefore resistive analysis is dependent on the slope of the baseline

and the slope of the transition region. It is possible that the differences between the

thermal and resistive account for this discrepancy. Another possibility is that difference

in thermomechanical processing, heat-treatments and or minor deviations from

stoichiometry may account for slight discrepancies.









It is known for binary NiTi and the NiTiPd systems that the transformation temperatures

are highly dependent on stoichiometry. Particularly, in some NiTi based shape memory

alloy slight deviations which make Ni rich NiTi alloys and Ni+Pd rich NiTiPd alloys

results in a strong depression in the transformation temperatures. The NiTiPt alloys

however differ from these systems in the amount of suppression by casting Ti defieient

alloys. A comparison of E430 and F3 11 alloys (Table 3-3) which both have 25% Pt

contents reveals the a 3% deficiency in Ti decrease the transformation temperatures from

only 29 degrees Celsius. Contrasting this to a comparable 3% Ti deficiency in the

NiTiPd alloys which results in a decrease of transformation temperatures of over 220

degrees Celsius, it is evident that the stoichiometric effects on transformation

temperatures are not as pronounced in the NiTiPt shape memory alloys. The details of

the effects of stoichiometry on NiTiPd shape memory alloy system are discussed in its

alloy development section later in this chapter.

Microstructure

As expected, all the stoichiometric alloys were essentially single phase.

Micrographs of these alloys were omitted as they did not provide any useful information

other than essentially confirming the single phase nature of the microstructure. Focusing

on the deviations from stoichiometry alloys with excess or defieient Ti ratios did not

exhibit an increase in transformation temperatures but rather a suppression of the

transformation occurred in non-stoichiometric alloys.

Conversely, most of the non-stoichiometric alloys, designated by the prefix F, were

found to contain a second phase. Figure 3-3 is a summary of SEM back scattered

electron micrographs of the microstructures for all the F-series alloys. Samples F303 and

F304 are essentially single-phase non-stoichiometric alloys, in that they do not contain a









second intermetallic phase. The spherical dark phase in these alloys is probably TiO2,

which is common to all the alloys. The microstructure for these two alloys is similar to

that observed in the E-designated alloys. F303 is comparable to the microstructure of the

lower Pt E-series alloys where the martensite is finely distributed and F304 is similar to

the higher Pt E-series alloys with a coarser martensitic distribution.

Most of the alloys contain second phases including alloys close to the line of

constant stoichiometry which indicates that the solubility of excess alloying additions on

either side of this line is very narrow. Only two non-stoichiometric alloys that did not

contain a second intermetallic phase were observed on the Ti rich side of this line of

constant stoichiometry. Both of these alloys contained Ti-rich interstitial containing

phases probably oxides. Therefore, any excess Ti could be tied up as TiO2 and the bulk

matrix phase was probably very close to a stoichiometric composition. Consequently this

signals that there is probably little solubility for excess solute on either side of

stoi chi ometry.

Both phase diagrams are recent assessments of the respective binary system. Based

on the compositional analysis and basic morphology of the microstructures shown in

Figure 3-3, there are basically two types of second phases observed in the non-

stoichiometric alloys. All the (Ni+Pt)-rich alloys contain a lathe like structure with a 2:3

Ti:(Ni+Pt) ratio. Given the nature of this phase, it would appear that it forms by

nucleation in the solid state. An intermetallic phase with the stoichiometry of Ti2(Ni,Pt)3

does not appear in either binary phase diagram, however, Ti2Ni3 is a metastable phase

that is observed in binary NiTi alloys.39 COnsequently, the Pt could stabilize this phase.

Or it is possible that it is a new phase, unique to the ternary phase diagram. It will take









additional quantitative x-ray diffraction or TEM analysis to determine the specific

structure of this phase.

The non-stoichiometric alloys on the Ti-rich side of the line of constant

stoichiometry have a phase that is more spherical or elliptical in morphology and usually

located along the grain boundaries. This would indicate that it is possibly a low melting

point phase that was last to form during solidification. This and the fact that the

composition has a 2: 1 Ti:(Ni,Pt) ratio, would indicate that it could be Ti2(Ni,Pt), which is

isostructural to Ti2Ni and has been previously identified by Garg.39 A small percentage

of other phases may also appear in these alloys.

Alloy Selection : NiTiPt

The results of the characterization of the ternary NiTiPt high-temperature shape

memory alloy system are summarized above and were presented and published in a

relevant conference proceeding.40 The alloy design phase was intended to build a

relationship between the compositional dependence of the transformation temperatures

and microstructure in the NiTiPt system. This system was chosen for a detailed alloy

study primarily due to evidence that Pt modified NiTi alloys exhibit the highest reported

transformation temperatures of any NiTi based alloy as well as the fact that there is a very

limited data available in the literature for this alloy system.2

The alloys were single or two phase and all contained a limited volume fraction of

interstitial containing phases. The formation of a second phase was evident even with

minor deviation in stoichiometry, therefore the solubility for excess components outside

of the iso-stoichiometric line between NiTi and TiPt is limited. The formation of

interstitial containing phases ties up Ti thus depleting the matrix by an amount which

depends on the bulk interstitial concentration. There for in order to insure that both the









alloys are not Ti defieient the Ti concentration should be increased slightly above the

expected interstitial impurity concentration.

Two alloys were selected for detailed thermomechanical testing from the NiTiPt

and NiTiPd alloy systems. Focusing on the selection of a composition in the Pt modified

NiTi alloy, the position along the stoichiometric line between NiTi and PtTi was selected

based on a compromise between elevated transformation temperatures and good

mechanical properties. As previously mentioned two baseline alloys (NiSO-xPtxTiSO

containing 20 and 30 at%/ Pt) along with a binary NiTi alloy (SM495 NiTi) supplied by

Nitinol were selected for detailed thermomechanical testing by colleagues at the NASA

Glenn research center.35

The alloys were prepared in a manner similar to that described in Chapter 2. The

materials were prepared by melting, extrusion, and subsequent machining of tensile dog

bone test specimens. Along with isothermal uniaxial tensile testing, load bias testing was

conducted on the 20 and 30 at% Pt extrusions along side the binary NiTi. Load bias

testing or constant-load, strain-temperature tests measure a materials ability to do work

against a constant biasing load which is a key design parameter in the development of

shape memory actuated deviceS35 and therefore must be considered in combination with a

high transformation temperatures in the selection of HTSMAs. Figure 3-6 shows the

results of a load bias examination in conjunction with results from a comparable test

documented in a characterization study of binary NiTi alloys.41

The applied tensile stress is related to a volume specific work output or work

density. SM495 NiTi followed by the stoichiometric and the Ti rich NiTi show the

highest specific work output yet the transformation temperatures are near room









temperature. The 20 and 30 at% Pt modified alloys however have much higher

transformation temperatures. (262C and 560 oC Ms temperatures respectively) but the

work output is lower. In the 20 at% Pt modified alloy the work output was marginally

lower that or the binary alloys at much higher transformation temperatures. In contrast

the 30 at% alloy exhibited the highest transformation temperature of all the alloys

encompassed in this study yet did not perform work under any biasing load. Therefore,

although this material exhibited high transformation temperatures, it is not directly useful

in application requiring actuation forces of any level.

Based on the findings of this study and those in reference [35]a new baseline

composition of Ni24.5Pt25Tiso.5 was selected for investigation in this study based on a

compromise between elevated transformation temperatures potential work output.

Although this alloy was not included in the NASA NiTiPt baseline alloy advanced

mechanical characterization study35 it lies within the compositional bounds of the

Ni20Pt30Ti5o and Ni20Pt30Tiso alloys. The transformation temperatures in the 30 at% Pt

alloy was high but the alloy exhibited no evident capacity for producing work while the

20 at% alloy exhibited a high work output at a much lower transformation temperature.

The Ni24.5Pt25Tiso.5 alloy lies halfway between these tests and therefore by extrapolation it

was assumed that is alloy would demonstrate some capacity for performing work at with

transformation temperatures centered around 450C.

Alloy Selection : NiTiPd

NiTi and PdTi, like the NiTiPt system, also form a continuous solid solution with a

high-temperature B2 phase that transforms to a Bl9 (orthorhombic) or Bl9' (monoclinic)

low-temperature martensite phase with transformation temperatures between those of the









binary alloys.42 Figure 3-7 is a map of the thermodynamically stable thermoelastic

transformation reactions as a function of composition and temperature for the NiTiPd

alloys.43 The Pt and Pd modified NiTi both follow similar transformation reactions at

similar alloying additions but with differing effectiveness of the ternary alloying

additions at increasing the transformation temperatures. Figure 3-8 (A) illustrates the

relationship between transformation temperature and alloying additions in both systems.16

As shown in this diagram it has been confirmed that the martensitic transformations over

the entire range of ternary Ni-Pd-Ti compositions are thermoelastic in nature and that the

alloys exhibited aspects of shape memory behavior similar to binary NiTi alloys.44 45

The level of Pt in the NiTiPt alloy was chosen on the basis of transformation

temperatures and the work output. To the best of the author' s knowledge, in the NiTiPd

shape memory alloy system no such tests results (load bias) are available in the literature.

However, measurements of the alloy's stress free strain recovery have been examined in

prior studies.l Through dilatometry and uniaxial ambient temperature tests the

compositional dependence of the shape recovery at specific initial strain intervals was

examined. The findings of this study are summarized in Figure 3-8 (b) where the

recovery strain is plotted as a function of alloy composition and initial plastic strain. For

all strain increments the maximum shape recovery is approximately at 30 at% Pd.

Additionally it has been shown that the shape memory behavior of NiTi-30OPd (at.%)

alloys can be quite good under unconstrained conditions with samples loaded to 2-4%

total strain levels in the martensitic condition recovering 100% of the strain while those

loaded up to 6% recovering 90% of the strain.36 46 47 Similar shape memory behavior has

been observed for samples deformed in compression 48 and torsion.49 Alloys containing









40 at.% or more Pd including the TiPd binary alloy could only recover about 0.5% strain

in tension to various strain levels.36 49 This poor shape memory performance has been

attributed to a low critical stress for slip (which is an irreversible process), such that the

maj ority of the deformation is accommodated by slip rather than the more typical twin

boundary motion or martensite reorientation (also referred to as detwinning).49

As a result of the Eindings listed above a base composition of Nil9.5Tiso.5Pd30 WAS

selected for the Pd modified alloy. The alloy is developed with a slightly Ti

(approximately .5 at%) rich composition in order to ensure high transformation

temperatures. Transformation temperatures are highly linked stoichiometry following the

relationship NiSO-xPdxTiSO. Compositional deviations veering into the Ti lean side of the

iso-stoichiometric line between TiNi and TiPd (NiSO-xPdxTiSO) strongly decreases

transformation temperatures in contrast to Ti rich compositions which have little effect on

the transformation temperatures. Figure 10 confirms the compositional dependence of

the transformation temperatures for off-stoichiometric alloys at Eixed Pd fractions

(Ni20+xPd20Tiso-x). The effects of stoichiometry on the transformation temperatures in the

(Ni20+xPd20Tiso-x) alloy is similar to those exhibited by binary NiTi (TiSO-x Niso) where

compositions crossing into the Ti deficient iso- stoichiometric line results in a sharp

decrease in transformation temperatures.'o

It is a well known fact that Ti has a high interstitial affinity. In the system under

study, the main interstitial impurity elements are C and O which enter the melts as

impurities during the alloy melting process. These interstitial elements react with Ti to

form titanium oxides and carbides thus depleting titanium from the alloy by tying it up in

interstitial compounds. By forcing the alloy to be Ti rich by a fraction greater than the









impurity concentrations results in a final alloy composition which remains Ti rich by an

amount proportional to the excess Ti not consumed in the formation of oxides. The goal

is to keep the overall matrix composition stoichiometric or slightly Ti-rich, thus

guaranteeing a high transformation temperature. Additionally Ti rich compositions yield

a fraction of the intermetallic phase Ti2(Ni,Pd) which is isostructural to Ti2Ni.39

Ti2(Ni,Pd) is an interstitial stabilized phase which has a high solubility for interstitial

oxygen. 5 The presence of this phase allows for further removal of interstitials from the

martensite or austenite matrix which has a lower interstitial solubility. This alloying

approach has been employed in previous studies of the Pt modified NiTi SMAs.35

Alloy Selection Summary

The NiTiPt alloy was selected based on the results from our study on elevated

transformation temperatures, effects of stoichiometry and the results of prior baseline

thermo-mechanical studies measuring work output. In the Pd modified NiTi no such

advanced thermo-mechanical studies existed thus the alloy selection was centered around

existing studies which demonstrated a composition dependent maximum in the

unconstrained or no load shape recovery. Consequently, the final selection of alloys for

this study are Nil9.5Pd30Tiso.5 for advanced thermo-mechanical testing in parallel to the

Ni24.5Pt25Tiso.5 for comparison of mechanical and thermoelastic properties. Each alloy is

developed with a slightly Ti (approximately .5 at%) rich composition to prevent Ti-loss

from the matrix resulting in a Ti-poor alloy with lower transformation temperatures.































Figure 3-1 Ternary plot of the Ti-Ni-Pt compositions studied. The composition of all
alloys was confirmed by spectrographic analysis

Table 3-2 Aim and measured compositions of all alloys investigated.

Aim Composition (at.:;.) Measured Con positions (at.:;..)
sample Ti Ni Pt Ti Ni Pt C N O

E426 50 50 0 49.96 49.73 0 0.05 D1 I 1 0 18
E427 503 45E 5 49 86 45.13 4.58 0.08 0 0 1 0 26
E428 50 4010 50.02 40.01 9.66 0.06 O 02 0.17
E429 50 30 20 49 43 29.52 20.64 0.07 O 02 0 25
E430 503 25 25 49 88 25.25 24.48 0.07 O 02 0 22
E431 50 20 30 49 77 20.03 29.81 0.08 0 03 0 21
E432 503 10 40 49 83 9 78 39.94 0.07 O 02 0 24
E433 50 5 45 49.94 5.04 44.53 0.11 0 03 0.23
E434 50 0 50 49 9 0 49.48 0.08 O 04 0.3
F301 48 31 21 48 02 31.01 20.45 0.12 O 08 0 28
F302 48 21 31 47 79 20.79 30.79 0.11 0 12 0 36
F303 52 29 19 52 03 2': 13 I 18.5 0.09 O 07 0.23
F304 52 19 29 51 58 19.38 28.52 0.13 O 09 0 25
F305 45 32 5 22.5 45 32.27 22.35 0.05 0 05 0 25
F306 45 22 5 32.5 44 84 22.76 31.87 0.076 0 05 0.3
F307 55 27.5 17.5 54.02 28.26 17.16 0.08 O 08 0.37
F308 55 17 5 27.5 54 46 18.23 26.73 0.1 0 01 0 23
F309 47 28 25 46 81 27.89 24.77 0.11 0 09 0 27
F310 47 25 28 47 05 25 27.43 0.11 0 08 0 25
F311 53 22 25 52 58 22.35 24.49 0.13 0 1 0 29
F312 53 25 22 52.59 25.08 21.68 0.12 O0.09 0.38
F313 48 21 31 38.79 25.24 35.54 0.09 O 07 0.2


61







Pm
E 2 0


50 E30 730 1 1
100 : rj1
TI o3 R to3 o do s o T
111~NITI
d~, 31~~05NI








62



Table 3-3 Transformation temperatures of alloy set
Sample
deinainAs oC Ar oC MsoC Mr (C
E426 60 86 57 35
E427 51 65 46 25
E428 86 94 87 76
E429 313 322 312 296
E430 449 484 436 394
E431 600 653 591 527
E432 838 917 811 751
E433 963 998 917 869
E434 1044 1069 1018 996
F301 284 352 307 239
F302 666 780 605 552
F303 268 302 259 212
F304 526 560 492 424
P305 444 511 459 379
F306 no peaks
F308 486 519 470 423
F309 449 507 446 382
F310 552 617 523 468
F311 420 453 402 363
F312 342 374 286 330


a.

E

S400


~ rooo
c
a
soo

~ soo-
5
~ 400
E
Q
E200


10 20 30 40 50
X (at. %/ Pt)


0 10 20 30
X (at% Pt)


40 50 SO


Figure 3-2 A. Effect of Pt on the Ms transformation temperatures for Ni50-xPtxTi50
alloys, including data from previous researchers B. Effect of Pt on all
transformation temperatures for Ni50-xPtxTi50 alloys


~ As
--- --0- --- Af
--t-- MB
----~---- Mf


























/t


63

























500 micron







S500 microns


Figure 3-3 SEM B SE micrographs of the non-stoichiometric alloys.







64




Table 3 -4 Semi-quantitative EDS analysis of the various phases observed.



Sampe ID Bulk Composition Region 1 martensitee) Region 2 (second phase
(at.%) (at.%) aricle) (at.%)
Ti Ni Pt Ti Ni Pt Ti Ni Pt
F301 48 31 20 46 29 25 39 39 22
F32 48 21 31 47 18 36 39 29 31
F303 52 29 19 49 28 22
F304 52 19 29 48 18 33
F30545 32 22 47 23 30 39 35 26
F306 45 23 32 46 15 39 38 27 34
F30855 18 27 47 22 31 63 9 28
F309 47 28 25 45 25 29 44 28 28
F310 47 25 27 46 21 33 45 24 31
F31 53 22 25 50 21 29 67 2 32
F312 53 25 21 49 27 24 67 2 30


Weight Percent Nickel


120 i 9



a-i


6000
0 20 40 60 80 100
Atomic percent platinum


Figure 3-4 Phase diagrams (A) NiTi binary phase diagram from reference37 (B) TiP1
binary phase diagram from reference38










II


c~sl
~tC



IELa
~~sa
p~rr
p~~olS
l~h3
~3 od~,~'SO
I


1000


10 20 30 40 50


at.% X

Figure 3-5 Effect of ternary alloying additions on the Ms (or Mp) temperature for NiTi-
based high-emperature shape memory alloy systems.2




25

~ r SM495j NiTi
c~20

S15
NisoTiso
O ,---- NASA CR-1433)


5 Ns~oieN495 0.5
-3 ,, (#454 CR-1433)

(1) Ni,,Pt,Ti3,


O 100 200 300
Stress


400
(M Pa)


500 600 700


Figure 3-6 Comparison of the specific work output for several conventional NiTi alloys,
SM495 NiTi, and the (Ni,Pt)Ti HITSMA35


















700t B 2


819


TiNi 10


TiPd


Pd at%


Figure 3-7 Phase diagram of TiPd+TiNi alloys.44





e Ti50 (Ni50-x) Pd,)
1100 -o Donkersloot and Vanl Vucht 1970
1000 -
900 -,
800 ,
700


6 00


Composition {x)


Composition (%)


Figure 3-8 Shape memory properties NiTiPd (A) Ms temperature resulting from ternary
alloy additions."l (B) Average shape recovery in Tiso (NiSO-x) Pdx.l







67




520-





Ia 440 -pA



-Ms

48 48.5 49 49.5 50 50.5 51
Ti /at%


Figure 3-9 Plots of martensitic transformation temperatures vs. composition for Tiso-
xPd inNi on 47. 47















CHAPTER 4
RESULTS AND DISCUSSION

Shape memory alloys have unique properties and related mechanical behaviors.

Principally, they have the ability act as a solid state device which performs mechanical

work against a biasing load. In order to examine this extraordinary behavior an in-depth

thermomechanical study of these two alloys is conducted. A comparison between the

two systems is carried throughout this chapter with the main focus on the Nil9.5Tiso.5Pd30

high temperature shape memory alloy in comparison to the Ni24.5Tiso.5Pt25 allOy.

Heat Treatment Optimization

Before testing the baseline Nil9.5Tiso.5Pd20 and NiTiPt alloys, it was necessary to

determine an optimum heat treatment for the annealing of the as-machined samples to

eliminate any residual effects due to the machining process that could affect the

transformation temperature or thermomechanical performance of the alloy. The

martensite phase in particular is susceptible to twinning/detwinning and or plastic

deformation of the near surface layers of the alloy. The function of the stress relief heat

treatment is to relieve any residual stresses in the samples due to the high speed

machining and the thermal expansion mismatch between the mild steel extrusion can and

the SMA. At elevated temperature interactions between dislocations generated during the

machining process allow for the formation of low energy structures by dislocation

movement. 52 As an indicator of the residual stresses in the material resistive

measurements are employed.









Resistivity is a stress sensitive property among other variables. A classic example

of this effect is the increase in resistivity as a function of maximum precipitation

hardening in aluminum 7XXX series alloys where the increase in resistivity is linked to

an increase in the materials internal stresses due to precipitate matrix interactions54

Additionally a materials stress state affects the relationship between temperature and

resistivity or the slope of the resistivity temperature curve. The combination of these

effects is the keystone for optimizing our stress relief heat treatments. For the purpose of

these measurements it was assumed that the linearity of the resistivity vs. temperature

curves is a function of the stress distribution in the sample. The underlying reasoning

behind this was that if a non uniform stress distribution exists within the material a

position dependent resistivity distribution will also exist. As the material is heated or

cooled through a thermal hysteresis a fraction of these stresses are relieved resulting in an

irregular resistivity temperature curve. A material which has been almost fully recovered

on the other hand will exhibit a linear resistivity temperature curve due to the absence of

the relief of residual stresses. Figure 4-10 shows the results of a heat treatment

optimization for the 25 at% Pt modified alloy. In this examination the samples were heat

treated for 1 hr at 500 oC and 600oC and heated through a thermal hysteresis with in-situ

resistivity measurements post heat treatment along side a no heat treatment sample for

comparison. The sample that did not have a heat treatment results in a resistivity

temperature curve which is highly irregular due to the irregular transient stress

distribution within the sample. The sample which was heat treated for 1hr at 500 oC (near

Af) showed a slightly more regular relationship due the partial stress relief during the heat

treatment. Finally the sample which was heat treated at 600 oC showed a smooth









resistivity temperature curve which is indicative of a stable stress distribution within the

material which may be linked to a fully recovered material.

Characterization

Materials Characterization

Each alloy was developed with a slightly Ti (approximately .5 at%) rich

composition in order to maintain a stoichiometric or slightly Ti-rich matrix composition

after formation of various interstitial containing phases. The alloying technique has been

employed in previous studies of the Pt modified NiTi SMAs.35 The resulting chemical

compositions of the extruded materials are Ni 19.5, Pd 30.0, Ti balance, O 0.30, C 0.50

and Ni 24.419, Pt 24.428, Ti balance, O 0.28, C 0.43. Detailed compositional analysis

of each extrusion is given in appendix C. SEM micrographs exhibit the phase contrast of

the resulting alloy is shown in Figure 4-11 revealing a small volume fraction of possible

carbide and oxide phases in a relatively homogeneous Ti Ni Pd and Ti Ni Pt matrix void

of macro cracks and porosity.

Properties and Transformation Temperatures

The no-load transformation strain of the undeformed material was measured by

dilatometric techniques. This technique measures macroscopic structural shape change

or uniaxial transformation strain of the extruded material as a function of temperature.

The thermal expansion coefficients of each phase austenitee and martensite) were

determined from the slope of the linear portions of the temperature vs. strain relationships

shown in Figure 4-12 and Figure 4-13. The samples were thermally cycled through

several hysteresis in order to allow the transformation temperatures and strains to

stabilize thus becoming reproducible on subsequent cycles. The forward and reverse

reaction in the undeformed material exhibited equal magnitude transformational strains









under no load conditions. The thermal expansion coefficient of each phase within each

alloy was approximately equal with the martensite having a slightly greater thermal

expansion than that of the austenite with the NiTiPd alloy exhibiting the greater thermal

expansion. These thermal physical properties are tabulated in appendix A.

Resistance measurement is a prominent characterization technique in the study of

SMAs due to the substantial differences in the electrical properties between the parent

and shear phases. This technique has been successfully applied to SMAs as a method of

determining transition temperatures in-situ to operating conditions. In addition, the

application design of SMA actuated active flow control devices requires an

understanding of the temperature dependence of the electrical resistance. Researchers

working on the Nitinol characterization study NASA CR-1433 utilized the resistance vs.

temperature behavior of unconstrained NiTi in a scheme to quantify the shape memory

performance of the material.53 In the current study this technique was exploited for the

determination of the transformation temperatures as well as the temperature dependent

electrical properties of the SMA under study in the stress free state.

The temperature dependence of the resistivity is plotted in Figure 4-12 and Figure

4-13 including marked isothermal test temperatures. As mentioned in the experimental

section the transformation temperatures are determined by constructing a tangent line to

the linear portion resistivity temperature plot of each phase and a tangent line for the

transformation intermediate region. The intersection of the tangent lines is determined as

the transformation start and finish temperatures.

A slight discrepancy exists between the transformation temperatures measured by

dilatometric and restive methods. This slight discrepancy was linked to the methods used










to instrument the samples with a thermocouple. In the dilatometer the thermocouple was

located a small distance from sample therefore this technique actually measures an

average temperature between the region of the furnace adjacent to the sample and the

sample. The transformation in the forward and reverse directions are highly exothermic

and endothermic therefore during the transformation the difference in temperature

between the sample and the furnace increases. This effect causes a slight error in the

temperature reading at the thermocouple. The resistive measurement on the other hand

circumvents this problem by having the thermocouple spot welded directly to the sample

and therefore is a more accurate measurement of the transformation temperatures. A

summary of the recorded transformation temperatures measured by resistive methods are

tabulated in appendix B.

Similarly to the thermal expansion results the temperature dependence of the

thermal resistivity coefficient of the martensite phase is greater than that of the austenite

phase. The resistivity of the austenite is greater than that of the martensite at all

temperatures within the experimental bounds. A metals resistivity is a stress and

structure sensitive property thus a greater resistivity could be an indicator of a higher

internal energy or stress state of a particular phase, which in this case is the austenitic

phase. Ti vs. Ni, Pt or Pd have a significant mismatch in atomic radii, and thus the

arrangement of these atoms into ordered B2 lattice results in internal lattice strains as

atoms are forced to reside at slight distances from their minimum energy distance. The

shear lattice on the other hand has greater distances between lattice positions resulting in

a more relaxed structure and thus a lower resistivity.









In the Pd modified alloy there is a small but significant peak in the resistivity for

the reverse reaction martensitee transforming to austenite). This peak occurs when a very

small volume fraction of martensite remains in an austenite matrix. Focusing on the

martensite/austenite interface and criteria for thermoelastic transformations which states

that this interface must be coherent and the maj ority of the transformation strain between

the phases must be accommodated elastically a correlation to a parallel well developed

system is made. A comparison of such an arrangement to the mechanism for

precipitation hardening which also initially has strains partially accommodated elastically

due to lattice mismatch can be made. It is feasible to imply that there could be a

significant elastic interaction in the Pd-containing SMA when very small fractions of

martensite reside in an austenite matrix. Precipitation hardened materials exhibit an

increase in resistivity as hardening resulting from elastic interactions between phases

increases.54 The evidence of this is the small peak in the resistivity curve and the

implication that elastic interactions between the austenite matrix and martensite produces

an increase in resistivity above that of the fully austenitic material. Comparing the mean

correspondent variant pair size of the Pt and Pd modified alloys it is evident that the

distribution is much coarser in the Pt than on the Pd alloy (Figure 4-11). As a result, it is

expected that this coarse distribution near the end of martensite to austenite

transformation results in widely spaced martensite packets which can not effectively

harden the remaining austenite or globally effect its electrical resistivity thus Pt alloy

does not exhibit the a resistivity peak.









Thermomechanical Testing

Isothermal Stress-Strain Behavior in Tension and Compression

To directly compare mechanical properties in tension and compression all the

isothermal curves are calculated in true stress and strain. The representative true stress

strain curves for tension and compression have been combined in to two sets of curves

for each material in each loading orientation making a total of eight multi-experiment

plots (Figure 4-6-Figure 4-18 and Figure 4-22-Figure 4-25). For each material, there is a

plot combining (Figure 4-6-Figure 4-18 for NiTiPd and Figure 4-22-Figure 4-25 for

NiTiPt) the stress strain relationship of the austenite and martensite well above or below

transformation temperatures. Tension and compression behavior for the NiTiPd alloy is

shown, respectively, in Figure 4-6 and Figure 4-7 for temperatures well away from the

transformation temperatures while Figure 4-17 and Figure 4-18 contain the results of

tensile and compressive isothermal tests near the transformation temperatures. The

NiTiPt alloy tested at isothermal temperatures well above or below the transformation

temperatures are shown in Figure 4-22 and Figure 4-23 while Figure 4-24 and Figure 4-

25 include the mechanical properties near the transformation temperatures.

The elastic loading region of each uniaxial compression test was analyzed for

linearity. Significant deviations from linearity in the elastic region are a good indicator

of slightly non-parallel loading surfaces which result from a combination of machining

imperfections, imperfect load bearing plenum faces and loading the sample slightly off

axis. An example this effect is shown in Figure 4-14 where the elastic loading region

below 3% engineering strain exhibits an experimentally induced deviation from linearity.

If analysis revealed such a deviation from linearity the uniaxial compression test was









corrected by extrapolation of a regression analysis of the well behaved (linear) portion of

the elastic loading region.

Although each isothermal compression test was carried out to over 30%

engineering strain the results are presented only to 20% strain. Analysis of the force

displacement curves revealed that friction between the load bearing plenums and the

sample induced barreling. During a constant strain rate isothermal compression test an

increase the slope of the force or engineering stress Vs displacement curve is indicative

of significant frictional forces and the onset non-uniform deformation. This effect is

exemplified by the monotonic compression test shown in Figure 4-14 at approximately

the 22% strain level. Beyond the 22% strain in this example the slope of the force strain

curve increases due to friction induced effects.

Isothermal stress-strain behavior in tension and compression NiTiPd

In the NiTiPd alloy well below the transformation temperatures in tension

(Figure 4-6) the classical behavior of a thermal elastic shape memory is exhibited. The

stress strain curve of the martensite at room temperature and at 200C exhibits three

identifiable regions in the stress strain curve. These regions are (1) elastic deformation at

lower stresses up until the lower yield stress or in the case of the martensite the

detwinning stress (2) further loading results in an almost linear inelastic deformation

resulting from detwinning of the martensite with continues until the variants with

favorable orientations are reoriented resulting in pseudo stress strain plateau (3) the work

hardening rate again increases, initially due to further elastic deformation of the re-

ordered variant structure followed subsequently by further deformation of the martensite

resulting in a more typical region of plastic deformation where the curve has a parabolic

appearance.









In tension the pseudo stress strain plateau (region 2) differs from the stress strain

plateau exhibited by classical SMA (NiTi) in that deformation does not occur at a near

constant stress. The positive slope in this region indicates that a mechanism is requiring

an increasing stress to continue deformation. Several mechanisms could produce such a

result in an SMA. As described earlier in the alloy development section, one of the

limitations of high temperature shape memory alloys is a low critical resolved stress for

the onset of non-recoverable slip processes at least relative to the detwinning stress. Slip

and the associated deformation motion and generation result in work hardening; thus

could explain why the combination of slip and detwinning requires a progressively

greater stress as the true stain increases. Although this is an important mechanism

examination of the no-load recovery tests results examined discussed a later section show

that it is not the dominant cause of the work hardening.

Another more plausible mechanism proposed here is based on the fact that there

exists an orientation distribution of correspondent variant pairs. As mentioned earlier, the

critical stress for detwinning, noted as the lower yield stress, is significantly lower than

the actual yield stress for macroscopic yielding of the martensite primarily through

dislocation motion (classical yield stress). Considering that in a typical multi-variant

martensitic structure, certain variants are oriented favorably for detwinning which means

the applied stress results in a maximum resolved shear stress. Variants which are

oriented such that the resolved shear is a maximum will detwin first at the lowest applied

uniaxial stress resulting in the bulk material exhibiting a yield stress. Variants which are

misaligned are simultaneously deformed elastically as the resolved shear stress is smaller

than the critical resolved shear stress for detwinning. As deformation increases at










increasingly greater stresses, more variants in other orientations are activated as the

applied stress produces a progressively increasing resolved shear stress. Thus although

detwinning is the main deformation mechanism in the pseudo stress-strain plateau the

orientation distribution of these variants requires different applied uniaxial stresses for

detwinning. However more work is necessary to fully confirm this mechanism.

In contrast, the isothermal deformation of polycrystalline martensite in a classical

SMA, such as NiTi, occurs at a constant or near constant stress. Here, similarly to the

NiTiPd extruded material, we also have a variant distribution which results in a

comparable situation yet the classic SMAs do not exhibit this behavior. A likely

explanation for this stems from the number of equivalent variants that can form from a

single parent austenite crystal. Recall that the martensite phase in the NiTi SMA has a

monoclinic structure in contrast the NiTiPd and NiTiPt alloys both of which transform

from a B2 to the orthorhombic structure. The orthorhombic structure is a higher

symmetry structure than the monoclinic. As a matter of fact a monoclinic structure in

SMAs is related to the orthorhombic structure by an additional non-basal shear. This

additional shear increases the number of equivalent variants by a factor of two which may

form from a given austenite crystal. Thus, any particular variant has twice as many

equivalent variants it can shear to under load. As a result, there is less of an orientation

dependence of the shear stress for detwinning in the monoclinic structure as there are

many more variants for a particular variant in a particular orientation to shear to thus in

SMAs with a monoclinic martensite exhibit detwinning at constant or near constant

uniaxial stress.









The mechanical properties of the NiTiPd martensite in compression (Figure 4-7)

are comparable up to the initial yield stress. Beyond the yield stress, however

(Figure 4-6) the deformation of the martensite does not exhibit a clear distinction

between the deformation mechanisms of detwinning and non-reversible slip processes.

Deformation of the martensite in compression after the yield stress results in a high work

hardening rate to strains of about 10%. Further deformation again results in a decreasing

work hardening rate similar to what is observed for a single phase alloy. The lower slope

is linked to deformation via non- recoverable slip processes and is comparable in both

tensile and compressive loading orientations. However, the isothermal behavior in

tension and compression deviates substantially beyond the yield stress of the martensite

phase. The tensile behavior exhibits a pseudo-stress plateau or low hardening rate at

stress levels starting at about 250 MPa which is linked to favored martensite variants

growing at the expense of the other martensite variants in contrast to the stress-strain

relationship for compression which does not show a clear stress strain plateau. This

indicates that the orientation of preferred variants requires an increasing stress suggesting

a low mobility interface and thus a deformation mechanism which is different from the

deformation mechanism in the stress plateau evident in the tensile loading mode.

Deformation by detwinning mechanisms is polar in nature, in contrast to slip

behavior, such that a reversal in the shear direction will not produce twin movement in

variants favorably aligned for operation in the forward direction. ,56 In Other words,

twins aligned favorably for operation in compression will not operate under a tensile

stress and vice versa. For monotonic testing of the martensite phase, the sample

microstructure was set prior to testing. Apparently the manner in which the material was










processed (extrusion in this case) resulted in a more favorable variant structure for

twinning/detwinning in tension than compression. Particularly if the variant orientation

distribution is such that the variants which predominate the microstructure are already in

the orientation at which the Schmid factor is at a maximum, deformation through

detwinning assisted variant reorientation is not possible since there are fewer variants to

shear over and j oin the growing variant. The evolution of a microstructure of such a

variant distribution results in a material which is essentially partially detwinned as the

product of the detwinning reaction is the formation of variants which already

predominate the microstructure.

Well above the transformation temperature the mechanical properties of the

austenite were comparable in tension and compression. The austenite exhibits a small

linear elastic stress strain region up until the yield stress of the material. After the yield

stress, the material exhibited a low work hardening rate which decreased with increasing

isothermal test temperature as evident in comparison of the 300 oC and 400 oC tensile

tests (Figure 4-15) and the 350 oC, 365 OC, 400 oC and 500 oC compression test

(Figure 4-16).

The 400 oC tensile test sample exhibited a maximum in the engineering stress strain

curve followed by a parabolically decreasing stress strain relationship. This is indicative

of non-uniform deformation of the gauge length. Optical comparator measurements of

the tensile sample confirmed that extensive necking occurred during deformation process

which led to the formation of a slightly diffuse neck. In this sample uniform deformation

reduced the cross-sectional diameter of the gauge length 4.72% from 3.81 mm to 3.63

mm while the neck region exhibited a 40.2% reduction and a minimum diameter of 2.28









mm. The true stress and true strain up to the point of necking was calculated from the

engineering stress and strain using the constant volume approximation. The true facture

stress (1522 MPa) and strain (102%) were calculated from the minimum cross-sectional

area of the necked region. The 400 oC curve combines the true stress and strain

calculated from the engineering values and an extrapolation between the value at the

onset of necking and the true fracture stress strain.

For the test results shown in Figure 4-17 and Figure 4-18, the samples were heated

to well above the austenite finish temperature (Af) and allowed to cool back to the test

temperature before loading the sample. The process of heating significantly beyond the

Af and subsequently cooling back to the desired test temperature ensured that the

resulting stable phase was that of austenite. In-situ resistivity measurements and the

occurrence of a well defined stress plateau, indicate that upon loading, a stress induced

martensite results. This behavior can be seen in detail in Figure 4-19, which is a

superimposed plot of resistivity, determined in-situ during tensile testing at 255 OC with

the stress strain curve. Both in tension and compression the relative amounts of stress

induced martensite as well as the stress at which the transformation occurred were

comparable. Finally, a ductility minimum was observed in the region where the stress

induced martensite occurred, similar to the behavior reported in NiTiPt high temperature

shape memory alloys.35

The combination isothermal stress strain resistivity curve shown in Figure 4-19

exhibits a clearly visible stress strain plateau which results from the parent austenite

phase isothermally transforming under stress to the martensitic shear structure. The

deformation mechanism in this case is not detwinning nor dislocation motion, rather









deformation occurs by the transformational shear strain associated with the thermoelastic

transformation. Initial deformation correlates an increase in resistivity with the elastic

loading of the austenite, after which a stress induced transformation begins. Recalling

the no-load resistivity hysteresis measurements (Figure 4-12) it was shown the resistivity

of the martensite is much lower than that of the austenite. Therefore as the austenite

transforms to martensite indicated by the stress plateau, the resistivity decreases with

increasing fractions of martensite. With continued strain elastic then plastic deformation

of the stress induced martensite and remaining austenite occurs and as expected results in

an increase in resistivity. A final change in the work hardening rate is accompanied by a

subsequent change in the slope resistivity strain relationship.

Another interesting feature of the stress strain plot (Figure 4-20) for stress induced

transformations in compression is that the stress at which the austenite is forced to

transform increases with increasing deviations from the no-load transformation

temperatures. This can be correlated to the fact the chemical driving force opposing the

non-chemical driving generated by the applied stress is greater at increasing deviation the

load free transformation temperatures. Additionally the extent of the stress induced

transformation and the resulting amount of strain generated by it decreases with

increasing deviation from the no-load transformation temperature. At temperatures much

higher than the no-load transformation temperature (T>>Md) it is not possible to form

stress induced martensite as the stress required for this to occur is higher than the yield

stress of the austenite which results in deformation by non-recoverable slip processes.

The temperature dependence of the yield stress in compression and tension shown

in Figure 4-20 is determined from the proportional limit from the above series of









isothermal stress strain curves. Generally, the yield stresses in tension and compression

were similar at a given temperature. The yield stress of the martensite decreased with

increasing temperature and reached a minimum at the transformation temperature where

the formation of stress induced martensite was possible. Beyond this minimum the alloy

was partially or fully austenitic and the yield strength increased significantly with

increasing temperature until a peak near 350 oC. At temperatures above 400 oC the

austenite weakened and the onset of slip occurred at lower stresses.

Dynamic elastic modulus determination NiTiPd

In order to have an accurate representation of the instantaneous, load free, elastic

modulus of these materials as a function of temperature, a dynamic modulus test was

conducted for the NiTiPd alloy. The temperature-modulus relationship is plotted in

Figure 4-21, which has three main approximately linear regions corresponding to the

fully martensitic or austenitic phase or a combination of both in the intermediate

temperature range bound by the reaction start and finish points. The dynamic modulus of

the martensite is about 10 GPa higher than that of the austenite. The martensite as the

sample temperature has an inverse relationship to elastic modulus. Converse to this

relationship the modulus increases with temperature for the fully austenitic material

suggesting that the internal energy resulting from atomic interactions of this phase is

increasing with increasing temperature up to 800 oC. The region bound by the

transformation temperature exhibits a stronger temperature dependence than does either

of the single phases which is proportional the reaction progression terminating at a

minimum at the transformation finish temperature. At this point the material is fully

austenitic and modulus value is that of the austenite at the transformation temperature.

An important characteristic of thermoelastic transformations is pre-martensitic elastic









softening. This effect is clearly exhibited by the NiTiPd alloy as evident in the elastic

modulus increasing with increasing temperature. Pre-martensitic elastic softening allows

for local transformation strains adj acent to the interface to be accommodated elastically at

a lower stress.

Isothermal stress-strain behavior in tension and compression NiTiPt

The NiTiPt alloy martensitic stress strain curves deformed isothermally well below

the transformation temperatures in tension and compression did not exhibit any of the

clearly distinguishable characteristics of a "typical" shape memory alloy. This alloy does

not exhibit a defined stress plateau. Initial loading both in tension (Figure 4-24) and

compression (Figure 4-25) resulted in a linear region as the alloy is elastically loaded

(linear elastic region). Similarly, upon loading in tension or compression past the yield

stress there is a change in the slope of the stress-strain curve. However, the tensile

samples fracture shortly after the yield stress. From no-load recovery experiments

detailed in a subsequent section on this alloy, we know that partial detwinning along with

slip is occurring at strains past the yield stress therefore, the second region will be

denoted as the detwinning region, following the nomenclature for conventional SMAs.

The ductility in tension at the test temperatures averages at about 2.5% while in

compression the samples either fractured slightly before 20% strains or were unloaded

when excessive bulging became evident. Similarly to the NiTiPd alloy in compression,

this alloy exhibited a second change in the work hardening rate in the 10% strain range.

Although a clear correlation to the deformation mechanism has not been determined it is

possible that changes in sample geometry from uniform deformation to sample bulging in

addition to slip result in such a change in slope.









As in the NiTiPd alloy, a ductility minimum was evident in tension at temperatures

in the range of the transformation temperatures. Furthermore the all the martensitic

alloys failed in a brittle manner without any visible necking. Most of the fractures

seemed to initiate at a surface defect (spot weld sites) or macroscopic inclusions visible to

the naked eye. Further microstructural characterization is required to fully characterize

the fracture mechanism in this alloy. Analysis of the fracture stress and fracture strain at

each isothermal test temperature (Figure 4-28) reveals two distinct regions. Starting at

200oC as the temperature is increased the facture stress and fracture strain decrease until a

minimum is reached. Above this point all the samples tested are fully austenitic and as

the isothermal test temperature is increased further the fracture stress decreased at a

slightly less negative rate while the fracture stress sharply increases. The fracture stress

in the austenitic phase decreases with increasing temperature due to dynamic recovery

becoming more prevalent while the fracture strain possibly increases due to the high

strain rate sensitivity of this phase as previously explored in Figure 4-26.

Isothermal tests for well above the transformation temperatures in the fully

austenitic state in tension (Figure 4-22) and compression (Figure 4-23) have similar

characteristics to the stress stain curves of NiTiPd alloy. Primarily there was elastic

loading of the austenite up to the yield stress followed by plastic deformation. At higher

temperatures dynamic recovery in the martensite was clearly evident. In order to explore

the effects of dynamic recovery two tests were conducted in tension at 550C, one test was

run at the standard strain rate while the second test was run at a strain rate increased by a

factor of 10, the results are shown in Figure 4-26. This test confirmed that dynamic

recovery was in fact a dominate mechanism affecting the mechanical properties of the









austenite and furthermore the austenite was quite strain rate sensitive although further

testing is required to quantify the strain rate sensitivity. This is important for subsequent

studies on the advanced thermomechanical processing of these alloys for actuation

applications.

The yield strength in both tension and compression in this alloy also overlap quite

well as shown in Figure 4-18. This indicates that the onset of the combination of

detwinning and yielding are similar in both tension and compression. Similar to the

NiTiPd alloy the yield strength drops at temperatures in the range of the transformation

temperatures followed by a sharp increase. A sharp decrease in the yield strength of the

austenite occurs at temperatures above the Af in which the yield strength of the austenite

drops over '/ its initial value. This is due to thermal energy helping overcome the

activation energy for dislocation motion. Contrasting to the yield strength vs.

temperature relationship of the NiTiPd SMA in this alloy the yield strength of the

martensite is less than 100MPa of the austenitic yield strength while in the NiTiPd alloy

this difference in strength is over 200MPa.

An important characteristic of viable SMA materials for actuation applications has

been linked to mechanical properties of the austenite vs. those of the martensite.

Primarily, experience has shown than for an SMA material to be a good candidate for

actuation applications the alloy the austenitic phase should be mechanically stronger than

the martensite in order to avoid non-reversible slip in the austenitic phase. This is clearly

the case in the NiTiPd alloy yet the strength of the martensite is comparable to the

strength of the austenite in the NiTiPt alloy. Furthermore, stress induced transformations

did not occur in this alloy at any temperature tested as confirmed by in-situ resistivity









measurements which is thought to be primarily because of the high shear strength of the

martensite.

Unconstrained Recovery Tests

Unconstrained recovery tests are used to characterize the ability of a shape memory

alloy to recover twinning-induced deformation that is introduced at temperatures below

the martensite finish temperature (Mf) by heating the material through its transformation

temperature. The shape recovery is commonly determined by measuring the amount of

strain introduced into a sample during deformation below the Mf, heating and cooling the

sample through a full hy steresi s b ack to room temperature.47-5 1 59-62

Unconstrained recovery tests NiTiPd

The Einal dimensions of the sample are measured and compared to the initial

dimensions to calculate the amount of strain recovered. However in monitoring the strain

changes continuously during this process as shown in Figure 4-29, it is clearly evident

that a number of different mechanisms are acting to contribute to the overall strain

recovery of NiTiPd alloys as first reported by Lindquist.l For comparison to prior work,

the total recovery of this alloy is plotted vs. initial plastic strain in Figure 4-30O Total

Recovery. The various contributions to the load free strain response of an alloy heated

from 100oC to 400oC then finally cooled back down to 100oC is shown in Figure 4-29.

Upon unloading the sample partially recovers a portion of the strain elastically often

termed the elastic spring back; however, although this region of the recovery curve is

often included in the analysis of an SMA's total recovery it in itself is not a true SMA

characteristic as the underlying mechanism is not a consequent of a thermoelastic

transformation. The strain recovery processes for the load free material occurs in three

distinct reactions. 1.) The thermal expansion coefficient of the deformed martensite




Full Text

PAGE 1

ADVANCED HIGH-TEMPERATURE SHAP E-MEMORY ALLOY DEVELOPMENT AND THERMOMECHANICAL CHARACTE RIZATION OF PLATINUM AND PALLADIUM MODIFIED NiTi BASED SMAs By ORLANDO RIOS A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Orlando Rios

PAGE 3

To my father and family

PAGE 4

iv ACKNOWLEDGMENTS I thank my mother and father. I thank my family for not letting distance separate us. I thank the great friends I have made he re. I would like to th ank Dr. Nathal and Dr. Ronald Noebe of the NASA Glenn Research Ce nters Advanced Stru ctures Division for their support, guidance, materials and proces sing and unlimited use of the divisions characterization and mechanical testing facilitie s. Without their suppo rt I would not have an acknowledgments section to write nor would I have as in teresting a study. I would like to thank the kind effort of my comm ittee members. I would like to thank Dr. Donnelly for her support and guidance at all times, and all of my colleagues here and afar.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT.....................................................................................................................xiii CHAPTER 1 INTRODUCTION AND BACKGROUND.................................................................1 Significance..................................................................................................................1 Background...................................................................................................................3 General Shape Memory Alloy Behavior...............................................................3 SMA Structural Characteristics.............................................................................4 SMA Mechanical Behavior.................................................................................10 Thermoelastic Shear Transformations.................................................................12 2 MATERIALS PROCESSING AND PROCEDURES...............................................22 Melting Procedures.....................................................................................................22 Arc Melting.........................................................................................................22 Arc Melt Machining............................................................................................22 Induction Melting................................................................................................23 Homogenization..................................................................................................23 Extrusion..............................................................................................................24 Stress Relief Heat Treatment...............................................................................25 Characterization Procedures.......................................................................................25 Dynamic Modulus...............................................................................................25 Compositional Analysis.......................................................................................26 Nitrogen, Oxygen, Carbon and Sulfur Analysis..................................................27 Thermal Analysis.................................................................................................28 Microstructural and Semi-Quantitative Compositional Analysis........................30 Dilatometry Measurements.................................................................................31 Resistivity Measurements....................................................................................31 Sample instrumentation................................................................................31 Resistivity apparatus....................................................................................32

PAGE 6

vi Thermomechanical Testing.................................................................................37 Thermomechanical instrumentation.............................................................37 Uniaxial isothermal mechanical tests...........................................................39 Load free strain recovery tests.....................................................................40 Load bias test................................................................................................40 3 ALLOY DEVELOPMENT........................................................................................50 Characterization of the NiTiPt SMA system..............................................................50 Bulk Compositional Analysis..............................................................................50 Transformation Temperature...............................................................................51 Microstructure.....................................................................................................53 Alloy Selection : NiTiPt.............................................................................................55 Alloy Selection : NiTiPd............................................................................................57 Alloy Selection Summary...........................................................................................60 4 RESULTS AND DISCUSSION.................................................................................68 Heat Treatment Optimization.....................................................................................68 Characterization..........................................................................................................70 Materials Characterization...................................................................................70 Properties and Transformation Temperatures.....................................................70 Thermomechanical Testing........................................................................................74 Isothermal Stress-Strain Behavi or in Tension and Compression........................74 Isothermal stress-strain behavior in tension and compression NiTiPd........75 Dynamic elastic modulus determination NiTiPd.........................................82 Isothermal stress-strain behavior in tension and compression NiTiPt.........83 Unconstrained Recovery Tests............................................................................86 Unconstrained recovery tests NiTiPd...........................................................86 Unconstrained recovery tests NiTiPt............................................................89 Constant-Load, Strain-Temperature Tests and Work Output..............................90 Constant-load, strain-temperature tests and work output : NiTiPd..............91 Constant-load, strain-temperature tests and work output:NiTiPt.................96 5 SUMMARY AND CONCLUSIONS.......................................................................116 Alloy Development...................................................................................................116 Characterization and Thermomechanical Testing....................................................117 Conclusions Relevant to Alloy Design.....................................................................120 Future Studies...........................................................................................................121 APPENDIX A NiTiPd HTSMA MATERIAL DATA SHEET........................................................123 Physical Properties....................................................................................................123 Electrical Resistivity..........................................................................................123 Thermal Coefficient of Resistivity....................................................................123

PAGE 7

vii Coefficient of Thermal Expansion....................................................................123 Shape Memory Properties.........................................................................................123 Transformation Temperatures...........................................................................123 Composition..............................................................................................................123 B NiTiPt HTSMA MATERI AL DATA SHEET.........................................................124 Physical Properties....................................................................................................124 Electrical Resistivity..........................................................................................124 Thermal Coefficient of Resistivity....................................................................124 Coefficient of Thermal Expansion....................................................................124 Shape Memory Properties.........................................................................................124 Transformation Temperatures...........................................................................124 Composition..............................................................................................................124 C CHEMICAL ANALYSIS OF EXTRUDED MATERIALS....................................125 LIST OF REFERENCES.................................................................................................126 BIOGRAPHICAL SKETCH...........................................................................................130

PAGE 8

viii LIST OF TABLES Table page 1-1 Several characteristics comm on to metallic thermal SMA......................................13 3-2 Aim and measured compositions of all alloys investigated.....................................61 3-3 Transformation temperatures of alloy set.................................................................62 3-4 Semi-quantitative EDS analysis of the various phases observed.............................64

PAGE 9

ix LIST OF FIGURES Figure page 1-1 Power-to-weight ratio versus weig ht diagram for common actuator types currently used in aeronautics....................................................................................16 1-2 Idealized plot of a property change vs. temperature................................................17 1-3 Structure of the parent phase (aus tenite) and shear pha ses (B19 and B19 martensite)................................................................................................................18 1-4 Thermoelastic transformation and tw in accommodated transformation strain........19 1-5 Two-dimensional lattice sche matic of monoclinic structures..................................19 1-6 TEM micrographs of twinned a nd untwined monoclinic martensite.......................19 1-7 Effects of thermal cycling thr ough the hysteresis on the transformation temperatures of several NiTi based shape memory alloys.......................................20 1-8 Deformation and shape recovery by de twinning (twins marked with arrows)........20 1-9 Isothermal stress strain behavior of a typical SMA strained in the fully martensitic state........................................................................................................21 1-10 Stress strain behavior showing the th ree main deformation regimes active in SMAs........................................................................................................................21 2-1 Induction melted Ni19.5Pd30Ti50.5 cast ingot with attach ed hot top on a quarter inch grid ...................................................................................................................41 2-2 Heat treatment and proc essing temperature schedule..............................................42 2-3 Hot extrusion press schematic..................................................................................42 2-4 Uniaxial sample (A) 5 X 10 mm comp ression sample (B) Threaded 17.4 mm long by 3.81 mm diameter gauge sample.................................................................43 2-5 Processing flow diagram of DSC, compression, and tensile samples......................43 2-6 The ICP using an Echelle type polychrometer.........................................................44

PAGE 10

x 2-7 Example of a DTA scan showing the exothermic and endothermic peaks characteristic of thermoel astic shape memory alloys...............................................44 2-8 Four-point probe resistivity configuration...............................................................45 2-9 Raw (blue) and conditioned (yello w) voltage signals for resistivity measurements during inductive heating...................................................................45 2-10 Resistivity vs. temperature pr ofile with regression analysis....................................46 2-11 High conductivity Pt wire resistivity vs temperature relationship demonstrating the repeatability of the du ring heating and cooling..................................................46 2-12 NIST (resistivity standard) resistivit y vs. temperature profile comparison of NIST measurements and the measurem ents by the resistivity apparatus.................47 2-13 Resistivity apparatu s data flow diagram..................................................................48 2-14 Materials Testing Systems (MTS) tensile frame fitted with high-temperature hot grips and induction heating confi gured for compressive testing..............................49 3-1 Ternary plot of the Ti-Ni-Pt compositi ons studied. The comp osition of all alloys was confirmed by spectrographic analysis...............................................................61 3-2 A. Effect of Pt on the Ms transforma tion temperatures for Ni50-xPtxTi50 alloys, including data from previous researcher s B. Effect of Pt on all transformation temperatures for Ni50-xPtxTi50 alloys....................................................................62 3-3 SEM BSE micrographs of th e non-stoichiometric alloys........................................63 3-4 Phase diagrams (A) NiTi binary phase diagram from reference (B) TiPt binary phase diagram from reference..................................................................................64 3-5 Effect of ternary alloying additions on the Ms (or Mp) temperature for NiTibased high-emperature shape memory alloy systems..............................................65 3-6 Comparison of the specific work output for several conventional NiTi alloys, SM495 NiTi, and the (Ni,Pt)Ti HITSMA................................................................65 3-7 Phase diagram of TiPdTiNi alloys.........................................................................66 3-8 Shape memory properties NiTiPd (A) Ms temperature resulting from ternary alloy additions. (B) Average shape recovery in Ti50 (Ni50-x) Pdx...........................66 3-9 Plots of martensitic transformati on temperatures vs. composition for Ti50xPd30Ni20+x 47.............................................................................................................67

PAGE 11

xi 4-1 Stress Relief Heat Treatment Optimization by Analysis of Resistivity Temperature Profiles (note the resistivity curves are offset for convenience on the same resistivity scale).........................................................................................98 4-2 SEM BSE image of extruded Ni19.5Ti50.5Pd30..........................................................99 4-3 NiTiPt Resistivity and Dilatometry Test Results.....................................................99 4-4 NiTiPd Resistivity and Dilatometry Test Results..................................................100 4-5 NiTiPd Force Strain Curve at 365 C......................................................................100 4-6 NiTiPd Alloy Uniaxial Isot hermal Tensile Tests at RT, 200 C, 300 C and 400 C (a) Engineering Stress Strain Curve (b) True Stress Strain Curve including correction for non-uniform deform ation of the 400C sample................................101 4-7 NiTiPd Alloy Uniaxial Isothe rmal Compression Tests at RT, 200 C, 350 C, 365 C, 400 C and 500 C (a) Engineering Stress Strain Curve (b) True Stress Strain Curve............................................................................................................101 4-8 NiTiPd Alloy Uniaxial Is othermal Tensile Tests at 225 C, 245 C, 255 C and 272 C......................................................................................................................102 4-9 NiTiPd Alloy Uniaxial Isot hermal Compression Test at 255 C, 272 C and 300 C......................................................................................................................102 4-10 Isothermal Uniaxial Stress Strain Cu rve with Resistivity Exhibiting a Stress Induced Transformations........................................................................................103 4-11 NiTiPd Yield Stress vs. Temper ature in tension and compression........................103 4-12 Temperature Dependent Dynamic El astic Modulus measured on heating............103 4-13 NiTiPt Alloy Uniaxial Is othermal Tensile Tests at 440 C, 470 C, 550 C and 600 C......................................................................................................................104 4-14 NiTiPt Alloy Uniaxial Isot hermal Compression Tests at 440 C, 470 C, 490 C, 550 C, and 600 C...................................................................................................105 4-15 NiTiPt Alloy Uniaxial Is othermal Tensile Tests at 200 C, 380 C, 400 C and 440 C......................................................................................................................105 4-16 NiTiPt Alloy Uniaxial Isot hermal Compression Tests at 200 C, 380 C, 400 C and 440 C...............................................................................................................106 4-17 Stress Strain Curve at 500 Cels ius at Low and High Strain Rates.........................106 4-18 Yield Stress vs. te mperature for NiTiPt.................................................................107

PAGE 12

xii 4-19 Fracture Stress and Strain vs. temperature for NiTiPt...........................................107 4-20 NiTiPd Unconstrained Recovery Test at 4 and 2 Percent Initial Strains...............108 4-21 Total Recovery Rate vs. Total Strain for NiTiPd...................................................108 4-22 Load Free Recovery Individual Compone nts of Total Recovery for NiTiPd........ 109 4-23 Temperature Dependent Load Free Rec overy Curve for Complete NiTiPt Test...109 4-24 Load Free Recovery Individual Com ponents of Total Recovery for NiTiPt......... 110 4-25 Load Bias in Tension (Sp ecific Work Output) for NiTiPd....................................110 4-26 Load Bias in Compression (Sp ecific Work Output) for NiTiPd............................111 4-27 Load Bias in Tension (Specific Wo rk Output) Complete Thermomechanical Path for NiTiPd......................................................................................................111 4-28 Specific Work vs. Biasing Load for NiTiPd..........................................................112 4-29 Transformation Strain (2nd cycle martensite to austenite) vs. Biasing Load for NiTiPd....................................................................................................................112 4-30 Open Loop Strain vs. Biasing Stress for NiTiPd...................................................113 4-31 Load Bias in Compression (Sp ecific Work Output) for NiTiPd............................113 4-32 Transformation Strain (2nd cycle martensite to austenite) vs. Biasing Load for NiTiPt.....................................................................................................................114 4-33 Specific Work vs. Biasing Load for NiTiPt...........................................................114 4-34 Open Loop Strain vs. Biasing Stress for NiTiPt....................................................115

PAGE 13

xiii Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science ADVANCED HIGH-TEMPERATURE SHAP E-MEMORY ALLOY DEVELOPMENT AND THERMOMECHANICAL CHARACTE RIZATION OF PLATINUM AND PALLADIUM MODIFIED NiTi BASED SMAs By Orlando Rios December 2006 Chair: Luisa Dempere Major Department: Materials Science and Engineering A series of Ti-Ni-Pt and Ti-Ni-Pd high temperature shape memory alloys (HTSMAs) have been examined in an effort to find alloys with a suitable balance of mechanical and physical properties for appl ications involving el evated temperature actuation. Initially, more than 20 Ti-Ni-Pt alloys were prepared by arc melting high purity materials followed by a homogenization heat treatment under va cuum. Each alloy was then characterized using optical and scanning electron microscopy and differential scanning calorimetry. A strong link to stoichiometry and transf ormation temperatures was not evident which indicates that a very limited solubil ity for off stoichiometry compositions exist with in the B2 and B19 structures. The result s from this study combined with the results of an advanced thermomechanical pro cessing study conducted by colleagues at the NASA Glenn Research Center we re used to select the Ni25Ti30Pt25 alloy for more

PAGE 14

xiv extensive investigations of their structure, property and processing relationships. For comparison, a Ti-Ni-Pd alloy, namely Ti50Ni30Pd20, was selected from literature because it was found to have a maximum in the unc onstrained recovery behavior. These materials were extruded characterized by advanced thermomechanical testing by measurement of the baseline mechanical properties and shape memory specific behaviors. In both alloys the work output reached a maximum as a function of applied stress (biasing load) as did the transformation stra in. Through thermomechanical testing it was evident that slip mechanisms were detrim ental to the performance of these alloys performance as actuator materials. In both al loys the resistance to slip under a biasing load in the temperature regions near the tr ansformation temperatures prevented complete recovery thus limiting the work performan ce of these alloys. A link between the difference in yield strength between the au stenite and martensite and the performance under a biasing load was confirmed which is a goo d indicator in further alloy selection.

PAGE 15

1 CHAPTER 1 INTRODUCTION AND BACKGROUND Significance A number of metallic alloys have been s hown to exhibit the shape memory effect. Wayman gives a broad definition of shape memory alloy (SMA) which encompasses the bulk of all thermal shape memory alloys. He defines a thermal shape memory alloy as an article that when deformed at a lower temp erature will regain its original shape when heated to a higher temperature.1 Shape memory alloys are used in multiple engineering applications2. The most common commercial system is NiTi based SMAs. Applications thus far for NiTi alloys include electrical switches, eyeglass frames brassiere underwires, cell phone antennas, appliance controllers, temperature sensitive valves, microactuators and countless medical and dental devices.3,4 In addition, the first large-scale co mmercial applications for shape memory alloys were made using NiFeTi and NiNbTi alloys with sub-room temperature transformation temperatures, for use as coupl ings for pipes, tubes, and electrical interconnects.5 These applications make use of NiTi alloys near room temperature. The main reason that commercial applications have been limited to near room temperature is that commercial NiTi SMAs have a maxi mum transformation temperature of about 100C. In addition, there are many c ontrol and actuation-type ap plications for materials exhibiting the shape memory effect at hi gher temperatures. High-temperature shape memory alloys (HTSMA) could be used in the aeronautic, automo tive, power generation,

PAGE 16

2 and chemical processing industries. While sp ecific applications have been identified based on some form of a HTSM A, no suitable materials have been developed. As is common in the materials field the development of applications for advanced materials is slightly ahead of the materials development itself, and such is the case for the development of high-temperature shape memory alloys. Integration of SMA actuators into aer onautic turbomachinery would result in several inherent benefits. Aeronautics clearl y emphasizes weight redu ction in all stages of engineering. Reduction in the net weight re sults in sizable gains in fuel efficiency. Additionally, SMA actuators decrease the number of subsystems as compared to standard pneumatic or more common hydraulic and motor-driven actuators, providing further reductions in weight and cost. Figure 1-1 shows the typical wei ght to power ratios of the more common commercial actuators curre ntly used by the aerospace industry.6 Minimizing weight and maximizing power resu lts in a performance index in which SMA actuators are clearly superior. The design and development of actively controlled SMA devices requires in-depth characterization of the mechan ical and shape memory specifi c properties. Past studies have accounted for compositional effects of transformation temperatures and, in some cases, load-free recovery, yet there is a comple te lack of data required for the application of shape memory alloys partic ularly in actuator-related appl ications. This study attempts to characterize these properties and correlate them to material composition and microstructure, which in turn can be used to identify possible areas for further alloy and process development.

PAGE 17

3 Background General Shape Memory Alloy Behavior SMAs are characterized by a set of temperatures at which a crystallographic structural change begins and ceases. The high-temperature aust enite or parent phase is a high symmetry phase usually ordered while th e lower temperature martensite phase is a lower symmetry structure which forms from the high symmetry parent phase by a diffusionless shear transformation. The various temperatures at which this transformation begins and ends on he ating and cooling are defined as As, Af, Ms and Mf. The austenite start temperature, As, is the temperature at which the transformation of the martensite to austenite phase begins on heating. Af is the temperature at which the transformation is completed and the material is 100% austenite. The martensite start, Ms, and martensite finish, Mf, temperatures are the temperatur es at which the transformation occurs on cooling.2 Figure 1-2 is a classical schematic pres ented by Wayman which shows a material property dependent change as a shape me mory alloy is cycl ed through a thermal hysteresis.1 A discontinuity in the material properties arises at the onset of the transformation on heating or c ooling, which is characterist ic of all SMA materials. Throughout the transformation the material exhib its a reversible stru ctural change that results in a measurable change in material properties. This property may be for example specific volume, electrical resis tivity, modulus, or other struct urally dependent property. The most common test methods for the determ ination of the transformation temperatures are thermal methods (DSC and DTA) dilatome tric methods, and resistive methods. The latter two result in similar hysteresis plots as exemplified in Figure 1-2. While this figure

PAGE 18

4 is idealized it is representative of many of the relevant characteristics of thermal shape memory alloys. SMA Structural Characteristics Shape memory alloys exhibit both a th ermodynamically and crystallographically reversible transformation. A crystallographically reversible transformation is most likely when the interface between the martensite and austenite is essentially coherent and the parent austenite phase is an ordered com pound. The high-temperature austenite phase is a higher-symmetry structure, usua lly ordered cubic (B2 structur e) as in the case of NiTi and NiTi modified alloys. The austenite pha se transforms without appreciable long range diffusion into a lower symmetry martensite stru cture at some lower temperature. Simple cells schematically illustrating the struct ures common in NiTi above and below the transformation temperatures are shown in Figure 1-3.2 The cubic B2 parent phase in NiTi based SMAs transforms to a number of different martensitic structures. The final structure depends on alloying additions, impurities, and processing history7, 8. The following transformation reactions have been identified. B2 B19 B2 B19 B2 B19 B19 B2 R B19 Each reaction is crystallo graphy reversible with the exception of the B2 R transformation. The R phase is attained by {100} elongation of the B2 structure resulting in a rhombohedral structure.9 The B19 phase is an orthorhombic structure which is formed from the B2 parent cr ystal in several steps, which consist of elonga tion about the a, b and c axes and shearing of the basal plane in the c direction which is normal to the b direction. The B19 phase is a monoclinic st ructure formed by addi tional shearing of the

PAGE 19

5 B19 non-basal plane, which is normal to the a direction in the c direction.10 The B19 phase is the primary shear structure which a ppears in binary NiTi leading to the shape memory effect. The cubic parent phase in NiTi SMA is stable at high temperature (above the Af). Multiple lattice variants of martensite can fo rm from each parent austenite grain. Each variant will follow a perfectly reversible path back to the parent austenite phase due to the ordered nature of the alloy. If this did not occur, the material would undergo a diffusional transformation in the chemical orderi ng of the original pare nt phases lattice. In the Pt and Pd modified NiTi SMAs th e B19 phase is of primary importance since alloying additions greater than about 10% results in the primary martensitic structure switching from the monoclinic B19 to the orthorhombic B19.11 The B19 and B19 are shear structures of the B2 and therefore e xhibit a lower symmetry, with the B19 having the lowest symmetry.12 The symmetry of the structur e is of importance as it is the underlying factor in the determ ination of the number of equi valent martensitic variants which may form from a parent B2 cubic structure. The B 19 has 12 equivalent variants that may form from a parent crystal. This tr anslates to 12 different ways to shear the B2 structure in the formation of the B19. Each equivalent B19 structure may then be sheared along the non-basal plane (001) in the positive and ne gative c directions to form a B19 martensite which results in 24 equivalent variants.13 In general when analyzing shear structures the number of equivalent variants increases as the symmetry of the shear structure decreases. This relationship will be examined in further detail as it relates to the deformation behavior and mechanisms in the shape memory alloys encompassed in thus study.

PAGE 20

6 The fundamental theory of martensite formation, which applies to both thermoelastic and thermoplastic transformations relates a high symmetry parent structure to a variant of the shear structure. Although th is theory is applicab le to many shear phase transformations for descriptive purposes we w ill focus on the structures shown in Figure 1-3. Three major deformations steps are required to relate to structures by a purely shear transformation.14 (1) Primarily there is a Bain distortion which attains its name historically from the distortion obser ved in the thermoplastic transformation which forms the metastable tetragonal martensite. The Bain distortion is simply the elongation of the parent phase which is shown in Figure 1-3 as the elongations of the a, b and c axis. (2) Secondly a shear deformation must occur in order to preserve the lattice symmetry which combined with the Bain distortion form s the undistorted plane or habit plane. (3) Finally there is a rotation which brings the undistorted plane into th e same orientation in both the parent and shear phase. The accommodation of arbitrary shearing of the lattice (noted as step 2) is a decisive factor defining whethe r a shear transformation is thermoelastic or thermoplastic. Thermoplastic transformations such as thos e common in steels are non-reversible and the majority of the transformation shear associated with step 2 is accommodated plastically. That is, there is the formati on of non-reversible defects, su ch as dislocation motion and generation. Thermoelastic transformations on the other hand accommodate the majority of the transformation shear elastically in co mbination with recoverable mechanisms. There are three main ways the transformation shear may be accommodated, two of which are reversible. The active shear mech anism depends in part on the mechanical properties of the austenite and martensite as well as the magnitude of the transformation

PAGE 21

7 shear. In both thermoelastic and thermoplas tic transformations a significant amount of shear is accommodated elastically which is dependent on the yield strength of the martensite and austenite adjacen t to the interface. If the tr ansformation strain results in an interface stress state which exceeds its local yield streng th transformation strains are accommodated irreversibly by plastic deformation or slip. Finally, if the shear stress at the interface required to initiate deformation twins is sufficiently low and higher than the yield strength the transformation strain is accommodated by twin formation. Twin accommodated strain, which is the keystone of shape memory alloys, forms during the transformation along th e twin planes in the shear (martensite) phase. It is important to note that the twinning plane is us ually a low index plane that is not parallel to the habit plane. Figure 1-3 is a schematic of th e nature of a thermoelastic transformation interface between parent and shear phase s in which the transformation strain is accommodated by twin formation. Th e macroscopic shear plane, which separates the cubic and shear structures is dependent on the structural relationships between the martensite and austenite which include Bain shear and rotation. The twinning plane however is based on the symmetry of the shea r structure as the deformation twin must only reorient the structure by a consorted move ment of atoms uniformly distributed over the volume separated by the twinning plan e. The transformation shear along the macroscopic shear plane is pa rtially accommodated by the shear associated with twinning along the twin planes. In thermoelastic tran sformations the net transformation shear in the martensite is accommodated in part by th e elastic deformation of the twinned and untwinned regions and partially by the forma tion of the deformation twins. If the deformation twins did not occur the stress due to the transformation strain would exceed

PAGE 22

8 the yield strength of the material and thus be accommodated plastically. Twins that form during thermoelastic transformations, in order to accommodate the shear along the macroscopic shear plane, have a twin related crystallographic relationship to one another. As exemplified in Figure 1-4 the sense of the shear asso ciated with a twin must alternate between twinned regions. This criterion for the accommodation of the transformation strain results in the formation of coup led pairs of twins referred to in the phenomenological theory of martensite fo rmation as correspondent variant pairs.15-20 Although many variant pairs may form from an austenite crystal each variant pair is equivalent, thus it is possible for the aust enite to transform to a single correspondent variant pair which accommodates the transfor mation strain by twin formation. This however does not occur in an un-biased (no exte rnal stress) sample. What is observed is that austenite transforms into a more or less random orientation di stribution of variant pairs. An addition mechanical constraint must be considered to examine the driving force for the observed distribution of variant pair s. The underlying mech anism driving such a distribution stems from the minimization of the macroscopic shape of the bulk material.15 The transformation product of the austenite is in different regions of the crystal transforms in such a manner that there is no macroscopic shape change. This behavior is referred to as self accommodation. Self accomm odation is an arrangement of martensitic variants such that the sum of their disp lacements within the boundary suffers no net displacement. It is possible to place a self accommodating arrangement within an austenitic matrix and not induce any macros copic strains. Self accommodation is a fundamental characteristic of all thermoelastic transforma tions as it minimizes interface stresses assuring interface coherence and el astic accommodation of strains. In other

PAGE 23

9 words, if self accommodation did not occur as the interface progresse s the stress in the interface would continue to rise and quickly su rpass its yield strength resulting in plastic deformation and a loss of coherency. Such is the case in twinned ferritic martensites, where although the twinning process is reversib le, the arrangement of twins is such that the interface stresses result in plastic deformation and thus a non-recoverable transformation. In shape memory alloys the thermoelas tic transformation results in a twinned structure. Figure 1-5 shows a simplified represen tation of two equivalent monoclinic variants separated by a twin boundary. Alt hough more strain coul d be accommodated by the additional translation associated with the formation of an incoherent twin the interfacial energy of a cohere nt twin boundary is on average an order of magnitude lower than the energy of an incohe rent twin thus additional energy is required to form incoherent twins.16 As a result the formation of coherent twin interfaces between martensite variants is thermodynamically favor able and additional en ergy is required to form incoherent twins. Figure 1-6 are TEM micrographs of a binary NiTi shape memory alloy.2 This figure shows the differences between a twinless martensitic structure (6.a.) and a finely twinned structure (6.b.). Both structures are monoclinic diffe ring only by the presence of a fine distribution of deformation twins that form during the phase transformation. The detwinned structure shown in micrograph 5.a resulted from deformation by detwinning of the martensite. Another fundamental aspect of shape memory all oys is deformation of the martensitic structure through detwinning.

PAGE 24

10 A completely reversible structural tr ansformation requires that the parentmartensite interface be glissile in the forw ard and reverse directions. In addition the thermal hysteresis must be small. Ling a nd Owen have shown that sessile dislocation loops and other defects in the matrix f acilitate the movement of the interface.17 Furthermore these sessile defects increase th e plastic flow stress of the matrix hence making the accommodation of strain by slip more difficult. Mechanically the matrix is effectively strengthened and the energy require d to move the parent martensite interface is lowered. This has been correlated to NiTi bases SMAs as well as other SMA systems. The density of these defects increases with thermal cycling up to a limit resulting in a decrease in transformation temperatures w ith increased thermal cycles. Decreasing transformation temperatures with thermal cyc ling has been observed experimentally as shown in Figure 1-7.18 SMA Mechanical Behavior Metals that exhibit a thermal shape me mory effect deform through twin boundary motion. Recoverable deformation of the mart ensite by twinning reac tions must occur at stresses lower than those for non-crystall ographically reversible reactions. Noncrystallographically reversib le reactions include disloc ation generation and motion. Structurally, twins are formed in the mart ensite during the forw ard reaction separating equivalent variants. Multiple martensite va riant formation is driv en by the minimization of the net transformational stresses. Hence, twins are present in the microstructure after transformation; therefore, nucleation by an appl ied shear is not necessary in contrast to standard deformation twinning. Deformation occurs by the growth of variants most favorably aligned with the largest principle shear component or Schmidt factor at the expense of those with the lo west component. This mechanism is commonly referred to

PAGE 25

11 as detwinning. Deformation by this mechan ism decreases the number of twins in the alloy. This is schematically shown in Figure 1-8. The arrow indicat es twin planes in this schematic. An aligned shear results in the twin boundary motion in the direction normal to the shear and the growth of a correspondi ng variant. Upon fully detwinning the alloy, the material theoretic ally exists as a si ngle variant although the extent of detwinning depends on crystal structure and the associated number of equi valent variants as well as the existing variant distribu tion prior to deformation. The typical macroscopic mechanical be havior of a shape memory alloy is represented in Figure 1-9. This figure is a schematic of the general stress strain curves exhibited in these systems below the Mf temperature. It should be noted that the alloys composition, and mechanical and thermal histor y may change this curve. It is also possible to have multiple active deformation mechanisms, which will affect the work hardening rate during the detw inning region of this curve. This figure represents the ideal case for the shape memory effect. The initial portion of the stress strain curv e is attributed to elastic deformation of the undeformed martensite. Upon reaching a cri tical stress, detwinning of the martensite begins. The detwinning stress is independent of twin density and therefore a region in the stress strain curve exists in which the st ress required to deform the material is independent of strain. A critical level is reached at the point where favorably oriented variants are most prevalent in the microstr ucture and thus reacta nts of the detwinning reactions that supply the growth of the favorably oriented tw ins are consumed. At this point twins with similar Schm idt factors may impede on each other. The result is an increase in stress-strain rela tionship that is attributed to elastically deforming the

PAGE 26

12 detwinned martensite. A second yield point is evident at whic h the critical stress for slip is reached. Non-reversible deformation mech anisms are active in this region thus the strains are not recoverable by shape memory processes. Figure 1-10 is a series of hypothetical stressstrain curves that graphically represent the three distinct deformation behaviors ex emplified by SMAs and the temperatures at which they may be active.19 At temperatures above the Af the shape memory alloy is austenitic and deformation occurs by elastic loading of the austenite followed by slip. Below the Mf temperature the shape memory alloy is fully martensitic and deformation occurs by the detwinning mechanisms de scribed above. SMAs demonstrate an extraordinary superelastic effect which occurs when the material is deformed above the Ms temperature and below the Md temperature (Figure 1-10). The Md temperature is defined as the temperature at which mech anical stresses can induce a martensitic transformation. Subsequent removal of the external stress results in a non diffusional reversion to the thermodynamically stable parent phase. Elastic strains attainable in NiTi shape memory alloys are 20X those of carbon spring steels. Thermoelastic Shear Transformations The thermodynamics of shape memory alloys and the relevant shear transformations is a complex subject, which involves a competition between the chemical and non-chemical driving forces. We have st ated that martensite forms from the parent phase by a purely diffusionless shear transfor mation. The transformation front progresses by shear atomic motions and the interface betw een the martensite and parent phase is coherent. Structurally the martensite result s in a net shape change of each equivalent variant. As in the case of NiTi addressed previously a cubic structure transforms to a monoclinic or orthorhombic. This net shape change results in an accommodation strain.

PAGE 27

13 The local strain around each variant can be accommodated plastically, elastically or as a mixture of both. This phenomena has been reviewed and the key thermodynamic parameters identified by Reed and Abbaschian.20 Table 1 was compiled from publications [1,16,28]. The listed characteristics are common to SMA systems. The structural a nd mechanical characteristics already have been briefly addressed. In addi tion to and as a result of thes e structural characteristics a specific set of thermodynamic properties arise. Table 1-1 Several characteristics common to metallic thermal SMAs. Structural characteristics of SMAs1 Ordered parent -> ordered martensite Martensitic transformation is thermoelastic Martensite is crystall ographically reversible Thermodynamic characteristics defined by Dunne and Wayman21 Small chemical driving force at Ms Small transformational volume and shape change High flow stress parent matrix Additional mechanical characteri stics defined by Ling and Owen 16 Parent-martensite interface must be gliss ile in both transformation directions Premartensite elastic softening A fundamental condition for the shape memo ry effect is that the transformation must occur reversibly. Accommodation of th e transformational strains adjacent to the interface could be plastic, elastic, or a mixtur e of both. In the case where the majority of the strain is accommodated elastically the inte rface is able to move in both directions referred to as a thermoelastic transformation. As a result the chemical driving force required to drive the reaction is small. This is the case in alloys exhibiting the shape memory effect, which are more precisely de fined as thermoelastic transformations. In thermoplastic transformations the transformation strain is accommodated plastically due to a low flow stress in the pa rent phase and a large transformational strain.

PAGE 28

14 In this case, the transformations on the fo rward and reverse direction occur at much higher chemical driving forces through the nucleation and rapid growth of the martensite. Typically individual shear plates are nucleat ed at defects and grow irreversibly. Subsequent thermal cycles result in new plates nucleating rather than reversible interface movement. A classic instance of such a case is the martensitic transformation of carbon steels where the thermal hysteresis is large and most of the transformational strains are accommodated plastically. Ortin and Planes elegantly treated the th ermodynamics of thermoelastic effects and systematically defined conditions for a thermoelastic energy balance.22 Thermoelastic transformations are driven by the chemical free energy. At equilibrium it would be expected that the transformation occurs wh en the chemical free energy of the parent phase is a small amount larger than that of the shear phase. This however has been shown not to be the case. In actuality, th e chemical driving force in thermoelastic transformations are opposed by non-chemical fo rces, thus the equilibrium transformation occurs when these forces are nearly e qual. Following the notation and approach presented by Ortin and planes, thermodynamic equilibrium is represented by the following equation. Gp..m is the molar free energy of transformation, Gch is the molar chemical free energy and Gnch is the molar non-chemical energy. At equilibrium the molar free energy of transformation is equal to zero thus the ch emical contributions are equal to the nonchemical contributions. G p..m G ch G nch 0

PAGE 29

15 The non-chemical contributions consist of several factors, the most prominent are the elastic free energy and work done against frictional forces. The elastic and frictional terms consist of se veral components. It was stated that a condition for thermoelastic transformations is that the majority of the transformational strains are accommodated for elastically. Th erefore adjacent to each interface we have a finite amount of stored elastic energy. In addition there exists an interfacial energy associated with the parent martensite interf ace as well as the interfaces between variants (twin boundaries). Both contributions are re versible therefore th ey have been grouped into the elastic term even t hough the interfacial ener gy is not truly an elastic contribution. This is in line with the convention set forth by Ortin and Planes. Frictional energy losses are non-reversib le losses primarily due to interface movement. This term may be treated as irreversible work done on the system. Three significant parts have been id entified by Olsen and Cohen.23,24 These include 1) frictional stresses required to move interfaces 2) irreversible free energy related to defects induced during the transformation 3) frictio nal stresses required to move interfaces. It is important to note that if all or most of the ac commodation occurs plastically the elastic term will be very small and the frictional energy loss term will be the main opposing energy to the chemical driving for ce. As a result a large non-reversible hysteresis will be evident as is the case in carbon steels as mentioned earlier. The underlying mechanisms and thermodynamics which control general behavior of shape memory alloys have been review. Principally the structural relationships G nch G el E friction G ch G nch G ch G el E friction

PAGE 30

16 between the parent in shear phases result in thermomechanical properties unique to shape memory alloys. Additionally, driving forces and the relating thermodynamic for thermoelastic transformations have been brie fly discussed and compared to thermoplastic transformations. A persistent problem that will be discussed in subsequent chapters points out that plastic accommodations by non -recoverable processes hinder the sought after characteristics of shape memory alloys, pr incipally the materials ability to do work. Essentially under certain conditi ons these shape memory alloys start to behave more like thermoplastic materials where a significant portion of the transf ormation strain is accommodated plastically. The combination of structure, mechanical properties and thermodynamics are used as tools to understand deformation and propose possible deformation mechanisms while identify possible target area for future alloy improvement. Weight (Kg) 0.010.1110100 Power/weight ratio (W/Kg) 1 10 100 1000 10000 Hydraulic ActuatorsPneumatic ActuatorsDC MotorsShape Memory Metals Figure 1-1 Power-to-weight ratio versus weight diagram for common actuator types currently used in aeronautics.6

PAGE 31

17 Figure 1-2 Idealized plot of a property change vs. temperature.1

PAGE 32

18 Figure 1-3 Structure of the parent phase (austenite) and shear phases (B19 and B19 martensite).2

PAGE 33

19 phenomenological theory of martensite fo rmation as correspondent variant pairs.1520 25,26,27,28,29,30. Figure 1-4 Thermoelastic transformation a nd twin accommodated transformation strain.14 Figure 1-5 Two-dimensional lattice schematic of monoclinic structures3. Figure 1-6 TEM micrographs of twinned and untwined monoclinic martensite.3 (a) (b)

PAGE 34

20 Figure 1-7 Effects of thermal cycling th rough the hysteresis on the transformation temperatures of several NiTi based shape memory alloys.18 Figure 1-8 Deformation and shape recove ry by detwinning (twins marked with arrows).15

PAGE 35

21 Figure 1-9 Isothermal stress strain behavior of a typical SMA strained in the fully martensitic state.3 Figure 1-10 Stress strain behavi or showing the three main deformation regimes active in SMAs.31

PAGE 36

22 CHAPTER 2 MATERIALS PROCESSING AND PROCEDURES Melting Procedures Arc Melting The experimental portion of the alloy deve lopment phase of this study was initiated with a cursory examination of a series of ove r twenty alloys along and on either side of a line of constant stoichiometry (constant Ti atomic fraction of 50%) between TiNi and TiPt in order to survey the basic properties of the Ni-Ti-Pt system in a region where the potential for shape memory behavior is likel y. The experimental alloys were produced by non-consumable-arc melting of high purit y starting components (99.95% purity Ti, 99.995% purity Pt, 99.98% purity Ni) using a wa ter-cooled copper crucible in a highpurity argon atmosphere. Due to the substa ntial density differences and melting points between the starting materials, it was difficu lt to melt the platinum completely in one step. Consequently, the buttons were turned ov er and remelted 4-6 times in an attempt to insure homogeneity. Arc Melt Machining Sectioning for thermal, microstructural, and hardness measurements was performed by wire EDM (electrical disc harge machining). For example, 10mm length by 5 mm diameter cylinders were EDMd for thermal analysis from the center of the arc melted buttons. Planar specimens for microstr uctural and hardness were fabricated by transversely sectioning the arc-melted bu tton followed by pressure mounting into phenolic bound thermosetting polymers mounts. The exposed planar section was then

PAGE 37

23 ground using SiC papers and polished usi ng diamond suspensions using standard metallographic preparation t echniques, yielding a highly po lished, low residual stress surface. Induction Melting The experimental alloy of a target composition of Ni19.5Pd30Ti50.5 was produced by vacuum, induction melting of high purity starting components (99.95 Ti, 99.995Pd, 99.98Ni) in a graphite crucible, which is ti lt poured into a 25.4 mm diameter by 102 mm in length copper mold. The mass of each ingot is approximately 450 grams including the hot top, which feeds the casting during the sh rinking associated with solidification (Figure 2-1). Induction melting was chosen over arc melting in order to circumvent problems inherent to arc melting materials with large density difference as in the case of Ti 4.506 g/cm3 and Pd 12.023 g/cm3 or Ni.8.908 g/cm3. Induction melting also induces a mixing action in the melt which assures a homogenous melt. Graphite does introduce a limited amount (approximately 0.5 at.%) of carbon during the melting process resulting in the formation of carbides with the excess off-stoichiometry Ti. Homogenization Each induction melted ingot or arc melted button was simultaneously sealed in a vacuum furnace and homogenized at 1050 C for 72 h. This was followed by a furnace cool as shown graphically in step 1 of Figure 2-2. This figure summ arized the thermal history of the extruded material. The arc melted material is homogenized (step 1) and as it is not extruded or mechanically machined a stress relief heat treatment is not required and stabilization of the transformation temperatures is accomplished by multiple thermal cycles during hysteresis measurements. The homogenization heat treatm ent is employed in order to remove fine

PAGE 38

24 atomic segregation and insure full reaction to form the ordered structures with a minimum amount of ordering defect and resul ting point defects. Uniformity in the hot zone is assured though multiple thermocouples and independent zone control though several calibrated temperature controllers and gradual temperature ramp rates. Extrusion After homogenization the ingots were indivi dually sealed in steel extrusion cans. The ingot was placed into the extrusion can and a vacuum cap was sealed by tungsten inert welding. The vacuum cap specifies a cylin drical cap with a fitted mild steel vacuum tube. Through this tube the sealed cav ity was evacuated following by a crimping and spot welding operation. At this point the ca nned ingot is ready for extrusion. Following the canning operation the sealed ingots were extruded with a 7:1 reduction ratio at 900C in a hydraulic press. A schematic of the extrusion operation is given in Figure 2-3. Prior work at the NASA Glenn Resear ch Center has demonstrated the feasibility of this technique and determined these are optim al conditions for the thermo-mechanical processing of comparable HTSMAs. The extrusions were X-rayed in order to have an accurate determination of the location of SMA cores start and finish position in the steel covered rod and the excess ends of the extrusion removed by abrasive cuttin g. The extrusion rods were then cut into various lengths. Compressive samples were fabricated by wire EDM methods and centerless grinding, yielding 5 mm diam eter by 10 long samples (Figure 2-4). Additionally 1 X 4 mm rectangular samples, x mm in length were sectioned by wire EDM from the core of the extruded bars for resistivity measurements Finally, a section of the extrusion is

PAGE 39

25 removed for ICPS (spell out) by which the bulk alloy composition was measured and documented. Tensile samples were fabricated by turn ing, using a computer controlled lathe (C&C). The canned extrusion was secti oned by EDM into 50.80 mm sections, which were individually center punched. The cut se ctions were loaded w ith the extrusion can still lining the sample into the C&C lathe. The entire machining procedure was performed in a cutting mode which comple tely excludes costly grinding operations. Successive cutting passes were made to yield threaded 17.4 mm long by 3.81 mm diameter gauge sections (Figure 2-4). A summary of all th e sample preparation steps and methods are graphically represente d in the flow diagram shown in Figure 2-5. Stress Relief Heat Treatment To complete the sample preparation phase, all samples are given a stress relief heat treatment at the Af plus 200 C for 1 hour followed by a furnace cool. The function of this heat treatment is to relieve any re sidual stresses on the surface of the samples resulting from the extensive machining ope rations during the fabrication stage. Structurally this heat treatment is perfor med in the austenite ph ase thus alloying for recovery in the high symmetry phase. Th is heat treatment was optimized based temperature resistivity measurements. Characterization Procedures Dynamic Modulus A dynamic modulus testing apparatus facili tated the determination of the modulus of elasticity as a function of temperature for each phase. A 34 mm bar was machined and fixed with an electrodynamic vibrator at one end a piezoelectric tran sducer at the other end. By locating the resonance frequency as a function of temperature it was possible to

PAGE 40

26 calculate the dynamic modulus as a function of temperature during heating. The sample was heated at 10 C/min in a convection furnace to a maximum temperature of 800 C. A detailed description of the melting, heat treatments, extrusion, and final machining of thermo-mechanical Ni19.5Pd30Ti50.5 samples is outlined here. As these alloys behave in a similar manner and in orde r to allow for direct comparison between the Pt and Pd modified alloys, the processing sc heme was intentionally followed as closely for both materials. The Pt modified alloy Ni24.5Pt25Ti50.5 selected for this study was processed in parallel following a comparable processing scheme as the Pd modified alloy described above and as outlined in Figure 2-5. The only signifi cant differences is that since the austenite has a lower flow stress in the Pd modified alloy the extrusion pressure is slightly lower. All melting and processing steps are comparable. Compositional Analysis The bulk alloy compositions were determ ined by inductively coupled plasma spectroscopy and the interstitial impurity con centrations were determined using standard LECO O/N and C/S determinators. For this analysis, samples were prepared by first cutting buttons into smaller sections of approximately 100mg followed by a petroleum ether rinse in order to remove any surface contamination. These sections were then air dried, reweighed and placed in a Teflon digestion vessel. 3 mL HCl, 1 mL HNO3, and 1 mL HF (all concentrated and trace metal grad e) were added to the vessel which was then placed in a block digester at 100-130 C until dissolution was complete (generally 1-2 hours). Finally, the dissolved sample was tr ansferred to a 100 mL volumetric flask for ICP analysis.

PAGE 41

27 The ICP solutions were analyzed for alloying elements and impurities using the Varian Vista-Pro Inductively Coupled Pl asma (ICP) Emission Spectrometer. The composition of the ingots was measured by inductively coupled plasma spectroscopy post heat treatment and thermo-mechanical pr ocessing to insure measurement of any additional contamination that may have b een picked up during these procedures. A schematic of the instrument is shown in Figure 2-6.32 In this process, the solution is injected by capillary action into an argon plasma. The sample is ionized and ex cited by the plasma, which results in each element emitting a characteristic wavelengt h. The emitted photons are allowed to pass into the system though the entrance slit and are diffracted by a fixed grating Echelle polychrometer, and finally the intensity of each wavelength is measured using a CCD (charge coupled device) detector. The detect or is capable of simultaneously detecting up to 73 different elements in the range of 167 785 nm. Several emission lines are selected for quantitative analysis of each element by extrapolation of measurements using calibrate d NIST-traceable solution standards. The emission intensity at each wavelength is propor tional to the concentr ation of the element present in the solution, which is determined fr om extrapolation of the calibration curves. The mean intensity over several peaks is used in calculating the repor ted concentration of each element. Nitrogen, Oxygen, Carbon and Sulfur Analysis Again, samples of approximately 100mg were analyzed for oxygen and nitrogen using the Leco TC-436 Nitrogen/Oxygen Determin ator. In this technique, a graphite crucible is baked out by a high current in order to expel any gas trapped within the crucible. The sample is then placed into the crucible and the chamber is evacuated.

PAGE 42

28 Resistive heating raises the cr ucible and sample temperatur e thus causing the sample to expel any nitrogen and oxygen. A gas solid reaction occurs between the oxygen and the carbon crucible which yields CO. Subsequent ly the CO is passed through a rare earth copper oxide reactant producing CO2 which is measured by an IR cell. The CO2 is separated from the remaining gas by as carite absorption which is a powerful CO2 absorber. Finally, nitrogen is measured by a thermal conductivity cell which measures the temperature difference between a heat so urce and a thermocouple which is protected from radiative heating. Samples parallel to those used for O/N analysis were analyzed for carbon and sulfur using the Leco CS-444LS Carbon/ Sulfur Determinator by the combustion instrumental method. In this method, the sample is combusted in the presence of oxygen yielding carbon dioxide and sulfur dioxide. The amounts of CO2 and SO2 are determined by measuring the absorption of specific IR wavelengths, wh ich are propor tional to the partial pressures of these gasses. Thermal Analysis Transformation temperatures were determ ined by differential scanning calorimetry (DSC) for low to intermediate temperature analyses and differential thermal analysis (DTA) was used for intermediate to high temperature transformations. DSC Data analysis was performed using the TA Universa l Analysis 2000 Software package. Each alloy was cycled through two full thermal hysteresis cycles (approximately Af +100 C to Ms-100 C) assuring the reproducibil ity of the thermoelastic transformations. It was experimentally determined that in thes e arc melted and homogenized samples, the transformation temperatures stab ilize after two cycles and ther efore remain constant with

PAGE 43

29 subsequent cycles. The acqui red data is formatted as th e temperature dependent heat flow as exemplified in Figure 2-7. A combination of t echniques was employed in order to circumvent temperature limitations of each instrument by maintaining the measurements within the designed temperatur e intervals thus assuring high precision measurements. For both techniques, a heati ng and cooling rate of 10 C per min was used. DSC and DTA are common techniques used in determining the thermal properties of materials with a high de gree of precision. Although both are differential methods, the DSC in contrast to the DTA calculates heat flow directly duri ng heating, cooling or isothermal holds by measuring the amount of energy required to maintain the specified temperature, while the DTA measures differences in temperature between a sample and a reference while heating both at a constant rate. Therefore, the main differences between the data acquired from a DTA and that of the DSC is that the determination of heat flow requires calculations from standards of known heat capacities similar to the unknown sample and the sample heating rate is not as precisely controlled. As a consequence of these differences, it is generally considered difficult to directly compare the thermodynamic properties measured without stringent calibration. The application of the DSC and DTA in the curren t SMA study is solely for the determination of the transformation temperatures without focusing on the heat capacities or the magnitude of the transformation enthalpy. Characterization of the transformati on temperatures through DSC and DTA measurements were made by the extrapolat ed onset method by which the transformation temperatures are recorded as the intersection of the base line and the best fit of the linear

PAGE 44

30 portions of the increasing and decreasing regions of the exothermic or endothermic peaks in the lambda type curve.33 A characteristic of the extrap olated onset method is that the relative amplitudes of the heats of transforma tion as a function of temperature determines the transformation start and stop temperatures independent of the absolute or actual transformation enthalpy, facilitating the comparison of transformation temperature measurements made on a DSC and DTA. Thermoelastic transformations, which are ch aracteristic of metallic shape memory alloys, involve displacive shear transformations. This type of transformation exhibits a strong temperature and stress dependence and rela tively fast transforma tion kinetics. Fast transformation kinetics result in the react ion reaching its temperature and stressdependent equilibrium rapidly in contrast to diffusional transformations, which require extended times. As a result, thermoelastic transformation temperatures are not highly time or more specifically heating rate de pendent unlike diffusional transformations. Consequently, although the heating or cooling ra te of a thermoelastic sample might vary slightly in a DTA where only the hot zones temperature is precisely controlled, the kinetics of the reaction are so fast that th e dependence of the transformation temperature on these slight variations are insignificant. Therefore it is feasible to compare the transformation temperature measurements of thermoelastic transformations in a DSC and DTA. Microstructural and Semi-Quantit ative Compositional Analysis The sectioned and polished alloys were examined using a JEOL 6400 scanning electron microscope (SEM) using backscat tered electron (BSE) mode. An annular detector was used in order to maximize the signal. The sample surfaces were kept normal to the 15 KeV beam and parallel to the dete ctors exposed face. This configuration

PAGE 45

31 maximizes phase contrast and minimizes any topographical effects by measuring the sum of the back scattered electrons around a 360 ring perpendicular to the electron beam and normal to the samples surface. All co mparable imaging was done at the same magnification in order to allow for simp le comparisons. Semi-quantitative ZAF corrected EDS analysis of the variou s phases was performed using the NiK TiK and PtM lines. Dilatometry Measurements The transformation temperatures were al so determined by m easuring the straintemperature response of the material under e ssentially zero load (s tress free condition) using a differential thermal dilatometer. This instrument compares the measured change in length of a test specimen to that of a standard as a function of temperature thus allowing for correction for any thermal expansion in the apparatus itself. In this measurement, the furnace heating rate is controlled at 10 C per minute and the sample temperature is measured directly by ther mocouple contact. A differential thermal dilatometer is used to measure the transfor mation strain in the longitudinal direction. Several cycles through the transforma tion hysteresis are performed until the transformation strain vs. temper ature relationship stabilizes. This technique is used to determine the transformation strains associat ed with the SMA reaction as well as the thermal expansion coefficient of the parent and austenite phases. Resistivity Measurements Sample instrumentation 1 X 4 X 20 mm rectangular samples were sectioned by wire EDM from the extruded bars described in a previous secti on and prepared for el ectrical resistivity measurements. The surface was cleaned by light polishing prior to th e stress relief heat

PAGE 46

32 treatment. The sample was instrumented with a four point probe and a K-type thermocouple as shown schematically in Figure 2-8. Two pairs of Pt or Ni drawn wires were spot welded to the sample functioning as the voltage sensing and current excitation leads. The current excitation leads were spot welded near the ends of the sample while the voltage sensing leads were spot welded further in from the ends between the two current excitation leads. The sample dime nsions (cross-sectiona l area) and distance between the voltage sensing leads was m easured using a vernier caliper. Resistivity apparatus Test specimens were heated through a thermal hysteresi s in an ATS 3200 series split tube furnace at 10 C/min followed by a subsequent furnace cool in an ambient atmosphere. Furnace control was achieved via a Eurotherm programmable PID (proportional integrating differe ntiating) temperature controll er. Thermal cycling though a thermal hysteresis was achieved by running a 3 leg program consisting of a ramp of 10 C/min to the final temperature followed by a step command which turns off the power to the furnace until the minimum specified te mperature was attained and finally an end command which signals the end of a thermal cycle. The independent programmable control loop in the resistivity apparatus is th e call for power from the furnace controller. Alternatively the resistivity apparatus is capable of sending control commands to the furnace controller allowing for external furn ace control and integration with moving hot zone cyclic furnaces. The stress-dependent transformation te mperatures and resulting electrical properties of a material were determined by high-resolution resistivity measurements. A virtual digital instrument that acquires hi gh-resolution, real time, temperature versus

PAGE 47

33 resistance profiles of materials was developed for this purpose. Add itional integration of this instrument with an MTS uniaxial testing apparatus facilitated the direct measurement of stress-induced transformations and the eff ects of external stresses on transformation temperatures. Two operational modes, level crossing acquisition or timed acquisition, depending on the function of the measurement (cyclic thermal measurements or stress, strain, resistivity profiles) were employed. Th is instrument was designed to be automated such that, once configured, it would allo w for the unattended measurement of many cycles. The components of the measurement appa ratus include a PC, National Instrument DAQ (data acquisition card) and a NI SCXI (signal conditioning ) chassis fitted with an analog signal amplification and filtering card. Current excitation was supplied by an Agilent power supply with internal shunt resistors while th e calibrated National Instruments DAQ system and analog signal co nditioner were used to measure voltage signals and linearize the thermocouple. The di gitized signals were us ed to calculate real time resistance values for each specimen, which in turn allows for the calculation of the samples resistivity, based on sample dimensions In addition, external calibrated devices have been integrated facilitating the calibrati on of the apparatus and the documentation of calibration prior to each measurement. In prin ciple, it was required by design criteria that the system measure accurate absolute values of resistivity as well as resolve transformation temperatures thus necessitati ng external calibrations. Finally, external digital and analog channels have been progra mmed into the instrument so that it could function as the controller for a cycl ic furnace or comparable device.

PAGE 48

34 Due to the high conductivity of the SMA alloys, the resulting voltages measured across the sample were on the order of fracti ons of a mV. Signals of this magnitude are very susceptible to noise thus necessitating several signal conditioning processes. Hardware and software low pass filters have been employed in this system in order to allow for measurements to take place while the sample was under an inductive load as well as filtering out cyclic noises above 4 Hz which may enter the signal via the extensive connecting wires or at the sample itself. Figure 2-9 shows the in-situ filter response for a sample that was heated by an inductive field. The signal in this type of instrument may be separated into two main components. The primary component is the DC signal of interest which results from the interaction of the material with the excitation source and a superimposed AC component which is cause d by induction heating of other external sources. The lower blue graph shows the unf iltered signal as measured by a documenting oscilloscope while the upper curve shows the filtered res ponse which demonstrated the efficiency of the filter system at is olating the DC compone nt of the signal.. A secondary benefit of using a powerful precession external power supply is the capability of resistively heating samples by driving high current densities along their length. Such an experiment may be used to monitor the power requirements of SMAs and SMA actuators. Routines for monitoring these parameters have been written into the instrument. The processed data is sample temperature, power, current, voltage across the sample and resistance. Again, these are all mon itored real time so that this data can be used to characterize the power requirements of SMAs in several geometries. It is possible to use this routine to char acterize model and real SMA actuators.

PAGE 49

35 As mentioned previously, the instrument was designed to be automated and acquire data in two modes (level cr ossing acquisition or timed acqui sition). The level acquisition mode is used for the generation of resistivit y vs. temperature profiles. Level acquisition refers to the monitoring of a trigger signal and acquiring a new set of data points or data object when the trigger is crossed. In th e case of the resistiv ity apparatus, the temperature signal is defined as the trigger wh ich implies that when the absolute value of the difference between the transient or trigger temperature and the previous trigger value, is greater than a sp ecified amount a new data point (t emperature and resistance) is recorded. Additionally, detection of a complete thermal hysteresis is required in order to allow the separation of data in to groups forming a complete hysteresis defined as a heat and cool cycle. This again was accomplis hed by level crossing triggers where the algorithm finds a maximum followed by a mini mum temperature before triggering a new hysteresis command, which records the data and clears the dynamic memory. Timed acquisition implemented in the in tegration of stress, strain, and resistivity measurements records data objects at pred efined clock intervals. Post processing measurements of the tran sformation start and finish temperatures for the forward and reverse shape memory react ions was also automated, thus facilitating the measurement of many cycles in a t ypical high cycle test. The temperature dependence of the resistivity of the material determined under zero load conditions is shown in Figure 2-10 as a representation of the method for determining transformation temperatures via resistive measurements The transformation temperatures are determined from this data by the constructi on of linear, polynomial curve best fits through the low temperature, intermediate te mperature, and high te mperature portions of

PAGE 50

36 the heating and cooling curves, respectively. In the intermediate temperature range where a mixture of the parent and martensite phases exist, the best linear fit is found by scanning a window of a specifi ed temperature width through the entire intermediate temperature range. An array of linear best fits and associat ed R values is constructed. The series of linear regressions are then ra nked by R values and the best fitting regression is used. Through the utilizati on of the equations for these best fits, start and finish temperatures were determined by in terrogating the inte rsection points. Figure 2-10 exemplifies why resistance measurements are a prominent characterization technique in the study of shape memory alloys as there is a substantial difference in the electrical resistivity between the parent B2 and marten site phases. A summary of the complete algorithm is summarized in the data flow diagram shown in Figure 2-13. The timed acquisition mode was employed wh en a continuous acquisition of data was required. There are two main functions for this mode. Timed acquisition may be used to optimize heat treatment times and te mperatures as a result of the effects of internal stresses, dislocation structure, grai n size and precipitate-matrix interactions and their respective kinetic s. In the current study, the timed acquisition mode was developed for integration of thermo-mechanical te sting methods with in-situ resistivity measurements. For this purpose, an exte rnal communication ch annel was developed, which continuously sends a signal to the MT S servo hydraulic contro ller proportional to the measured resistivity values thus allowing for a resistivity record on the same system that is acquiring thermo-mechanical data. This allows for a point by point correlation between the measurements ultimately lead ing to the simultaneous documentation of resistivity, stress, stra in, and temperature.

PAGE 51

37 Experiments were conducted on a Pt wire (Figure 2-11) and a NIST steel resistivity (Figure 2-12) standard34 in order to determine the accuracy of the instrument and calibration. The resistivity vs temperature relationships fo r both materials are shown in the figures below. For the platinum wire, the resistivity vs. temperature relationship should be linear in the temperature range of RT to 900 C. The measurement shown in Figure 2-11 which confirms the linearity of this relationshi p indicating the accuracy of the resistivity apparatus in both heating and cooling modes. The NI ST steel resistivity standard was measured in order to verify the calibration of the absolute values of resistivity as a function of temperat ure. These results are shown in Figure 2-12 and confirm a high degree of accuracy between these resistivity measurements and the measurements made by NIST. Figure 2-12 compares NIST measurements with our results via level crossing acquis ition in a sample configured w ith a 4 point probe and spot welded thermocouple. A low and high range ex ist in analog amplifiers which allow for higher precision measurements in lower resistivity samples. Figure 2-12 includes measurements for the low and high range both of which are within the margin of error specified by NIST for the resistivity standard. Thermomechanical Testing Thermomechanical instrumentation Mechanical testing was performed on an MTS servo-hydraulic test frame equipped with an MTS 484 controller and MTS soft ware. MTS 646.10B hydraulic collet grips with a modified 680 LCF grip set were used to grip the threaded specimens. A 20 kip load cell was used and strain measurements in tension were taken with an MTS Model 632.51B-04 extensometer using a 12.7 mm gage length. This extensometer is equipped with 85mm long quartz probes with a v-chisel ed ge having a maximum range of

PAGE 52

38 +20/-10% strain. Strain m easurements during compression te sting were acquired using a laser extensometer. Specimens were inducti on heated using an Am eritherm Novastar 7.5 power supply. Resistivity measurements were integrated with the uniaxial mechanical tests, facilitating measurement of transformation temperature, determination of the matrix phase as well as determine if a phase change occurred during the test while simultaneously determining mechanical properti es. Additionally this test technique was used to verify the materials phase fracti ons. The tensile or compressive sample was instrumented with a four point probe and a K-type thermocouple. A four point probe configuration was again chosen in order to e liminate the effects of contact resistance. Nickel wires were spot welded to the sample functioning as the volta ge sensing leads. Current excitation was supplie d through the hot grips. This instrument has been developed at the NASA Glenn Research Center advanced metallics branch. Figure 2-14 with key components labeled shows this wo rking configuration on a Materials Testing Systems (MTS) tensile frame fitted with high temperature hot grips and induction heating. K thermocouples were spot welded to th e sample. Temperature gradients across the gauge were to within +/-0.5% of the test temperature by calibration on a control sample fitted with three thermocouples wh ile actual test samples contain a single centrally located thermocouple. A problem inhe rent to spot welding is the formation of stress raisers during the rapid melting and solidification of a narrow region adjacent to the spot weld base and the wire (weld nugget). Th e radius of the nugget is a critical factor in determining the stress concentration at the we ld and impacts the fracture stress and strain

PAGE 53

39 of a sample. During a spot weld a predefined amount of energy is stored in a capacitive system and subsequently discharged through a welding probe following the path of least resistance. As the discharge passes from th e wire to the sample or base metal a small region melts first. This event leads to the contact resistance in this very narrow region dropping significantly which results in the remaining discharge to focus through this path. The result is a very sharp defect on the samples surface. This problem was minimized by using a multistep spot-welding procedure. Initially multiple low energy spot welds are made at closely spaced distin ct locations, which a ssures numerous wide wire to sample contact points. Although thes e wide contact points are not mechanically strong enough to withstand the stress during sample handling, loading, and deformation they provide multiple wide low resistance interf aces. Step two consists of a second pulse that is approximately 10X great er in discharge energy than th ose employed in step one. The secondary pulse forms a strong weld nugge t with a wide radius thus reducing the stress concentration at the weld. Uniaxial isothermal mechanical tests Tensile specimens were strained to failure in strain control at a rate of 1 x 10-4 sec-1. If a specimen reached the 20% limit of the extensometer, the sample was unloaded under strain rate control un til reaching a 0N trigger after wh ich the control mode was switched to load control. The specimen was then a llowed to cool and then unloaded. Compression tests were run in displacement rate control at approximately th e same strain rate. Use of displacement control was necessa ry due to limitations in the maximum scan rate of the extensometer and the acquisition rate required by the MTS controller for proper and safe PID (proportional integrating diffe rentiating) control. Analysis of the stress strain curves

PAGE 54

40 and determination of the proportional limit wa s facilitated by pr oprietary NASA Glenn Research Center Advanced St ructures Division software. Load free strain recovery tests In the unconstrained or load free strain recovery tests, tensile specimens were deformed in strain rate control. The ma ximum strain level trigger was set, which upon crossing reverses the strain rate to a compressive 1 x 10-4 sec-1 until the sample is fully unloaded. When the load reached O N the contro ller was switched to load control. Load was held at 0 N while the specimens were thermally cycled to a temperature in the range of 400 C. Heating rates were maintained at 10 C/min after which the samples were allowed to air cool to well below the Mf before further loading. The recovery rate was determined by monitoring the samples strain during the thermal cycling. Load bias test Load bias testing measures an SMAs ability to perform work. This is accomplished by measuring strain under a constant load. Three modes of load bias test were employed in this study. The primary tensil e load bias tests were run in a series of progressive loads on the same sample. Specime ns were deformed in strain rate control (of 1 x 10-4 sec-1) near room temperature to the predef ined holding load. At this point the controller was switched to load control holding a constant load. Specimens were finally thermally cycled twice from room temperature to about 100 C above the austenite finish temperature. Heating rates were maintained at 10 C/min followed by air cooling. An auxiliary fan was turned on after the sa mple temperature dropped below the Mf, providing additional cooling. In this test specimens were unloaded at near room temperature and then strained again to the ne xt higher load level. This procedure was repeated for each load. The work output was calculated by measuring the resultant

PAGE 55

41 change in strain during the martensite-t o-austenite transforma tion during the second heating cycle and multiplying by the applied stre ss. Additionally a series of tests were run on samples with an identical procedure except that the samp les were unloaded hot while austenitic or above the Af. Compression load bias tests were run in a parallel procedure in displacement rate control rather than strain ra te control with near room temperature unloading. In summary several test modes are utilized each targ eting the measurement of specific material and shape memory propertie s. Resistivity measurements targeted determination and stability of the no load tr ansformation temperatures as well as quantify the temperature dependence of the resistivity of the austenite and martensite phases. Isothermal uniaxial tests comb ined with data from dynamic modulus tests were used to determine the baseline mechan ical properties of each phase as well as the mechanical behavior of the alloys near the transformation temperatures Furthermore, load free recovery experiments measured the effectiven ess of the alloys in recovering elastic and plastic strains while the load bias testing was used to determine the alloys specific work output. Finally the cool at lo ad test measured the stress de pendence of the transformation temperatures as well as the transformational strains under load. Figure 0 Figure 2-1 Induction melted Ni19.5Pd30Ti50.5 cast ingot with attach ed hot top on a quarter inch grid

PAGE 56

42 Figure 2-2 Heat treatment and pr ocessing temperature schedule. Figure 2-3 Hot extrusion press schematic

PAGE 57

43 Figure 2-4 Uniaxial sample (A) 5 X 10 mm compression sample (B) Threaded 17.4 mm long by 3.81 mm diameter gauge sample. Figure 2-5 Processing flow diagram of DSC, compression, and tensile samples.

PAGE 58

44 Figure 2-6 The ICP using an Echelle type polychrometer32 Figure 2-7 Example of a DTA scan showi ng the exothermic and endothermic peaks characteristic of thermoelastic shape memory alloys

PAGE 59

45 Figure 2-8 Four-point probe resistivity configuration Figure 2-9 Raw (blue) and conditioned (yel low) voltage signals for resistivity measurements during inductive heating

PAGE 60

46 y = 0.1245x + 115.26 y = -0.025x + 176.8801002003000100200300400500600 Temperature (C)Resistivity (micro ohm cm ) Figure 2-10 Resistivity vs. temperature profile with regression analysis 0 0.01 0.02 0.03 0.04 0.05 0.06 01002003004005006007008009001000Temperature (C)Resistivity (micro ohm Cm) Figure 2-11 High conductivity Pt wire resi stivity vs. temperature relationship demonstrating the repeatability of the during heating and cooling

PAGE 61

47 0 10 20 30 40 50 60 70 80 90 100 0100200300400500600700800900Temperature (C)Resistivity (micro ohm cm) NIST measurment Standard test low range Standard test high range Figure 2-12 NIST (resistivity standard) resist ivity vs. temperature profile comparison of NIST measurements and the measurements by the resistivity apparatus.

PAGE 62

48 Figure 2-13 Resistivity appara tus data flow diagram.

PAGE 63

49 Figure 2-14 Materials Testing Systems (MTS) tensile frame fitted with high-temperature hot grips and induction heating conf igured for compressive testing.

PAGE 64

50 CHAPTER 3 ALLOY DEVELOPMENT The alloy development section of this st udy consists of two parts addressing the selection of the Pt and Pd modi fied NiTi alloys studied in de tail in Chapter 4. The NiTiPt alloy was selected based on the results from the study of over 20 NiTiPt alloys, which focused on elevated transformation temp eratures, effects of stoichiometry on microstructure, and the results of prior ba seline thermo-mechanic al studies showing good work output. The NiTiPd all oy was selected based on studies others have performed, which have focused on transformation temper atures and no-load recovery tests. In Chapter 4, the thermomechanical properties of the chosen baseline Ni19.5Pd30Ti50.5 alloy will be presented. The majo rity of the thermo-mechanical tests performed on the Pd modified alloy were also performed in a parallel effort on the chosen Pt-modified Ni24.5Pt25Ti50.5 alloy. Data from investigations on the Pt alloy will be drawn upon for comparison to the Pd alloy, which exhi bits similar structural transformations yet has quite different mechanical propertie s and shape memory performance. The underlying motivation of such a study is to compare the mechanical properties and materials performance of the two HTSMA system s, thus allowing for the identification of areas where further alloy development may im prove the materials performance. Characterization of th e NiTiPt SMA system Bulk Compositional Analysis Selection of a baseline Pt-modified all oy was accomplished by a cursory study of the effect of near stoichiometric Pt alloyi ng additions on transformation temperatures and

PAGE 65

51 microstructure of NiTi alloys. Stoichiome tric alloys refers to alloys which lie approximately along the Ni50-xPtxTi50 iso-stoichiometric line between the binary NiTi and NiPt intermetallics. This term will be us ed throughout for the NiTiPt and NiTiPd (Ni50xPdxTi50) alloy. Prior studies have shown that ternary Pt additions produce elevated transformation temperatures well above the tr ansformation temperatur es of NiTi. Thus one would expect to find potential high-te mperature shape memory behavior in this system. As a matter of fact, Pt additions are superior to all other current alloying modifications to NiTi at increasing transforma tion temperatures, even so, this system has received little attention in the form of ch aracterization studies or alloy development. Three groups of alloys were chosen for charact erization: stoichiometric alloys, Ti rich alloys in which the fraction of Ti in the all oy is greater than 50 at.% and Ti deficient alloys which have atomic fracti ons of Ti less than 50 at.%. A ternary compositional plot of the experimental compositions selected for this alloy development study is shown in Figure 3-1. The stoichiome tric alloys are denoted with a E### designation and Ti rich and deficient alloys are designated with an F### designation. Subsequent to th is study two stoichiometric alloys of 20 and 30 at% Pt denoted as baseline alloys as designated in Figure 3-1 were selected for detailed thermomechanical testing and characterization.35 Transformation Temperature Transformation temperatures as determ ined by DSC or DTA are given in an inclusive list tabulated in Table 3-3. It was determined that the transformation temperatures in Ni50-xPtxTi50 alloy system are strongly de pendent on Pt content along the Ni50-xPtxTi50 iso-stoichiometric line between the binary NiTi and TiPt intermetallics. The transformation temperatures in the alloys of stoichiometric compositions designated with

PAGE 66

52 an E prefix are in good agreement w ith several other reported values.36-3836,37,38 Figure 3-2 (A) compares several reported values of the Ms temperature as a function of Pt for Ni stoichiometric alloy additions while the comp lete set of transformation temperatures are documented in this study are presented in Figure 3-2 (B). Minor al loying additions of Pt resulted in a depression in transformation temperatures up to approximately 10 at% Pt after which further alloy additions resulted in a potent increase in transformation temperature with additional Pt content. A slight discrepancy exists between the transformation temperature measurements taken by Lindquist and Wayman and the results of the current study. This discrepancy is evident primarily in the compositional region where the transformation temperatures are depressed by alloying additions. The curr ent study and the measurements found in (37,38) determine the transformation temperat ures by calorimetric methods (DSC or DTA) while Lindquist and Wayman employed el ectrical resistivity measurements to determine the transformation temperatures. As described previously, transformation temperature analyses by thermal methods ar e based on the extrapolated onset method where the baseline is forced to have a nearly flat slope. Resistivity measurements, on the other hand, rely on deviations in the slope of the vs. T curves determination of transformation temperatures the base line is formed by the temperature dependence of the samples resistivity. Therefore resistive analys is is dependent on the slope of the baseline and the slope of the transition region. It is possible that the differences between the thermal and resistive account for this discrepanc y. Another possibility is that differenced in thermomechanical processing, heat-tre atments and or minor deviations from stoichiometry may account for slight discrepancies.

PAGE 67

53 It is known for binary NiTi and the NiTiPd systems that the transformation temperatures are highly dependent on stoich iometry. Particularly, in some NiTi based shape memory alloy slight deviations which make Ni rich NiTi alloys and Ni+Pd rich NiTiPd alloys results in a strong depression in the transf ormation temperatures. The NiTiPt alloys however differ from these systems in the am ount of suppression by casting Ti deficient alloys. A comparison of E430 and F311 alloys (Table 3-3) which both have 25% Pt contents reveals the a 3% deficiency in Ti decrease the transforma tion temperatures from only 29 degrees Celsius. Contrasting this to a comparable 3% Ti deficiency in the NiTiPd alloys which results in a decrease of transformation temperatures of over 220 degrees Celsius, it is evident that the stoichiometric effects on transformation temperatures are not as pronoun ced in the NiTiPt shape memo ry alloys. The details of the effects of stoichiometry on NiTiPd shape memory alloy system are discussed in its alloy development section later in this chapter. Microstructure As expected, all the stoichiometric a lloys were essentially single phase. Micrographs of these alloys we re omitted as they did not pr ovide any useful information other than essentially confirming the single pha se nature of the microstructure. Focusing on the deviations from stoichiometry alloys with excess or deficient Ti ratios did not exhibit an increase in transf ormation temperatures but ra ther a suppression of the transformation occurred in non-stoichiometric alloys. Conversely, most of the non-stoichiometric alloys, designated by the prefix F, were found to contain a second phase. Figure 3-3 is a summary of SEM back scattered electron micrographs of the microstructures fo r all the F-series alloys. Samples F303 and F304 are essentially single-phase non-stoichiomet ric alloys, in that th ey do not contain a

PAGE 68

54 second intermetallic phase. The spherical da rk phase in these alloys is probably TiO2, which is common to all the alloys. The micros tructure for these two alloys is similar to that observed in the E-designated alloys. F303 is comparable to the microstructure of the lower Pt E-series alloys where the martensite is finely distributed and F304 is similar to the higher Pt E-series alloys with a coarser martensitic distribution. Most of the alloys contain second phases including alloys close to the line of constant stoichiometry which indicates that the solubility of excess alloying additions on either side of this line is very narrow. On ly two non-stoichiometric alloys that did not contain a second intermetallic phase were observed on the Ti rich side of this line of constant stoichiometry. Both of these allo ys contained Ti-rich in terstitial containing phases probably oxides. Therefore, a ny excess Ti could be tied up as TiO2 and the bulk matrix phase was probably very close to a st oichiometric composition. Consequently this signals that there is probably little sol ubility for excess solute on either side of stoichiometry. Both phase diagrams are recent assessments of the respective binary system. Based on the compositional analysis and basic morphology of the microstructures shown in Figure 3-3, there are basically two type s of second phases observed in the nonstoichiometric alloys. All the (Ni+Pt)-rich al loys contain a lathe lik e structure with a 2:3 Ti:(Ni+Pt) ratio. Given the nature of this phase, it would appear that it forms by nucleation in the solid state. An intermetallic phase with the stoichiometry of Ti2(Ni,Pt)3 does not appear in either bina ry phase diagram, however, Ti2Ni3 is a metastable phase that is observed in binary NiTi alloys.39 Consequently, the Pt coul d stabilize this phase. Or it is possible that it is a new phase, unique to the ternary phase diagram. It will take

PAGE 69

55 additional quantitative x-ray diffraction or TEM analysis to determine the specific structure of this phase. The non-stoichiometric alloys on the Ti -rich side of the line of constant stoichiometry have a phase that is more s pherical or elliptical in morphology and usually located along the grain boundaries. This w ould indicate that it is possibly a low melting point phase that was last to form during so lidification. This and the fact that the composition has a 2:1 Ti:(Ni,Pt) ratio would indicate that it could be Ti2(Ni,Pt), which is isostructural to Ti2Ni and has been previ ously identified by Garg.39 A small percentage of other phases may also a ppear in these alloys. Alloy Selection : NiTiPt The results of the characterization of th e ternary NiTiPt high-temperature shape memory alloy system are summarized above and were presented and published in a relevant conference proceeding.40 The alloy design phase was intended to build a relationship between the compositional depende nce of the transformation temperatures and microstructure in the NiTiPt system. Th is system was chosen for a detailed alloy study primarily due to evidence that Pt modifi ed NiTi alloys exhibi t the highest reported transformation temperatures of any NiTi based a lloy as well as the fact that there is a very limited data available in the lite rature for this alloy system.2 The alloys were single or two phase and all contained a limited volume fraction of interstitial containing phases The formation of a second phase was evident even with minor deviation in stoichiometry, therefore the solubility for excess components outside of the iso-stoichiometric line between NiTi and TiPt is limited. The formation of interstitial containing phases ties up Ti thus depleting the matrix by an amount which depends on the bulk interstitial concentration. Th ere for in order to in sure that both the

PAGE 70

56 alloys are not Ti deficient th e Ti concentration should be increased slightly above the expected interstitial impu rity concentration. Two alloys were selected for detailed th ermomechanical testing from the NiTiPt and NiTiPd alloy systems. Focusing on the se lection of a composition in the Pt modified NiTi alloy, the position along the stoichiometric line between NiTi a nd PtTi was selected based on a compromise between elevated transformation temperatures and good mechanical properties. As previously mentioned two baseline alloys (Ni50-xPtxTi50 containing 20 and 30 at% Pt) along with a bi nary NiTi alloy (SM495 NiTi) supplied by Nitinol were selected for detailed thermom echanical testing by colleagues at the NASA Glenn research center.35 The alloys were prepared in a manner sim ilar to that described in Chapter 2. The materials were prepared by melting, extrusion, and subseque nt machining of tensile dog bone test specimens. Along with isothermal uni axial tensile testing, lo ad bias testing was conducted on the 20 and 30 at% Pt extrusions along side the binary NiTi. Load bias testing or constant-load, strain -temperature tests measure a materials ability to do work against a constant biasing load which is a key design parameter in the development of shape memory actuated devices35 and therefore must be considered in combination with a high transformation temperatures in the selection of HTSMAs. Figure 3-6 shows the results of a load bias examination in conj unction with results from a comparable test documented in a characterization study of binary NiTi alloys.41 The applied tensile stress is related to a volume specific work output or work density. SM495 NiTi followed by the stoichio metric and the Ti rich NiTi show the highest specific work output yet the transformation temp eratures are near room

PAGE 71

57 temperature. The 20 and 30 at% Pt modi fied alloys however have much higher transformation temperatures (262C and 560 C Ms temperatures respectively) but the work output is lower. In the 20 at% Pt m odified alloy the work output was marginally lower that or the binary alloys at much hi gher transformation temper atures. In contrast the 30 at% alloy exhibited the highest tran sformation temperature of all the alloys encompassed in this study yet did not perfor m work under any biasing load. Therefore, although this material exhibited high transformation temperatures, it is not directly useful in application requiring actuat ion forces of any level. Based on the findings of this study and those in reference [35]a new baseline composition of Ni24.5Pt25Ti50.5 was selected for investiga tion in this study based on a compromise between elevated transformati on temperatures poten tial work output. Although this alloy was not included in th e NASA NiTiPt baseline alloy advanced mechanical characterization study35 it lies within the compositional bounds of the Ni20Pt30Ti50 and Ni20Pt30Ti50 alloys. The transformation te mperatures in the 30 at% Pt alloy was high but the alloy exhibited no ev ident capacity for producing work while the 20 at% alloy exhibited a high work output at a much lower transformation temperature. The Ni24.5Pt25Ti50.5 alloy lies halfway between these test s and therefore by extrapolation it was assumed that is alloy would demonstrate some capacity for performing work at with transformation temperatur es centered around 450C. Fire 0 Alloy Selection : NiTiPd NiTi and PdTi, like the NiTiPt system, also form a con tinuous solid solution with a high-temperature B2 phase that transforms to a B19 (orthorhombic) or B19 (monoclinic) low-temperature martensite phase with transf ormation temperatures between those of the

PAGE 72

58 binary alloys.42 Figure 3-7 is a map of the thermodynamically stable thermoelastic transformation reactions as a function of composition and temperature for the NiTiPd alloys.43 The Pt and Pd modified NiTi both fo llow similar transformation reactions at similar alloying additions but with differi ng effectiveness of the ternary alloying additions at increasing the tr ansformation temperatures. Figure 3-8 (A) illustrates the relationship between transformation temperatur e and alloying additions in both systems.16 As shown in this diagram it has been confirme d that the martensitic transformations over the entire range of ternary Ni-Pd-Ti compositi ons are thermoelastic in nature and that the alloys exhibited aspects of shape memory behavior similar to binary NiTi alloys.44,45 The level of Pt in the NiTiPt alloy wa s chosen on the basis of transformation temperatures and the work output. To the be st of the authors knowledge, in the NiTiPd shape memory alloy system no such tests results (load bias) are availabl e in the literature. However, measurements of the alloys stress fr ee strain recovery have been examined in prior studies.11 Through dilatometry and uniaxia l ambient temperature tests the compositional dependence of the shape recovery at specific initial strain intervals was examined. The findings of this study are summarized in Figure 3-8 (b) where the recovery strain is plotted as a function of alloy composition a nd initial plastic strain. For all strain increments the maximum shape r ecovery is approximately at 30 at% Pd. Additionally it has been shown that the shape memory behavior of NiTi-30Pd (at.%) alloys can be quite good under unconstrained conditions with samples loaded to 2-4% total strain levels in the martensitic conditi on recovering 100% of the strain while those loaded up to 6% recovering 90% of the strain.36,46,47 Similar shape memory behavior has been observed for samples deformed in compression 48 and torsion.49 Alloys containing

PAGE 73

59 40 at.% or more Pd including the TiPd binary alloy could only recover about 0.5% strain in tension to various strain levels.36,49 This poor shape memory performance has been attributed to a low critical stress for slip (whi ch is an irreversible process), such that the majority of the deformation is accommodated by slip rather than the more typical twin boundary motion or martensite reorientat ion (also referred to as detwinning).49 As a result of the findings liste d above a base composition of Ni19.5Ti50.5Pd30 was selected for the Pd modified alloy. The alloy is developed w ith a slightly Ti (approximately .5 at%) rich composition in order to ensure high transformation temperatures. Transformation temperatures are highly linked stoichiometry following the relationship Ni50-xPdxTi50 Compositional deviations veering in to the Ti lean side of the iso-stoichiometric line between TiNi and TiPd (Ni50-xPdxTi50) strongly decreases transformation temperatures in contrast to Ti rich compositions which have little effect on the transformation temperatures. Figure 10 confirms the compos itional dependence of the transformation temperatures for off-stoi chiometric alloys at fixed Pd fractions (Ni20+xPd20Ti50-x). The effects of stoichiometry on th e transformation temperatures in the (Ni20+xPd20Ti50-x) alloy is similar to those exhibited by binary NiTi (Ti50-x Ni50) where compositions crossing into the Ti deficient is ostoichiometric line results in a sharp decrease in transformation temperatures.50 It is a well known fact that Ti has a high interstitial affinity. In the system under study, the main interstitial impurity elemen ts are C and O which enter the melts as impurities during the alloy melting process. These interstitial elements react with Ti to form titanium oxides and carbides thus depleti ng titanium from the alloy by tying it up in interstitial compounds. By forcing the alloy to be Ti rich by a frac tion greater than the

PAGE 74

60 impurity concentrations results in a final al loy composition which remains Ti rich by an amount proportional to the excess Ti not consumed in the formation of oxides. The goal is to keep the overall matrix composition st oichiometric or sli ghtly Ti-rich, thus guaranteeing a high transformation temperature. Additionally Ti ri ch compositions yield a fraction of the intermetallic phase Ti2(Ni,Pd) which is isostructural to Ti2Ni.39 Ti2(Ni,Pd) is an interstitial stabilized phase which has a high solubility for interstitial oxygen.51 The presence of this phase allows for further removal of interstitials from the martensite or austenite matrix which has a lo wer interstitial solubility. This alloying approach has been employed in previous studies of the Pt mo dified NiTi SMAs.35 Alloy Selection Summary The NiTiPt alloy was selected based on the results from our study on elevated transformation temperatures, effects of stoich iometry and the results of prior baseline thermo-mechanical studies measuring work out put. In the Pd modified NiTi no such advanced thermo-mechanical studies existed t hus the alloy selecti on was centered around existing studies which demonstrated a composition dependent maximum in the unconstrained or no load shape recovery. Cons equently, the final sel ection of alloys for this study are Ni19.5Pd30Ti50.5 for advanced thermo-mechanical testing in parallel to the Ni24.5Pt25Ti50.5 for comparison of mechanical and thermoelastic properties. Each alloy is developed with a slightly Ti (approximately .5 at%) rich composition to prevent Ti-loss from the matrix resulting in a Ti-poor all oy with lower transformation temperatures.

PAGE 75

61 Figure 3-1 Ternary plot of the Ti-Ni-Pt compositions studied. The composition of all alloys was confirmed by spectrographic analysis Table 3-2 Aim and measured compositions of all alloys investigated.

PAGE 76

62 Table 3-3 Transformation temperatures of alloy set Figure 3-2 A. Effect of Pt on the Ms tr ansformation temperatures for Ni50-xPtxTi50 alloys, including data from previous researchers B. Effect of Pt on all transformation temperatures for Ni50-xPtxTi50 alloys

PAGE 77

63 Figure 3-3 SEM BSE micrographs of the non-stoichiometric alloys.

PAGE 78

64 Table 3-4 Semi-quantitative EDS analys is of the various phases observed. Figure 3-4 Phase diagrams (A) NiTi bi nary phase diagram from reference37 (B) TiPt binary phase diagram from reference38 Sample ID Bulk Composition (at.%) Region 1 (martensite) (at.%) Region 2 (second phase particle) (at.%) Ti Ni Pt Ti Ni Pt Ti Ni Pt F301 48 31 20 46 29 25 39 39 22 F302 48 21 31 47 18 36 39 29 31 F303 52 29 19 49 28 22 F304 52 19 29 48 18 33 F305 45 32 22 47 23 30 39 35 26 F306 45 23 32 46 15 39 38 27 34 F308 55 18 27 47 22 31 63 9 28 F309 47 28 25 45 25 29 44 28 28 F310 47 25 27 46 21 33 45 24 31 F311 53 22 25 50 21 29 67 2 32 F312 53 25 21 49 27 24 67 2 30

PAGE 79

65 Figure 3-5 Effect of ternary alloying additions on the Ms (o r Mp) temperature for NiTibased high-emperature shape memory alloy systems.2 Figure 3-6 Comparison of the specific work out put for several conventional NiTi alloys, SM495 NiTi, and the (Ni,Pt)Ti HITSMA35

PAGE 80

66 Figure 3-7 Phase diagram of TiPdTiNi alloys.44 Figure 3-8 Shape memory properties NiTiPd (A) Ms temperature resulting from ternary alloy additions.11 (B) Average shape recovery in Ti50 (Ni50-x) Pdx.11

PAGE 81

67 Figure 3-9 Plots of martensitic transfor mation temperatures vs. composition for Ti50xPd30Ni20+x 47. 47

PAGE 82

68 CHAPTER 4 RESULTS AND DISCUSSION Shape memory alloys have unique propertie s and related mechanical behaviors. Principally, they have the ability act as a solid state device which performs mechanical work against a biasing load. In order to exam ine this extraordinary behavior an in-depth thermomechanical study of these two alloys is conducted. A comparison between the two systems is carried throughout this chapter with the main focus on the Ni19.5Ti50.5Pd30 high temperature shape memory alloy in comparison to the Ni24.5Ti50.5Pt25 alloy. Heat Treatment Optimization Before testing the baseline Ni19.5Ti50.5Pd20 and NiTiPt alloys, it was necessary to determine an optimum heat treatment for the annealing of the as-machined samples to eliminate any residual effects due to the machining process that could affect the transformation temperature or thermomech anical performance of the alloy. The martensite phase in particular is suscep tible to twinning/detw inning and or plastic deformation of the near surface layers of the alloy. The function of the stress relief heat treatment is to relieve any residual stresses in the samples due to the high speed machining and the thermal expansion mismatch between the mild st eel extrusion can and the SMA. At elevated temperature interactio ns between dislocations generated during the machining process allow for the formation of low energy structures by dislocation movement.52 As an indicator of the residual stresses in the material resistive measurements are employed.

PAGE 83

69 Resistivity is a stress sensitive property among other variables. A classic example of this effect is the increase in resis tivity as a function of maximum precipitation hardening in aluminum 7XXX series alloys wh ere the increase in resistivity is linked to an increase in the materials internal stresses due to precipitate matrix interactions54. Additionally a materials stress state affects the relations hip between temperature and resistivity or the slope of the resistivity temperature curve. The combination of these effects is the keystone for optimizing our stre ss relief heat treatments. For the purpose of these measurements it was assumed that the linearity of the resistivity vs. temperature curves is a function of the stress distribu tion in the sample. The underlying reasoning behind this was that if a non uniform stress distribution ex ists within the material a position dependent resistivity dist ribution will also exist. As the material is heated or cooled through a thermal hysteresis a fraction of these stresses are reli eved resulting in an irregular resistivity temperature curve. A material which has been almost fully recovered on the other hand will exhibit a linear resistivity temperature curve due to the absence of the relief of residual stresses. Figure 4-10 shows the results of a heat treatment optimization for the 25 at% Pt modified alloy. In this examination the samples were heat treated for 1 hr at 500 oC and 600oC and heated through a therma l hysteresis with in-situ resistivity measurements post heat treatment along side a no heat treatment sample for comparison. The sample that did not have a heat treatment resu lts in a resistivity temperature curve which is highly irregular due to the irregular transient stress distribution within the sample. The samp le which was heat treated for 1hr at 500 oC (near Af) showed a slightly more regu lar relationship due the partia l stress relief during the heat treatment. Finally the sample which was heat treated at 600 oC showed a smooth

PAGE 84

70 resistivity temperature curve which is indicativ e of a stable stress di stribution within the material which may be linked to a fully recovered material. Characterization Materials Characterization Each alloy was developed with a sli ghtly Ti (approximately .5 at%) rich composition in order to maintain a stoichiome tric or slightly Ti-rich matrix composition after formation of various interstitial contai ning phases. The alloying technique has been employed in previous studies of the Pt modified NiTi SMAs.35 The resulting chemical compositions of the extruded materials are Ni 19.5, Pd 30.0, Ti balance, O 0.30, C 0.50 and Ni 24.419, Pt 24.428, Ti balance, O 0.28, C 0.43. Detailed compositional analysis of each extrusion is given in appendix C. SE M micrographs exhibit the phase contrast of the resulting alloy is shown in Figure 4-11 revealing a small volume fraction of possible carbide and oxide phases in a re latively homogeneous Ti Ni Pd and Ti Ni Pt matrix void of macro cracks and porosity. Properties and Transformation Temperatures The no-load transformation strain of th e undeformed material was measured by dilatometric techniques. This technique m easures macroscopic structural shape change or uniaxial transformation strain of the extruded material as a function of temperature. The thermal expansion coefficients of each phase (austenite and martensite) were determined from the slope of the linear portions of the temperature vs. strain relationships shown in Figure 4-12 and Figure 4-13. The samples were thermally cycled through several hysteresis in order to allow the tr ansformation temperatures and strains to stabilize thus becoming reproducible on subs equent cycles. The forward and reverse reaction in the undeformed material exhibite d equal magnitude transformational strains

PAGE 85

71 under no load conditions. The thermal expansio n coefficient of each phase within each alloy was approximately equal with the mart ensite having a slightly greater thermal expansion than that of the austenite with th e NiTiPd alloy exhibiti ng the greater thermal expansion. These thermal physical prope rties are tabulated in appendix A. Resistance measurement is a prominent ch aracterization techni que in the study of SMAs due to the substantial differences in the electrical properties between the parent and shear phases. This technique has been su ccessfully applied to SMAs as a method of determining transition temperatures in-situ to operating conditions. In addition, the application design of SMA actuated activ e flow control devices requires an understanding of the temperature dependence of the electrical resistance. Researchers working on the Nitinol characterization st udy NASA CR-1433 utilized the resistance vs. temperature behavior of uncons trained NiTi in a scheme to quantify the shape memory performance of the material.53 In the current study this t echnique was exploited for the determination of the transformation temperat ures as well as the temperature dependent electrical properties of the SMA unde r study in the stress free state. The temperature dependence of th e resistivity is plotted in Figure 4-12 and Figure 4-13 including marked isothermal test temperat ures. As mentioned in the experimental section the transformation temperatures are determined by construc ting a tangent line to the linear portion resistivity temperature pl ot of each phase and a tangent line for the transformation intermediate regi on. The intersection of the tange nt lines is determined as the transformation start and finish temperatures. A slight discrepancy exists between the transformation temperatures measured by dilatometric and restive methods. This slight discrepancy was linked to the methods used

PAGE 86

72 to instrument the samples with a thermocoupl e. In the dilatometer the thermocouple was located a small distance from sample theref ore this technique actually measures an average temperature between the region of th e furnace adjacent to the sample and the sample. The transformation in the forward a nd reverse directions are highly exothermic and endothermic therefore during the tran sformation the difference in temperature between the sample and the furnace increases. This effect causes a slight error in the temperature reading at the thermocouple. The resistive measurement on the other hand circumvents this problem by having the thermoc ouple spot welded directly to the sample and therefore is a more accurate measuremen t of the transformation temperatures. A summary of the recorded tran sformation temperatures measured by resistive methods are tabulated in appendix B. Similarly to the thermal expansion resu lts the temperature dependence of the thermal resistivity coefficient of the martensite phase is greater than that of the austenite phase. The resistivity of the austenite is gr eater than that of the martensite at all temperatures within the experimental bounds A metals resistivity is a stress and structure sensitive property thus a greater resi stivity could be an indicator of a higher internal energy or stress state of a particular phase, which in this case is the austenitic phase. Ti vs. Ni, Pt or Pd have a significant mismatch in atomic radii, and thus the arrangement of these atoms into ordered B2 la ttice results in internal lattice strains as atoms are forced to reside at slight distan ces from their minimum energy distance. The shear lattice on the other hand has greater dist ances between lattice pos itions resulting in a more relaxed structure and t hus a lower resistivity.

PAGE 87

73 In the Pd modified alloy there is a small but significant peak in the resistivity for the reverse reaction (martensite transforming to austenite). This peak occurs when a very small volume fraction of martensite remains in an austenite matrix. Focusing on the martensite/austenite interface a nd criteria for thermoelastic transformations which states that this interface must be c oherent and the majority of th e transformation strain between the phases must be accommoda ted elastically a correlation to a parallel well developed system is made. A comparison of such an arrangement to the mechanism for precipitation hardening which also initially ha s strains partially accommodated elastically due to lattice mismatch can be made. It is feasible to imply th at there could be a significant elastic interaction in the Pd-cont aining SMA when very small fractions of martensite reside in an austenite matrix. Precipitation hardened materials exhibit an increase in resistivity as hardening resulti ng from elastic interactions between phases increases.54 The evidence of this is the small peak in the resistivity curve and the implication that elastic intera ctions between the austenite matrix and martensite produces an increase in resistivity above that of the fully austenitic material. Comparing the mean correspondent variant pair size of the Pt and Pd modified alloys it is evident that the distribution is much coarser in the Pt than on the Pd alloy (Figure 4-11). As a result, it is expected that this coarse distribution n ear the end of martensite to austenite transformation results in widely spaced ma rtensite packets which can not effectively harden the remaining austenite or globally eff ect its electrical resi stivity thus Pt alloy does not exhibit the a resistivity peak.

PAGE 88

74 Thermomechanical Testing Isothermal Stress-Strain Behavior in Tension and Compression To directly compare mechanical proper ties in tension and compression all the isothermal curves are calculated in true stre ss and strain. The repres entative true stress strain curves for tension and compression have been combined in to two sets of curves for each material in each loading orientation making a total of eight multi-experiment plots (Figure 4-6-Figure 4-18 and Figure 4-22-Figure 4-25). For each material, there is a plot combining (Figure 4-6-Figure 4-18 for NiTiPd and Figure 4-22-Figure 4-25 for NiTiPt) the stress strain rela tionship of the austenite and ma rtensite well above or below transformation temperatures. Tension and co mpression behavior for the NiTiPd alloy is shown, respectively, in Figure 4-6 and Figure 4-7 for temperatures well away from the transformation temperatures while Figure 4-17 and Figure 4-18 contain the results of tensile and compressive isothermal tests ne ar the transformation temperatures. The NiTiPt alloy tested at isot hermal temperatures well above or below the transformation temperatures are shown in Figure 4-22 and Figure 4-23 while Figure 4-24 and Figure 425 include the mechanical properties n ear the transformation temperatures. The elastic loading region of each uniax ial compression test was analyzed for linearity. Significant deviations from linear ity in the elastic region are a good indicator of slightly non-parallel load ing surfaces which result from a combination of machining imperfections, imperfect load bearing plenum faces and loading the sample slightly off axis. An example this effect is shown in Figure 4-14 where the elastic loading region below 3% engineering strain exhibits an expe rimentally induced devi ation from linearity. If analysis revealed such a deviation from linearity the uniaxial compression test was

PAGE 89

75 corrected by extrapolation of a regression anal ysis of the well beha ved (linear) portion of the elastic loading region. Although each isothermal compression test was carried out to over 30% engineering strain the results are presented only to 20% strain. Analysis of the force displacement curves revealed that friction between the lo ad bearing plenums and the sample induced barreling. During a constant strain rate isothermal compression test an increase the slope of the force or engineering stress Vs disp lacement curve is indicative of significant frictional forces and the onset non-uniform deformation. This effect is exemplified by the monotonic compression test shown in Figure 4-14 at approximately the 22% strain level. Beyond th e 22% strain in this example the slope of the force strain curve increases due to fri ction induced effects. Isothermal stress-strain behavior in tension and compression NiTiPd In the NiTiPd alloy well below the transf ormation temperatures in tension (Figure 4-6) the classical beha vior of a thermal elastic shap e memory is exhibited. The stress strain curve of the martensite at room temperature and at 200C exhibits three identifiable regions in the stress strain curve. These regions are (1) elastic deformation at lower stresses up until the lower yield stre ss or in the case of the martensite the detwinning stress (2) further loading results in an almost linear inelastic deformation resulting from detwinning of the martensi te with continues until the variants with favorable orientations are reorie nted resulting in pseudo stress strain plateau (3) the work hardening rate again increases initially due to further el astic deformation of the reordered variant structure followed subsequently by further deformation of the martensite resulting in a more typical re gion of plastic deformation wh ere the curve has a parabolic appearance.

PAGE 90

76 In tension the pseudo stress strain plateau (region 2) differs from the stress strain plateau exhibited by classical SMA (NiTi) in that deformation does not occur at a near constant stress. The positive slope in this region indicates that a mechanism is requiring an increasing stress to continue deformati on. Several mechanisms could produce such a result in an SMA. As described earlier in the alloy development section, one of the limitations of high temperature shape memory alloys is a low critical resolved stress for the onset of non-recoverable slip processes at le ast relative to the detwinning stress. Slip and the associated deformation motion and ge neration result in work hardening; thus could explain why the combination of slip and detwinning requires a progressively greater stress as the true stain increases. Although this is an important mechanism examination of the no-load rec overy tests results examined di scussed a later section show that it is not the dominant cau se of the work hardening. Another more plausible mechanism proposed here is based on the fact that there exists an orientation distribution of corresponde nt variant pairs. As mentioned earlier, the critical stress for detwinning, noted as the lower yield stress, is significan tly lower than the actual yield stress for macroscopic yi elding of the martensite primarily through dislocation motion (classical yield stress). Considering that in a typical multi-variant martensitic structure, certain variants are or iented favorably for detwinning which means the applied stress results in a maximum reso lved shear stress. Variants which are oriented such that the resolved shear is a ma ximum will detwin first at the lowest applied uniaxial stress resulting in the bulk material exhibiting a yiel d stress. Variants which are misaligned are simultaneously deformed elastically as the resolved shear stress is smaller than the critical resolved shear stress for detwinning. As deformation increases at

PAGE 91

77 increasingly greater stresses, more variants in other orientations are activated as the applied stress produces a progressively increas ing resolved shear st ress. Thus although detwinning is the main deformation mechanis m in the pseudo stress-strain plateau the orientation distribution of thes e variants requires different applied uniaxial stresses for detwinning. However more work is necessary to fully confirm this mechanism. In contrast, the isothermal deformation of polycrystalline martensite in a classical SMA, such as NiTi, occurs at a constant or near constant stress. Here, similarly to the NiTiPd extruded material, we also have a variant distribution which results in a comparable situation yet the classic SMAs do not exhibit this behavior. A likely explanation for this stems from the number of equivalent variants that can form from a single parent austenite crystal. Recall that the martensite phase in the NiTi SMA has a monoclinic structure in contra st the NiTiPd and NiTiPt al loys both of which transform from a B2 to the orthorhombic structure. The orthorhombic structure is a higher symmetry structure than the monoclinic. As a matter of fact a monoclinic structure in SMAs is related to the orthorhombic structur e by an additional non-basal shear. This additional shear increases the number of equiva lent variants by a factor of two which may form from a given austenite crystal. Thus any particular variant has twice as many equivalent variants it can shear to under load. As a result, there is less of an orientation dependence of the shear stress for detwinni ng in the monoclinic structure as there are many more variants for a particul ar variant in a particular orie ntation to shear to thus in SMAs with a monoclinic martensite exhibit detwinning at constant or near constant uniaxial stress.

PAGE 92

78 The mechanical properties of the NiTiPd martensite in compression (Figure 4-7) are comparable up to the initial yield stress. Beyond the yield stress, however (Figure 4-6) the deformation of the marten site does not exhibit a clear distinction between the deformation mechanisms of detw inning and non-reversible slip processes. Deformation of the martensite in compression after the yield stress results in a high work hardening rate to strains of about 10%. Further deformation again results in a decreasing work hardening rate similar to what is obser ved for a single phase alloy. The lower slope is linked to deformation via nonrecoverable slip processes and is comparable in both tensile and compressive loadi ng orientations. However, th e isothermal behavior in tension and compression deviates substantiall y beyond the yield stress of the martensite phase. The tensile behavior exhibits a pse udo-stress plateau or low hardening rate at stress levels starting at a bout 250 MPa which is linked to favored martensite variants growing at the expense of the other martensite variants in contrast to the stress-strain relationship for compression which does not sh ow a clear stress strain plateau. This indicates that the orientation of preferred variants requires an increasing stress suggesting a low mobility interface and thus a deformati on mechanism which is different from the deformation mechanism in the stress plateau evident in the tensile loading mode. Deformation by detwinning mechanisms is pol ar in nature, in contrast to slip behavior, such that a reversal in the shear direction will not produ ce twin movement in variants favorably aligned for ope ration in the forward direction.55,56 In other words, twins aligned favorably for operation in co mpression will not operate under a tensile stress and vice versa. For monotonic test ing of the martensite phase, the sample microstructure was set prior to testing. Apparently the manner in which the material was

PAGE 93

79 processed (extrusion in this case) resulted in a more favorable variant structure for twinning/detwinning in tension than compression. Particularly if the variant orientation distribution is such that the variants which predominate the microstructure are already in the orientation at which the Schmid fact or is at a maximum, deformation through detwinning assisted variant reor ientation is not possible since there are fewer variants to shear over and join the growing variant. Th e evolution of a micros tructure of such a variant distribution results in a material whic h is essentially partially detwinned as the product of the detwinning reaction is the formation of variants which already predominate the microstructure. Well above the transformation temperatur e the mechanical properties of the austenite were comparable in tension and compression. The austenite exhibits a small linear elastic stress strain region up until the yield stress of the material. After the yield stress, the material exhibited a low work ha rdening rate which decreased with increasing isothermal test temperature as evident in comparison of the 300 C and 400 C tensile tests (Figure 4-15) and the 350 C, 365 C, 400 C and 500 C compression test (Figure 4-16). The 400 C tensile test sample exhibited a maximum in the engineering stress strain curve followed by a parabolically decreasing stress strain relati onship. This is indicative of non-uniform deformation of the gauge lengt h. Optical comparator measurements of the tensile sample confirmed that extensive necking occurred during deformation process which led to the formation of a slightly diffuse neck. In this sample uniform deformation reduced the cross-sectional di ameter of the gauge length 4.72% from 3.81 mm to 3.63 mm while the neck region exhibited a 40.2% reduction and a minimum diameter of 2.28

PAGE 94

80 mm. The true stress and true strain up to the point of necking wa s calculated from the engineering stress and strain using the cons tant volume approximation. The true facture stress (1522 MPa) and strain (102%) were cal culated from the minimum cross-sectional area of the necked region. The 400 C cu rve combines the true stress and strain calculated from the engineering values and an extrapolation between the value at the onset of necking and the true fracture stress strain. For the test results shown in Figure 4-17 and Figure 4-18, the samples were heated to well above the austenite finish temperature (Af) and allowed to cool back to the test temperature before loading the sample. Th e process of heating significantly beyond the Af and subsequently cooling back to the de sired test temperature ensured that the resulting stable phase was that of austenite. In-situ resistivity measurements and the occurrence of a well defined stress plateau, i ndicate that upon load ing, a stress induced martensite results. This behavi or can be seen in detail in Figure 4-19, which is a superimposed plot of resistivity, determined in-situ during tensile te sting at 255 C with the stress strain curve. Both in tension and compression the relative amounts of stress induced martensite as well as the stress at which the transformation occurred were comparable. Finally, a ductility minimum was observed in the region where the stress induced martensite occurred, similar to the be havior reported in NiTi Pt high temperature shape memory alloys.35 The combination isothermal stress st rain resistivity curve shown in Figure 4-19 exhibits a clearly visible stress strain pl ateau which results from the parent austenite phase isothermally transforming under stress to the martensitic shear structure. The deformation mechanism in this case is not detwinning nor dislocation motion, rather

PAGE 95

81 deformation occurs by the transformational shear strain associated with the thermoelastic transformation. Initial deformation correlates an increase in resistivity with the elastic loading of the austenite, after which a stre ss induced transformati on begins. Recalling the no-load resistivity hysteresis measurements (Figure 4-12) it was shown the resistivity of the martensite is much lower than that of the austenite. Therefore as the austenite transforms to martensite indicated by the stress plateau, the resis tivity decreases with increasing fractions of martensite. With con tinued strain elastic then plastic deformation of the stress induced martensite and remaining austenite occurs and as expected results in an increase in resistivity. A final change in the work hardening rate is accompanied by a subsequent change in the slope re sistivity strain relationship. Another interesting feature of the stress strain plot (Fi gure 4-20) for stress induced transformations in compression is that the stress at which the austenite is forced to transform increases with increasing devi ations from the no-load transformation temperatures. This can be correlated to th e fact the chemical driving force opposing the non-chemical driving generated by the applied st ress is greater at increasing deviation the load free transformation temperatures. Add itionally the extent of the stress induced transformation and the resulting amount of strain generated by it decreases with increasing deviation from the no-load transfor mation temperature. At temperatures much higher than the no-load tran sformation temperature (T>>Md) it is not possible to form stress induced martensite as the stress require d for this to occur is higher than the yield stress of the austenite which results in de formation by non-recoverable slip processes. The temperature dependence of the yield stress in compression and tension shown in Figure 4-20 is determined from the propor tional limit from the above series of

PAGE 96

82 isothermal stress strain curves. Generally, the yield stresses in tension and compression were similar at a given temperature. The yi eld stress of the martensite decreased with increasing temperature and reached a minimu m at the transformation temperature where the formation of stress induced martensite wa s possible. Beyond this minimum the alloy was partially or fully austenitic and the yield strength increased significantly with increasing temperature until a peak near 350 C. At temperatures above 400 C the austenite weakened and the onset of slip occurred at lower stresses. Dynamic elastic modulus determination NiTiPd In order to have an accurate representation of the instantaneous, load free, elastic modulus of these materials as a function of temperature, a dynamic modulus test was conducted for the NiTiPd alloy. The temper ature-modulus relations hip is plotted in Figure 4-21, which has three main approxima tely linear regions corresponding to the fully martensitic or austenitic phase or a combination of both in the intermediate temperature range bound by the reaction start an d finish points. The dynamic modulus of the martensite is about 10 GPa higher than th at of the austenite. The martensite as the sample temperature has an inverse relations hip to elastic modulus. Converse to this relationship the modulus increases with temp erature for the fully austenitic material suggesting that the internal energy resulting fr om atomic interactions of this phase is increasing with increasing temperature up to 800 oC. The region bound by the transformation temperature exhibits a stronge r temperature dependence than does either of the single phases which is proportional th e reaction progression terminating at a minimum at the transformation finish temperatur e. At this point the material is fully austenitic and modulus value is that of the austenite at the transformation temperature. An important characteristic of thermoelastic transformations is pre-martensitic elastic

PAGE 97

83 softening. This effect is clearly exhibited by the NiTiPd alloy as evident in the elastic modulus increasing with increasi ng temperature. Pre-martens itic elastic softening allows for local transformation strains adjacent to the interface to be accommodated elastically at a lower stress. Isothermal stress-strain behavior in tension and compression NiTiPt The NiTiPt alloy martensitic stress strain curves deformed isothermally well below the transformation temperatures in tension and compression did not exhibit any of the clearly distinguishable characte ristics of a typical shape me mory alloy. This alloy does not exhibit a defined stress plateau. Initial loading both in tension (Figure 4-24) and compression (Figure 4-25) resulted in a linear region as the alloy is elastically loaded (linear elastic region). Sim ilarly, upon loading in tension or compression past the yield stress there is a change in the slope of th e stress-strain curve. However, the tensile samples fracture shortly after the yield st ress. From no-load recovery experiments detailed in a subsequent sect ion on this alloy, we know that partial detwin ning along with slip is occurring at strains past the yiel d stress therefore, the second region will be denoted as the detwinning region, following th e nomenclature for conventional SMAs. The ductility in tension at th e test temperatures averag es at about 2.5% while in compression the samples either fractured sli ghtly before 20% stra ins or were unloaded when excessive bulging became evident. Sim ilarly to the NiTiPd alloy in compression, this alloy exhibited a second change in the work hardening rate in the 10% strain range. Although a clear correlation to the deformati on mechanism has not been determined it is possible that changes in sample geometry fr om uniform deformation to sample bulging in addition to slip result in such a change in slope.

PAGE 98

84 As in the NiTiPd alloy, a ductility minimu m was evident in tension at temperatures in the range of the transformation temperat ures. Furthermore the all the martensitic alloys failed in a brittle manner without a ny visible necking. Most of the fractures seemed to initiate at a surface defect (spot weld sites) or macroscopic inclusions visible to the naked eye. Further micros tructural characterization is required to fully characterize the fracture mechanism in this alloy. Analysis of the fracture stress and fracture strain at each isothermal test temperature (Figure 4-28) reveals two distinct regions. Starting at 200oC as the temperature is increased the facture stress and fracture strain decrease until a minimum is reached. Above this point all the samples tested are fully austenitic and as the isothermal test temperature is increased further the fracture stress decreased at a slightly less negative rate while the fracture stress sharply increases. The fracture stress in the austenitic phase decreas es with increasing temperat ure due to dynamic recovery becoming more prevalent while the fracture strain possibly increases due to the high strain rate sensitivity of this phase as previously explored in Figure 4-26. Isothermal tests for well above the tr ansformation temperatures in the fully austenitic state in tension (Figure 4-22) and compression (Figure 4-23) have similar characteristics to the stress st ain curves of NiTiPd alloy. Primarily there was elastic loading of the austenite up to the yield stre ss followed by plastic deformation. At higher temperatures dynamic recovery in the martensite was clearly evident. In order to explore the effects of dynamic recovery two tests were conducted in tension at 550C, one test was run at the standard strain rate while the sec ond test was run at a stra in rate increased by a factor of 10, the results are shown in Figure 4-26. This test confirmed that dynamic recovery was in fact a dominate mechanism affecting the mechanical properties of the

PAGE 99

85 austenite and furthermore the austenite was quite strain rate se nsitive although further testing is required to quantify the strain rate sensitivity. This is important for subsequent studies on the advanced thermomechanical processing of these alloys for actuation applications. The yield strength in both tension and comp ression in this alloy also overlap quite well as shown in Figure 4-18. This indicate s that the onset of the combination of detwinning and yielding are similar in both tension and compression. Similar to the NiTiPd alloy the yield strength drops at temp eratures in the range of the transformation temperatures followed by a sharp increase. A sharp decrease in the yield strength of the austenite occurs at temperatures above the Af in which the yield st rength of the austenite drops over its initial value. This is due to thermal energy helping overcome the activation energy for dislocation motion. Contrasting to the yield strength vs. temperature relationship of th e NiTiPd SMA in this alloy the yield strength of the martensite is less than 100MPa of the austen itic yield strength while in the NiTiPd alloy this difference in strength is over 200MPa. An important characteristic of viable SMA materials for actuation applications has been linked to mechanical properties of th e austenite vs. those of the martensite. Primarily, experience has shown than for an SMA material to be a good candidate for actuation applications th e alloy the austenitic phase shoul d be mechanically stronger than the martensite in order to avoid non-reversible sl ip in the austenitic phase. This is clearly the case in the NiTiPd alloy yet the strength of the martensite is comparable to the strength of the austenite in the NiTiPt alloy. Furthermore, stress induced transformations did not occur in this alloy at any temperatur e tested as confirmed by in-situ resistivity

PAGE 100

86 measurements which is thought to be primarily because of the high shear strength of the martensite. Unconstrained Recovery Tests Unconstrained recovery tests are used to characterize the ability of a shape memory alloy to recover twinning-induced deformation that is introduced at temperatures below the martensite finish temperature (Mf) by heating the material through its transformation temperature. The shape recovery is comm only determined by measuring the amount of strain introduced into a sample during deformation below the Mf, heating and cooling the sample through a full hysteresis back to room temperature.47-51,59-6245,46,48,49,57,58,59,60 Unconstrained recovery tests NiTiPd The final dimensions of the sample are measured and compared to the initial dimensions to calculate the amount of strain recovered. However in monitoring the strain changes continuously during th is process as shown in Figure 4-29, it is clearly evident that a number of different mechanisms are ac ting to contribute to the overall strain recovery of NiTiPd alloys as first reported by Lindquist.11 For comparison to prior work, the total recovery of this alloy is plotted vs. initial plastic strain in Figure 4-30 Total Recovery. The various contribu tions to the load free strain response of an alloy heated from 100oC to 400oC then finally cooled back down to 100oC is shown in Figure 4-29. Upon unloading the sample partially recovers a portion of the strain elastically often termed the elastic spring back ; however, although this region of the recovery curve is often included in the analysis of an SMAs total recovery it in itself is not a true SMA characteristic as the underlying mechanism is not a consequent of a thermoelastic transformation. The strain recovery processes for the load free material occurs in three distinct reactions. 1.) The thermal expans ion coefficient of the deformed martensite

PAGE 101

87 shown originally in Figure 4-12 was smaller than that of the undeformed martensite. Continuous shape recovery begins on heating up to the As temperature and the amount of recovery is estimated from the differences in the slope (thermal expa nsion coefficient) of the deformed and undeformed martensite expa nsion curves. The mechanism behind this difference may be the that due to the rear rangement of twins dur ing detwinning as the material is deformed. Consequently, th e expansion characteristics along this one direction may no longer be the same as measured in an undeformed sample with a more randomly oriented martensite (Figure 4-12). Further micros tructural and experimental work is needed to confirm this mechanism. 2.) The majority of the strain is recovered in a stepwise reaction which occu rs during heating from the As to the Af temperatures. The strain response in this region is attributed to two competing mechanism. First and foremost is the recovery of strain by the tr ansformation of the detwinned martensite to austenite. Opposing this is th e transformation strain associat ed with transformation from martensite to austenite (depicted in Figure 4-12) which measured along the deformation axis. 3.) There is a small amount of strain re covered as the sample is cooled between the Ms and Mf. This strain is linked to the aust enite to martensite transformational strain however it is not equal to this tr ansformation strain. As shown in Figure 4-29 this portion of the recovery is dependent on the initial st rain applied to the martensite. A possible mechanism proposed here is based on nucleatio n of a preferred orient ation of martensite on cooling due to the dislocation structure wh ich forms during deformation. Initially the martensite is strained to low levels and the majority of the deformation is accommodated by detwinning which is reversible. As the amount of strain in the martensite is increased there is increasing amount of plastic deform ation along with detwi nning. Since at lower

PAGE 102

88 strain levels the martensite is detwinned with little plastic deformation the majority of the strain is recovered in the martensite to austenite reaction. Upon cooling through the austenite to martensite reacti on a small amount of strain is recovered due to the formation of a random distribution of correspondent vari ant pairs (CVP). When strain levels are sufficiently large to produce significant am ounts of plastic deformation, detwinning is recovered during the martensite to austenite reaction yet plastic deformation associated with dislocation motion is not recovered and is carried over into the austenitic structure. This remnant dislocation network could aff ect the distribution of correspondent variant pair such that particular orie ntations are preferred resulting in a lower amount of recovery in comparison to a structure with a lowe r dislocation density and a more random distribution of CVP. Further microstructural examinations of samples thermally cycled after varying amounts of st rain could possibly confirm this mechanism. The unconstrained recovery tests were conduc ted at several strain levels and the recovery levels were analyze d. A summary of the results ar e plotted in Figure 4-30 Total Recovery and Figure 4-31. The relative amount of recovery for th e individual steps in the process are plotted vs. the to tal initial plastic strain whic h is defined as the amount of strain remaining after the load is remove d. Four curves are shown in Figure 4-31 representing the three shape memory recovery components described above as well as the percent total recovery which is calculated from the initial plastic strain and the strain after cycling through a complete hyste resis. The values of recovery strain used in the calculations are corrected for thermal expans ion data attained by dilatometric methods. The conventional shape memory performan ce in this alloy is average. This material exhibits full recovery up to total st rains of 2% and 98% r ecovery at 3% initial

PAGE 103

89 strain which drops off linearly to 86% reco very at 5% total st rain (Figure 4-30). Although it is useful to analyze th e total recovery in order to characterize an alloys shape recovery several contributing mechanisms are act ive therefore in this study an analysis of the temperature dependent recovery was of inte rest. Analysis of these results shown in Figure 4-31 reveals that for low initial plastic strain the alloy exhibits complete recovery (curve A) with a significant portion of the recovery occurr ing during the cooling reaction, evident in curve C in addition to thermal r ecovery during the heati ng of the martensite (curve D). As the initial plastic strain is in creased the total recovery decreased (curve A) as does the recovery from the cooling reac tion (curve C). The thermal recovery and recovery during the austenite to martensite reaction shown in curve C and D respectively both exhibit an inverse relations hip to the initial plastic st rain which may be linked to increasing amounts of non-recoverable slip. This relationship (curve C) indicates that the randomness in variant orientations is d ecreasing with increasing strains providing evidence of a preferred orientation resulting fr om the formation of dislocation networks. This effect in part contributes to the training of shape memory alloys where the generation of sessile dislocation networks cau se the formation of a preferred martensitic variant orientation in the austen ite to martensite reaction. Unconstrained recovery tests NiTiPt The NiTiPt alloy behaved similarly th e Pd modified alloy in unconstrained recovery tests. The overall performance in shap e recovery is comparable to that of the Pd alloy at lower strains and is sl ightly lower as initial stain le vels increase. Again there are three main identifiable regions attributed to several mechanisms contributing to the total shape recovery as shown in Figure 4-33. One distinguishable difference in this alloy is the sharpness of the phase transformation. The NiTiPd alloy like traditional shape

PAGE 104

90 memory alloy have a narrow hysteresis (Figure 4-32) in contrast to the NiTiPt alloy in which the recovery attributed with the forw ard and reverse shear phase transformations occurs over approximately 100 degrees Celsius. An interesting behavior in this alloy is that at low initial strains there is no recovery associated with the martensite to austenite transformation yet the material totally recovers after a full hysteresis. Approximately 80% of shape recovery is attained during the austenite to martensite transformation which is significan tly greater than that in the Pd alloy. Similarly to the Pd modified alloy this eff ect quickly drops off as slip becomes more prevalent. What makes this result interes ting is that, although it is known that at low strain levels the martensite is being deform ed completely by detwinning as is evident in the total recovery observed, there is no significant recovery associated with the martensite to austenite reaction. A qualitative explanation proposed here links this effect to the simultaneous recovery of the detwi nned martensite which is counteracted by the transformation strain which was measured by dilatometric techniques. Constant-Load, Strain-Temperature Tests and Work Output A significant amount of research has been performed on unconstrained shape recovery of NiTi-30Pd alloys, yet no investigations on the materials ability to perform work have been reported. This shape memo ry alloy specific material property is a fundamental design parameter in actuation applications. The primary application identified for high-temperature shape memory alloys such as NiTi Pd and NiTiPt would be for use as solid state actuator material s where work output (o r the ability of the material to recover strain ag ainst some biasing force) is the primary consideration. Therefore, characterization of the work output is required for these alloys to advance from the development stage to real world app lications. Consequentl y, load bias testing

PAGE 105

91 was performed on the NiTiPd and NiTiPt alloys in order to quantify the work output as a function of the applied stress. Constant-load, strain-temperature te sts and work output : NiTiPd Load bias testing was applied to the Ni TiPd alloy as a method of measuring its specific work output as a function of the appl ied stress and mode of application (uniaxial tension or compression). A series of cons tant load tests were conducted in tension (Figure 4-34) and compression (Figure 4-35). Strains are plotted as a function of temperature for each applied stress. During each test, the resistivity of the sample was continuously monitored and recorded although th ese results will not be included here as analysis revealed that this portion of the measurement provided no additional information during load bias testing. Essent ially, the temperature vs. resis tivity relationship paralleled the temperature vs. strain relationship. Measurements of the transformation strain used to calculate the work output are taken from the second thermal cycle through a complete hysteresis under load, due to the formation of a preferred orient ation distribution of corresponden t variant pairs that forms. In order to describe this e ffect and why measurements ar e taken on the second cycle the details of Figure 4-36 are described here, which is representative of all the high temperature shape memory alloys included in this study. Figure 4-36 shows the temperature vs. strain relationship for the entire thermomechanical path which includes loadi ng, the first two thermal cycles and, finally, unloading. Initially, the sample was loaded to the test stress at r oom temperature which is identified as region 1 in Figure 4-36. During this phase of the experiment the martensite material is composed of a n early random orientation distribution of correspondent variant pairs and deforma tion is accommodated by detwinning the

PAGE 106

92 martensite. The sample was then heated unde r load marked as region 2. During the first heating cycle, as the alloy transforms from martensite to austenite there is a small increase in strain. After reaching the maximu m test temperature the sample is allowed to cool through the transformation temperatur es (region 3). During cooling through the transformation temperatures (austenite to martensite) a large transformation strain occurs which mostly results from the formation of a highly directional distribution of correspondent variant pairs under the applied st ress. This is an important mechanism because the orientation distribution in the marten site is not the same as the distribution in the undeformed material. Under the applied st ress, the variants orient themselves such that their alignment increases the transformation strain. This is comparable to the product of detwinning a random martensite distribut ion where the applied stress causes certain variants to be favorable over others. The ma terial is essentially detwinned with one significant difference. Principally, the stress re quired to produce a de twinned structure is much lower when transforming the material und er load than by deforming the martensite isothermally. This is clearly evident by co mparing the amount of deformation attained during the isothermal loading of the marten site denoted by region 1 to the amount of strain which results during the phase transf ormation under load. During the first thermal cycle, the strain in the sample due to the initial applied load at room temperature was smaller than when the sample was cooled th rough the martensite temperature to room temperature under load. During the second heating cycle (marked as re gion 4), the martensite is transformed to austenite which results in the recovery of a significant portion of the transformation strain which occurs during the forward reaction (r egion 3). It is important to note that the

PAGE 107

93 transformation strain during th e second heating cycle (region 4) is significantly greater than the same reaction on the first heating cycl e (region 2). As the martensite transforms to austenite the material exhibits a change in strain against the bi asing load, therefore, work is performed and the strain measured on the second heating cycle is used to calculate the transformation strain. Additionally, these alloys do not exhibit full recovery of the transformation strain as evident from comparison of the change in strain marked in regions 3 and 4. The difference in strain is termed the open loop st rain. This is due to non-recoverable slip processes which occur alongside the shear transformation. Referring back to the isothermal uniaxial thermal mechanical te st (Figure 4-20), it was clear that the martensitic yield stress exhibits a minimu m in the vicinity of the transformation temperatures. Apparently, the martensite is deforming partially by slip during the austenite to martensite transformation as ev ident in the differences in transformation strain. One possible reason as to why these a lloys seem to slip during the transformations stems from the fact that the martensite aust enite interface contains a high interfacial stress resulting from the shear transformation. Adjacen t to the interface, the material exists at a higher stress, therefore, application of extern al stresses can cause th is interface to surpass the stress level required to in itiate plastic deformation and the material slips. During thermoelastic recovery, the plastic deformation is not recovered and is carried over to the next cycle. Referring back to Figure 4-34 and Figure 4-35, the biasing lo ad and specific work output are given next to their respective te mperature strain plot recorded during the second heating and cooling cycle. Note that the engineering strain values are offset in

PAGE 108

94 order to include the entire data set on a single strain axis. Generally, the transformation strain is a function of the biasing stress. Additionally, the thermodynamics thermoelastic transformations force the equili brium transformation temperatur es to change with applied stress which is evident in both figures. In compression at stresses greater than about 400MPa, thermally activated slip becomes mo re prominent and the material deforms continuously at temperatures past the Af which is exemplified in Figure 4-35 at 687MPa. A plot of the biasing stress dependent specific work output and transformation strain are shown in Figure 4-37 and Figure 4-38. The work outputs in tension and compression as a function of stress were sim ilar. The specific work output increased with increasing stress level, with a maximum work output of about 12 J/cm3 attained by interpolation at approximat ely 500 MPa, and then began to decrease with further increases in stress. This peak is evident fr om the compression data since the tensile data reached a maximum just before the peak but was actually limited by fracture of the sample during testing. The st ress dependent behavior of the specific work output stems from the transformation strain which also initially increased with increasing applied stress up to a maximum before decreasing. The transformation of high symmetry cubic phase to a lower symmetry orthorhombic phase has the freedom to produce many possible equivalent variants whereas the detwinning of martensite can occur through a limited number of detwinning reactions producing fewer equivalent variants61. Thus, whereas the detwinning of marten site is strongly isotropic, the formation of equivalent martensite variants from the austenite phase is dependent on the applied external stresses. The applied stresses has a distinct effect on the distribution of vari ant orientations that form during the austenite to martensite reac tion. Tests conducted in tension contained a

PAGE 109

95 variant structure that was optimized during the phase transformation for tensile loading while tests conducted in compression containe d a structure optimized for compressive loading. The work output stress behavior of the Ni TiPd was similar to that observed in conventional NiTi alloys. Conventional NiTi alloys also exhibit a peak in the work output stress behavior with maximum values of 10 20 J/cm3.41,46 This peak in work output is due to competing factors. Primaril y, as the applied stress increase it prevents complete recovery of the transformation strain during the martensite to austenite reaction which causes the transformation strain to reach a maximum. Therefore, even though the applied stress is increasing, the transformati on strain during the heating cycle begins to decrease. The product of the applied stress and transformation strain is work output which reaches a maximum at a particular stress level. The mechanism as to why the transforma tion strain reaches a maximum has not been clearly identified yet tw o significant contributions have been found. Initially, as the biasing stress increases it forces the tr ansforming austenite to form a less random distribution of variants and thus the transformation strain increases. Second, as the biasing stress further increases, it also cau ses non-reversible slip to occur during the forward and reverse transformations. During the austenite to martensite transformation slip occurs which results in an increasing amount of strain duri ng the transformation. Similarly, in the reverse reaction (martensite to austenite) the oriented martensite variants transform to the parent austenitic phase which results in the decrease in the total sample strain. As this transformation occurs, the al loy may simultaneously slip which results in a decrease in the transformati on strain against the biasing load. Both slip processes

PAGE 110

96 contribute to the total open loop strain by increasing the transformation strain associated with the austenite to martensite reaction a nd decreasing the recovery strain associated with the austenite to martensite reaction. A summary plot of the open loop strain as a function of applied biasing stress is given in Figure 4-39 and confirms these observations. The open loop strain increases with increasing load thus causing the transformation strain and thus the work output to reach a maximum. Constant-load, strain-temperature tests and work output:NiTiPt Load bias testing was also performed in compression on the NiTiPt alloy shown in Figure 4-40. Compression was chosen since the results for the NiTiPd load biasing tests in tension and compression showed that the work output in both loading schemes was comparable. This alloy behaved similarly to the NiTiPd alloy, with the exception that the work output was significantly lowe r at all biasing stresses. In line with the results for the NiTiPd alloy the NiTiPt alloy exhibited a maximum in the transformation strain and stress which are shown in Figure 4-41 and Figure 4-42, respectively, which implies that similar deformation mechanisms occur in the NiTiPt alloy. The a pplied stresses which correlate to the maximum transformation stra in and work output closely match the stress at which the maximums were obs erved in the NiTiPd alloy. In contrast to the NiTiPd alloy at very high stresses the transformation strain and work output essentially drop to zero. Although we know the transformation is not completely suppressed by the applied stress which was determined by resistive measurements and by monitoring the power requir ed to maintain a constant heating rate through the endothermic transformation, there was no significant recovery strain during the forward and reverse reactions. A plot of the open loop strain is shown in Figure 4-43 which is similar to the NiTiPd alloy at low st resses. Further increases in the load biasing

PAGE 111

97 stress causes the open loop strain to decrea se signaling that the material has work hardened sufficiently such that the majority of the stress is be ing carried by sessile dislocation networks and not elastically by the martensite or austenite. Further experimental work is required to confirm th is implication which is based on the following argument. A shape memory alloy which is loaded below the stress required to initiate slip carries most of the load elastically. Even if the material is detwinned the majority stress is carried elastically. As the material shears from one structure to another there is a strain change associated with the orientation and displacem ent of the new structure. Now, if the alloy is loaded well above the yield strength where a significant amount of plastic deformation has resulted in work ha rdening, the load is be ing carried mostly by dislocation interactions. Micr ostructural studies have show n that dislocations remain sessile in the parent and shear phases in thermoelastic transf ormations. Therefore, since there is little dislocation m ovement during the transformati on a transformation strain is not observed. Prior studies have shown for a shape memo ry alloy to perform with a high work output under a biasing load, the strength of th e austenite should be significantly lower than the detwinning stress of the martensite as measured by isothermal uniaxial testing methods35. The current study confirms this resu lt by the comparison of two alloys in similar alloy systems. In the NiTiPd all oy which performed bette r under a biasing load, the difference in strength between the martensi te and austenite is greater than in the NiTiPt alloy. Comparing the alloy explored here with a Ni20Ti50Pt30 alloy also subjected to load biasing in a prior study, the strength of the martensite was greater than that of the

PAGE 112

98 austenite and the material exhibited no capacity to perform work. This is an important confirmation of a critical obser vation in the advancement of shape memory alloy design. Temperature (C) 250300350400450500550600650 Resistvitiy (micro ohm cm) 0 20 40 60 80 100 120 140 160 1hr 500C 1hr 600C No heat treament Figure 4-10 Stress Relief Heat Treatment Optimization by Analysis of Resistivity Temperature Profiles (note the resistivity curves are offset for convenience on the same resistivity scale).

PAGE 113

99 Figure 4-11 SEM BSE image of extruded Ni19.5Ti50.5Pd30 Figure 4-12 NiTiPt Resistivity and Dilatometry Test Results Temperature (oC) 200300400500600700 Strain (%) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Resistivit y ( micro-ohm*cm ) 20 40 60 80 100 120 140 160 Dilatometry Resistivity25Ni 50Ti 25Pt

PAGE 114

100 Figure 4-13 NiTiPd Resistivity and Dilatometry Test Results Engineering Strain (%) 051015202530 Force (lb) 0 1000 2000 3000 4000 5000 6000 365CCompression Figure 4-14 NiTiPd Force Strain Curve at 365 C 20Ni 50Ti 30Pd Temperature (oC) 0100200300400500 Resistivity (micro-ohm*cm) 40 60 80 100 120 140 Strain ( % ) 0.0 0.2 0.4 0.6 0.8 1.0 Dilatometry Resistivity20Ni 50Ti 30Pd

PAGE 115

101 Engineering Strain (%) 05101520 Engineering Stress (MPa) 0 500 1000 1500 2000 Tension200C RT 300C 400C Figure 4-15 NiTiPd Alloy Uniaxial Isothermal Te nsile Tests at RT, 200 C, 300 C and 400 C (a) Engineering Stress Strain Curve (b) True Stress Strain Curve including correction fo r non-uniform deformation of the 400C sample Engineering Strain (%) 05101520 Engineering Stress (MPa) 0 500 1000 1500 2000 2500 CompressionRT 200C 400C 350C 365C 500C Figure 4-16 NiTiPd Alloy Uniaxial Isothermal Co mpression Tests at RT, 200 C, 350 C, 365 C, 400 C and 500 C (a) Engineering Stress Strain Curve (b) True Stress Strain Curve True Strain (%) 05101520 True Stress (MPa) 0 500 1000 1500 2000 2500 CompressionRT 200C 400C 350C 365C 500C 60 7 0 8 0 9 0 0 0 0 2 0 (a) (b) (a) (b) True Strain ( % ) 05101520 True Stress (MPa) 0 500 1000 1500 2000 Tension200C RT 300C 400C

PAGE 116

102 Figure 4-17 NiTiPd Alloy Uniaxial Isothermal Tensile Tests at 225 C, 245 C, 255 C and 272 C Figure 4-18 NiTiPd Alloy Uniaxial Isothermal Compression Tests at 255 C, 272 C and 300 C

PAGE 117

103 True Strain 0510152025 True Stress (MPa) 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Corrected Resistivity (arb. unit) 70 75 80 85 90 95 100 105 Stress Resistivity Figure 4-19 Isothermal Uniaxial Stress Strain Curve with Resistivity Exhibiting a Stress Induced Transformations Temperature (C) 0100200300400500600 Yield Stress (MPa) 150 200 250 300 350 400 450 500 Compression Tension Figure 4-20 NiTiPd Yield Stress vs Temperature in tension and compression

PAGE 118

104 True Strain (%) 0.00.51.01.52.02.53.0 0 200 400 600 800 True Strain (%) 0.00.51.01.52.02.53.0 0 200 400 600 800 Figure 4-22 NiTiPt Alloy Uniaxial Isothermal Tensile Tests at 440 C, 470 C, 550 C and 600 C Temperature (C) 0200400600 Dynamic Youngs Modulus (GPa) 0 20 40 60 80 100 120 Figure 4-21 Temperature Depende nt Dynamic Elastic Modulus measured on heating Temperature (C) 200300400500600 60 80 100 120 140 Tension

PAGE 119

105 True Strain (%) 05101520 True Stress (MPa) 0 500 1000 1500 2000 2500 3000 470C 600C 440C 550C Figure 4-23 NiTiPt Alloy Uniaxial Isothermal Compression Tests at 440 C, 470 C, 490 C, 550 C, and 600 C True Strain (%) 0.00.51.01.52.02.53.0 0 200 400 600 800 200C 380C 400C 440C Figure 4-24 NiTiPt Alloy Uniaxial Isothermal Tensile Tests at 200 C, 380 C, 400 C and 440 C 200300400500600 60 80 100 120 140 Compression Tt(C) 200300400500600 0 0 0 0 0 Tension

PAGE 120

106 True Strain (%) 05101520 True Stress (MPa) 0 500 1000 1500 2000 2500 3000 200C 380C 400C 440C Figure 4-25 NiTiPt Alloy Uniaxial Isothermal Compression Tests at 200 C, 380 C, 400 C and 440 C Temperature (C) 200300400500600 60 80 100 120 140 200C 380C 400C 440C True Strain (%) 0.00.51.01.52.02.53.0 True Stress (MPa) 0 100 200 300 400 500 600 550C low strain rate 550C high strain rate (10X low rate) Figure 4-26 Stress Strain Curve at 500 Celsius at Low and High Strain Rates Compression

PAGE 121

107 Figure 4-27 Yield Stress vs. temperature for NiTiPt Temperature ( o C) 100200300400500600700 TrueYield Stress (MPa) 0 100 200 300 400 500 600 Compression Tension Temperature ( o C) 100200300400500600700 Engineering Strain (%) 0 1 2 3 4 5 6 7 UTS (MPa) 200 400 600 800 1000 1200 Failure Strain Failure Stress Figure 4-28 Fracture Stress and Stra in vs. temperature for NiTiPt

PAGE 122

108 Tem p erature ( o C ) 50100150200250300350400450 Strain (%) 0 1 2 3 4 5 6 T h e r m a l E x p a n s i o n A u s t e n i t e T h e r m a l E x p a n s i o n A u s t e n i t e Figure 4-29 NiTiPd Unconstrained Recovery Te st at 4 and 2 Percent Initial Strains. Total Strain (elastic + residual) 0123456 Recovery Rate (recovery strain/(recovery strain + plastic strain) 84 86 88 90 92 94 96 98 100 102 Figure 4-30 Total Recovery Rate vs. Total Strain for NiTiPd

PAGE 123

109 Tem p erature ( oC ) 100200300400500600700 Strain (%) -1 0 1 2 3 4 5 Figure 4-32 Temperature Depende nt Load Free Recovery Curve for Complete NiTiPt Test Inital Plastic Strain (%) 0.00.51.01.52.02.53.03.54.0 Percent Recovery (%) 0 20 40 60 80 100 120 140 Total Recovery after a complete hysteresis Austenite recovery (As to Af) Martensite recovey (Ms to Mf) Themal martensite Recovery (min temp As) A B C D Figure 4-31 Load Free Recovery Individual Co mponents of Total Recovery for NiTiPd

PAGE 124

110 Temperature (C) 100200300400 Strain (%) 2 4 6 8 10 12 98.6 MPa ( 1.5 J/cm 3 ) 295.1 MPa ( 7.52 J/cm 3 ) 196.5 MPa ( 4.3 J/cm 3 ) 393 MPa ( 9.6 J/cm 3 ) Figure 4-34 Load Bias in Tension (S pecific Work Output) for NiTiPd Initial Plastic Strain (%) 0.00.51.01.52.02.53.03.5Percent Recovery 0 20 40 60 80 100 120 Total Recovery after a complete hysteresis Austenite recovey (As to Af) Martensite recovey (Ms to Mf) Themal martensite Recovery Figure 4-33 Load Free Recovery Individua l Components of Total Recovery for NiTiPt

PAGE 125

111 Temperature ( o C) 0100200300400500 Engineering Strain (%) -14 -12 -10 -8 -6 -4 -2 0 2 687 MPa (6.41 J/cm 3 ) 517 MPa (8.53 J/cm 3 ) 393 MPa (9.19 J/cm 3 ) 295 MPa (7.7 J/cm 3 ) 197 MPa (4.5 J/cm 3 ) 99 MPa (1.55 J/cm 3 ) Figure 4-35 Load Bias in Compression (Specific Work Output) for NiTiPd Temp (C) 0100200300400 Strain (%) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Region 1 Region 2 Region 3 Region 4 Figure 4-36 Load Bias in Tension (Specific Work Output) Complete Thermomechanical Path for NiTiPd

PAGE 126

112 Stress (MPa) 0100200300400500600700800 Specific Work (joule/cm 3 ) 0 2 4 6 8 10 12 Figure 4-37 Specific Work vs. Biasing Load for NiTiPd Stress (MPa) 0100200300400500600700800 Transforamtion Strain (%) 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Figure 4-38 Transformation Strain (2nd cycle martensite to austenite) vs. Biasing Load for NiTiPd

PAGE 127

113 Stress (MPa) 0100200300400500600700800 Open Loop Strain (%) 0 1 2 3 4 5 Tesion Compression Figure 4-39 Open Loop Strain vs Biasing Stress for NiTiPd Tem p erature ( o C ) 300400500600 Engineering Strain (%) 0 2 4 6 8 10 12 14 458 MPa ( 1.44 J/cm 3 ) 345 MPa (6.72 J/cm 3 ) 690 MPa (1.03 J/cm 3 ) 255 MPa (5.52 J/cm 3 ) 176 MPa (2.62 J/cm 3 ) 172 MPa (2.83 J/cm 3 ) 86 MPa (0.72 J/cm 3 ) Figure 4-40 Load Bias in Compression (Specific Work Output) for NiTiPd

PAGE 128

114 Stress (MPa) 0100200300400500600700800 Transforamtion Strain (%) 0.0 0.5 1.0 1.5 2.0 2.5 Figure 4-41 Transformation Strain (2nd cycle martensite to austenite) vs. Biasing Load for NiTiPt Stress (MPa) 0100200300400500600700800 Specific Work (joule/cm 3 ) 0 1 2 3 4 5 6 7 8 Figure 4-42 Specific Work vs. Biasing Load for NiTiPt

PAGE 129

115 Stress (MPa) 0100200300400500600700800 Open Loop Strain (%) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Figure 4-43 Open Loop Strain vs Biasing Stress for NiTiPt

PAGE 130

116 CHAPTER 5 SUMMARY AND CONCLUSIONS Alloy Development Of the several candidate alloy systems for high temperature shape memory actuation, the NiTiPt and NiTiPd systems were se lected. Pt or Pd additions to binary NiTi above approximately 10%, satisfying the Ni50-xPtxTi50 or Ni50-xPtxTi50 compositional relationship, raise the thermo elastic transformation temperat ures significantly. Although both alloying additions to NiTi resulted in elevated transformation temperatures and comparable ordered orthorhombic crystal stru ctures, each system had quite different mechanical properties and shape memory attributes. Therefore, a multi system comparison was explored in an attempt to draw parallels to these differences in mechanical properties and the shape memory specific properties of each alloy. Two alloys, one from each alloy system were selected for thermo-mechanical testing and comparison of shape me mory specific properties. The Ni30Ti50Pt20 alloy was selected based on the results from an initial screening study of over 20 NiTiPt alloys that focused on elevated transformation temperatur es and the effects of deviations from stoichiometry (i.e., deviations from (Ni+Pt)/Ti = 1). Analysis of this compositional study in combination with the results of a ba seline thermo-mechanical studies on the Ni30Pt20Ti50 and Ni20Pt30Ti50 facilitated the selection of a Pt modified NiTi shape memory alloy based on transformation temperatures, microstructure and sp ecific work output. The NiTiPd alloy was selected based on prio r studies which focused on the effect of stoichiometric Pd for Ni alloying additions on transformation temperatures and no-load

PAGE 131

117 recovery tests. Particularly the baseline composition of 30Pd showed a maximum in the no-load recovery therefore it seemed proba ble that this alloy was a good candidate for high temperature shape memory actuation. The final selection of the Ni30Pt25Ti50 and Ni30Pd30Ti50 was made slightly Ti rich by .5% in order to insure elevated transformati on temperatures. This yields a small volume fraction Ti rich phase since bot h the iso-stoichiometric lines between NiTi-PtTi and NiTi and PdTi are strong line compounds with little solubility for excess Ti. This Ti rich phase has a high affinity for in terstitial elements and theref ore reacts with any available carbon impurities acquired from the graphite cr ucible to form carbides. The excess Ti has a high affinity for carbon and oxygen thus assures that very little carbon or oxygen goes into solid solution but rather forms a small volume fraction of oxide and or carbide phases. Characterization and Thermomechanical Testing The transformation temperatures where determined by electrical resistivity techniques and dilatometric techniques. The change in resistivity and transformation strain were measured along side the temperat ure coefficient of resistivity and thermal expansion. These are important parameters in the design of shape memory alloy devices. A unique instrument was developed which simultaneously measure mechanical and electrical properties. This instrument wa s employed in order to accurately determine what the initial structure is and where th e test temperature lies relative to the transformation temperatures. Additionally th e structure during deformation and loading schemes is determined by analysis of the samples electrical resistivity. Isothermal mechanical test were conduc ted above below and at the transformation temperatures which developed the baseline m echanical properties of the austenitic and

PAGE 132

118 martensitic phases. Analysis of the temper ature dependent yield stress showed that the difference in yield strength between the high and low temperature phases was significantly different in the NiTiPt and NiTiPd alloys. The NiTiPt alloy exhibited a smaller change in yield strength between the martensite and the austenite. Recalling that the yield strength in the martensite is indi cative of the stress required for detwinning while the yield strength in the austenite is related to slip processes therefore for a shape memory alloy to be a good candidate for actua tion applications the martensite must be weaker than the austenite. Additionally the ma rtensite in the NiTiPt alloy fractured in a brittle manner at low strains. It was possibl e to stress induce martensite in the NiTiPd alloy while no detectable stress induced transf ormations occurred in the NiTiPt alloy. Dynamic modulus measurements provided c onfirmation of pre-martensitic elastic softening in the NiTiPd alloy which is an important characteristic of thermoelastic transformations. Analysis of the fracture stress and strain at various temperatures revealed a change in fracture mechanisms (ductile to brittle be havior) near the transformation temperatures of the NiTiPt alloy. Particul arly the stress and strain bo th decreased with increasing temperature up to the transformation temp eratures after which the fracture stress decreased with increasing temperature as th e fracture strain increased. Isothermal uniaxial tests at different st rain rates verified that dyna mic recovery is a prevalent mechanism in the austenite phase. This is a feasible mechanism which explains the increase in transformation strain and associ ated decrease in transformation stress as temperature increases.

PAGE 133

119 Load free recovery tests were performe d on each alloy incrementing the initial strain levels until fracture. The strain was monitored continuously during the deformation and recovery process. Ther mal cycling through the hysteresis revealed several stages of recovery pr ocesses which contribute to th e total recovery. The three main contributions consisted of thermal rec overy, martensite to austenite recovery and austenite to martensite recovery. The anal ysis of these measurements revealed the fraction of strain which is accommodated by reversible (detwinning) and non-reversible processes (slip) as well as the strain a ssociated with the formation of correspondent variant pairs. Combined with the isotherm al martensitic stress-strain curves it was possible to correlate regions of the stress st rain curves with the underlying deformation mechanisms. Particularly the pseudo stress strain curve, for low strain values, was determined to be deforming completely by de twinning yet the rearra ngement of twins did not occur at a constant stress. Load bias test were conducted in tens ion and compression for the NiTiPd alloy. These test measure a shape memory alloys ab ility to do work (specific work output). Thermal cycling through the hysteresis wa s performed under a constant load. The transformation strains associated with the fo rward (austenite to martensite) and reverse (austenite to martensite) reactions were m easured. The specific work was calculated from the product of the engineering stress and th e transformation strain or recovery strain during the heating cycle upon which the sample does work against the biasing load. The results were similar in tension and comp ression. A maximum in the work output vs. biasing stress relationship was seen in bot h tension and compression. This maximum was linked to a maximum in the recovery strain which results from two components, mainly

PAGE 134

120 non-recoverable slip and the formation of an oriented correspondent variant pair structure. Conclusions Relevant to Alloy Design This study is a section of a larger study geared towa rd the development of high temperature shape memory actuation materi als. The further development of high temperature shape memory alloys to the poi nt were they are feasible for commercial applications is limited by the limite d knowledge database on advanced alloy development, thermomechanical processing and mechanical test procedures and results. Several goals were sought in this portion of the study. Unique instrumentation and testing methods were explored and devel oped to examine the baseline mechanical properties in combination with shape memory alloy specific properties. Additionally, this study was to identify areas in high temper ature shape memory alloys where targeted alloy development and thermomechanical pr ocessing could improve the shape memory characteristics of the alloy, specifically the wo rk output. In order to accomplish this two comparable alloys were developed from a re latively well known system (NiTiPd) and a system which, although superior in high temp erature transformation temperatures, has received little advanced thermomechanical studies. These systems parallel each other structurally, yet have very different shape memory performances. Therefore a comparison of these alloys could give some insight as to where future work should focus. Throughout this study it seemed that two main properties where inhibiting the performance of these alloys. Pr imarily the alloys resistance to slip under a biasing load in the temperature regions near the transfor mation temperatures prevented complete recovery and thus a maximum in the recovery strain. Additionally a link between the difference in yield strength between the au stenite and martensite and the performance

PAGE 135

121 under a biasing load was confirmed. During the transformation the stress state in the interface is high, therefore, in or der to prevent slip in a region of the material which is already highly stressed a permanent strengthen ing mechanism which is stationary as the interface passes must be present. Second phase strengthening and dispersion hardening are both methods which may accomplish this. Additionally a sessile dislocation network which is formed by the movement of the in terface itself could prove useful in forming entanglements for mobile dislocations. Alt hough a direct link to the prior statement has not been made, repeated movements of the in terface under load (training) has been shown to decrease the amount of slip under a biasing load. This would be feasible for the NiTiPd alloy and not the NiTiPt alloy as dynami c recovery sets in heavily just above the transformation temperatures. Future Studies Although mechanisms explaining the deform ation behavior were proposed here, significant gaps remain in their experimental confirmation. In order to address these a series of experiments could be conducte d which analyze the extruded and deformed structures. This investigation consists of three main parts. Primarily a set of martensitic samples will be strained in into the different regions of the stress strain curves then unloaded. The as extruded and deformed st ructures will then be analyzed by TEM and the results compared. This could be used to determine prevalent detwinning reactions by the comparison of the starting and deformed structures. A similar study has been performed on NiTi thin films62. A second part of the investigation will anal yze the structure of the stress induced martensite which consists of highly oriented correspondent variant pairs, in comparison to the thermal martensite which should be more randomly oriented. Although not

PAGE 136

122 included in this work preliminary investigations showed that at particular test temperatures the stress induced martensite rema ins in the material even after the applied stress is removed. Upon unloading a two phases exist in the alloy. These are the stress induced martensite and the austenite. Now if the sample is allowed to cool the remaining austenite transforms thermally to martensite. It has been verified in preliminary test that these structures are indeed different by DTA analysis which showed two peaks in a sample which contains stress induced martensi te while a sample which does not exhibits one peak. The orientation relationship betw een the stress induced martensite and the martensite which transforms under no-load coul d be exploited to compare the formation a random distribution and an oriented one. A significant question in thermoelastic tran sformations and in the development of shape memory alloys ask how is the strain be accommodated. If it is largely plastic, then the deformation is not reversible in contra st to elastic accommodation and detwinning. The final phase of the proposed study app lies the thermodynamics of thermoelastic transformations as a tool to measure the am ount of residual elastic stress remnant after deformation. This portion of the stu dy models the relationship between the transformation temperatures and applied stress. Prior work conducted here has developed a test method which measures the shift in transformation temperatures and has successfully conducted the necessary test to build a baseline relationship between the transformation temperatures and the material s stress state. Now by taking the alloys deformed for TEM analysis and recovering th em in a DTA it is possible to determine what the internal stress state of the material is by correlating the shift in transformation temperature to the materials stress state.

PAGE 137

123 APPENDIX A NiTiPd HTSMA MATERIAL DATA SHEET ( This material is suitable for shape me mory applications with transformation temperatures greater than 200 C) Physical Properties Density 7.6 g/cm3 Electrical Resistivity Martensite (@ AS) 93 ohm-cm Austenite (@ MS) 114 ohm-cm Thermal Coefficient of Resistivity Martensite 12 x 10-2 ohm-cm /C Austenite 18 x 10-3 ohm-cm /C Coefficient of Thermal Expansion Martensite 16 x 10-6/C Austenite 11 x 10-6/C Shape Memory Properties Transformation Temperatures Austenite Start Temperature (As) 255 C Austenite Finish Temperature (Af) 260 C Martensite Start Temperature (Ms) 249 C Martensite Finish Temperature (Mf) 239 C Composition Nickel (nominal) 19.5 at.% Palladium (nominal) 30.0 at.% Titanium (balance) balance Oxygen (maximum) 0.30 at.% Carbon (nominal) 0.50 at.%

PAGE 138

124 APPENDIX B NiTiPt HTSMA MATE RIAL DATA SHEET ( This material is suitable for shape me mory applications with transformation temperatures greater than 400 C) Physical Properties Density 9.9 g/cm3 Electrical Resistivity Martensite (@ AS) 110 ohm-cm Austenite (@ MS) 135 ohm-cm Thermal Coefficient of Resistivity Martensite 85 x 10-3 ohm-cm /C Austenite 91 x 10-4 ohm-cm /C Coefficient of Thermal Expansion Martensite 99 x 10-7/C Austenite 96 x 10-7/C Shape Memory Properties Transformation Temperatures Austenite Start Temperature (As) 446 C Austenite Finish Temperature (Af) 491 C Martensite Start Temperature (Ms) 458 C Martensite Finish Temperature (Mf) 417 C Composition Nickel (nominal) 19.5 at.% Platinum (nominal) 25.0 at.% Titanium (balance) balance Oxygen (maximum) 0.30 at.% Carbon (nominal) 0.50 at.%

PAGE 139

125 APPENDIX C CHEMICAL ANALYSIS OF EXTRUDED MATERIALS Ext. ID Ti Ni Pt Pd C N O 20 50.4 19.3 29.4 0.446 0.014 0.334 21 50.5 19.4 29.3 0.437 0.014 0.305 22 50.4 19.4 29.5 0.436 0.014 0.268 23 50.6 19.6 29.0 0.400 0.014 0.270 24 50.6 19.3 29.4 0.421 0.014 0.263 25 50.6 19.2 29.4 0.482 0.014 0.266 30 50.4 24.4 24.4 0.431 0.018 0.277 31 50.3 24.4 24.3 0.645 0.018 0.306 32 50.3 24.4 24.5 0.504 0.012 0.283 33 50.4 24.4 24.4 0.449 0.018 0.269 34 50.4 24.4 24.4 0.503 0.015 0.280 35 50.3 24.4 24.5 0.468 0.012 0.299

PAGE 140

126 LIST OF REFERENCES 1. C. M. Wayman MRS Bulletin April (1993) 49. 2. R.D. Noebe, T. Biles, and S.A Padula, in Advanced Structural Materials: Properties, Design Optimization, and App lications, W.O. Soboyejo, ed., Taylor & Francis Group, Boca Raton, FL. 2005. p. 141. 3. C.M. Wayman and T.W. Duerig, in Engineering Aspects of Shape Memory Alloys, ed., Butterworth-Heinemann, Boston, 1990. p. 3. 4. J. V. Humbeeck, Maters. Sci. Eng. A 273 (1999) 134. 5. L. McD. Schetky, Maters. Sci. Forum 327 (2000) 9. 6. C. Mavroidis, Res. Nondestr. Eval., 14 (2002) 1. 7. C.M. Hwang and C.M. Wayman, Scripta Metall. 17 (1983) 38. 8. S. Miyazaki and K. Otsuka, Metall. Trans. 17A (1986) 53. 9. K. Otsuka and X. Ren, Intermetallics 7 (1999) 511. 10. K. Hane and T. Shield, Ac ta Materialia 47 (1999) 2603. 11. P.G. Lindquist, Ph.D. Dissertation, University of Illinois (1988). 12. T. Hara, T. Ohba, and K. Otsuka Mater. Trans. JIM 38 (1997) 11. 13. L.L. Meisner, V.P. Sivokha Physica B 344 (2004) 93. 14. R. Reed-Hill, R. Abbaschian, in Phys ical Metallurgy Principles, ed. PWS Publishing, 1992. p. 267. 15. R. J. COMSTOCK, Acta Met. 44 (1996) 3505. 16. D. S. Lieberman, M. S. Wechsler, and T. A. Read, J. Appl. Phjx. 26 (1955) 473. 17. M. S. Wechsler, D. S. Lieberman, and T. A. Read, Trans. AIME 197 (1953) 1503. 18. J. S. Bowles and J. K. Mackenzie, Acta Metall. 2 (1954) 129.

PAGE 141

127 19. C.M. Wayman, in Introd uction to the Crystallography of Martensitic Transformations, ed. Macmillan, NY 1964. p. 84. 20. K. Hane and T. Shield Philosophical Magazine A. 78 (1998) 1215. 21. K. Bhattacharya, in Microstructure of Martensite, ed. Oxford University Press 2003. p.144. 22. D. A. Porter and K. E. Easterling, in Ph ase Transformations in Metals and Alloys, ed.vChapman & Hall, London, 1991. p. 122. 23. H. C. Ling and W. S. Owen, Acta Metall. 29 (1981) 1721. 24. J. Uchil, K. Ganesh Kumara, and K. K. Mahesh, Mater. Sci. Eng. A, 332 (2002) 25. 25. http://www.sma-inc.com/html/_shape_memory_alloys_.html Jan 2005. 26. L. Zhao, Ph.D. Thesis, University of Twente, Enschede, The Netherlands (1997). 27. R. Reed-Hill, R. Abbaschian, in Phys ical Metallurgy Principles, Ed. PWS Publishing, 1992. p. 580. 28. 28 D. P. Dunne and C. M. Waym an, Metall. Trans. 4 (1973) 147. 29. J. Ortin and A. Planes, Acta Met. 36 (1988) 1873. 30. G. B. Olsen and M. Cohen, Scripta Metall. 9 (1975) 1247. 31. G. B. Olsen and M. Cohen, Scripta Metall. 11 (1977) 345. 32. http://home.paran.com/babocpu/_jwhbbs/12849882/1102477889_icpms.ppt Nov 2006. 33. G.W.H. Hohne, W.F. Hemminger, and H.J. Flammersheim, in Differential Scanning Calorimetry, ed. Springer, Berlin, 2003. p. 116. 34. J.G. Hust and A.B Lankford National Bureau of Standards Report of Investigation Research Materials 8420 and 8421. 35. R. Noebe, D. Gaydosh, S. Padula II, A. Garg, T. Biles, M. Nathal, SPIE Conf. Proc.:Smart Structures and Materi als 2005 San Diego, CA.; 5761 (2005) 364. 36. P.G. Lindquist and C.M. Wayman, in E ngineering Aspects of Shape-Memory Alloys, ed. T.W. Duerig, K.N. Melton, D. Stockel, and C.M. Wayman, ed. Butterworth-Heinemann, Boston 1990. p. 132. 37. H.C. Donkersloot and J.H. Van Vuch t, J. Less-Common Mets. 20 (1970) 83.

PAGE 142

128 38. T. Biggs, M.B. Cortie, M.J. Witcomb, a nd L.A. Cornish, Metall. Maters. Trans. A 32 (2001) 1881. 39. K. Otsuka and X. Ren, Intermetallics 7 (1999) 511. 40. T. Biggs, L.A. Cornish, M.J. Witcomb, and M.B. Cortie, J. Alloys Compds. 375 (2004) 120. 41. A. Garg, Unpublished Research, NASA Glenn Research Center, 2004. 42. O. Rios, R. Noebe, T. Biles, A. Garg, A. Palczer, D. Scheiman, H. J. Seifert, M.Kaufman, SPIE Conf. Proc.: Smart St ructures and Materials 2005, San Diego, CA, 5761 (2005) 376. 43. W.B. Cross, A.H. Kariotis, and F.J. Stimler, NASA CR-1433, (1969). 44. N.G. Boriskina and E.M. Kenina, in Titanium 80, Science & Technology, Proceedings of the 4th Internati onal Conference on Titanium 1980. p. 2917. 45. N. Matveeva, Y. Kovneristyi, A. Savinov, V. Sivokha, and V. Khachin, J Phys 4312 (1982) C4-249. 46. P.G. Lindquist and C.M. Wayman, in E ngineering Aspects of Shape-memory Alloys, ed. T.W. Duerig, K.N. Melton, D. Stockel and C.M. Wayman, ButterworthHeinemann, London, 1990. p.58. 47. V.N. Khachin, Revue Phys. Appl. 24 (1989) 733. 48. Q. Tian and J. Wu, SPIE Conf. Proc. 5116 (2003) 710. 49. S. Shimizu, Y. Xu, E. Okunishi, S. Tanaka K. Otsuka and K. Mitose, Maters. Lett. 34 (1998) 23. 50. W.S. Yang and D.E. Mikkola, Scripta Metall. Mater. 28 (1993) 161. 51. K. Otsuka, K. Oda, Y. Ueno, M. Piao, T. Ueki, H. Horikawa, Scripta Metall. Mater. 29 (1993) 1355. 52. K. Otsuka and X. Ren, Prog. Maters. Sci. 50 (2005) 511. 53. M.V. Nevitt, Trans. Metall. Soc. AIME 218 (1960) 327. 54. R. Reed-Hill, R. Abbaschian, in Phys ical Metallurgy Principles, Ed. PWS Publishing,1992 p. 230. 55. W. B. Cross, A. H. Kariotis, and F. J. Stimler NASA CR-1433, 1970. 56. M. J. ek, M. Slmov, and M. Hje k, Journal of Alloys and Compounds, 378 (2004) 316.

PAGE 143

129 57. K. Gall, H. Sehitoglu, Y. I. Chumlyakov, and I. V. Kireeva Acta, mater. 47 (1999) 1203. 58. H. Sehitoglu, I. Karaman, X. Zhang, A. Viswanath, Y. Chumlyakov and H. J. Maier Acta Mater. 49 (2001) 3621. 59. V.N. Kachin, N.M. Matveeva, V.P. Sivokha D.B. Chernov, and Yu. K. Koveristyi, Doklady Akad. Nauk SSSR 257 (1981) 195. 60. Y. Suzuki, Y. Xu, S. Morito, K. Otsuka and K. Mitose, Maters. Lett. 36 (1998) 85. 61. D. Goldberg, Y. Xu, Y. Murakami, S. Morito, K. Otsuka, T. Ueki, and H. Horikawa, Scripta Metall Mater. 30 (1994) 1349. 62. W. Cai, S. Tanaka, and K. Ots uka, Maters. Sci. Forum 327 (2000) 279. 63. K. F. Hane and T. W. Shield, Ph ilosophical Magazine A, 78 (1998) 1512. 64. J.X. Zhang M. Sato, and A. Ishi da, Acta Materialia 54 (2006) 1185.

PAGE 144

130 BIOGRAPHICAL SKETCH Orlando Rios was born in Elizabeth, NJ in April 18, 1978. His parents are Jose and Marta Rios both of which were both born in Cuba and are current ly US citizens. Orlandos primary language were English and Sp anish as this is what was spoken in his household and neighborhood. Orlando was raised in Miami, Fl where he had his primary schooling. At the age of eight he began attending ka rate classes. This continued until he began college at the age of 19. He was awarded a black be lt as well as receiving high marks and recognition in several competitions. Orlando is also an established scuba diver and has completed advanced training in scuba diving including night, wreck, blue water, search and recovery and deep diving. Upon completion of high school Orlando was educated as automotive technician at Linsy Hopkins in Miami, Fl. After which he worked on rotary engines for two years while attending Miami Dade Community Colleg e where he received an associate in arts in engineering sciences. He then attended the University Of Florida where received a bachelors degree in materials science and e ngineering. Afterwards Orlando completed his Masters at the University of Florida on a NASA funded GSRP three year fellowship. He spent significant time working at the NAS A Glenn research cente rs advanced metallic division where he focused on instrumentati on, instrument development and design, alloy development and advanced thermomechanical testing of high temperature shape memory alloys. Orlandos work experience includes retail wo rk in Eckerd Drugs Pharmacy and at Rose auto parts. He also worked as a rota ry engine mechanic at CL motor sports in Miami, Fl. He also worked as a researcher at a Christian-Albrechts University of Kiel

PAGE 145

131 Germany in the Materials engineering depart ment. There he had experience working with III-V semiconductors and porous sili con. His experience there included SEM analysis, precision electrochemical etching a nd electrochemical impedance spectroscopy.