Structure and Properties of Titanium Tantalum Alloys for Biocompatibility

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Daniel E. Huber

Graduate Program in Materials Science and Engineering

The Ohio State University

2016

Committee:

Dr. Hamish L. Fraser, Advisor

Dr. David McComb

Dr. Stephen Niezgoda

Copyrighted by

Daniel E. Huber

2016

Abstract

In this thesis, the phase stability and elastic modulus of Ti–Ta simple binary alloys as well as alloys with small additions of ternary elements have been studied. The binary alloy from a nominal 8 to 28 wt.% Ta was first explored using a combinatorial approach. This approach included Laser Engineered Net Shape (LENS™) processing of materials and subsequent characterization by instrumented indentation and site specific

Transmission Electron Microscopy (TEM). The composition range of 15 to 75 wt.% Ta was further explored by more traditional methods that included vacuum arc melting high purity elements, X-Ray Diffraction (XRD) and modulus measurements made by ultrasonic methods. Beyond the simple binary, alloys with low levels of ternary elements, oxygen, aluminum, zirconium and small additions of rare earth oxides were investigated.

The crystal structure with space group Cmcm was chosen for it applicability with

P63/mmc and !"3! sub group / super group symmetry. This provides a consistent crystal structure framework for the purpose of studying the α to β transformation pathway and associated α’ and α’’ martensitic phases. In this case, the pathway is defined by both the lattice parameters and the value of the parameter “y”, where the parameter “y” describes the atomic positions of the [002]α’’ plane. It was found that the lattice

ii parameter changes in the Ti–Ta binary alloys are similar to structures reported for compositions in the Ti–Nb system of similar atomic percentages.

Although samples produced by the LENS™ process and characterized by instrumented indentation demonstrated the correct trends in modulus behavior, absolute agreement was not seen with modulus values published in literature. Alloys of the binary

Ti–Ta system produced from high purity materials do indeed show close agreement with literature where there exist two minima of modulus near the compositions of Ti–28Ta wt.% and Ti–68Ta wt.%. These two minima occur at the discreet boundary between α’ /

α’’ and α’’/ β respectively.

The role of oxygen as an alloying addition was studied as it relates to the stability of α’ and α’’ martensite, here it was found that oxygen will stabilize α’ yet cause an increase in the Young’s modulus. Rare earth additions to getter interstitial oxygen in the high purity materials show no further reduction in modulus. Conversely, additions of another α stabilizer, Al, proved to lower the α’ stability, with one composition exhibiting a modulus as low as 53 GPa. Zirconium being a neutral element regarding α and β stability slightly changed the structure and lattice parameter, while making a little or no difference in the observed modulus.

Observations by TEM of quenched specimens indicate the rise in modulus observed between the two minima is not caused the appearance of ω. Rather weak ω reflections were observed in Ti–65Ta wt.% in the as arc-melted condition and on annealing for 450°C for 24 hours. Precipitates of ω were not clearly identified by dark-

iii field TEM imaging. High Resolution Scanning Transmission Electron Microscopy

(HRSTEM) of the aged specimen indicated that ω might exist as 3-5nm particles.

Dedication

This document is dedicated to my family and especially my younger brother Dennis.

Acknowledgments

I would like to sincerely thank my advisor Dr. Hamish L. Fraser for his encouragement, patience and tireless support of my academic efforts. I would also like to thank all the members of the Fraser research group past and present for their support and mentorship.

Vita

June 1996 ...... Centerville High School

June 2003 ...... B.S. Materials Science and Engineering,

Ohio State University

June 2007 ...... M.S. Materials Science and Engineering,

Ohio State University

June 2007 to June 2013 ...... Research Engineer, Department of Materials

Science, The Ohio State University

June 2013 to present ...... Lead Instrumentation Scientist, Center for

Electron Microscopy and Analysis, The

Ohio State University

Publications

J.K. Jensen, B.A. Welk, R.E.A. Williams, J.M. Sosa, D.E. Huber, O.N. Senkov, G.B.

Viswanathan, H.L. Fraser: Characterization of the microstructure of the compositionally complex alloy Al1Mo0.5Nb1Ta0.5Ti1Zr1.

vii

J. M. Sosa, J. K. Jensen, D. E. Huber, M. A. Gibson, H. L. Fraser: Three-dimensional characterization of microstructure of high entropy alloy using STEM/HAADF tomography. Materials Science and Technology 04/2015; 31(10).

DOI:10.1179/1743284715Y.0000000049

Y. Liu, P. Samimi, I. Ghamarian, D. A. Brice, D. E. Huber, Z. Wang, V. Dixit, S. Koduri,

H. L. Fraser, P. C. Collins: Discovery via Integration of Experimentation and Modeling:

Three Examples for Titanium Alloys. JOM: the journal of the Minerals, Metals &

Materials Society 12/2014; 67(1). DOI:10.1007/s11837-014-1197-3

R. E. A Williams, D. Huber, J. Sosa, H. L. Fraser: 15 Years of Characterizing Titanium

Alloys' Microstructure by DBFIB. Microscopy and Microanalysis 08/2014; 20(S3).

DOI:10.1017/S143192761400333X

J.M. Sosa, D.E. Huber, B. Welk, J.K. Jensen, R.E.A. Williams, S. Lambert, H.L. Fraser:

3D ChemiSTEM™ Tomography of Nano-scale Precipitates in High Entropy Alloys.

Microscopy and Microanalysis 08/2014; 20(S3). DOI:10.1017/S1431927614005546

Binbin Deng, Camila M. Freria, Robert E.A. Williams, Daniel Huber, John Sosa, Philip

G. Popovich, David W. McComb: 3D Visualization of Motor-Neurons in Mice Spinal

Cord Using FIB\SEM Tomography. Microscopy and Microanalysis 08/2014; 20(S3).

viii

DOI:10.1017/S1431927614008733

John M Sosa, Daniel E Huber, Brian Welk, Hamish L Fraser: Development and application of MIPAR: A novel software package for two- and three-dimensional microstructural characterization. 04/2014; 3(1). DOI:10.1186/2193-9772-3-10

R. A. Causey, D. F. Cowgill, R. Doerner, R. Kolasinski, B. Mills, D. Morse, J.

Smugeresky, W. R. Wampler, R. Williams, D. Huber: Deuterium retention in tungsten at elevated temperatures. Journal of Nuclear Materials 08/2011; 415(1).

DOI:10.1016/j.jnucmat.2010.10.057

D Huber, H L Fraser, D O Klenov, H S Von Harrach, N J Zaluzec: Relative Sensitivity of

XEDS vs EELS in the AEM.

HO Colijn, DE Huber, PC Collins, HL Fraser: Practical Remote Microscopy Using KVM over IP. Microscopy and Microanalysis 07/2010; 16. DOI:10.1017/S1431927610057843

A. Blankemeier, D. E. Huber, H. L. Fraser, W. Goodson, Rea Williams, H. O. Colijn:

Characterization of Pseudomonas Fluorescens Bacteria on Polyurethane Using DB-FIB,

SEM and STEM. Microscopy and Microanalysis 07/2010; 16(S2).

DOI:10.1017/S1431927610062045

Rea Williams, A. Genc, D. Huber, H. L. Fraser: Sample Surface Preparation For

Traditional EBSD Collection and 3D EBSD Collection. Microscopy and Microanalysis

07/2010; 16. DOI:10.1017/S1431927610062574

Y. Murayama, S. Sasaki, R. Srinivasan, D. Huber, H. Kimura, A. Chiba, H.L. Fraser:

Mechanical properties and phase stability of Ti-Cr system alloys.

A Genç , D Huber , D Basile , HL Fraser , P Fischione : Sample Preparation for

Aberration Corrected Microscopy. Microscopy and Microanalysis 08/2008; 14(S2).

DOI:10.1017/S1431927608083669

D. Hill, R. Banerjee, D. Huber, J. Tiley, H. L. Fraser: Formation of equiaxed alpha in TiB reinforced Ti alloy composites. Scripta Materialia 03/2005; 52(5).

DOI:10.1016/j.scriptamat.2004.10.019

Fields of Study

Major Field: Materials Science and Engineering

Table of Contents

Abstract ...... ii

Dedication ...... v

Acknowledgments ...... vi

Vita ...... vii

Publications ...... vii

Table of Contents ...... xi

List of Tables ...... xvi

List of Figures ...... xvii

Chapter 1: Introduction and Objectives ...... 1

Chapter 2: Background and Critical Literature Review ...... 5

2.1 Titanium Metallurgy ...... 5

2.1.1 Titanium Alloy Classifications ...... 10

2.1.2 Intermetallic compounds ...... 13

2.1.3 Eutectoid Decomposition ...... 14 xi

2.2 Processing ...... 17

2.2.1 Primary Titanium Production ...... 17

2.2.2 Ingot Metallurgy ...... 17

2.2.3 Powder Metallurgy ...... 19

2.2.4 Effects of Oxygen and Other Impurities ...... 20

Chapter 3: Low Modulus Ti-Ta Binary Alloys; A Combinatorial Approach ...... 23

3.1 Introduction ...... 23

3.2.1 LENS™ Deposition Procedure ...... 24

3.2.2 Instrumented Indentation ...... 26

3.2.3 Scanning Electron Microscopy ...... 29

3.2.4 Focused Ion Beam / Scanning Electron Microscopy ...... 30

3.2.5 Transmission Electron Microscopy ...... 31

3.3 Results ...... 31

3.4 Discussion and Conclusions ...... 34

Chapter 4: Structure and Properties of Low Modulus Ti-Ta alloys ...... 39

4.1 Introduction ...... 39

4.2 Experimental Procedures ...... 40

4.2.1 Vacuum Arc Melting ...... 40

4.2.2 Ultrasonic Elastic Modulus Measurement ...... 41

xii

4.2.3 X-Ray Diffraction ...... 42

4.2.4 Focus Ion Beam Specimen Preparation ...... 44

4.2.5 Transmission Electron Diffraction ...... 44

4.3 Results ...... 45

4.4 Conclusions and Discussion ...... 47

4.4.1 Observations related to Modulus ...... 47

4.4.2 Observations related to structure ...... 48

4.5 Conclusion ...... 50

Chapter 5: Oxygen and Modulus in the Ti-Ta alloy system ...... 56

5.1 Introduction ...... 56

5.2 Procedure ...... 57

5.3 Results ...... 58

5.4 Discussion and Conclusions ...... 59

Chapter 6: Rare Earth additions to Ti-Ta alloy system ...... 63

6.1 Introduction ...... 63

6.2 Procedures ...... 65

6.3 Results ...... 65

6.4 Discussion and Conclusions ...... 66

Chapter 7: Ti-Ta-Al ternary alloy system ...... 70

xiii

7.1 Introduction ...... 70

7.2 Procedures ...... 71

7.3 Results ...... 72

7.4 Discussion and Conclusions ...... 72

Chapter 8: Ti-Ta-Zr alloy system ...... 75

8.1 Introduction ...... 75

8.2 Procedures ...... 75

8.3 Results ...... 76

8.4 Discussion and Conclusions ...... 76

Chapter 9: Beta à ω and meta-stability in Ti–Ta alloy system ...... 80

9.1 Introduction ...... 80

9.2 Procedures ...... 84

9.3 Results ...... 85

9.4 Discussion and Conclusions ...... 87

10. Summary Discussion ...... 99

10.1 Lattice Parameters and Atomic Positions ...... 99

10.2 Effects of Oxygen on Modulus and Atomic Positions ...... 100

10.3 Effect of Symmetry on Modulus ...... 102

10.4 Effects on Modulus and Atomic Positions ...... 103

xiv

10.5 Final Thoughts and Future Work ...... 105

References ...... 116

List of Tables

Table 1 Ultrasonic elastic modulus of arc melt specimens...... 53

Table 2 Modulus measurements, Poisson's ratio and density of Ti–xTa alloys containing low, CP grade levels and high levels of oxygen. Modulus measurements made by ultrasonic techniques...... 61

Table 3 Modulus, lattice parameters and structure of Ti-xTa +Y specimens...... 67

Table 4 Modulus measurements, Poisson’s ratio, density of binary Ti x-Ta alloys containing approximately 2wt% Al...... 73

xvi

List of Figures

Figure 1 Schematic representations of Hexagonal Close Packed (HCP) α and Body

Centered Cubic (BCC) β...... 6

Figure 2 Schematic representation of a pseudo-binary β isomorphous phase diagram[10]

...... 12

Figure 3 Manufacturer’s image of commercial LENS™ system (a), schematic of LENS™ system (b), deposition in process (c), and schematic of laser deposit build and parameters

(d)...... 25

Figure 4 MTS NanoIndenter XP (left), Berkovich indenter (center), example loading and unloading plot (right)...... 29

Figure 5 Instrumented indentation of LENS™ gradient specimen in as deposited, solution treated and water quenched condition, oxidized by solution treating in air and air cooling and water quenching...... 35

Figure 6 SEM BSE image near deposit base (a), un-melted particles and layer of high Ta concentration (b), high Ta concentration layer (c) highest Ta concentration layer (d), with

BSE intensity profile of LENS™ (e). Arrows indicate the location of the image with respect to the gradient specimen...... 36

xvii

Figure 7 Elastic modulus measured by instrumented indentation of LENS™ specimen in the solution treated and water quenched condition. Measurement performed by

UNICAMP in Brazil...... 37

Figure 8 Bright Field TEM image of martensite in high Ta concentration near indent location #95 (a), simulation of α" martensite in [100]α’’ zone axis diffraction pattern, where color is used to indicate the predicted relative intensity (b), zone axis diffraction pattern (c) ...... 38

Figure 9 Ultrasonic inspection equipment, Fluke Scopemeter 190-204 (a), Olympus pulse processor (b), transducers for longitudinal (c) and shear wave (d) velocity measurements.

...... 42

Figure 10 Rigaku SmartLab X-Ray Diffractometer...... 43

Figure 11 TEM images and diffraction patterns from Ti–25Ta, Ti–30Ta, Ti–35Ta, Ti–

45Ta, Ti–55Ta recorded near [110]β direction...... 51

Figure 12 TEM images and [111]β diffraction patterns of Ti–65 wt.% (a) and Ti–75Ta wt.% (b)...... 51

Figure 13 XRD patterns from Ti–xTa wt.% (x=15, 20, 30, 35, 40, 45, 55, 65, 75)...... 52

Figure 14 Young's modulus of Ti–Ta binary alloys and published results from Zhou et al.

[41, 51] and Fedotov et at. [9]...... 53

Figure 15 Lattice parameters, sonic velocity, Poison's ratio, density and unit cell volume for high purity Ti–Ta binary alloys...... 54

Figure 18 Ti–xTa +1.6 wt.% O Elastic modulus values plot with results from chapter 4 and Zhou et al. [41]...... 62

xviii

Figure 19 Ti–xTa+1.6O wt.% Lattice Parameters, sonic velocities, density and unit cell volumes...... 62

Figure 20 Ti-xTa +Y Alloy series elastic modulus...... 67

Figure 21 Ti-xTa+Y specimen XRD patterns and simulations...... 68

Figure 22 Ti-xTa will the addition of Y for gettering dissolved oxygen...... 69

Figure 23 Lattice parameters, ultrasonic velocity, Poisson’s ratio, density and unit cell volume for alloys of Ti-x Ta + 2Al wt.%...... 73

Figure 24 Modulus of Ti-xTa +2Al wt.% alloy series. Plotted with values from Zhou et al. [41]...... 74

Figure 25 Ti-25Ta-2Al (a) and Ti-55Ta-2Al (b) bright field TEM images...... 74

Figure 26 Lattice parameters, ultrasonic velocity, Poisson’s ratio, density and unit cell volume for alloys of Ti–xTa + 5Zr wt.%...... 78

Figure 27 Modulus values of Ti–xTa+5Zr alloys. Plot against results of from chapter 4 and modulus values for the Ti–xTa binary system by Zhou et al. [41]...... 78

Figure 28 TEM image from Ti–35Ta–5Zr wt.% and [0001] zone axis pattern...... 79

Figure 29 Combined nudged elastic band and density functional theory study of β → (α

'', ω) in Ti–Ta alloys published by Chakraborty et al. [29]...... 89

Figure 30 XRD patterns from Ti–65Ta wt.% as cast, aged at 255°C for 12hr, 350°C for

4hrs, 450°C for 2hrs and 450°C for 24hrs...... 90

Figure 31 TEM images and diffraction patterns from Ti–65Ta in as cast (a), aged for

255°C for 12hr (b), 350°C at 4hr. (c), 450°C for 2hr. (d) and 450°C for 24hrs...... 91

xix

Figure 32 Extraction of intensity profiles from [100] zone axis electron diffraction patterns (left). Intensity profiles extracted from ⅓ [110], ½ [110] and ⅔ [110] reflections, indicating the relative diffracted intensity of α'' and ω from each aging step (right)...... 92

Figure 33 STEM HAADF image of α precipitates in Ti–65Ta wt.% (a), selected area for

EDS mapping, red line indicates location of line profile (b), Map of Ta Kα intensity (c), map of Ti Kα intensity (d), line profile across α precipitate where wt.% were calculated from Kα intensities...... 93

Figure 34 Dark Field TEM image of Ti-α precipitates in Ti–65Ta wt.% aged at 450°C for

24 hours. Inset is image of [113]β zone axis. Circle indicates the location of objective aperture...... 94

Figure 35 Plots of Ti–Ta binary alloy lattice parameters, lattice parameters reported by

Dobromyslov et al. [38] and relaxed lattice parameters used in combined nudge elastic band and density functional theory calculations [29]...... 95

Figure 36 Ultra high resolution STEM HAADF image along [110]β (left), Fourier

Transform of real space image (right)...... 96

Figure 37 Ultra high resolution STEM HAADF image along [111]β. Dark area in upper right corresponds to α, were mottled contrast surrounding contains ω and β...... 97

Figure 38 Dark field TEM image of TNZT alloy with a Young’s modulus of 50 GPa.

Inset image of [110]β diffraction pattern showing weak ω and α’’ orthorhombic reflection intensity...... 98

Figure 39 Atomic position y of [002]α" in Ti–xTa binary alloys. Value of 0.166 corresponds to the position necessary to create hexagonal symmetry and 0.25 corresponds to the location for BCC...... 107

Figure 40 Experimentally determined atomic position y of [002]α" in Ti–xTa, Ti–xNb [1] binary alloys and predicted positions for Ti–xTa and Ti–xNb published by Li et al. [63].

...... 108

Figure 41 XRD patterns of Ti–40Ta wt.% and Ti–40Ta+1.6O wt.% and simulations of

Cmcm crystal structure where lattice constants are similar and y position is 0.2 and 0.166 respectively...... 109

Figure 42 XRD patterns for Ti–25Ta, Ti–25Ta+Y and CP Ti, where the Y containing sample shows very close agreement with CP Ti. Slight loss of symmetry HCP symmetry is apparent in Ti–25Ta where [132] and neighboring [221] may show splitting...... 110

Figure 43 XRD patterns of CP Ti, Ti–15Ta, Ti–20Ta and Ti–25Ta wt.%. Reduced intensity and broadening of peaks such as [113]α’’ indicate symmetry may be less Cmcm and less than P63/mmc...... 111

Figure 44 TEM images and diffraction patterns from Ti–65Ta (a), Ti–55Ta+1.6O (b), Ti–

55Ta+2Al (c), Ti–55Ta (d), Ti–31Nb–6Zr–4Ta (e). TNZT shows a low modulus of 50

GPa and weak ω and orthorhombic reflections. Weak ω streaking was also observed in

Ti–65Ta...... 112

Figure 45 HRSTEM image of Ti–65Ta wt.% in aged condition 450°C for 24hrs [110]β (a) and [111]β (b). Particles of ω may exist as 3-5nm precipitates, though well-formed particles were not observed...... 113

xxi

Figure 46 HRSTEM image of [100]β, FFT filtered to remove background contrast (a) inset, schematic of transverse shuffle motion as described by Bönisch et al. [50], FFT kernel filter of [110]β HRSTEM image. Intensity derived from frequencies corresponding to 1/2[110]β (b). Region related to high fit, orthorhombic domain (b) inset.

...... 114

Figure 47 XRD patterns for Ti–xTa+Al (x=25,35,45,55) where the [200]α'' reflection being parallel to [110]β is not present...... 115

xxii

Chapter 1: Introduction and Objectives

Titanium alloys present an optimal balance of corrosion resistance, high strength and low density for the use as a biocompatible implant material. Yet there are still opportunities to improve the performance of these materials for implant materials. For instance the effects of “stress shielding” can cause a loss of bone surrounding implants such as femoral hip implants. Where the disparity between the elastic modulus of pure titanium, 110 GPa, and human cortical bone, 10-40GPa leads to the mechanical load being carried by the implant rather than the surrounding bone, leading to bone atrophy

[2]. It has been proposed that implant materials with a lower elastic modulus would reduce the stress shielding effect and improve longer term implant performance [3, 4].

The creation of a new titanium alloys for the purposes of medical implantation is a growing field of research in materials engineering. Implant materials require the use of elements that are non-toxic and also promote a high level of biocompatibility. Alloys containing Titanium, Niobium, Tantalum and Zirconium such as TNZT[5, 6] have shown to have high levels of bio compatibility and offer low elastic modulus in the range of 50 to 60 GPa [7]. Yet the origin of the low modulus and the role of oxygen in the alloy are not completely understood.

Variation in elastic modulus has been reported in many simple binary Titanium systems, Such as Ti–Nb, Ti–Mo, and Ti–Cr, Ti–Mn, Ti–Fe, Ti–Co in work published by

Fedotov et al. [8]. The lowest elastic modulus in each system is reported in the quenched condition near the composition boundary between α’ and α’’ martensite, as described in chapter 2, and again at the boundary between α” and ß. Similarly, literature published by

Fedotov et.al [9] indicated that the low modulus of approximately 60 GPa was observed in the Ti–Ta binary system near 28 wt. % Ta and a second similar minima is present near

68 Ta wt.%. The 28 Ta wt.% composition was shown to be near the boundary of α’ and

α’’ martensite stability in the solution treated and water quenched samples. The second minima appears near the boundary between α’’ and retained β.

Titanium alloys, such as TNZT, offer a low modulus where they contain either orthorhombic α’’ martensite or metastable BCC ß crystal structures [6]. This presents the possibility there may exist low modulus alloys of lower alloy content near the boundary of α’ and α’’ martensite stability. Alloys of lower alloy content could benefit implant materials by having lower density as well as lower formulation costs.

The purpose of the following work is to investigate lowering the elastic modulus of Ti-Ta binary alloys, by increasing the stabilization of α’ with the addition of oxygen, a potent α stabilizer. Aluminum and oxygen are both potent α stabilizers. However, aluminum is known to be a substitutional element in titanium while oxygen behaves as an interstitial element [10]. The benefit of using oxygen as a stabilizer of α’ is twofold.

First, oxygen would represent higher biocompatibility than aluminum. Secondly, oxygen is a common impurity in titanium and low oxygen grades of titanium are costly.

Furthermore, advances in 3D printing and near-net shape processing could potentially benefit advanced implant manufacturing. But these processing routes rely on powder processing and are subject to oxygen contamination.

The following chapters describe a series of experiments to explore the stability of

α’ and α’’ martensite, ß and the related effects on crystal structure and elastic modulus.

Chapter 3 describes the production of a Laser Engineered Net Shape LENS™ deposition of Ti–Ta composition gradient and the application of the combinatorial approach to characterize the range of compositions. Chapter 4 investigates the elastic modulus and structure of specimens produced of high purity charge materials spanning the compositional range of 15 to 75 wt.% Ta, where an effort is put forth to reproduce the modulus and structure of alloys described in literature. The alloys are then characterized and compared with recent work published in the Ti–Nb binary system. In chapter 5 the effect of increasing oxygen content on the modulus and crystal structure are compared to high purity and commercially pure grades of titanium. The shift in the stability of α’, α’’ and elastic modulus with the addition of oxygen was observed. Chapter 6 describes the production of alloys containing small amounts of yttrium, a rare earth element. The addition of rare earth elements are used to react with any dissolved interstitial elements such as C, O and N. Thus high purity alloys combined with small amounts of yttrium were used to produce specimens of very low interstitial content. Experiments in Chapter

7 describe where a modest amount of Al is added to Ti-Ta binary alloys to observe the effect on the stability of α’ and α’’ martensite and resulting changes in crystal structure.

Similarly, the exploration of the effect of zirconium in Ti-Ta alloys is presented in

3 chapter 8. Finally, chapter 9 involves the precipitation of ω in Ti-Ta alloys.

Observations regarding the precipitation of ω in quenched as well as aged conditions were made.

Chapter 2: Background and Critical Literature Review

2.1 Titanium Metallurgy

Titanium is a transition metal that is lustrous with a silvery appearance. Pure titanium exists in two allotropic forms, where above 882°C ß, the body centered cubic

(BCC) phase, having space group symmetry !"3!, is stable. At temperatures below

882°C, the α phase with the hexagonal close packed (HCP) crystal structure, with space group !6!/!!", is stable. Titanium may also exist in two common metastable phases,

α’ and α’’, where α’ hexagonal, similar to α but not at an equilibrium composition and α’’ is orthorhombic with Cmcm symmetry. Schematic representations of α and ß are shown in Figure 1. Elements such as C, O, N, Al, Sn increase the temperature at which α transforms to β, termed the β transus temperature. Elements that increase the transformation temperature of the HCP phase to BCC are called α stabilizers. The elements Mo, V, Nb, and Ta are isomorphous ß stabilizers, meaning that high concentrations of these elements will maintain a BCC structure. Mn, Fe, Cr, Co, Cu and

H are eutectoid type ß stabilizers that will ultimately form intermetallic compounds [10,

11]. Thus, alloys of titanium may use one or more elements in combination to alter the stability of the α and ß phases to obtain desirable balance of properties, such as strength, modulus and corrosion resistance. For instance, stabilizing the β phase lowers the β transus temperature, thus improving the rolling and workability. Reduced processing temperatures are only one factor making β and near β alloys attractive for component fabrication. High hardenability, high strength and fatigue strength are all benefits of β and near β alloys [10, 12, 13]. The added benefit of lower working temperatures and high hardenability must be balanced against the higher formulation costs and density associated with the high β stability.

Figure 1 Schematic representations of Hexagonal Close Packed (HCP) α and Body

Centered Cubic (BCC) β.

Titanium alloys can have strengths comparable to steels while having roughly half the density. Likewise Titanium alloys, while having a higher density than aluminum alloys, often have much higher strength and stiffness. The higher stiffness combined with high strength can offer components with comparable dimensions to steel, but smaller and more rigid than aluminum components [14]. Thus, titanium is an ideal choice for many applications where lightweight, strong and durable components are needed as in the case of aerospace and medical applications.

Medical implants require high performance components that are lightweight, durable and corrosion resistant. These requirements can be met with new classes of titanium alloys that are being specifically designed for biocompatibility. These applications require, in addition to high strength and corrosion resistance, a low elastic modulus to reduce stress shielding effects [2]. A key component of biocompatibility is elimination of alloying elements that may be toxic such as Al and V as found in the ubiquitous aerospace titanium alloy Ti-6Al-4V. Thus, contemporary efforts seek to create new alloys specifically for biocompatible applications.

Structural applications for titanium, especially for components that experience elevated temperature, are typically used in an aged condition where the alloy is treated to produce near equilibrium ratios of α and β phases. This aging ensures that the alloy will remain stable during service and the properties will not degrade over time. However, low modulus titanium alloys such as TNZT are used in metastable β condition as α’’ martensitic conditions which also exhibit lower strength.

Titanium alloys are rarely used for structural applications in a solution treated and quenched state. The resulting metastable β or α’/ α’’martensite phases are relatively low strength and must be hardened by aging treatments that allow the martensite to decompose in to an α + β structure closer to equilibrium. Alloys that are high in β stabilizing elements and retain β on quenching are generally too soft for structural applications. On quenching, titanium can form several phases such as α’, α’’ or ωathermal.

Martensites typically form on sufficiently fast cooing from a temperature above the β transus. At low concentrations of ß stabilizers the martensite will maintain a hexagonal symmetry, nearly indistinguishable from α, and is denoted as α’. The concentration of β stabilizers can increase to a level where the martensite can lose hexagonal symmetry in favor of an orthorhombic structure [15]. The orthorhombic martensite structure is denoted as α’’ with a space group of Cmcm. These findings are well documented and supported by various sources [15-17].

Other phases often reported in titanium include ωisothermal and β’. Though β’ is generally only reported in older literature, prior to the acceptance of ω as a hardening mechanism. The hexagonal phase ω is metastable and can be observed in β alloys and α

+ β alloys with sufficiently high concentrations of β stabilizing elements. Hickman et al.

[18] describes three conditions where omega is expected to form. The first is where ω forms on cooling from temperatures above the β transus. The composition must be sufficiently rich in β stabilizing elements where the formation of martensite is suppressed yet metastable β is not fully retained. This method of forming athermal ω exists only over a narrow range of compositions. The second mode of ω formation, isothermal,

8 occurs by aging metastable β at temperatures generally below 500°C. The third mode of

ω formation according to Hickman [18] is by deformation at room temperature. The paper continues to cite various sources where ω had either occurred or volume fraction increased as a result of mechanical deformation. Williams et al. [15] later describe their work characterizing ω by electron diffraction, though they make no mention of deformation induced ω. They noted that early work on ω relied on X-Ray diffraction and hardness measurements to detect the presence of ω. Likewise, Banerjee et al. [17] describe two modes to form ω, quenching known as ωathermal and aging which forms

ωisothermal. However, they also chronicle a body of work related to the pressure induced α

→ ω transformation. This mode of ω formation should be noted but is not addressed in the current effort.

Duerig et al. [19], attempted to clarify differences and describe similarities between

ωisothermal and ωisothermal. The stated motivation relates to resolving differences between materials scientists, which attempt to study ω, and engineers who view ω as a source of embrittlement and “search for ways to avoid it”.

The state of these new biocompatible alloys on quenching are often α” or β.

Considering the state of alloy on quenching is a common basis for the classification of all titanium alloys. There have been various different definitions and attempts to create distinct classifications. The following section describes the common system classifications of Titanium alloys.

2.1.1 Titanium Alloy Classifications

Titanium alloys are generally separated into three main categories, α, α+ β and β.

Alloys that contain α stabilizing elements and transform to α on annealing at a temperature below the β -transus are called α alloys. It is worthwhile to note commercially pure (CP) grade titanium is often alloyed with very small amounts of Fe, where the very limited solubility of Fe in the α phase leads to the formation of very small

β precipitates that inhibit grain coarsening during processing. Thus, the small amount of

Fe stabilizes a very limited amount of the β phase. The small amounts of β act to maintain a fine grain size during processing by pinning grain boundaries, preventing grain coarsening and thus improve mechanical properties. Thus most grades of CP-Ti would be considered α alloys, yet will contain very low volume fractions of β when annealed below the β transus temperature.

The definition of a β alloy has been accepted as an alloy where the concentration of

β stabilizing elements is such that on rapid cooling to room temperature the formation of martensite is suppressed and metastable β retained [10, 12]. Williams et al. [10] provides a subtle, yet useful distinction, beyond the distinction provided by Boyer et al. [12], that concentrations of β stabilizers where 100% β phase would be stable at room temperature are not commonly used in commercial applications due to their low strength. It is further discussed that alloys used commercially are only metastable at room temperature and that metastable β alloys are most commonly referred to as simply ß alloys or near ß alloys in practice.

The range of alloys that provide the composition that exist between α and β, such as the well-known Ti - 6Al - 4V are called α+ß alloys. The α+ β alloys will readily transform to α’ or α’’ martensite when cooled sufficiently fast from the single-phase β region. Moiseev et al. [20] rather clearly makes a distinction between alloys VT14,

VT16 and VT15 where VT14 will form α’, VT16 forms α’’ while VT15 retains metastable ß on cooling. The paper does not make a distinction between β and “near β” or metastable β alloys. Later Moiseev [16] wrote in greater detail describing conditions defining: α alloys, pseudo α alloys, α + β alloys of the martensitic type, alloys of the transition class, pseudo β alloys, and β alloys. These multiple classifications generally coincide with descriptions given in [17]. McClintick et al. [21], in 1955 noted in that there had existed “considerable confusion resulting from the nomenclature applied to titanium alloys”. They define, α alloys as those that contain α + β regardless of heat treatment. Alloys that form α’ on quenching were called martensitic alloys while alloys that form metastable β on quenching were called metastable β alloys.

Figure 2 Schematic representation of a pseudo-binary β isomorphous phase diagram[10]

Each β stabilizing element has an effectiveness in terms of stabilizing β.

Molybdenum equivalency is an attempt to quantify each element’s ability to stabilize martensite on quenching to room temperature in binary alloys with respect to Mo. The

qualitative equivalent ß stability in a multicomponent is as follows: [Mo]eq = [Mo] +

0.2[Ta] + 0.28[Nb] + 0.4[W] + 0.67[V] + 1.25[Cr] + 1.24[Ni] + 1.7[Mn] + 1.7[Co] +

2.5[Fe]. The [Mo]eq relationship as reported in [10] is often cited in literature as a quantitative measure to compare the relative β stability of Ti alloys. The relationship is

12 useful for predicting the constitution of an alloy composition, despite the obvious utility

“it should be used with caution”, as the author notes [10].

2.1.2 Intermetallic compounds

The prior sections described several classifications of alloys and crystal structures that are random solid solutions with either substitutional or interstitial elements.

However alloying titanium with a large concentration of aluminum can for instance produce an ordered intermetallic compound such as Ti3Al. Intermetallic compounds can have very high strengths as well as high melting points, making these compositions attractive for high temperature lightweight structural applications. Titanium can contain several ordered phases such as α2, β2, γ and O as well as various inter-metallic compounds[17, 22]. According to [10, 22], α2 has the chemical formula Ti3Al. The compound γ is closely related to α2 with the formula TiAl. Bendersky et al. [22] studied these compounds and their crystal symmetry as they exist as BCC at high temperatures and hexagonal at lower temperature with varying degrees of ordering. These alloys have been extensively studied for their high temperature strength and lightweight. However they typically exhibit low ductility and poor fracture toughness. Thus their use in structural applications is very limited [10]. Alloys of the type α2 containing Ti3Al may also contain concentrations of β stabilizing elements. Thus alloys of Ti2AlNb can produced where the Nb can be interchanged for Ti in the ordered structure. The bcc

13 structure can be stabilized, though according to [10] it is always ordered and thus should be referred to as β2. It is further noted in [10] that literature contains inconsistent terminology regarding this phase. For example in [10], [17] and [15] the decomposition of the β phase is described as β → β1 + β 2, where the β phase separates into a rich and lean solute content, βrich and βlean. The β1 is what was referred to earlier as β’ while in this case β2, the βrich phase, refers the BCC β and not the ordered β2 (B2 crystal structure) structure associated with the α2 phase. Subtle differences in the nomenclature should be noted as reviews of industrial literature such as [21] from the 1950’s had not fully adopted ω as a hardening mechanism and instead referred to “β ’ hardening”.

2.1.3 Eutectoid Decomposition

Unlike elements such as V, Mo, Nb and Ta, which ultimately have stable equilibrium BCC structures at very high alloy concentrations. Elements such as Cr, Mn, and Fe had been identified early as eutectoid forming elements [21]. Meaning at elevated temperatures and extended times the precipitation of a eutectoid intermetallic compound may be observed. A comprehensive study was made by Franti et al. [23], where a survey of 10 eutectoid elements was conducted. The purpose of the study was to identify the morphology of the eutectoid reaction products pearlite, which is lamellar, or bainite, which does not form a lamellar structure. The eutectoid forming elements investigated by [23] were Bi, Co, Cr, Cu, Fe, Mn, Ni, Pb, Pd and Pt. The authors

14 observed a pearlitic reaction forming in Ti-Fe only after 100’s or 1000’s of hours aging at temperatures 50°C and 100°C below the eutectoid temperature. The compound was found to form “more frequently” at impinged proeutectoid α plates [23]. Franti et al. also noted that the eutectoid reaction in Fe and Mn was only observed after times that were 3 to 5 orders of magnitude longer than required in other eutectoid forming elements.

Transmission Electron Microscopy (TEM) was used to make the observation of the compound as the authors noted that TEM inspection was “prohibitively tedious for TTT diagram determination”. Lee and Aaronson [29] observed the eutectoid decomposition in

Ti - 5.2Fe by TEM inspection and commented that the product was bainitic. However, the main focus of their work related to the study of the Ti-Cr system.

Mössbauer spectroscopy was used in studying the decomposition of β in Ti - 7.1%

Fe by Stupel et al. [24]. The procedural section describes the experiment in detail. The experiments reveal the formation of the β-Ti (Fe) solid solution and the formation ω. The paper makes no mention to the observation of Ti–Fe intermetallic and no metallography is provided. This is in stark contrast to research published by [25] where Ti–Fe is given

“as an extreme example of strongly ordering alloys”. It was further concluded, “the hypothetical disordered phase of Ti-Fe would be a bcc substitutional disordered alloy.”

The conclusion was reached “that it may be of interest to make a randomized Ti-Fe solid solution by irradiation with neutrons or electrons.” Kale et al. [26], reported when diffusion couples of pure Ti and Fe or even incremental couples are annealed that no intermetallic components are observed. This finding by Kale [26] was reiterated by

Banerjee in [17]. Ferrante et al. [27], referenced diffusional data from Kale et al. [26].

However, in their work diffusion bonding Ti–6Al–4V to 316L stainless steel they reported observing TiFe, α-Ti, Fe2Ti and β-Ti. They only characterized fracture surfaces of the phases and identified them by composition energy dispersive spectroscopy and x- ray diffraction. They cited prior work where transmission electron microscopy had been performed and stated that the “results were essentially confirmed”.

R. Ray et al. [28] published a paper regarding ordering in Ti–Fe alloys. The work was rather thorough as splat quenching produced 12 compositions ranging from 10 to 50 atom percent Fe. The resulting specimens were analyzed by X-Ray diffraction to measure lattice parameters, and plotted these against the Fe concentration. The splat- quenched sample lattice parameters were compared with, and reasonably matched the lattice parameters published on bulk samples solution-treated and quenched from the β phase field. It was concluded that the distinction between β -Ti solid solution and TiFe

(Ti) was not distinguishable by X-ray diffraction “due to the lack of superstructure lines”.

They conclude that due to the continuous nature of the change in lattice parameter as the concentration of Fe increases that there must be a continuous phase change between the random disordered β -Ti (Fe) and the partially ordered TiFe (Ti) non-stoichiometric solution. The Ti-Ta system however, is not known to easily form intermetallic compounds. Calculations performed by Chakraborty et al. [29] found that ordering energies for Ti-Ta were very low.

2.2 Processing

2.2.1 Primary Titanium Production

Titanium is produced from ores rutile TiO2 and Ilmenite FeTiO3. The ores are processed to eliminate Fe and produce a synthetic or pure TiO2. The pure TiO2 is then chlorinated to produce TiCl4 and distilled for further purification. The TiCl4 is reduced to form metallic titanium. The final product is a sponge-like material of pure titanium.

According to [27], during most refining processes, the iron found in Ilmenite FeTiO3 is lost as an oxide or hydroxide that is discarded at significant cost. The loss of iron is regarded as a “significant financial and environmental issue”.

2.2.2 Ingot Metallurgy

Titanium for modern structural applications has traditionally been processed from ore to a sponge product then vacuum re-melted to form an ingot. Early production of titanium was produced by the decomposition of titanium iodide vapor, forming very high purity titanium crystals. Sponge processed titanium often contains low levels of impurities such as iron and oxygen.

Boyer et al. [12] have stated existing problems regarding the use of β alloys in general, most notably melt related segregation. Williams et al. [19] describe eutectoid

17 forming elements such as Fe, C, Mn, Ni and Cu as depressing the melting temperature, leading to elemental segregation on ingot solidification. The resulting local enrichment of β stabilizing elements has detrimental effects on properties such as fatigue. This type of defect is often called “beta fleck” due to its characteristic appearance on visual inspection [30]. Mitchell et al. [31] describes how the solidification defects can form as two types of defects, macro segregation and micro segregation. Macro segregation forms during solidification “under high thermal gradients” while micro segregation can be found where the solidification interface is dendritic. Mitchell describes that Fe will tend to segregate to the top and center of an ingot while interestingly he comments that oxygen will tend to be of lower concentration in center and “head regions”. Mitchell [31] also describes newer, more modern ingot processing techniques for controlling segregation. These new techniques have presumably made alloys such as Ti-10-2-3 (Ti -

10V - 2Fe - 3Al), Ti–LCB (Ti - 4.5Fe - 6.8Mo - 1.5Al) and Ti-62S (Ti - 6Al - 2Fe) commercially feasible. The problems related to the processing of Fe and other eutectoid elements are pervasive in the literature and are not under dispute. While processing alloys containing Mo, Ta and Nb can be difficult due to solute segregation, high melting points and slow rates of homogenization.

2.2.3 Powder Metallurgy

Powder processing of metals and alloys can have many benefits. Powder metallurgy (PM) can save significant cost in that components can often be made near net shape, saving energy and cost related to producing fabricated components. In contrast, the high cost of alloyed Ti powders limits their potential use. The process for producing

Ti powder from a sponge source can and often does introduce atmospheric contamination

[32]. Considerable cost can be added to produce pre-alloyed powder without increasing atmospheric contamination. The most common method of producing pre-alloyed titanium powder involves a plasma arc rotating electrode process (PREP). There are currently many new processes under development for the production of low cost titanium powders. One such new process, the Armstrong/International Titanium Powder (ITP), promises to produce alloy powder in an efficient and continuous process [33]. The MHR process for example uses calcium hydride to directly form alloy powders from a mixture of oxides [33].

Powder processing can eliminate the problems associated with segregation associated with the use of β eutectoid elements such as Fe. Thus the powder processing industry may not be limited by the same compositional constraints the current titanium industry has with regards to alloy chemistry.

2.2.4 Effects of Oxygen and Other Impurities

Common impurities in titanium are oxygen, nitrogen, carbon, hydrogen and iron.

Regarding the current work at hand oxygen is not an impurity but a deliberate alloy addition used to stabilize the α’ phase. A residual content of oxygen and Fe remain as contaminants in primary Ti production and will be discussed later. Hydrogen is known to stabilize the β phase and reactions between hydrogen and titanium are known to be reversible [34]. Thus hydrogen is not generally a troublesome impurity and will no be discussed further.

Jaffee and Ogden at Battelle [35] performed extensive work regarding carbon, oxygen and nitrogen as interstitial in titanium. The work however, was performed in the

1950’s when many modern characterization techniques were not yet available. They made extensive use of thermodynamic data and ternary phase equilibrium however; the experimental work was limited to mechanical test data and trends related to hardness measurements. A brief summary of their findings is as follows. Oxygen, nitrogen and carbon are α phase stabilizers in β titanium and stabilizers of β in liquid titanium. Carbon has limited solubility in titanium and tends to form TiC in the form of “stringers” in ingot processes. The stringers, a lenticular grouping of casting defects, are typically broken up and re-dispersed through mechanical working. Oxygen and nitrogen have much higher solubility in titanium and like carbon tend to expand the lattice parameter and distort the

20 c by a ratio of hexagonal α titanium. Carbon, nitrogen and oxygen can also contribute significantly to the strength of titanium. From the ternary phase equilibrium they concluded that oxygen and nitrogen should segregate strongly from the β phase in favor of the α phase.

The diffusion constants for oxygen, nitrogen and carbon in α and β titanium were also reported [35]. The data shows that the diffusivity constant Do for oxygen is 2 orders of magnitude larger than Do nitrogen. Convincing data was reported relating the increasing interstitial content with increasing hardness in iodide titanium. The increase in hardness was countered by the interstitials detrimental effect on toughness. Liu et al.

[36] studied the effect of oxygen on the mechanical properties of only two alloys, Ti -

6Al - 2V and Ti - 2Al - 16V. They noted that oxygen caused the α alloy Ti - 6Al - 2V to become very brittle while the β alloy Ti - 2Al - 16V lost ductility with increasing oxygen but not to the extent of embrittlement. The α and β samples were reportedly heat treated for as much as 50 to 100 hours. Thus it is plausible and reasonable the level of contamination may have increased during the experimental heat treatment.

The argument for ordering of the oxygen in the α phase was made on the basis of observation of apparent dislocation pair formation. While seemingly contradictory statements were also made regarding the lack of dislocation pairs when diffraction evidence indicated the presence of Ti3Al. Yamaguchi et al. [37] presented convincing evidence for the existence of oxygen ordering at much higher concentrations of oxygen,

Ti2O and Ti3O. The samples were also given long heat treatments of one week at 400°C to promote ordering. In this case contamination is not expected to be a factor given the

21 high concentration of oxygen in the starting material. Neutron and electron diffraction experiments by [37] show oxygen ordering on octahedral sites in the HCP structure on alternating interstitial planes while the metal atoms are slightly displaced in the c direction away from the oxygen atom.

It is stated that in [10] that oxygen strongly promotes the formation of α2 (Ti3Al).

The α2 will positively effect the yield stress but due to a moving dislocation’s ability to cut the particles “intense” planar slip leads to easy crack formation and reduced tensile ductility. Williams et al. [10] makes no mention of oxygen ordering, though similarly cites the ability of oxygen to change the deformation behavior from wavy to planar slip.

Chapter 3: Low Modulus Ti-Ta Binary Alloys; A Combinatorial Approach

3.1 Introduction

Titanium alloys, such as TNZT, offer a low elastic modulus by combining

Titanium, Niobium, Zirconium and Tantalum. Work published by Besse et al. [7] and

Satio et al. [5] have shown oxygen as a beneficial addition to low modulus ß or near ß alloys. TNZT, “Gum Metal” can exhibit “super” properties, high ductility, and high strength after extreme cold working [5]. Cold working and forming are contrary to the efforts to produce materials in near net shape by 3D printing techniques such as LENS™, where it is desired to create a net shape component with little or no post deposition processing. Simple binary alloys have been reported to also show low modulus behavior without the need for cold working. While the variation in elastic modulus has also been observed in many simple binary Titanium systems [8, 38]. These alloys are typically solution treated above the β transus, and water quenched. The compositions of these alloys produce a β or near β alloy, where the crystal structure on quenching after the heat treatment produces a α’’ or β crystal structure. However, the mechanism underlying the low elastic modulus is not fully understood. It is observed in literature that the elastic

23 modulus decreases while the content of tantalum is increased, producing two pronounced minima. The first minima in modulus coincides with a concentration where a boundary between α’ and α” is observed. The second minima in elastic modulus exists near the phase stability boundary between α” and β. Thus, it is hypothesized that lowering the elastic modulus of the first minima further could be realized by increasing the stabilization of α’ by the addition of oxygen, a potent α stabilizer. The benefit of oxygen is two fold. First oxygen is often regarded as an impurity and is costly to eliminate.

Powder metal having a high surface area to volume ratio can be prone to oxygen contamination. Applications such as 3D printing of custom implant components could benefit alloys where oxygen is a desirable addition. Secondly, oxygen could provide better biocompatibility over aluminum, another common α stabilizer used to strengthen titanium alloys [10]. Given that production of many specimens of various compositions can be time consuming and labor intensive it was proposed that a 3D printing technology such as the LENS™ be utilized to rapidly create a specimen of graded composition spanning a range of low Ti-Ta compositions. The mechanical properties and crystal structure would then be characterized by site-specific methods. The production and analysis of the specimen is as follows.

3.2.1 LENS™ Deposition Procedure

Alloy specimens from the Titanium Tantalum alloy system were produced using an

Optomec LENS™ laser deposition system. The LENS™ system is equipped with two powder feeders which enable the fabrication of compositionally graded specimens by delivering powdered metal to the active melt pool created by a 750 W Nd:YAG laser.

The powder feeders were filled with commercially pure Ti and an elemental blend of Ti and Ta powders, respectively. During deposition of the 2 cm tall cylindrical specimen, the Ti and Ta feed rates were varied such that the target composition ranged from Ti + 8 wt.% Ta to Ti + 30 wt.% Ta from bottom to top along the length of the build. Figure 3 shows schematically the LENS™ 3D laser deposition system.

Figure 3 Manufacturer’s image of commercial LENS™ system (a), schematic of LENS™ system (b), deposition in process (c), and schematic of laser deposit build and parameters (d).

The sample was produced as a right cylinder deposited on Grade 2 CP titanium plate stock. The deposit was cut from the substrate and sectioned longitudinally by wire electric discharge machining (EDM). The deposit produced four samples. The first was

25 characterized in an as-deposited condition. The second sample was briefly solution treated at 1000°C and water quenched. The third and fourth samples were held at

1000°C in air for 30 minuets then water quenched and furnace cooled respectively.

3.2.2 Instrumented Indentation

Instrumented indention is a mechanical testing technique where an indenter tip, typically diamond, is applied to a material with a high resolution actuator while the force is recorded by a high sensitivity load cell. The actuator recodes the indentation depth hc.

The high precision indenter tip is manufactured with a particular shape where the projected area of the indent is known for a given indent depth. The indenter tip used in the following experiment was a diamond Berkovitch type, as shown in Figure 4 where the shape factor ß was equal to 1.034. From the shape factor and indentation depth the projected area of the indent can be determined. Since during loading, the projected area of the indent is increasing continuously as the depth increases and the material deforms plastically and elastically. The slope of the load displacement curve is calculated on the initial portion of unloading. An example of the loading and unloading curves are shown in Figure 4. The hardness of a material is given by the depth and load applied over the projected area. The reduced modulus, Er can be measured as a function of applied load and projected area as shown in Equation 1.

Equation 1. Resolved modulus is a function of projected area and applied load.

The reduced modulus, observed from the indention experiment depends on the modulus of the specimen, Poisson’s ratio and the modulus of the indenter tip. Thus the modulus of the specimen can be determined by the following relationship shown below in Equation

Equation 2. From the resolved modulus the specimen modulus is related to the indenter modulus and Poisson’s ratio, however an estimation of the specimen Poisson’s ratio must be made.

Provided the modulus of the indenter tip is known, the Young’s modulus of the specimen can be determined by making an estimate of the Poisson's ratio. For instance 0.3 is typical for many metals.

Instrumented indention was used to measure the elastic modulus across the

27 compositional gradient. Thus, the variation in elastic modulus was observed as a function of position and thus composition. The elastic modulus of LENS™ gradient samples, as well as samples produced by vacuum arc melting, were measured by instrumented indentation. Sample were first sectioned and polished through 25µm SiC grinding papers, and final polished with 0.04µm colloidal silica on Buehler ChemoMet™ cloth. The sample testing was performed on an Agilent Technology nano-indenter XP,

Figure 4. The testing methods included modulus at depth and continuous stiffness method measurements. Continuous stiffness method measures the modulus of the specimen as a function of indentation depth by modulation of the tip depth at typically

45Hz. Thus from the instantaneous loading and unloading the modulus can be calculated. Average modulus values were calculated during each indent experiment by averaging the continuous stiffness values collected from indentation depth of 400 to

1200nm. Young’s modulus reported for the Indentation profile of the LENS™ produced samples was calculated from the average of the prior and post observations as a moving average (average of n-1, n, n+1). Instrumented indentation of the gradient LENS™ deposited material was also conducted by independent measurements made in collaboration with UNICAMP Universidade Estadual de Campinas in Brazil. Each of the

100 reported values of Young’s modulus spanning the composition gradient was made from an average of 100 indents made transverse to the compositional gradient.

Figure 4 MTS NanoIndenter XP (left), Berkovich indenter (center), example loading and unloading plot (right).

3.2.3 Scanning Electron Microscopy

Microstructures of specimens produced by LENS™ were imaged using an FEI XL

30, or Teneo™ Field Emission Gun scanning electron microscope (SEM). A solid-state backscatter electron (BSE) detector in combination with low, 2-10kV, accelerating voltages was used to collect images of the martensite structures. Typically, BSE signal intensity is approximately proportional to the square of the atomic number. Due to the lack of solute segregation in martensite typically only weak contrast may be observed due to electron channeling. Ta containing particles show bright contrast in BSE imaging due to their high atomic number with respect to Ti. The composition of specimens was also checked via SEM based Energy Dispersive Spectroscopy (EDS).

3.2.4 Focused Ion Beam / Scanning Electron Microscopy

Specimens for Transmission Electron Microscopy (TEM) were prepared in either

FEI Nova 600 or FEI Helios 600 DualBeam™ Focus Ion Beam / Scanning Electron

Microscope (FIB/SEM). Samples were first prepared by grinding and polishing through

1200 grit SiC carbide papers and then fine polished by 0.04 µm colloidal silica. The area of interest was protected by a layer of FIB deposited Pt prior to the trench milling operation. Trenching was completed at a current of 21nA and acceleration voltage of

30kV. The samples were removed from the bulk in-situ via OmniProbe AutoProbe™

200 micromanipulator and placed on a Cu OmniProbe ‘three post” TEM grid. FIB milled lamella were then thinned at a beam current of 3000pA to roughly 200nm in thickness.

Then final thinning was performed at 100-500pA with a final thickness of approximately

100nm. Areas of each foil were intentionally prepared such that were slightly thicker and thinner regions of interest. The variation in thickness is beneficial to facilitate practical

TEM characterization tasks. The navigation of Kikuchi space benefits from thicker samples where as thinner areas are preferred for imaging.

3.2.5 Transmission Electron Microscopy

Electron diffraction patterns were recorded on an FEI CM-200 TEM with LaB6 thermionic electron source carried out at an accelerating voltage of 200kV. Scanning

Transmission Electron microscopy was performed on an FEI Tecnai TF20 TEM using the

High Angle Annular Dark Field Detector (HAADF). HAADF imaging was performed where images deriving contrast from atomic number “z contrast” were necessary.

Compositional analysis was performed in an FEI Titan 60-300 TEM with ChemiSTEM™ high solid angle Energy Dispersive Spectrometer. EDS collection was performed at

300kV with a beam current of approximately 1nA.

3.3 Results

Instrumented indention was used to measure the elastic modulus across the

LENS™ prepared compositional gradient. The change in modulus was observed as a function of position and thus composition. LENS™ gradient specimens showed the expected trend where the elastic modulus would decrease with the increase in Ta concentration. Modulus measurements were made on samples in the four different conditions, i.e., as deposited, water quenched, oxidized / air cooled, and oxidized / water

31 quenched. All specimen conditions show similar values and trends in modulus given slight inter and intra layer compositional variation. Results of the modulus measurements, collected for the four processing conditions are shown in Figure 5. It was noted that the modulus of the gradient specimen did not decrease monotonically layer by layer. This is attributed to a variation in composition since layer-to-layer variations are expected from depositions based on blends of elemental powders. Small areas of un- melted tantalum powder particles were observed Figure 6 (a, b). This was especially true near the substrate, where the heat sink effect of the substrate and the lower alloy melting point contributed to the inclusion of un-melted Ta particles. Layers away from the substrate where the melting point of the alloy is higher produced layers of higher homogeneity Figure 6 (c, d). The layer-by-layer change in composition was investigated by quantification of the BSE intensity spanning the 17mm deposit. An image intensity profile, Figure 6 (e), was created spanning the LENS™ build where a continuous montage of images was stitched to form a single micrograph. Despite the variability in layer-to-layer composition, the expected trend in modulus was still observed, i.e., increasing concentration of Tantalum corresponded to a decrease in elastic modulus over the range of 8 wt.% to 28 wt.% Ta as verified by EDS. However the observed modulus of the LENS™ deposited sample was higher than expected when compared with the work published by Fedotov et al. [9]. The solution treated and water quenched specimen was sent to Universidade Estadual de Campinas (UNICAMP) for additional nanoindentation testing. Results of the instrumented indentation performed in Brazil are shown below in

Figure 7.

The small size of the LENS™ specimens prohibited the use of X-Ray Diffraction for the determination of lattice parameters. Characterization of the crystal structure of the

LENS™ deposited material was performed by site specific FIB / TEM electron diffraction. A site-specific TEM specimen was prepared by FIB near the location where the modulus was lowest and SEM EDS indicated a composition of Ti-28Ta wt.%. Figure

8 shows a TEM bright field image recorded near the [100] beam direction of α’’ martensite, where the [100] α’’ is parallel to [0001]α and [110]β. There exist a high density of fine scale martensite plates in the high Ta content section of the gradient deposit. The small lenticular martensite plates interpenetrate obscuring their boundaries making crystallographic analysis difficult. Electron diffraction patterns recorded of the LENS™ deposited specimens were compared with simulated patterns. Candidate lattice parameters and crystal structure for simulation were chosen from experimentally observed structures as will be discussed in chapter 4. The experimentally observed pattern matched well with simulated patterns of Cmcm crystal structure with lattice parameters a = 3.018Å, b = 5.016Å and c = 4.6963Å.

3.4 Discussion and Conclusions

Work published by Fedotov et al. [9] shows low modulus values for Ti-Ta alloys where the lowest observed moduli were approximately 60 GPa at concentrations of 28 and 66 wt.%. The work was based on high purity iodide titanium processed by electron beam melting [9]. Though the oxygen content was not reported, the materials and processing was likely to have produced alloy materials of very low oxygen content [39].

Fedotov et al. [8] reported the use of an “Elastomat device” used to measure the elastic modulus. Though e-mail correspondence with Victor Samarov, of Laboratory of New

Technology, Inc. Garden View CA, in May of 2015, it was learned that the Elastomat device is a similar technique to ultrasonic velocity measurements where electromagnetic excitation, rather than a piezoelectric transducer, is used to measure resonant frequencies of a cylindrical specimen. Matlakhova et al. [40] in 2005 also published data collected by “Elastomat equipment”. Thus to reproduce results as reported in literature it was proposed to produce high purity alloy ingots by vacuum arc melting, then perform modulus measurements by ultrasonic methods.

Figure 6 SEM BSE image near deposit base (a), un-melted particles and layer of high Ta concentration (b), high Ta concentration layer (c) highest Ta concentration layer (d), with BSE intensity profile of LENS™ (e). Arrows indicate the location of the image with respect to the gradient specimen.

Figure 7 Elastic modulus measured by instrumented indentation of LENS™ specimen in the solution treated and water quenched condition. Measurement performed by UNICAMP in Brazil.

Chapter 4: Structure and Properties of Low Modulus Ti-Ta alloys

4.1 Introduction

Chapter 3 describes the use of a combinatorial approach to create a specimen of graded composition. The specimen was produced by the Optomec LENS™ deposition system. The Young’s modulus was measured by small scale testing using instrumented indentation. The local microstructure was characterized by site specific TEM in conjunction with FIB prepared specimens. The resulting microstructure was analyzed along the gradient, thus as a function of composition. Modulus values from the LENS™ deposit were not in agreement with values published in literature. Fedotov et al. [9], however, reported using iodide titanium and electron beam melting for production of alloy that demonstrated the low modulus. Matlakhova et al. [40] stated specification for iodide titanium where the purity was 99.99%. Thus, it was proposed that an effort to reproduce the published results should be made, where specimens produced from high purity materials by vacuum arc melting would be characterized by bulk characterization techniques such as X-Ray Diffraction (XRD) and Ultrasonic inspection. The results of this effort are described below.

4.2 Experimental Procedures

4.2.1 Vacuum Arc Melting

Specimens of Ti-Ta binary alloys were produced by vacuum arc melting using a non-consumable tungsten electrode and water-cooled copper hearth. The chamber was evacuated, flushed with Ar and re-evacuated. Then the chamber was back filled with a partial atmosphere of Ar to facilitate melting. The charge material was melted on the surface of the water-cooled copper hearth to form ingots. The ingots were produced from high purity tantalum slug (Alfa Aesar 99.95%) and high purity titanium granules (Alfa

Aesar 99.99%). The ingots were melted, flipped and re-melted many times to produce a homogeneous specimen. Compositions of 15, 20, 22, 30, 40, 65 and 75 wt.% Ta were produced as 15 to 25g ingots.

Master alloy ingots were made of composition Ti 25, 35, 45 and 55 wt.%. Ta. The

100g master alloys were sectioned into four 25g pieces. These four samples were processed as sets, were further processed in the following conditions:

1) Briefly solution treated at 1280°C and water quenched.

2) Analyzed in an as melted condition.

3) Remelted with the addition of 1.5g Yttrium. (See Chapter 6)

4) Alloyed with approximately 2 wt.% Al and tested in an as-melted condition.

(See Chapter 7)

4.2.2 Ultrasonic Elastic Modulus Measurement

Samples were first sectioned and parallel surfaces for ultrasonic measurements were prepared by mechanical grinding with SiC abrasive on an Allied MultiPrep precision metallographic polishing system (MultiPrep). Elastic modulus was calculated from sonic velocity measurements made using an Olympus 5072PR pulse processor, and a Fluke 190-204 Scopemeter, longitudinal Olympus transducer (part no. V110), a transverse Olympus transducer (part no. V156) shown in Figure 9. Measurements were also performed and results confirmed with collaborations with UNICAMP Universidade

Estadual de Campinas.

Figure 9 Ultrasonic inspection equipment, Fluke Scopemeter 190-204 (a), Olympus pulse processor (b), transducers for longitudinal (c) and shear wave (d) velocity measurements.

4.2.3 X-Ray Diffraction

The samples produced by vacuum arc melting were characterized by XRD. XRD scans were performed on a Rigaku Smart Lab™ diffractometer with Cu Kα x-ray source

(λ =1.5406 nm) as shown in Figure 10. XRD scans were performed at a 0.02° step size and scanning rate of 1°/min with source accelerating voltage of 40kV and current of

44mA. The Bragg-Brentano 2θ scans were performed using parafocusing and utilized a

Kβ filter. Samples were first prepared by sectioning and parallel surfaces prepared by mechanical grinding with Allied MultiPrep precision sample metallographic polishing system. Grinding steps included SiC abrasives though 1200 grit and final polishing was

42 performed using 0.04 µm colloidal silica. Simulation of XRD patterns was performed using commercially available software packages Single Crystal, Crystal Maker and

Crystal Diffract. X-Ray diffraction patterns from candidate crystal structures were compared to experimental data to determine measurements of lattice parameters and atomic positions.

Figure 10 Rigaku SmartLab X-Ray Diffractometer.

4.2.4 Focus Ion Beam Specimen Preparation

Specimens for TEM characterization were prepared according the methods described in section 3.2.4. The top surface of the alloy ingots was fine polished and TEM foils were excised from near the center of each specimen.

4.2.5 Transmission Electron Diffraction

Transmission Electron Microscopy was used to characterize martensite specimens. Specimens were prepared in the DualBeam™ FIB as described in the prior section 3.2.4. Electron diffraction patterns were produced in an FEI CM 200 TEM with

LaB6 thermionic electron source operating at an accelerating voltage of 200kV. The patterns and images were recorded on a CCD camera rather than electron film. While film can provide a high dynamic range for imaging and diffraction, the use of CCD cameras can provide speed and instantaneous feed back to the operator. The use of the camera allows the operator quickly adjust exposures to capture diffraction patterns of weak and strong reflections quickly.

4.3 Results

The moduli of the arc-melted specimens were initially measured by instrumented indentation. The moduli of the Ti 25, 35, 45 and 55Ta wt.% specimens were found to be

95, 102, 108 and 109 GPa by instrumented indentation and 58, 75, 85 and 81 GPa by ultrasonic techniques. Modulus measurements were then made of all specimens by ultrasonic methods. The results from the vacuum arc melt alloy ingots of compositions

15, 20, 22, 25, 30, 35, 40, 45, 55, 65 and 75 wt.% Ta are summarized in Table 1. Values of modulus found in the current study by ultrasonic methods are shown concurrently with the results of Fedotov et al. and Zhou et al. [9, 41] in Figure 14. The results from all three studies show close agreement, where the minimum in modulus is observed in a very narrow composition range and the magnitude of the elastic modulus can change rapidly over a few weight percent.

Characterization of the crystal structure of the each composition was performed by XRD. The lattice parameters determined for each composition are shown in Figure

15. From the lattice parameters observed the crystal structure was determined to be either α’, α’’ or β. The phase of each alloy composition is noted in Table 1. TEM specimens were prepared from each specimen and the results from 25, 35, 45 and 55Ta wt.% are shown below in Figure 11.

XRD measurements of lattice parameters agree and confirm observed changes in lattice parameter as reported in prior studies such as Fedotov et al. [9], Bywater et al.

[42], and others [43]. Recent literature published by Bönisch et al. [1] has sought to determine the atomic positions in Ti-xNb martensite materials. Bendersky et al. [22] proposed the application of an “intersection” space group to study the transformation of

BCC to HCP in Ti-Al-Nb alloys. The low symmetry Cmcm space group, having the

Wyckoff positions 4(c), with various levels of distortion can produce the higher symmetry BCC and HCP space groups. The BCC having the !"3! and the HCP with space group P63/mmc can thus be related by Cmcm symmetry where the atomic positions

of the [002] α’’ are shifted along, [110] <110>, the glide plane in the Y-axis from y = 0.166 to 0.25. The “y” position represents, atomic positions (0, y, 1/4), (0, -y, 3/4), (1/2, 1/2+y,

1/4), (1/2,1-y+1/2, 3/4) and defines the relative location of the atomic positions in the

[002] α’’ plane. A position of y = 0.166 represents the location of the atoms in a structure equivalent to P63/mmc. Conversely the position where y = 0.25 creates the subgroup symmetry of !"3! [22].

The modulus values were compared with measurements made in material produced from a prior study at OSU, Ti alloy TNZT (Ti–31Nb–6Zr–4Ta). The modulus for the TNZT alloy was found to be 50.5 GPa by ultrasonic methods, while the lowest modulus observed in the current study was at Ti–25 Ta wt.%, the value was still higher at

58 GPa. Zhou et al. [41] reported Ti–25 Ta wt.% with a modulus of 64 GPa was of the highest biocompatibility. However Fedotov et al. [9] show the lowest modulus near Ti-

28 wt.% with a modulus of 56 GPa. Given the close agreement between the three studies the lowest modulus is likely near Ti-28Ta wt.%.

4.4 Conclusions and Discussion

4.4.1 Observations related to Modulus

The origin of discrepancies observed between the measured modulus by instrumented indentation and ultrasonic methods was not determined definitively.

However, results published by Majumdar et al. [44] indicate that modulus measurements from instrumented indentation and ultrasonic methods were comparable, although ultrasonic methods were likely more accurate. It should be noted that the calculation of

Young’s modulus by instrumented indentation requires an estimate of the Poisson’s ratio

[45] and assumes the material obeys Hooke’s Law. From the experimentally observed

“reduced modulus”, Young’s modulus of the specimen is computed. The reduced modulus is related to the indenter tip modulus and Poisson’s ratio combined and the specimen Modulus and Poisson’s ratio, the latter being specified by the user [45].

Although, as stated above, the source of the discrepancy is not completely understood, it may be due to local plastic deformation as the diamond indenter tip is applied to the specimen, plastically forming the indent. Furthermore, martensitic materials can exhibit non-linear stress versus strain behavior and thus do not necessarily obey Hooke’s Law.

Supporting this argument, literature published by Kumar et al. [46] imply techniques that do not involve plastic strains are best suited for the characterization of materials with high Poisson’s ratio and low modulus. In this study it was seen that, in agreement with

47 literature, Poisson’s ratio tends to increase while Young’s modulus decreases. Indeed the minimum of the modulus was seen to correspond to the maximum in the values of

Poisson’s ratio as shown in Figure 15.

4.4.2 Observations related to structure

The increase in Ta content up to 25 Ta wt.% show a α’ martensite structure consisting of P63/mmc hexagonal symmetry. Increases in Ta content beyond 25 wt.% but less than 75 Ta wt.% were observed to have less symmetry than P63/mmc and are α’’ martensite with Cmcm space group symmetry. The alloy of Ti-75Ta wt.% was observed to have a BCC β crystal structure with a space group !"3!. The microstructure of each alloy up to 55Ta wt.% was seen to consist of fine plates of martensite with a high degree of twinning. As the alloy content increased toward 55Ta wt.% the degree of twinning increased and size of the martensite decreased. The trends observed in microstructure agree with those published by Bywater et al. [42] and Dobromyslov et al.[38]. Alloy compositions beyond 55 Ta wt.%, containing 65Ta wt.% and 75Ta wt.% show few or no twins respectively. XRD patterns as shown in Figure 13 from the Ti-65 wt.% show a very near β structure where the broad peaks and peak shape indicate the crystal structure has less than !"3! space group symmetry. The XRD pattern from Ti-75Ta wt.% indicated that the structure is indeed BCC.

The structure of martensite in the Ti-xNb system was first reported by Brown et al.

48 and Bagariatskii et al. [47, 48] in the 1950’s. Follow-on work of several other groups such as Dobromyslov et al. [38], Banuathy et al. [43] and Pathak et al. [49] have since expanded on this work. Recent work published by Bönish et al. [1] and Banumathy et al.

[43] on the Ti-Nb system show and concur that changes in lattice parameter as well as

shifts in atomic positions in the [002]α’’‘as a function of Nb concentration. The studies show a systematic shift as a function of Ta composition. The lattice parameters observed in the current study show Ti-Ta binary system as measured by XRD show close agreement with the Ti-Nb when compared as atomic percentages. Furthermore, through simulation of the x-ray diffraction patterns of the Cmcm structure with various magnitudes of the “y” parameter, it can be seen in Figure 16 that the [002] α’‘ shifts in Ti-

Ta appear to be similar. In terms of the atomic percentages the α’ and α” phase boundary of both Ti-Ta and Ti-Nb are very similar. The boundaries of the α’ and α” phase stability also agree with the ranges published in literature [10]. The boundary between α’ and α” can be difficult to determine as was noted by both Zhou et al. [41] and Bönish et al. [12] in the Ti-Nb system. Bönish et al. reported that x-ray diffraction peaks broadened by small crystal size and strain led to the misidentification of the co-existence of α’ and α”

[50]. It can be difficult to discriminate α’ compared with α”, because the change in lattice parameters is only slight. The boundary between the two phases in Ti-Ta binary appears to be abrupt in agreement with the finding in Ti-Nb published in [1].

4.5 Conclusion

Characterization of martensite in Ti-Ta binary alloys presents several challenges.

First, changes in crystal structure must be determined by careful measurement of lattice parameters. This can be readily accomplished by XRD on small ingot specimens.

Second, accurate measurements of Young’s modulus require techniques that do not induce plastic strains in the materials [46]. The small size of the LENS™ gradient specimen prevented the measurement of the Elastic Modulus from ultrasonic velocities.

Thus the elastic modulus was measured by instrumented indentation for the LENS™ gradient only. Measurements of elastic modulus made by both instrumented indentation of arc-melted specimens. It was found the two techniques do not agree well with the measurements made by ultrasonic methods. Values of modulus from the ultrasonic methods do agree well with the values published by both Fedotov et al. [9] and Zhou et al. [51]. Each group used various different dynamic resonance techniques to measure modulus. In each case the modulus measurements were made by, either torsional experiments in the Elastomat device or beam bending techniques respectively. Both techniques can be used to produce measurements of Young’s modulus E, and Shear modulus G, from which the value of Poisson’s ratio can be calculated.

Figure 11 TEM images and diffraction patterns from Ti–25Ta, Ti–30Ta, Ti–35Ta, Ti– 45Ta, Ti–55Ta recorded near [110]β direction.

Figure 12 TEM images and [111]β diffraction patterns of Ti–65 wt.% (a) and Ti–75Ta wt.% (b).

Figure 13 XRD patterns from Ti–xTa wt.% (x=15, 20, 30, 35, 40, 45, 55, 65, 75).

Figure 14 Young's modulus of Ti–Ta binary alloys and published results from Zhou et al. [41, 51] and Fedotov et at. [9].

Phase Alloy ρ (g/cc) ν E (GPa) (XRD) Ti 15 Ta (wt%) 5.04 0.359 94 α’ Ti 20 Ta 5.32 0.377 86 α’ Ti 22 Ta 5.36 0.387 74 α’ Ti 25 Ta 5.44 0.405 58 α’ Ti 30 Ta 5.80 0.402 68 α’’ Ti 35 Ta 5.99 0.380 75 α’’ Ti 40 Ta 6.36 0.379 82 α’’ Ti 45 Ta 6.66 0.372 85 α’’ Ti 55 Ta 7.61 0.388 80 α’’ Ti 65 Ta 8.62 0.417 63 α’’ Ti 75 Ta 10.28 0.392 87 β TNZT 5.53 0.436 51 near β Table 1 Ultrasonic elastic modulus of arc melt specimens.

Figure 15 Lattice parameters, sonic velocity, Poison's ratio, density and unit cell volume for high purity Ti–Ta binary alloys.

Figure 16 Plot of the observed "y" parameter in Ti–xTa alloys.

Figure 17 Plot of the lattice parameters of the Ti-Ta binary system and lattice parameters published by Bönish et al. [1].

Chapter 5: Oxygen and Modulus in the Ti-Ta alloy system

5.1 Introduction

Alloys produced by LENS™ processing as described previously in chapter 4 show a trend in Young’s modulus, where increasing the Ta content from 0 to 30wt% Ta decreased the modulus from that of Ti at approximately 110 GPa to approximately 80

GPa (note: modulus of pure Ta is 230 GPa). In the literature however, the minimum modulus reported is approximately 60 GPa. The minimum value of the observed Young’s

Modulus in the LENS™ experiments is higher than values often reported in literature

[12-14]. It is then hypothesized that the discrepancy may be due to either slight change in interstitial content, where the primary concern would be oxygen or in measurement of modulus by nanoindentation. Publications such as [44, 49] indicate that ultrasonic methods and nanoindentation may be comparable, however techniques that do not involve plastic strains yield more accurate measurements of Young’s modulus.

Alloys produced by Fedotov et al. [9] were produced from iodide grade titanium, where iodide grade titanium is generally of very high purity and of low interstitial content. For example McClintick et al. [21] cited work performed at Armour Research

Foundation indicating that β in Ti – 30 Mo may decompose during testing at elevated temperatures, while “Ti – 30% molybdenum produced from iodide grade titanium may be a true stable β alloy”. Thus, the decomposition of β in Ti–30Mo was being attributed to impurity content, and more specifically oxygen content. Cotton et al. [39] have commented in the literature that iodide grades of titanium are of “sufficient purity” for studies of phase transformation in Ti alloys. An experiment was then proposed to observe the effect of oxygen content on the observed Young’s modulus and phase stability of alloys of the Ti–Ta alloy system.

5.2 Procedure

In the molten state, titanium can readily dissolve small rutile pellets. Thus, by adding small amounts of rutile to high purity charge materials in the hearth of the vacuum arc furnace, the oxygen content can be varied. Samples were produced by vacuum arc melting with a non-consumable tungsten electrode as described in chapter 4, where the chamber was evacuated, flushed with Ar of 99.998% purity and evacuated. The chamber was then back filled with half of one atmosphere of Ar. The charge material was melted on a water-cooled copper hearth to form 15 to 25g ingots. The ingots were melted, flipped and re-melted multiple times as necessary to produce a homogeneous ingot.

Compositions of Ti–xTa (x=20, 30, 40, 55) were produced as high purity ingots. These samples were divided in half and re-melted with a measured addition of rutile pellets to

57 produce specimens with approximately 1.6 wt.% oxygen. Regarding the use of CP Grade

2 materials in LENS™ deposits, a comparison was made of one ingot produced from

Grade 2 CP Titanium and high purity tantalum slug with a Ta content of 30 wt.%.

Alloy ingots were mechanically polished on a Struers MultiPrep parallel polishing system. Young’s modulus of specimens was measured by ultrasonic techniques and the crystal structure and lattice parameters were measured by XRD. Chemical composition analysis of the alloy ingots was obtained using a LECO Inductively Coupled Plasma

(ICP) in Brazil in collaboration with Universidade Estadual de Campinas (UNICAMP).

5.3 Results

The ICP results show that the oxygen content of the material produced from high purity 99.99% Ti granules contained only 0.08 wt.% oxygen. Specimens produced from commercial purity Grade 2 Ti contained 0.23 wt.% oxygen. Materials made from CP materials contain a level of oxygen consistent with the Grade 2 ASTM material specification, where Grade 2 would have less than 0.25 wt.% O (Note: it is worthwhile to mention the grade also allows for up to 0.30 wt.% Fe). The specimens produced with the high purity material with added rutile contained 1.68 wt% Oxygen. The modulus of the individual samples was measured by ultrasonic pulse echo method at OSU and

UNICAMP.

The elastic modulus was measured at OSU by ultrasonic methods and the results

58 summarized in Table 2. Ultrasonic methods show increasing the oxygen content from

0.08 wt.% to 0.23 wt.% and then up to 1.6 wt.% oxygen increased the modulus from 68

GPa, 72 GPa to 90 GPa respectively. The modulus observed in the high purity material agreed well with the findings of Fedotov et al. [9] where the modulus was observed was in the range of 58 GPa to 85 GPa.

X-ray diffraction was used to characterize the crystal structure and lattice parameter of the alloys produced by vacuum arc melting. The diffraction results were compared with diffraction pattern simulations and predictions of the atomic positions. Results of the lattice parameter measurements are shown in Figure 19. The results agree well with the findings of Fedotov et al. [9] where the 25 Ta specimen appears to have α’ martensitic structure and 35, 45 and 55 Ta show an α’’ martensite structure. The alloys of, Ti–20Ta, Ti–30, Ti–40Ta wt.% containing additional oxygen all show the α’ crystal structure. This is consistent with oxygen acting to promote the α’ structure over α’’ and demonstrates that the use of an α stabilizer can promote α’ however the modulus was observed to increase due to the additional oxygen.

5.4 Discussion and Conclusions

Results of this effort were observed to be consistent with the work performed by

Fedotov et al. [8] where the starting material was also of high purity. X-ray diffraction shows as expected, oxygen likely stabilizes the α’ phase. Increasing the Ta content in the

59 high purity material stabilizes the α’’ phase. XRD results show broad peaks that indicate the presence of small crystallite size and/or some degree of strain in the crystal lattice.

Little if any structure change was observed as oxygen was increased from 0.23 and 1.68 wt.%.

The moduli of the samples were measured by ultrasonic methods at OSU and

UNICAMP. Results from both groups were in close agreement where by ultrasonic methods the CP grade Ti-30Ta wt.% was measured at 72GPa and 70GPa at OSU and

UNICAMP respectively. The sample with 1.6 wt.% oxygen was observed in both instances at 90GPa. The results of the ultrasonic modulus tests performed at OSU are shown in Table 2. Ultrasonic methods show comparable results from both groups, where the increased oxygen content increased the observed modulus from 72 GPa to 90 GPa.

The low oxygen specimen produced a modulus of 68 GPa agreeing nearly exactly with

[11] and inline with the results published in [9] thus, the ability of the molten titanium to dissolve the rutile and create an alloy of desired oxygen content has been demonstrated.

Second, the ability of oxygen to increase the elastic modulus in the arc-melted alloy ingots is also clearly evident.

It was observed that oxygen indeed stabilizes α’ over α’’ as was evidenced in alloys of Ti–30Ta that contain 0.23 and 1.68 wt % oxygen show a α’ martensite structure, where the Ti–30Ta specimen of high purity and low oxygen content exhibits a α” structure.

Since the materials with oxygen content greater than 0.08 wt.% show a large increase in the observed modulus. The 0.23 wt.% oxygen content would likely increase the modulus to a level higher than that of alloys of low interstitial content. Thus binary alloys created

60 from CP Grade 2 materials are not likely to produce the lowest Young’s modulus.

From the results in the current study it was hypothesized that further reductions in oxygen could reduce the Young’s modulus in the Ti–Ta binary system. ICP tests of materials produced in the current study from starting materials of high purity did not indicate there was excess oxygen contamination from processing. Thus to realize the lowest interstitial content, specimens would be re-melt with an oxygen scavenger such as yttrium where any dissolved oxygen would react to form an in soluble rare earth oxide.

The preparation of such samples is described below.

Phase Alloy ρ (g/cc) ν E (GPa) (XRD) Ti 20 Ta + 1.6 O 5.28 0.346 107 α’ Ti 30 Ta (0.08 O) 5.80 0.402 68 α’’ Ti 30 Ta + 0.3 O 6.05 0.403 72 α’ Ti 30 Ta + 1.6 O 5.63 0.376 90 α’ Ti 40 Ta + 1.6 O 6.27 0.373 92 α’ Ti 55 Ta + 1.6 O 7.52 0.376 100 α’’ Table 2 Modulus measurements, Poisson's ratio and density of Ti–xTa alloys containing low, CP grade levels and high levels of oxygen. Modulus measurements made by ultrasonic techniques.

Figure 18 Ti–xTa +1.6 wt.% O Elastic modulus values plot with results from chapter 4 and Zhou et al. [41].

Figure 19 Ti–xTa+1.6O wt.% Lattice Parameters, sonic velocities, density and unit cell volumes.

Chapter 6: Rare Earth additions to Ti-Ta alloy system

6.1 Introduction

Samples produced from high purity staring materials as discussed in chapter 4 showed low modulus behavior in agreement with results published in literature [9, 41,

51]. However, considering the strong relationship between oxygen content and Young’s modulus observed in chapter 5, it is logical that a reduction in oxygen could lead to a possible further reduction of Young’s modulus. The removal of oxygen can be realized in Ti alloys by the addition of an aggressive oxygen scavenger such as yttrium, erbium dysprosium or gadolinium [52]. Adding a small amount of, in this case, yttrium to the charge materials, the rare earth metal would react with dissolved oxygen and reduce the content of interstitials and lower the Young’s modulus further. Metallic yttrium and its oxide are insoluble in the solid state and should not positively affect the modulus of the specimens.

Khorev [52] published results that indicate that Rare Earth Oxides (REO) reduce grain size and can promote higher strength in structural alloys such as VT15, VT23 and

VT38. Research published by Khorev [52] also notes three factors that may benefit efforts to produce specimens by 3D printing techniques such as the LENS. First, as stated previously, the negative effects of oxygen in Ti alloys can be reduced by the addition of rare-earth elements. Second, REO particles can act as nucleation sites during solidification and refine microstructure. Third, “Rare-earth metals improve weld seams” and the REO can further stabilize the passive oxide layer to improve high temperature oxidation resistance.

Specifically regarding the elastic modulus of titanium, two additional sources are found in literature. First, Whittsett et al. [53] observed that small additions of Y and Er may slightly lower the β transus temperature of Ti-6Al-4V, being consistent with the removal of oxygen, an α stabilizer. Secondly, it was observed that the grain size of the alloys containing REO was reduced and continued to have reduced grain sizes on subsequent annealing. This would be consistent with reduction of dissolved oxygen in the alloys and oxide particles residing on, and pinning grain boundaries. Secondly, results published by Hieda et al. [54] show that there was not an observed increase in modulus where rare earth oxide dispersions were investigated for their possible use in oxide dispersion strengthen β-type alloys. Hieda et al. [54] observed that small amount of Y, 0.5 wt.% made small changes in Young’s modulus. However, the changes were attributed to possible change in the texture of test specimens.

6.2 Procedures

The yttrium content was selected to react with dissolved oxygen in the alloys to produce the lowest effective interstitial content. With the high purity starting materials it was assumed the concentration of oxygen was not higher than 0.23 wt.% and thus a modest amount of yttrium, 0.5g, was necessary to react with dissolved oxygen in the molten ingots.

Samples were prepared from high purity Ti master alloys as described in earlier in

Chapter 4. Ingots were re-melted with Y by vacuum arc melting producing alloy ingots of Ti–xTa+REO (x = 25, 35, 45, 55). Specimens were parallel polished on an Allied

Multiprep instrument through 1 micron SiC. Young’s modulus of specimens was measured by ultrasonic methods as described in section 4.2.2. XRD characterization was performed as described in section 4.2.3.

6.3 Results

The results of the modulus measurements are shown in Table 3 and plotted against values published by Zhou et al. [41] in Figure 20. The results show that the difference in the observed modulus between the high purity alloys specimens and the samples containing REO are negligible, with the exception of Ti-25Ta wt.% where the modulus

65 was slightly increased. The resulting analysis by XRD is shown in the following Figure

21. The shift of the [110] along the <110> can be clearly seen from the comparison of simulated XRD patterns to the experimentally observed patterns. As the value of “y” increases it can bee seen by the (131) peak gains in intensity vs. the reduction in intensity of (113). Simultaneously, as the degree of orthorhombicity increases the (112) and (022) reflections split and the ratio of the shifts from (112) dominating at low values of y and

(022) being dominant at higher values of y.

6.4 Discussion and Conclusions

From the ultrasonic measurements of modulus it is seen that the changes in modulus from the addition of Y were negligible. This is further supported in the observation of the structure as seen in XRD where the lattice parameters and structure are similar. However, it was noted that the XRD patterns from simulation match well with the experimental observations. The dispersion of REO may act to refine the grain size of these alloys and as the experimental patterns are nearer to that of intensity ratios expected from powder diffraction conditions. The better than expected agreement of the simulated patterns supports this observation.

Phase Alloy ρ (g/cc) ν E (GPa) (XRD) Ti 25 Ta + Y 5.62 0.384 71 α’ Ti 35 Ta + Y 6.04 0.383 75 α’’ Ti 45 Ta + Y 6.74 0.366 87 α’’ Ti 55 Ta + Y 7.46 0.391 80 α’’ Table 3 Modulus, lattice parameters and structure of Ti-xTa +Y specimens.

Figure 20 Ti-xTa +Y Alloy series elastic modulus.

Figure 21 Ti-xTa+Y specimen XRD patterns and simulations.

Figure 22 Ti-xTa will the addition of Y for gettering dissolved oxygen.

Chapter 7: Ti-Ta-Al ternary alloy system

7.1 Introduction

The experiments described in chapters 4 and 5 show that in the binary Ti-Ta system oxygen effectively stabilized the α’ martensite structure at concentrations below

35 Ta wt.% and caused an increase in the observed Young’s modulus. Since the initial effect of increasing Ta content in the Ti binary is a reduction in modulus while α’ is stable. Then with further increases in Ta content the modulus reached a minimum and increased as α” became stable. It was hypothesized that increasing stability of α’ to higher concentrations of Ta could yield a lower minimum modulus. Since, oxygen is known as an interstitial α stabilizer [10] and the modulus increased. There is still a possibility that increased α’ stability could lower the modulus through the use of another

α stabilizer such as Al.

Compared to oxygen, aluminum will also stabilize α but substitutes for Ti atoms in the lattice, rather the locating in interstices. Similar studies on the alloy effect of Al in the Ti-Nb system have been performed in by Matlakhova et al. [40]. The author noted

70 that the modulus in general increased with increasing Al content from 2 wt.% to 15 wt.%, but decreased with increasing Nb content. Where the lower concentrations of Al were effective, concentrations of 3 wt.% and 5 wt.% promoted the orthorhombic distortion in

Ti-Nb alloys and at sufficiently high concentrations stabilized the bcc β phase. It is proposed that an experiment similar to that described in chapter 5 be performed, where a limited amount of Al is added to effect the stability of α’ and possibly increase the stability of α’ over α” and produce a lower minimum modulus.

7.2 Procedures

Specimens were produced from sections of the master alloy materials Ti-xTa

(x=25, 35, 45, 55 wt.%) as described in chapter 4. This Al series was subsequently expanded with the addition of ingots of Ti-15Ta +2Al and 20 Ta +2Al wt.%., produced as 20g and 15g ingots respectively. Specimens were polished through 1 micron SiC, by parallel polishing on a Struers Multiprep polishing system. Young’s modulus measurements were obtained by ultrasonic methods and the crystal structure and lattice parameters measured by XRD as described in prior chapters.

7.3 Results

The results of the modulus measurements are shown in Figure 24 and plotted against values published by Zhou et al. [41] for comparison. The values of elastic modulus, density and structure are list in Table 4. Measurement of the lattice parameters, density, Poisson’s ratio and unit cell volume are shown in Figure 26. In a similar trend as seen in other alloys described in prior chapters, the Poisson’s ratio is high in the Ti-xTa-

2Al when the modulus is observed to be a minimum. Microstructure investigation of the

Ti-xTa-2Al alloys by TEM shows that the degree of twinning and size scale of the microstructure is different from the simple binary alloys shown in chapter 4 Figure 11.

Very fine scale lath martensite was observed in the Ti-55Ta-2Al wt.% specimen.

7.4 Discussion and Conclusions

From the results shown in Figure 24, it is seen that the lower concentrations of Ta alloys have higher modulus than the simple binary. However, it was seen at higher concentrations, Ti -55 Ta -2Al wt.%, that the modulus was quite low at 53GPa. The results from Matlakhova et al. [40] in Ti-Nb +5wt.% Al show a similar trend observed in

Ti-Ta at only 2 Al wt.%.

Phase Alloy ρ (g/cc) ν E (GPa) (XRD) Ti 15 Ta + 2 Al 5.02 0.333 118 α’ Ti 20 Ta +2 Al 5.17 0.356 97 α’ Ti 25 Ta + 2 Al 5.42 0.371 94 α’ Ti 35 Ta + 2 Al 6.03 0.375 90 α’’ Ti 45 Ta + 2 Al 6.46 0.387 83 α’’ Ti 55 Ta + 2 Al 6.84 0.415 53 α’’

Table 4 Modulus measurements, Poisson’s ratio, density of binary Ti x-Ta alloys containing approximately 2wt% Al.

Figure 23 Lattice parameters, ultrasonic velocity, Poisson’s ratio, density and unit cell volume for alloys of Ti-x Ta + 2Al wt.%..

Figure 24 Modulus of Ti-xTa +2Al wt.% alloy series. Plotted with values from Zhou et al. [41].

Figure 25 Ti-25Ta-2Al (a) and Ti-55Ta-2Al (b) bright field TEM images.

Chapter 8: Ti-Ta-Zr alloy system

8.1 Introduction

Alloys such at TNZT show low modulus values and often contain relatively high contents of oxygen. Work published by Besse et al. [7] shows that TNZT alloys Ti–

23Nb–0.7Ta–2Zr and Ti–23Nb–0.7Ta–2Zr–1.2O exhibit Young’s Modulus values of

55GPa and 60 GPa respectively. The results given in chapter 3 show that the behaviors of the Ti -Ta and Ti -Nb system are indeed similar. It was proposed that the effect of small additions of Zr to the Ti-Ta binary system be investigated. Thus a series of Ti–Ta alloys containing 5 wt.% Zr were produced and analyzed in a similar manner to work performed in chapter 7.

8.2 Procedures

Specimens were produced from high purity materials, Alfa Aesar Ti starting materials 99.99% and Ta 99.95%, Zr 99.9+%. Alloys produced were Ti–xTa (x=15, 20,

25, 35, 55 wt.%) by arc-melting techniques as described in chapter 4. This Zr series produced as small buttons ingots ranging in size from 9 to 22g. Specimens were polished through 1 micron SiC, by parallel polishing on a Struers Multiprep polishing system.

Young’s modulus measurements were obtained by ultrasonic methods and the crystal structure and lattice parameters measured by XRD as described in chapters 4.

8.3 Results

The results of the modulus measurements are shown in Figure 27. The lattice parameters, Poisson’s ratio, density and unit cell volumes are shown in Figure 26. A slight shift can be seen toward α’’ stability in the alloys containing 5 Zr wt.% as compared to the high purity Ti–Ta alloy specimens. TEM image shown in Figure 28 show a very fine scale martensite structure.

8.4 Discussion and Conclusions

The results of the modulus measurements are similar to the values observed in the

Ti–Ta binary alloys. Although, there seems to be a slight shift toward the increase in α’

76 stability, a decrease in the modulus was not observed. However, there may still exist a small window of composition space at Ti–30Ta–5Zr or Ti–40Ta–5Zr where a lower modulus may be obtainable. However, there appears to be evidence in any case that Zr at low concentrations will not cause large changes in modulus or crystal structure as compared to the Ti–Ta binary system.

There exists a wide range of compositions in the Ti–Nb–Zr–Ta compositional space that produce low moduli as a partial list published by Mohammed et al. [6] and others [4, 55]. However, most alloys are of the β-type, which suggests there may exist an additional class of alloy where the lowest modulus is obtained more near the boundary of

α’ and α’’ martensite stability. Futhermore, while there may be benefits of greater specific strength and modulus as shown in the current and prior chapters high alloy contents also increase the alloy density, negating the benefits of titanium as a light weight alloy.

Figure 26 Lattice parameters, ultrasonic velocity, Poisson’s ratio, density and unit cell volume for alloys of Ti–xTa + 5Zr wt.%..

Figure 27 Modulus values of Ti–xTa+5Zr alloys. Plot against results of from chapter 4 and modulus values for the Ti–xTa binary system by Zhou et al. [41].

Figure 28 TEM image from Ti–35Ta–5Zr wt.% and [0001] zone axis pattern.

Chapter 9: Beta à ω and meta-stability in Ti–Ta alloy system

9.1 Introduction

The preceding chapters have discussed, primarily, the desire to produce an alloy of sufficiently low modulus as to reduce the stress shielding effect in implant applications

[3]. This presents several challenges, as the implied design limitations require that a high degree of biocompatibility is maintained and potentially toxic elements are excluded.

However, the low modulus behavior is one of three main research interests that have driven the study of Ti–Ta, Ti–Nb and similar alloys over the past 60 years. Early efforts sought to produce high damping materials for structural applications. James [56] wrote a review published in 1969 specifically discussing the uses and technical aspects of high damping materials, termed “hidemets”. Where metals with high damping capability could reduce noise, vibration and fatigue in components. Compared with other vibration reduction strategies, such as increasing mass to improve stiffness is not often advantageous in aerospace applications that favor lighter weight solutions.

With regard to the techniques applied in early studies, Fedotov et al. [8] utilized the “Elastomat device” used applied torsional oscillations to a cylindrical specimen. It is interesting to note James [56] also mentions specifically that high damping materials can limit the amplitude of torsional vibrations. Bagariatskii [48] utilized microhardness measurements and both groups performed characterization by XRD. Where is it was observed by XRD in [48] alloys such as Ti–Mo, Ti–V and Ti–W would form ω in the as quenched condition. From the increases in the observed hardness and the precipitation of

ω on aging at 400°C for 1hr, it was concluded that Ti–Ta must also form ω on quenching.

Fedotov et al.[9] observed in Ti–Mo and Ti–V alloys of an intermediate alloy content where the rise in Young’s modulus and shear modulus corresponded with the formation of ω observed by XRD. The author then concluded that the similar rise in modulus in the

Ti-Ta 28-50 wt.% range must also be related to the formation of small amounts of ω as well, though it was not observed by XRD.

The third research interest for alloys such as Ti-Ta, Ti-Nb and Ti-Ni is related to the unique deformation behavior, such as the shape memory and super elastic effect [57-

61]. The super elastic properties for instance are seen in the commercially available Ti-

50 Ni at.% alloy Ni where stress induced martensitic transformation allows the material to deform to a higher degree that is typical in other metal alloys. The shape memory effect is driven by the reversible transformation between two crystal structures, α’’ and β, as in the case of Ti–Nb and Ti-Ta[57, 59]. Ti–Ta and Ti–Nb show promise in forming higher temperature shape memory alloys that may outperform the NiTi where the Ms temperature is only 100°C, thus limiting the useful service temperature to 100°C or

81 less[59]. Again, efforts seek to avoid the formation of ω as it is undesirable in shape memory alloys. The formation of ω in a shape memory alloys degrade the effective reversible transformation and can ultimately inhibit the formation martensite completely.

In the context of alloys with low elastic modulus, the findings of Bagariatskii et al. [48] attributed the rise in hardness in the range of 40 to 50 Ta wt.%, where α’’ is stable, to the existence of ω. This range represents a slight discrepancy between the results in [48] and [9] where the reported range of α’’ stability was reported between 28 and 65Ta wt.%. The latter being in agreement with the finding reported above in chapter

4. This may be attributed to the use of Ti powder rather than iodide titanium as used in

[9]. Nonetheless, literature also lacks evidence of ω precipitation on the Ti-Ta binary system in an as quenched condition. The precipitation of ω has been reported on annealing at relatively low temperatures for long periods of time, in the range of 24 to 60 hours[60]. The current study as presented in prior chapters, ω was not seen in the alloys in the quenched condition by XRD and only weak ω reflections were observed in the alloy of Ti-65Ta wt.% by TEM by electron diffraction. Thus there is a discrepancy between early work where ω was assumed to form but was not observable by XRD and more current work such as [42] where ω was not observed by TEM in compositions up to

22Ta at.% ( ~52wt.%) by Dobromoslov et al. [38]

Recent work published by Chakraborty et al. [29] utilized density functional theory calculations and nudged elastic band calculations to determine minimum energy paths between β → α’’, and β → ω in Ti–Ta binary alloys. The calculations were performed at compositions of Ti–xTa (x = 0, 25, 33, 50, 75, 100 at.%). With respect to

82 the current experimental efforts, these compositions are approximately 0, 56, 65, 79, 92 and 100 wt.%. The calculations utilized “optimized” lattice parameters and a y parameter for the construction of unit cells for simulation proposes. These values for the unit cell deviate slightly from the experimentally determined values from chapter 4. Plots of the calculations given in [29] are shown in Figure 29. From these calculations it was shown in [29] that at the composition of Ti–65Ta wt.% at 0°K, the transformation pathway from

β to α’’ was just favorable and near the boundary between β and α’’ stability, Ti-33Ta at.% or approximately Ti-65Ta wt.%. The similar calculation was performed for the transformation between β and ω. The result of this calculation was such that the composition was just slightly favorable and there exist only a slight energy barrier to the transformation progression.

References cited in [29] included results published by Buenconsejo et al. [60] where, after heat treatments of 1 hour at 300°C it was shown that ω reflections were present, but clear images of the ω were not provided. Other researchers have reported ω on much longer aging times. For instance, observations made by Zhou et al.[62] in Ti–50 wt.% Ta show the precipitation of ω in on annealing quenched specimens at 450°C for

86.4ks (24 hr.). On annealing for longer times 24hrs and up to 6 days (518.4ks) ω and α were reported to be present.

Cotton et al. [39] report that Ti-Ta alloys can form ω on room temperature annealing over the period of several months. The current author can find no other reports of room temperature precipitation are mentioned elsewhere in literature. The long term precipitation of ω will not be addressed in the current study. Similar precipitation was

83 also reported in heat treatments of 400°C for 25 hours consistent with the findings of

Zhou. Thus it was suggested that a series of heat treatments be performed on the Ti-65Ta wt.% specimen to attempt to precipitate ω. The results of the proposed experiment and characterization are as follows.

9.2 Procedures

A specimen prepared from high purity Ti granules, Alfa Aesar 99.99% and high purity Ta of at least 99.99% purity. The alloy prepared by non-consumable vacuum arc- melting under a inert atmosphere of 99.998% Ar as described in section 4.2.1. The specimen was parallel polished on an Allied MultiPrep and modulus measurement taken by ultrasonic techniques. XRD scan were performed as described in section 4.2.3. The lattice parameters were measured and the structure compared with simulations of XRD patterns from candidate structures and atomic positions as was described in chapter 4.

The specimen of 65 wt.% Ta was annealed under vacuum of <1x10^-5 Torr for the following times and temperatures.

1. 250°C for 12 hours 2. 350°C of 4 hours 3. 450°C for 2 hours 4. 450°C for 24 hours.

Given the observed sensitivity of the structure and modulus to the concentration of Ta it was chosen to perform the heat treatments in sequence and extract site-specific TEM foils from the same local region of the specimen. After each heat treatment step XRD was performed on each specimen to determine the crystal structure and compared by observations in the TEM. Procedures used for sample preparation, ultrasonic modulus measurements, TEM, FIB and XRD procedures were used as described in chapters 3 and

4. Scanning Transmission Electron Microscopy (STEM) and EDS were performed on a

FEI Titan 60-300 operated at 300kV. Ultra high resolution STEM was performed on specimens an FEI Titan 80-300 operated at 300kV with third order Cs corrector.

DualBeam™ FIB sample preparation for ultra high resolution STEM included an additional “cleaning” step using an approximately 100pA beam with an accelerating voltage of 5kV.

9.3 Results

XRD patterns from each condition, as arc-melted, aged at 255°C for 12hr, 350°C for 4hrs, 450°C for 2hrs and 450°C for 24 hrs. are shown in Figure 30. It can be seen that the reflections related to the orthorhombic α’’ such as [012] are present in the as arc- melted condition. The reflection is then seen in diminished intensity as the specimen is aged. Finally in the 450°C age for 24hr condition, the diffraction pattern is nearly commensurate with a BCC structure. The modulus measurements performed by

85 ultrasonic techniques were 63, 67, 66, 69 and 91 GPa respectively. Thus in conjunction with the associated change in crystal structure the modulus of the material abruptly increased. TEM diffraction patterns from the [110] zone axis from each condition are shown in Figure 31. The diffraction patterns indicate there exists an orthorhombic α’’ structure in the as cast condition and weak ω reflection streaking is visible in the ⅓ [110] positions. Intensity profiles from each set of reflections are shown in Figure 32. On initial ageing the intensity of the reflections in the ½ [110]β representing the α’’ orthorhombic structure decrease in intensity. On aging at 450°C for 2 hours only diffuse

ω reflections are still visible. However, aging at 450°C for 24 hours produce more well defined ω reflections, yet ½ [110]β representing the α’’ orthorhombic structure are still visible. Dark field imaging attempts to selectively image ω are shown in Figure 31 (e).

Though, ω reflections are clearly observed in the diffraction patterns, well-defined ω particles were not observed as is typical in other alloy systems. For instance, clear ellipsoidal ω particles can be seen in results published by Kim et al. [57] in Ti–Nb binary alloys. The particles were roughly 50 nm in diameter in an alloy of Ti-26 at.% aged at

400°C for 10 hours. However, as shown in Figure 34 (a) limited number of α lath were visible confirming the observations of Zhou et.al [62]. Reflection seen in the diffraction pattern indicates only one variant of α was present. When observed in dark imaging the morphology of the α precipitates seem to confirm that only one variant of α was present.

The specimen further imaged by Scanning Electron Microscopy by using High Angle

Annular Dark-Field imaging. Precipitates were confirmed to be α by EDS analysis as shown in Figure 33. EDS mapping was utilized to visualize the distribution of Ti and Ta

86 in and near the α precipitates. A line profile was recorded where solute segregation between α and surrounding matrix can be observed.

Ultra high resolution STEM images were recorded parallel to the [110]β of the specimen aged at 450°C for 24hrs shown in Figure 36. Additionally an image was recorded parallel to the [111]β and is shown in Figure 37. Lattice images recorded in

[110]β and [111]β directions do not show well formed ω. However, there exist 3-5nm modulations in intensity that corresponds to weak ω dark field images shown in Figure 31

(e).

9.4 Discussion and Conclusions

From the experimental results shown above it is seen that a relatively large change in modulus can be observed while only modest change is observed in the structure and lattice parameter of Ti-65 wt.% Ta. Heat treatment altered the alloy constitution of

α’’, metastable β, α and ω in Ti-65 wt.% Ta binary alloy. In agreement with the conclusions made by Bagariatskii et al. [48] and Fedotov et al. [9], there may exists very small amount of ω in Ti-Ta alloys of sufficiently high Ta concentration. However, these small amounts of ω are not likely the cause of increased elastic modulus observed in the range of 28 to 55 wt.% Ta. The Ti–65Ta wt.% alloy with a modulus of 62GPa is close in magnitude to the lowest reported in [9]. Furthermore, on aging it was seen that the intensity of the α’’ reflection in TEM and XRD patterns were relaxed initially while only

87 slight changes were observed in the modulus. Similarly, in the TNZT alloy where the

Young’s modulus was observed to be only 50 GPa there exist both weak reflections of

α’’ and ω. In agreement with the calculations published by Chakraborty et al. [29], aging at 450°C for 24 hours was sufficient to initiate the precipitation of α. This would indicate that the metastable alloy was approaching some degree of equilibrium, and neither α’’ nor ω was completely stabilized. Thus, the calculations predicting that at a composition of Ti-33 at.% there is little driving force to produce α’’ or ω from β would appear to be correct. Figure 35 shows the results of lattice parameters reported chapter 4 as well as the lattice parameters published by Dobromyslov et al. [38]. The experimental lattice parameters, according [29] were ”optimized” for cell shape and volume. The

“optimized” lattice parameters are plotted along with the current and published experimental lattice parameters in Figure 35. Though the lattice parameters a and b are similar to the experimental observation, the lattice parameter c seems to differ significantly, such that the change in the unit cell volume appears to be consistent with experimental observations.

TEM and XRD patterns indicate that on annealing the structure of Ti–65 wt.% Ta is near BCC, yet contains a small degree of α’’ and ω distortions. The intensity of α’’ reflections seen in the as arc-melted materials by both diffraction techniques are in agreement. Lattice images recorded of Ti–65 wt.% Ta after annealing at 450°C for 24 hours show evidence of a small-scale disturbances in the BCC structure. Morphological features of the α’’ microstructure were absent after annealing. The modulus of 91GPa in the Ti-65Ta wt.% specimen on annealing was similar to the Ti-75 wt.% Ta specimen in

88 the as arc-melted condition at 89 GPa. Thus on annealing the crystal structure and elastic modulus became similar to that observed in Ti–75 wt.% Ta however, ω reflections and α were not observed in the Ti–75 wt.% specimen. Similarly TNZT, shown in Figure 38, with modulus of 50 GPa was observed to contain similar weak α’’ orthorhombic reflections, ω streaking and a similar microstructure. Thus is can be concluded that ω is likely not the cause of the increase in modulus but rather the precipitation of α or the enrichment of the β phase due to the rejection of solute from the precipitation of α and ω or both.

Figure 29 Combined nudged elastic band and density functional theory study of β → (α'', ω) in Ti–Ta alloys published by Chakraborty et al. [29].

Figure 30 XRD patterns from Ti–65Ta wt.% as cast, aged at 255°C for 12hr, 350°C for 4hrs, 450°C for 2hrs and 450°C for 24hrs.

Figure 31 TEM images and diffraction patterns from Ti–65Ta in as cast (a), aged for 255°C for 12hr (b), 350°C at 4hr. (c), 450°C for 2hr. (d) and 450°C for 24hrs.

Figure 33 STEM HAADF image of α precipitates in Ti–65Ta wt.% (a), selected area for EDS mapping, red line indicates location of line profile (b), Map of Ta Kα intensity (c), map of Ti Kα intensity (d), line profile across α precipitate where wt.% were calculated from Kα intensities.

Figure 34 Dark Field TEM image of Ti-α precipitates in Ti–65Ta wt.% aged at 450°C for 24 hours. Inset is image of [113]β zone axis. Circle indicates the location of objective aperture.

Figure 35 Plots of Ti–Ta binary alloy lattice parameters, lattice parameters reported by Dobromyslov et al. [38] and relaxed lattice parameters used in combined nudge elastic band and density functional theory calculations [29].

Figure 36 Ultra high resolution STEM HAADF image along [110]β (left), Fourier Transform of real space image (right).

Figure 37 Ultra high resolution STEM HAADF image along [111]β. Dark area in upper right corresponds to α, were mottled contrast surrounding contains ω and β.

Figure 38 Dark field TEM image of TNZT alloy with a Young’s modulus of 50 GPa. Inset image of [110]β diffraction pattern showing weak ω and α’’ orthorhombic reflection intensity.

10. Summary Discussion

10.1 Lattice Parameters and Atomic Positions

Binary Ti alloys have been studied extensively as they offer unique properties such as super elasticity, shape memory, high damping and low elastic modulus. Modern efforts to study theses alloys and predict their behavior have included first-principles calculations and density functional theory. These computational methods require a detailed understanding of the underlying crystal system, as the structure is highly dependent on composition as has been shown in prior chapters and elsewhere in the literature. The structure of Ti–Ta and Ti–Nb binary alloys can be studied under a single consistent crystal system with space group symmetry Cmcm, by defining the lattice parameters and the Wyckoff 4c atomic position along the glide plane “y”. These lattice parameters and atomic positions are strongly dependent on composition as demonstrated in prior chapters. Li et al. [63] noted that there is little published regarding the atomic positions in the structure of binary alloy martensite and lattice parameters are rarely reported.

Early work by Brown et al. [47] indicated the lattice parameters for Ti–20 Nb at.%

99 to be a = 3.166Å, b = 4.854Å and c = 4.652Å and the position y = 0.2 for the [002]α’’.

These findings were later expanded on by Banumathy et al. [43] for the Ti–Nb system.

Later in 2014 Bönisch et al. [1] published in detail the lattice parameters and atomic positions of the Ti–Nb system. Results shown in chapter 4 demonstrate clear agreement between values of elastic modulus as well as lattice parameters published in literature.

Further analysis of the intensity of the XRD reflections seen revealed that the position of

“y” could be determined and is shown in Figure 39. The results show similar behavior when compared with the experimental results published in [1]. Figure 40 shows, in terms of atomic percentage, the variation of y as determined experimentally and as predicted by

Li et al. [63]. The magnitude of the atomic shuffle for both Ta and Nb appear to be similar with regard to the respective concentrations in terms of atom percent yet, the calculations predict a slight variation in the magnitude.

10.2 Effects of Oxygen on Modulus and Atomic Positions

Binary alloys of Ti–Ta show two distinct minima in the Young’s modulus near the boundaries of α’/α’’ and α’’/β [8, 9]. Similar trends in modulus are also observed in other binary alloys such as Ti–Mo and Ti–Nb. Given the significant decrease in modulus as Ta content increases while α’ remains stable, it was hypothesized in chapter 5 that stabilization of α’ with oxygen could produce a lower modulus by allowing the hexagonal structure to be maintained at higher Ta concentrations. It was then shown in chapter 5

100 that the addition of oxygen corresponded to a higher modulus as well as the stabilization of α’ in Ti–30Ta wt.% while in the high purity condition the similar Ta content was observed to be α’’. The observed shuffle positions were also seen to correspond with y =

0.166 at the Ti–30Ta wt.% concentration, indicating that the shuffle from the BCC to

HCP direction was complete. XRD patterns collected of Ti–40Ta wt.% and Ti–

40Ta+1.6O wt.%, shown in Figure 41, represent a higher degree of orthorhombic distortion. The diffraction pattern of the sample containing oxygen shows broad reflections indicating a significant distortion of the structure. The relative position of

[112] and [022] peaks indicate that the addition of oxygen changed the y shuffle as the ratio is more representative of y = 0.166 rather than y = 0.2 as observed in the high purity material. Alloys of lower Ta concentrations with 1.6 O wt.% are consistent with the HCP crystal structure although the observed diffraction patterns from Ti–40 Ta+1.6O show less agreement with the simulated patterns. The intensity and positions of low order reflections agree well with predictions of the diffraction pattern with the given lattice constants, as shown in Figure 41. The higher order reflections however, only show limited agreement and may indicate the atomic positions are not well described by the

Cmcm space group and belong to another crystal system. Work published by Yamaguchi et al. [37] showed that oxygen in high concentrations had a high probability of locating at octahedral sites. The resulting ordering led to the displacement of Ti atoms away from the ideal HCP sites in the c direction of α. Similar to this result, it was shown in chapter

5, Figure 15, that the c lattice constant of α’’ was also slightly dilated with respect to that of the high purity materials.

101

10.3 Effect of Symmetry on Modulus

The addition of yttrium to the binary alloys as described in chapter 6 changed the structure and modulus values little. The exception was however, Ti–25Ta+Y wt.% where the modulus increased slightly. Close inspection of the XRD patterns shown in Figure 42 show that the alloy containing the addition of Y produced a diffraction pattern which nearly matched CP Ti with α’ structure and hexagonal symmetry. The alloy of Ti–25Ta wt.% shows some slight degree of distortion when compared with CP Ti. Comparison of the Ti-25Ta XRD pattern was made with patterns for CP Ti, Ti–15Ta and Ti–20Ta as shown in Figure 43. The pattern for Ti–25Ta wt.% shows slight broadening of diffraction peaks and a loss of relative intensity in [113] peak where P63/mmc symmetry may be lost. Thus the increase in modulus from 57 GPa to 71 GPa may in fact be due to the slight change in lattice parameters and increase in symmetry. This result agrees with the similar observation in chapter 9 that on annealing Ti–65 Ta wt.% symmetry was gained by changing from α’’ to β, similar to Ti–75Ta wt.%. The resulting modulus of 91

GPa on annealing is similar to the level observed in Ti–75 Ta wt.% at 87 GPa. These results indicate that the change in modulus is closely related to both crystal structure and distortions leading to decrease in symmetry. In the case of alloys that are near the α’/α’’,

α’’/β stability boundaries, transitions between crystal symmetries appears to be abrupt as reported by others. The modulus is then, in this case, not necessarily dictated by a rule of

102 mixtures as suggested by Zhou et al. [62].

10.4 Effects on Modulus and Atomic Positions

Work published by Bagariatskii et al. [48] described studies of many binary alloys of Ti, including W, V, Re, Mo, as well as Nb and Ta. The work included XRD and hardness measurements of as a function of alloy content. Bagariatskii found a common trend in the results that indicated a minimum in hardness may be attributed to the presence of α’’ and again in β. Between the two hardness minima, there was a significant maximum in hardness that was attributed to the formation of ω. Although the ω phase was not observed in the Ta containing alloys in the quenched condition, since its formation was observed in other alloys, it was concluded that ω must exist in the Ti–Ta binary. Later work in 1965 on Ti binary systems of Cr, Mn, Fe, Co, V and Ni by Fedotov et al. [8] shows a similar decrease in elastic modulus and then a sharp increase that was also attributed to the formation of ω. The elastic modulus of the binary Ti–Ta system was not studied until 1984 by Fedotov et al. [9]. It was again concluded that the formation of ω led to the increase in modulus observed in the intermediate compositions.

Although ω was not observed in XRD studies, the change in mechanical properties led the authors to conclude that only small quantities must form. In the current study ω was not observed in the intermediate compositions though weak ω reflection streaking was observed in the [110]β diffraction patterns of Ti–65 Ta wt.% as shown in Figure 44. This

103 composition, though it contains weak ω reflections, displayed a low value of modulus at

62 GPa. Likewise a low modulus of 50 GPa was observed in the TNZT alloy, Ti–31 Nb–

6 Zr–4 Ta wt.%, and also weak ω reflections were observed as seen in Figure 44 (e).

HRSTEM was used to image the lattice of the Ti–65Ta wt.% specimen in the [110] and

[111] directions as shown in Figure 45. Well-formed ω particles were not observed as has been reported in Ti–Nb alloys. Small particles of approximately 3-5nm were resolved in the [110]β direction, Figure 45 (a).

The weak ω streaking was observed in two low modulus conditions and not observed in the Ti–55Ta+2Al specimen. The Al containing specimen was imaged further in HRSTEM. An image recorded along [110]β shown in Figure 46 (a) shows that the y shuffle can be directly observed as predicted by XRD observations. Bönisch et al. [1] predict that the structure of martensite should exist in well defined domains bound by

“anti-phase” like boundaries. This type of domain structure should be visible in the image formed with the electron beam parallel to [110]β. Imaging the same area in the

[100] β beam direction reveals no clear domain structure, thus the image was further processed by passing an FFT kernel filter over each pixel of the image where the intensity of the pixels represents the measured ½ [110] frequency intensity sampled by the kernel. Intensity from [110] can be contributed in this case by either Cmcm symmetry or !"3!. The intensity at ½ [110] should in this case only result from the

Cmcm structure. Thus the domain structure can be visualized by quantification of the degree of fit to the local FFT intensity. The result of the kernel filter processing is shown in Figure 46 (b). The processed image shows apparent regions that contain various

104 degrees of fit however there are no well-defined domains or sharp domain boundaries observed. This result is consistent with the observations of the XRD patterns that show broad diffraction peaks resulting from a distorted structure that contains considerable variation in planar spacing. Close inspection of the (110)β structure along the [001]β direction in Figure 46 shows that the alignment of every second plane is necessarily perfect. The registry of the (110)β along the [001] β is represented as the [200]α’’ in the

XRD data. Close inspection of the [200]α’’ reflection from Ti–25Ta+2Al reveals that the peak in not visible in the pattern and thus the miss alignment seen in HRSTEM may be consistent with the bulk structure, implying that again this structure may have less than

Cmcm symmetry.

10.5 Final Thoughts and Future Work

The work presented above in combination with the existing published literature indicates that there are many combinations of Ti with alloying elements that produce low modulus. The low modulus may exist in two distinct composition ranges. Given that the majority of low modulus, biocompatible alloys reported in literature are of the β-type or near β alloys [6], there may exist another class of alloys that exhibit a much lower specific modulus, producing an alloy of much lower density and thus a low specific modulus may be desirable.

There are concerns that β-type or near β alloys, being only metastable, may not

105 maintain consistent properties over the lifetime of an implant device. Report of room temperature aging and precipitation of ω in Ti–Ta as reported in [39] is concerning.

However as has been demonstrated, very small amounts of ω do not necessarily increase the modulus of the materials.

The addition of oxygen appears to increase the elastic modulus of α’ and α’’ martensitic alloys. Even low concentrations of oxygen that exist in commercial purity grade 2 Ti materials and Ti powder would most likely not produce the ultimate low modulus in binary alloy materials for practical applications. Likewise, alloys of Ti–Ta at a concentration of 65 wt % would be too dense and add un-desirable weight in implant components. A better understanding is therefore needed to determine the origin of the low modulus phenomena so that compositions of new alloys may be predicted. It is proposed that a set of alloys is produced containing various concentrations of Ti–Ta and an ternary alloy element such as Mo and Fe where there are reports in literature that indicate the similar changes in lattice parameters and α’ and α’’ stability may occur [10,

48]. The benefits of adding these elements are expected since they have a stronger influence on lattice parameters than Nb and Ta and thus could form low modulus alloys of lower density than heavily alloyed Ti–Ta alloys. Work published by Chaves et al. [64] shows, in the Ti-xNb-3Fe wt.% system, a reduced elastic modulus of 77, 71, 67 and 65

GPa at Nb concentrations of 10, 15, 20 and 25 wt.%. Though no lattice parameters are given, phases reported are combinations of β and β + ω. Due to the high coefficient given Fe in the Mo equivalency equation, there is a possibility of obtaining improved properties at low concentrations of only1, 2 and 3 wt.%. Since intermetallic compounds

106 may form and eutectoid decomposition is possible, careful studies would need to be completed. The development of new alloys may be expected in the Ti - (Nb,Ta) –

(Mo,Fe) alloy system where the high β stability provided by rather small additions Mo or

Fe could reduce the specific modulus.

Figure 39 Atomic position y of [002]α" in Ti–xTa binary alloys. Value of 0.166 corresponds to the position necessary to create hexagonal symmetry and 0.25 corresponds to the location for BCC.

107

Figure 40 Experimentally determined atomic position y of [002]α" in Ti–xTa, Ti–xNb [1] binary alloys and predicted positions for Ti–xTa and Ti–xNb published by Li et al. [63].

108

Figure 41 XRD patterns of Ti–40Ta wt.% and Ti–40Ta+1.6O wt.% and simulations of Cmcm crystal structure where lattice constants are similar and y position is 0.2 and 0.166 respectively.

109

110

Figure 43 XRD patterns of CP Ti, Ti–15Ta, Ti–20Ta and Ti–25Ta wt.%. Reduced intensity and broadening of peaks such as [113]α’’ indicate symmetry may be less Cmcm and less than P63/mmc.

111

Figure 44 TEM images and diffraction patterns from Ti–65Ta (a), Ti–55Ta+1.6O (b), Ti– 55Ta+2Al (c), Ti–55Ta (d), Ti–31Nb–6Zr–4Ta (e). TNZT shows a low modulus of 50 GPa and weak ω and orthorhombic reflections. Weak ω streaking was also observed in Ti–65Ta.

112

113

114

Figure 47 XRD patterns for Ti–xTa+Al (x=25,35,45,55) where the [200]α'' reflection being parallel to [110]β is not present.

115

References

1. Bönisch, M., et al., Composition-dependent magnitude of atomic shuffles in Ti–Nb martensites. Journal of Applied Crystallography, 2014. 47(4): p. 1374-1379. 2. Sumner, D.R. and J.O. Galante, Determinants of Stress Shielding: Seding Versus Materials Versus Interface. Clinical Orthopaedics and Related Research, 1992. 274: p. 202-212. 3. Huiskes, R., H. Weinans, and B. Reitbergen, The relationship between stress shielding and bone resorption around total hip stems and the effects of flexible materials. Clinical Orthopaedics and Related Research, 1992. 274: p. 124-134. 4. Niinomi, M., Low Modulus Titanium Alloys for Inhibiting Bone Atrophy, in Biomaterials Science and Engineering, P.R. Pignatello, Editor 2011, InTech,. 5. Saito, T., et al., Multifunctional Alloys Obtained via a Dislocation-Free Plastic Deformation Mechanism. Science 2003. 300(5618): p. 464-467. 6. Mohammed, M.T., Z.A. Khan, and A.N. Siddiquee, Beta Titanium Alloys: The Lowest Elastic Modulus for Biomedical Applications: A Review. International Journal of Chemical, Nuclear, Materials and Metallurgical Engineering, 2014. 8(8). 7. Besse, M., P. Castany, and T. Gloriant, Mechanisms of deformation in gum metal TNTZ-O and TNTZ titanium alloys: A comparative study on the oxygen influence. Acta Materialia, 2011. 59(15): p. 5982-5988. 8. Fedotov, S.G. and E.P. Sinodova, Specific features of martensitic transformation in titanium alloys. Metal Science and Heat Treatment, 1965. 7(4): p. 233-238. 9. Fedotov, S.G., et al., Phase Structure, Critical Points Ms and As of Martensitic Transformation and Elastic Properties of Metastable Alloys of the Ti–Ta System. Physics of Metals and Metallography, 1985. 60: p. 139-143. 10. Williams, J.C. and G. Luetjering, Titanium. 2 ed2007: Springer. 11. Collings, E.W., The physical metallurgy of titanium alloys. ASM series in metal processing1984, Metals Park, OH: American Society for Metals. xxii, 261 p. 12. Boyer, R.R. and R.D. Briggs, The Use of β Titanium Alloys in the Aerospace Industry. Journal of Materials Engineering and Performance, 2005. 14(6): p. 681- 685. 13. Brewer, R., R.K. Bird, and T.A. Wallace, Titanium alloys and processing for high speed aircraft. Materials Science and Engineering A, 1998. 243(1-2): p. 299-304. 116

14. Williams, J.C. and J. Edgar A. Starke, Progress in structural materials for aerospace systems. Acta Materialia, 2003. 51(19): p. 5775-5799. 15. Williams, J.C., B.S. Hickman, and D.H. Leslie, The Effect of Ternary Additions on the Decompositon of Metastable Beta-PhaseTitanium Alloys. Metallurgical Transactions, 1971. 2: p. 477-484. 16. Moiseev, V.N., Modern Structural Titanium Alloys. Metal Science and Heat Treatment, 1980. 22(7): p. 489-494. 17. Banerjee, S. and P. Mukhopadhyay, Phase transformations : examples from titanium and zirconium alloys. Pergamon materials series2007, Amsterdam ; Oxford: Elsevier/Pergamon. xxii, 813 p. 18. Hickman, B.S., The Formation of Omega Phase in Titanium and Zirconium Alloys: A Review. Journal of Materials Science, 1969. 4: p. 554-563. 19. Duerig, T.W., G.T. Terlinde, and J.C. Williams. The ω-phase Reaction in Titanium Alloys. in Titanium '80 Science and Technology: Fourth International Conference on Titanium. 1980. Kyoto, Japan: The Metallurgical Society of AIME. 20. Moiseev, V.N. and L.V. Geras'kova, Variation Of The Structure And Properties Of Titanium Alloys With The Heat Treatment Conditions. Metal Science and Heat Treatment, 1965. 7(4): p. 280-285. 21. McClintick, R.J., W.G. Bauer, and L.S. Busch, Physical Metallurgy and Heat Treatment of Titanium Alloys1955: Mallory Sharon Titanium Corperation,,. 22. Bendersky, L.A., A. Roytburd, and W.J. Boettinger, Phase Transformations in the (Ti, Al)3 Nb Section of the Ti-Al-Nb System--I. Microstructural Predictions Based on a Subgroup Relation Between Phases Acta Materialia, 1994. 42(7): p. 2323- 2335. 23. Franti, G.W., J.C. Williams, and H.I. Aaronson, A Survey of Eutectoid Decomposition in Ten Ti-X Systems. Metallurgical Transactions A, 1978. 9A: p. 1641-1649. 24. Stupel, M.M., M. Ron, and B.Z. Weiss, Formation of the ω-Phase During the Aging of ß Ti (7.1Wt Pct Fe)-- A Mössbauer Study. Metallurgical Transactions A, 1978. 9A: p. 249-252. 25. Silva, E.Z., et al., Order and disorder in iron-titanium. Physical Review B, 1987. 36(6): p. 3015-3025. 26. Kale, G.B., R.V. Patil, and P.S. Gawade, Interdiffusion studies in titanium-304 stainless steel system. Journal of Nuclear Materials, 1998. 257(1): p. 44-50. 27. Ferrante, M. and E.V. Pigoretti, Diffusion bonding of TI-6AL-4V to AISI 316L stainless steel: mechanical resistance and interface microstructure. Journal of Materials Science, 2002(37): p. 2825-2833. 28. Ray, R., B.C. Giessen, and N.J. Grant, The Consitiution of Metastable Titanium- Rich Ti-Fe Alloys: An Order-Disorder Transition. Metallurgical Transactions, 1972. 3: p. 627-629. 29. Chakraborty, T., J. Rogal, and R. Drautz, Martensitic transformation between competing phaes in Ti-Ta alloys: a solid-stae nudges elastic band study. Journal of Physics: Condensed Matter, 2015. 27. 117

30. Williams, J.C., R. Taggart, and D.H. Polonis, The Morphology and Substructure of Ti-Cu Martensite. Metallurgical Transactions, 1970. 1: p. 2265-2270. 31. Mitchell, A., Melting, Casting, and Forging Problems in Titanium Alloys. Journal of Materials Science, 1997: p. 40-46,69. 32. Imam, A.M. and F.H. Froes, Low Cost Titanium and Developing Applications. Journal of Materials Science, 2010. 62(5): p. 17-20. 33. Moxson, V.S., O.N. Senkov, and F.H. FROES, Innovations in Titanium Powder Processing. Journal of Materials Science, 2000: p. 24-26. 34. Maykuth, D.J., H.R. Ogden, and R.I. Jaffee, The Effects of Alloying Elements in Titanium. 1960. A. Constitution. 35. Ogden, H.R. and R.I. Jaffee, The Effects of Carbon, Oxygen, and Nitrogen on The Mechanical Properties of Titanium And Titanium Alloys, 1955, Batelle Memorial Institute: Columbus, Ohio. 36. Liu, Z. and G. Welsch, Effects of Oxygen and Heat Treatment on the Mechanical Properties of Alpha and Beta TitaniumAlloys. Metallurgical Transactions A, 1987. 19A: p. 527-542. 37. Yamaguchi, S., Interstitial Order-Disorder in the Ti-O Solid Solution. I. Ordered Arrangement of Oxygen. Journal of the Physical Society of Japan, 1969. 27(1): p. 155-163. 38. Dobromyslov, A.V., et al., Phase and structural transformations in Ti-Ta alloys. The Physics of Metals and Metallography, 2009. 107(5): p. 502-510. 39. Cotton, J.D., Microstructure and Mechanical Properites of Ti-40 Wt Pct Ta (Ti- 15 At. Pct Ta). Metallurgical And Materials Transactions A, 1994. 25A: p. 461- 472. 40. Matlakhova, L.A., et al., Properties and structural characteristics of Ti-Nb-Al alloys. Materials Science and Engineering A, 2005. 393: p. 320-326. 41. Zhou, Y.-L. and M. Niinomi, Ti–25Ta alloy with the best mechanical compatibility in Ti–Ta alloys for biomedical applications. Materials Science and Engineering: C, 2009. 29(3): p. 1061-1065. 42. Bywater, K.A. and J.W. Christian, Martensitic transformations in titanium- tantalum alloys. Philosophical Magazine, 1972. 25(6): p. 1249-1273. 43. Banumathy, S., R.K. Mandal, and A.K. Singh, Structure of orthorhombic martensitic phase in binary Ti–Nb alloys. Journal of Applied Physics, 2009. 106(9): p. 093518. 44. Majumdar, P., S.B. Singh, and M. Chakraborty, Elastic modulus of biomedical titanium alloys by nano-indentation and ultrasonic techniques—A comparative study. Materials Science and Engineering: A, 2008. 489(1-2): p. 419-425. 45. Oliver, W.C. and G.M. Pharr, An improved technique for determining hardness and elastic modulus using displacement sensing indentation experiments. Journal of Materials Research, 1992. 7(6): p. 1564-1583. 46. Kumar, A., et al., Correlation between ultrasonic shear wave velocity and Poisson’s ratio for isotropic solid materials. Acta Materialia, 2003. 51: p. 2417- 2426.

118

47. Brown, A.R.G., et al., The Titanium-Niobium System. Nature, 1964. 201: p. 914- 915. 48. Bagariatskii, I.A., G.I. Nosova, and T.V. Tagunova, Factors in the Formation of Metastable Phases in Titanium-Base Alloys. Soviet Physics Doklady, 1958. 3. 49. Pathak, A., et al., Orthorhombic martensitic phase in Ti–Nb alloys: A first principles study. Computational Materials Science, 2014. 83: p. 222-228. 50. Bönisch, M., et al., Thermal stability and phase transformations of martensitic Ti- Nb alloys. Science and Technology of Advanced Materials, 2013. 14. 51. Zhou, Y.L., M. Niinomi, and T. Akahori, Effects of Ta content on Young’s modulus and tensile properties of binary Ti–Ta alloys for biomedical applications. Materials Science and Engineering: A, 2004. 371(1-2): p. 283-290. 52. Khorev, A.I., Alloying Titanium Alloys with Rare-Earth Metals. Russian Engineering Research, 2011. 31(11): p. 1087-1094. 53. Whittsett, C.R., et al., Influence of Rare-Earth Additions on Properties of Titanium Alloys, 1977, Department of The Navy: St. Louis, Missouri 63166. 54. Hieda, J., et al., Effect of Oxide Particles Formed through Addition of Rare-Earth Metal on Mechanical Properties of Biomedical ^|^beta;-Type Titanium Alloy. Materials Transactions, 2013. 54(8): p. 1361-1367. 55. Meng, Q., et al., A β-type TiNbZr alloy with low modulus and high strength for biomedical applications. Progress in Natural Science: Materials International, 2014. 24(2): p. 157-162. 56. James, D.W., High Damping Metals for Engineering Applications. Materials Science and Engineering, 1969. 4(1): p. 1-8. 57. Kim, H.Y., et al., Martensitic transformation, shape memory effect and superelasticity of Ti–Nb binary alloys. Acta Materialia, 2006. 54: p. 2419-2429. 58. Hao, Y.L., et al., Effect of Zr and Sn on Young’s modulus and superelasticity of Ti–Nb-based alloys. Materials Science and Engineering A, 2006. 441: p. 112-118. 59. Buenconsejo, P.J.S., H.Y. Kim, and S. Miyazaki, Novel β-TiTaAl alloys with excellent cold workability and a stable high-temperature shape memory effect. Scripta Materialia, 2011. 64: p. 1114-1117. 60. Buenconsejo, P.J., et al., Shape Memory behavior of Ti-Ta and its potential as a high-temperature shape memory alloy. Acta Materialia, 2009. 57: p. 1068-1077. 61. Buenconsejo, P.J.S., H.Y. Kim, and S. Miyazaki, Effect of ternary alloying elements on the shape memory behavior of Ti–Ta alloys. Acta Materialia, 2009. 57: p. 2509-2515. 62. Zhou, Y.-L. and M. Niinomi, Microstructures and mechanical properties of Ti– 50mass% Ta alloy for biomedical applications. Journal of Alloys and Compounds, 2008. 466(1-2): p. 535-542. 63. Li, C.-X., et al., Lattice parameters and relative stability of α'' phase in binary titanium alloys from first-principles calculations. Solid State Communications, 2013. 159: p. 70–75. 64. Chaves, J.M., et al., Influence of phase transformations on dynamical elastic modulus and anelasticity of beta Ti–Nb–Fe alloys for biomedical applications.

119

Journal of the Mechanical Behavior of Biomedical Materials, 2015. 46: p. 184- 196.

120