The Pennsylvania State University

The Graduate School

College of Engineering

EVALUATING THE IN VITRO CORROSION BEHAVIOR AND

CYTOTOXICITY OF VAPOR DEPOSITED ALLOYS

A Thesis in

Engineering Science

by

John D. Petrilli

© 2009 John D. Petrilli

Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science

May 2009

The thesis of John D. Petrilli was reviewed and approved* by the following:

Barbara A. Shaw Professor of Engineering Science and Mechanics Thesis Advisor

Mark W. Horn Associate Professor of Engineering Science and Mechanics

Elzbieta Sikora Research Associate of Engineering Science and Mechanics

Judith A. Todd Professor of Engineering Science and Mechanics P.B. Breneman Department Head of Engineering Science and Mechanics

*Signatures are on file in the Graduate School

iii ABSTRACT

Magnesium alloys are emerging as a promising class of bioabsorbable implant materials due to magnesium’s biocompatibility and propensity for corrosion. These alloys are useful for implants like cardiac stents. Stents only need to support a diseased blood vessel while it heals, but their permanent presence has been linked with medical complications. Bioabsorbable magnesium stents have therefore been developed, but clinical trials with stents made of magnesium alloy WE43 have shown that they corrode before the healing process is complete.

This thesis therefore proposes physical vapor deposition (PVD) as a means of fabricating

nonequilibrium magnesium alloys with improved corrosion resistance. XRD analysis of PVD

alloys containing and titanium showed preferential Mg (002) orientation, but alloying

elements remained in solid solution. Electrochemical methods were used to test both PVD

magnesium alloys and commercial WE43 under in vitro conditions in 37ºC Hanks Balanced Salt

Solution (HBSS). Results showed that the PVD alloys exhibited superior corrosion resistance to

WE43, especially after aging for several months to one year. Experimental evidence did not

indicate that the surface oxide was responsible for the increased corrosion resistance after aging,

so it was attributed to stress relief annealing at room temperature. In addition, transformed A549

cells were cultured on magnesium alloys to assess the alloys’ in vitro cytotoxicity. Cells grown

on WE43 were observed to demonstrate indeterminate or apoptotic morphologies, but cells grown

on PVD alloys with low (< 15 μA/cm2) corrosion rates were observed to be viable. These cells

demonstrated healthy morphologies and grew to form partial monolayers over the PVD alloys’

surfaces. An aged PVD alloy containing 2 wt% titanium was observed to demonstrate a lower

corrosion rate, comparable passive current density, higher breakdown potential, larger passive

region, and lower cytotoxicity than commercial alloy WE43.

iv TABLE OF CONTENTS

LIST OF FIGURES ...... vi

LIST OF TABLES...... x

ACKNOWLEDGEMENTS...... xii

Chapter 1 Introduction ...... 1

1.1 Drug-Eluting Stents and Bioabsorbable Magnesium...... 3 1.2 Objectives...... 4

Chapter 2 Background ...... 6

2.1 Magnesium Alloys As Bioabsorbable Implants...... 6 2.1.1 Cardiovascular Uses for Bioabsorbable Magnesium ...... 8 2.1.2 Othopedic Uses for Bioabsorbable Magnesium...... 10 2.1.3 Cytotoxicity Experiments with Magnesium...... 11 2.2 Corrosion of Magnesium...... 12 2.2.1 Nature of the Passive Film ...... 15 2.2.2 Impurities and Tolerance Limits ...... 16 2.2.3 The Negative Difference Effect ...... 18 2.3 Alloying to Improve Corrosion Resistance...... 19 2.3.1 Electrochemical and Atomic Considerations for Magnesium Alloys ...... 21 2.3.2 Physiological Considerations for Magnesium Alloys ...... 26 2.3.3 Electrochemical Studies of Magnesium Alloys in Simulated Body Fluid ...... 28 2.4 Magnesium Alloys Produced Via Physical Vapor Deposition ...... 30 2.4.1 Morphology of Vapor Deposited Alloys...... 32 2.4.2 Corrosion Studies of PVD Magnesium Alloys ...... 34 2.5 Summary ...... 36

Chapter 3 Experimental ...... 37

3.1 Electron Beam Physical Vapor Deposition...... 37 3.1.1 Dual-Gun Deposition ...... 37 3.1.2 Industry-Scale Deposition...... 40 3.2 Chemical Characterization...... 43 3.2.1 Energy Dispersive X-Ray Spectroscopy ...... 43 3.2.2 Inductively Coupled Plasma Atomic Emission Spectroscopy ...... 43 3.3 Structural Characterization...... 45 3.3.1 Profilometry ...... 46 3.3.2 Optical Microscopy...... 46 3.3.3 Scanning Electron Microscopy ...... 47 3.3.4 X-Ray Diffraction ...... 48 3.4 Corrosion Characterization ...... 49 3.4.1 Mass Loss Measurements...... 52 3.4.2 Open Circuit Potential...... 55 3.4.3 Linear Polarization ...... 55

v 3.4.4 Potentiodynamic Polarization...... 58 3.4.5 Electrochemical Impedance Spectroscopy...... 59 3.5 Magnesium Cytoxicity Characterization...... 60 3.5.1 Cell Cultures...... 60 3.5.2 Cell Fixation...... 62

Chapter 4 Results ...... 64

4.1 Chemical Analysis of EB-PVD Mg Alloys...... 64 4.2 Structural Characterization of WE43 and EB-PVD Mg Alloys...... 66 4.2.1 Alloy Thicknesses ...... 66 4.2.2 Microstructure and Surface Morphology ...... 67 4.3 Corrosion of Magnesium Alloys...... 84 4.3.1 Corrosion Rates of WE43 from Mass Loss...... 84 4.3.2 Corrosion Rates from Linear Polarization and EIS...... 87 4.3.3 Potentiodynamic Polarization...... 94 4.3.4 Mode of Corrosion Attack...... 99 4.4 Cytotoxicity of Magnesium Alloys...... 104 4.4.1 Effects of Elemental Concentrations in Growth Media ...... 104 4.4.2 Effects of Corrosion on Cell Morphology and Viability...... 106

Chapter 5 Summary and Discussion ...... 110

5.1 Experimental Error...... 110 5.2 Factors Influencing Corrosion Rate ...... 112 5.2.1 Microstructure and Surface Morphology ...... 112 5.2.2 Passive Film Formation...... 122 5.3 Corrosion Behavior of Magnesium Alloys ...... 124 5.3.1 Electrochemical Corrosion Rates ...... 124 5.3.2 Mass Loss Corrosion Rates for WE43 ...... 126 5.3.3 The Effect of Time ...... 126 5.4 Factors Influencing Cytotoxicity...... 131

Chapter 6 Conclusions and Future Work...... 133

6.1 Conclusions...... 133 6.2 Future Work ...... 135

References...... 137

Appendix A Biological and Medical Terminology...... 147

Appendix B Corrosion Rates Converted into Penetration Rates ...... 150

vi LIST OF FIGURES

Figure 1-1: Biotronik’s experimental magnesium stent...... 2

Figure 2-1: Perrault’s Pourbaix diagram for magnesium...... 13

Figure 2-2: Theoretical domains of magnesium’s Pourbaix diagram...... 14

Figure 2-3: Tolerance limits for binary magnesium alloys...... 17

Figure 2-4: Pourbaix diagrams for yttrium, neodymium, and zirconium ...... 23

Figure 2-5: Pourbaix diagram for titanium ...... 24

Figure 2-6: Hydrogen exchange current densities of various elements ...... 25

Figure 2-7: Structured PVD magnesium...... 32

Figure 2-8: Thornton’s structure zone model for vapor deposits...... 33

Figure 3-1: Picture and shematic of the dual gun EB-PVD system designed at Penn State...38

Figure 3-2: Industrial prototype Sciaky EB-PVD system...... 40

Figure 3-3: Layout of the Mg-Y-Ti alloys made using the Sciaky EB-PVD system...... 42

Figure 3-4: Tencor P-10 Surface Profiler ...... 46

Figure 3-5: Motic Images 2000 computer interface...... 47

Figure 3-6: Sintag X2 Theta-Theta Powder Diffractometer ...... 48

Figure 3-7: Example of a samples masked with Apiezon® Wax W ...... 50

Figure 3-8: The corrosion cell used for electrochemical tests ...... 52

Figure 3-9: Corrosion product on a magnesium sample ...... 54

Figure 3-10: Mass loss samples of WE43 cleaned according to ASTM G1...... 54

Figure 3-11: Sample Rp measurement from a linear polarization curve...... 56

Figure 3-12: Sample Rp measurement from a Nyquist plot ...... 59

Figure 4-1: Etched WE43 grains...... 67

Figure 4-2: EB-PVD pure Mg microstructures and surface morphologies...... 68

Figure 4-3: Grazing angle XRD scans of bulk and EB-PVD Mg...... 69

vii Figure 4-4: EB-PVD Mg-Y alloy microstructures and surface morphologies...... 70

Figure 4-5: Grazing angle XRD scans of fresh EB-PVD Mg-Y and commercial Mg...... 71

Figure 4-6: (311) peaks...... 71

Figure 4-7: EB-PVD Mg-Ti alloy structure...... 72

Figure 4-8: Grazing angle XRD scans of fresh EB-PVD Mg-Ti and commercial Mg...... 73

Figure 4-9: Aged EB-PVD Mg-Ti alloy ...... 74

Figure 4-10: EB-PVD Mg-Y-Ti surface morphology...... 75

Figure 4-11: EB-PVD Mg-Y-Ti alloy microstructures...... 75

Figure 4-12: EB-PVD Mg-Y-Ti alloy with columnar microstructure and porous surface .....76

Figure 4-13: Fracture surfaces of EB-PVD Mg-Y-Ti alloys ...... 76

Figure 4-14: Grazing angle XRD scans of fresh EB-PVD Mg-Y-Ti and commercial Mg.....77

Figure 4-15: Aged EB-PVD Mg-Y-Ti alloy surface morphologies...... 78

Figure 4-16: Aged EB-PVD Mg-Y-Ti alloy microstructures ...... 79

Figure 4-17: Structures of thick Mg-Y-Ti alloys...... 80

Figure 4-18: Defects in thick Mg-Y-Ti alloy T1 ...... 81

Figure 4-19: Surface morphologies of thick Mg-Y-Ti alloys ...... 82

Figure 4-20: Grazing angle XRD scans of aged EB-PVD T15 and commercial Mg ...... 83

Figure 4-21: Aged thick Mg-Y-Ti alloys...... 84

Figure 4-22: Localized corrosion on WE43 immersed in 37ºC HBSS for 1 week...... 86

Figure 4-23: Pitting corrosion on WE43 immersed in 37ºC HBSS for 1 week ...... 86

Figure 4-24: Undermining of grains on WE43 immersed in 37ºC HBSS for 1 week ...... 87

Figure 4-25: Mg-Y-Ti sample before and after linear polarization and EIS tests...... 87

Figure 4-26: Corrosion rates determined by linear polarization as a function of time ...... 91

Figure 4-27: Corrosion rates determined by EIS as a function of time ...... 92

Figure 4-28: LP corrosion rates for thin alloys as a function of time ...... 93

viii Figure 4-29: EIS corrosion rates for thin alloys as a function of time...... 93

Figure 4-30: Mg-Y-Ti sample before and after a potentiodynamic polarization test ...... 94

Figure 4-31: IR compensation for potentiodynamic polarization scans of WE43...... 95

Figure 4-32: Potentiodynamic scans of aged Mg-Y and Mg-Ti EB-PVD alloys ...... 96

Figure 4-33: Potentiodynamic scans of aged thick alloys...... 97

Figure 4-34: Potentiodynamic scans comparing selected EB-PVD alloys to WE43...... 98

Figure 4-35: Pitting observed on a Mg-Y-Ti alloy exposed to HBSS for 40 minutes...... 99

Figure 4-36: Aged alloy T3 after potentiodynamic testing...... 100

Figure 4-37: Unaffected area of aged alloy T3 after potentiodynamic testing ...... 101

Figure 4-38: EB-PVD Mg alloy surfaces after linear polarization and EIS testing...... 102

Figure 4-39: Thick alloy surfaces after being submerged in 37ºC HBSS for 16 hours ...... 103

Figure 4-40: A549 cells grown on the glass control specimen ...... 105

Figure 4-41: A549 cells grown on the glass WE43 control specimen...... 106

Figure 4-42: A549 cells grown on WE43 ...... 107

Figure 4-43: A549 cells grown on fresh thick alloy T15...... 107

Figure 4-44: A549 cells grown on aged alloy Mg98 Ti2...... 108

Figure 4-45: Proliferation and viability of A549 cells grown on aged alloy Mg89 Y9 Ti2 ...109

Figure 4-46: Attachment and morphology of A549 cells grown on aged alloy Mg86 Y11 Ti3...... 109

Figure 5-1: Surface defects correlated with corrosion rate ...... 113

Figure 5-2: Striations on the rear surface of a thick EB-PVD alloy ...... 114

Figure 5-3: Pitting on EB-PVD Mg-Y-Ti and Mg-Ti EB-PVD alloys...... 115

Figure 5-4: Effects of substrate rotation on microstructure and surface morphology ...... 116

Figure 5-5: Sculptured magnesium deposited at an oblique angle to the incoming vapor flux ...... 117

Figure 5-6: Changes in thick alloy microstructure with growth ...... 118

ix Figure 5-7: Comparison of surface morphologies of EB-PVD alloys ...... 120

Figure 5-8: XRD scans of EB-PVD alloys showing preferential (002) orientation...... 121

Figure 5-9: Surface morphologies of fresh and aged thin Mg-Y-Ti alloys...... 128

Figure 5-10: Surface morphologies of fresh and aged thick Mg-Y-Ti alloys...... 128

Figure 5-11: EIS curves comparing both fresh and aged thin and thick EB-PVD Mg...... 130

Figure 5-12: Comparison of A549 cells grown on Mg alloys ...... 132

x LIST OF TABLES

Table 2-1: Summary of findings of the PROGRESS-AMS study...... 9

Table 2-2: Comparison of mechanical properties of natural bone with Mg and Ti alloy materials...... 11

Table 2-3: ASTM designation for alloying elements in magnesium...... 21

Table 2-4: Considerations for choosing alloying elements likely to impart passivity to Mg...... 26

Table 2-5: Solubility data for binary magnesium alloys...... 31

Table 3-1: Composition of 99.98% pure Mg ingot used to produce EB-PVD alloys (%)...... 39

Table 3-2: Deposition variables for Mg alloys deposited with the dual-gun EB-PVD system ...... 39

Table 3-3: Composition of Ti and Y used to produce thick alloys ...... 41

Table 3-4: Sample calculation to convert ICP results into alloy composition...... 45

Table 3-5: Composition of HBSS and human blood plasma ...... 51

Table 3-6: Composition of Magnesium Elektron WE43 ...... 53

Table 3-7: Sample calculation for equivalent weight, density, and volume fraction...... 58

Table 4-1: ICP-AES chemical analysis of selected EB-PVD magnesium alloys ...... 65

Table 4-2: ICP-AES chemical analysis of selected EB-PVD magnesium alloys ...... 65

Table 4-3: Thicknesses of selected EB-PVD magnesium alloys...... 66

Table 4-4: Corrosion rates for WE43 samples immersed in 37ºC HBSS for 1 week ...... 85

Table 4-5: Average OCPs of fresh and aged Mg alloys in 37ºC HBSS...... 88

Table 4-6: Average linear polarization corrosion rates of fresh and aged Mg alloys in 37ºC HBSS less than one hour after submersion ...... 89

Table 4-7: Average EIS corrosion rates of fresh and aged Mg alloys in 37ºC HBSS less than one hour after submersion...... 90

Table 4-8: Average OCP, Eb, passive region, and ipass for selected Mg alloys in 37ºC HBSS...... 98

xi Table 4-9: EDS analysis of the surface elements of alloy Mg89 Y9 Ti2 after potentiodynamic testing ...... 100

Table 5-1: Surface composition (at%) of alloys Mg98 Ti2, Mg89 Y9 Ti2, T3, and WE43 after potentiodynamic testing in 37ºC HBSS...... 123

Table 5-2: Composition (at%) of corrosion product on WE43 after 15 days in 37ºC HBSS compared with Ref. [123] ...... 123

Table 5-3: Electrochemical comparison of the best EB-PVD Mg alloys with WE43 ...... 125

xii ACKNOWLEDGEMENTS

First of all, I would like to thank my adviser, Dr. Barbara Shaw, for trusting me and supporting me throughout the course of my research. She has helped me realize the value of encouraging your employees (and graduate students!). The long hours would have been much longer without her support. Dr. Mark Horn was another instrumental figure in my academic and personal development. I am thankful for both his insight into experimental design and for his friendship. His warm smile and positive attitude constantly point me to the bigger joy in life.

Much thanks to Dr. Ela Sikora and Dr. Eliza Montgomery. Eliza taught me what questions to ask (and then helped me answer them). Ela waded through the data with me to identify patterns. I always left our conversations with more clarity and understanding than when I went into them. Her insight and encouragement are worth more than newspapers!

I would also like to thank Nick Pytel, both for his friendship and for helping me get my feet on the ground as I entered the research group. I have fond memories him with artificial sea water and army panels. Special thanks to Sean Pursel, who fabricated most if not all of the EB-

PVD alloys. Sean is the best I know at what he does. He helped me understand the “re-” aspect of re-search, and his level attitude encouraged me when experiments did not go as planned.

Thanks to Aaron Todd. He made me look forward to coming to lab—work is so much more enjoyable with friends. Our conversations provided a much needed break from research.

Scott Kralik of Engineering Science, Missy Hazen of the Huck Institute of Life Sciences,

Henry Gong and Nichole Wonderling of MCL, Dr. Wallace Greene of Hershey Medical Center, and Doug Wolfe of ARL provided training, materials, and equipment. Keith Stern and Jackie

Van Pelt helped obtain data. Johnson & Johnson and NSF Grant #0606294 provided funds.

This may sound crazy to some, but I owe everything, including the capacity to do this work, to Jesus Christ. You are my king, my rescuer, and my treasure. Soli Deo Gloria.

Chapter 1

Introduction

By now many Americans are aware that the number one killer of men and women in the

United States is not cancer, but coronary heart disease. Coronary heart disease, which is characterized by deficient blood flow to cardiac muscle and the surrounding tissue, kills 500,000

Americans and 7.2 million worldwide every year [1]. The most familiar form of the disease, atherosclerotic coronary heart disease, occurs when cholesterol infiltrates the coronary artery endothelium* and underlying connective tissue, where it accumulates with white blood cells to

form arterial plaque. The body triggers neointimal hyperplasia (smooth muscle cell proliferation)

to cover the plaque; however, the plaque can rupture the smooth muscle layer and initiate a

thrombus (blood clot) in the coronary artery, which in turn may obstruct blood flow and

commence events that to heart attack and possibly death. Other forms of coronary artery

disease also exist, and most are characterized by stenosis, or a narrowing of the arterial lumen.

One of the major breakthroughs in treatment of coronary stenosis came in 1977 when

Andreas Grüntzig [2] used percutaneous transluminal coronary angioplasty (PTCA) to treat an

obstructed coronary artery. Unfortunately, 50-60% of treated patients reported complications due

to restenosis. As the name suggests, re-stenosis is the process by which an arterial vessel narrows

again after medical treatment, and it has come to be used in association with the adverse effects of

both stent procedures and PTCA. Hunter [3] provides a general definition for restenosis as “a

* As a general note, the nature of this discussion requires biological and medical terminology that

may not be familiar to the reader. The author makes an effort to define terminology in the text when it is

first used, but Appendix A is provided as a general reference for subsequent uses.

2 complex biological cascade resulting in the overproduction of scar tissue in response to the injury

created by compressing the atherosclerotic plaque against the arterial wall during balloon

dilatation and stent placement.”

Stents are small metal cages used in conjunction with balloon angioplasty to open and

support diseased blood vessels. An example of an experimental magnesium stent is shown in

Figure 1-1, but most stents are made out of metals like titanium. The device is inserted into a

coronary artery through a process that begins with a small incision in a peripheral artery, usually

in the groin. A doctor then inserts a guidewire into the incision and uses x-ray imaging to thread

the wire up through the aorta to the heart. Next, the doctor slides a catheter tube over the wire up

to the heart, where a special dye is injected through the catheter to visualize the coronary arteries.

Figure 1-1: Biotronik’s experimental magnesium stent [4]. SEM microscopy reveals the detailed structure of Biotronik’s (Bülach, Switzerland) WE43 alloy, Magic coronary stent (prior to inflation).

The stent is radio-opaque (visible by x-rays) and crimped onto a balloon catheter, so the doctor can use a combination of dye and x-ray imaging to position the stent in the occluded vessel. Once in position, the balloon is inflated to open the stent, the balloon is deflated, and the

3 catheter is removed while leaving the stent behind. Sigwart [5] described use of the first coronary stent in 1987 as an attempt to treat restenosis of arteries opened with balloon angioplasty. Stents do decrease the occurrence of restenosis, but problems continue to persist. One prominent study

[6] found in-stent restenosis (ISR) in approximately 25% of patients treated with a bare metal stent (BMS).

1.1 Drug-Eluting Stents and Bioabsorbable Magnesium

As a response to ISR, companies like Boston Scientific and Johnson & Johnson (J&J) began developing the drug-eluting stent (DES), which is a metal stent coated with a drug designed to inhibit neointimal hyperplasia and prevent restenosis. Early studies fueled the hype surrounding them. A one-year follow-up study [7] involving the Cypher® stent showed that

16.6% of patients receiving a BMS required target lesion revascularization (TLR—the percentage of patients receiving a stent as a repeat procedure to open a stented vessel occluded by restenotic tissue), whereas only 4.1% of patents receiving a DES required TLR. A comparable study [8] involving the Taxus® stent revealed similar results: 15.1% for the BMS group and 4.4% for the

DES group.

Drug-eluting stents were hailed as the solution to ISR, but concerns are arising as clinical trials continue to reveal new results. Polymer coatings have been shown to provoke an increased inflammatory response [9] and contribute to late stent thrombosis [10], a condition involving a blood clot in the stented region long after the stent has been placed. Furthermore, drug-eluting stents reduce restenosis by inhibiting neointimal hyperplasia, which also delays the normal healing response beyond the 3-month period required for a BMS [11, 12]. Neointimal and endothelial revascularization prevent thrombogenic events by separating the stent and underlying tissue from the bloodstream [13], but the DES delays this process, so prolonged anti-platelet

4 regimens are recommended for persons receiving a DES [12]. These revelations, combined with the cost of a DES—$3000, about three times that of a BMS, which has implications for the patient as well as the healthcare industry [14-16]—have burst the initial euphoria associated with the effect of a DES on in-stent restenosis.

Several negative effects of the DES, however, also pertain to the BMS and, more generally, to permanent implants. Erne et al. [17] give several drawbacks inherent to permanent metallic stents: thrombogenicity, permanent physical irritation, chronic inflammatory local reactions, mismatches in vasomotion (mechanical behavior) between stented and nonstented areas, inability to adapt to growth, and nonpermissive or disadvantageous characteristics for later surgical revascularization. Bioabsorbable materials have been proposed as a way of circumventing the problems associated with permanent implants, particularly with stents [18].

Both metallic and polymeric materials have been proposed, but the biocompatibility and mechanical integrity of polymers are questionable. Therefore, metals, magnesium and its alloys in particular, are emerging as the most promising materials for bioabsorbable stents [17].

1.2 Objectives

This research is part of an ongoing study of vapor deposited magnesium alloys [19, 20], and it seeks to build on previous results by tailoring magnesium alloys for bioabsorbable implants. A host of issues remain to be explored with regard to bioabsorbable magnesium implants—mechanical properties, alloying compositions, electrochemical properties, surface characteristics, etc.—but this thesis will focus specifically on exploring nonequilibrium Mg alloy compositions to improve their corrosion resistance in physiological environments. Conventional alloying methods have failed to develop a magnesium alloy with the desired corrosion resistance, so electron beam physical vapor deposition (EB-PVD) will be employed to produce

5 nonequilibrium alloys. Potential variables include alloying elements, deposition rate, substrate temperature, substrate rotation, and alloy thickness. These variables affect alloy properties such as crystallization, second phase formation, film stress, porosity, surface morphology, and alloy chemistry. Consequently, they have an indirect effect on an alloy’s corrosion properties (such as formation and stability of the passive film at various potentials, rate of hydrogen evolution, etc.) and biocompatibility (cytotoxicity, effect on physiology, etc.).

A variety of techniques will be used to assess the relationship between alloy chemistry, corrosion rate, and the effect of magnesium corrosion on cells cultured in vitro. Crystallization and second phase formation will be evaluated using x-ray diffraction (XRD). Porosity and morphology can be explored using light microscopy and scanning electron microscopy (SEM).

Comparative corrosion rates, which are an indirect assessment of the effect of alloy chemistry and microstructure on corrosion, will be determined using electrochemical techniques. Precise alloy chemistries will be obtained using inductively coupled plasma-atomic emission spectrometry

(ICP-AES) analysis. In vitro cell cultures will be used to explore the effects of alloy corrosion on

A549 cell morphology and viability.

This research was funded in part by Johnson & Johnson and by the National Science

Foundation. Its goal was to identify a corrosion-resistant alloy chemistry that could be used to

produce bulk-like magnesium alloys for bioabsorbable implants, particularly stents.

6

Chapter 2

Background

Some background information is necessary in order to better understand the scope of the problem and the methods used to explore it. This work will begin with a look at the literature involving magnesium alloys as bioabsorbable implants (2.1), followed by magnesium corrosion

(2.2), alloying to improve corrosion resistance (2.3), and magnesium alloys produced via physical vapor deposition (2.4).

2.1 Magnesium Alloys as Bioabsorbable Implants

Whenever a material is going to come into contact with the human body, one must

consider its biocompatibility. Magnesium has thus been chosen as a candidate for bioabsorbable

implants because of its benefits to human physiology. Magnesium is the fourth most common

mineral salt in the human body and the second most common intracellular cation (Mg2+) [21].

The majority of physiological magnesium is located in bone, where it is stored as a reserve to

buffer extracellular Mg2+ concentration. If the concentration rises to a toxic level, the kidneys can

easily eliminate extra magnesium [22].

Toxicity, however, is generally not a concern since most Americans consume less than

the recommended daily allowance for magnesium (for men and women 31-50 years of age, 420

and 320 mg/day, respectively) [23]. Lower than normal dietary intake of Mg has been linked to

neural and neuromuscular hyperexcitabilty and physiologic damage due to oxidative stress [24],

as well as hypertension, cardiac arrhythmias, ischemic heart disease, atherosclerosis, and sudden

cardiac death [25]. The effects of magnesium on the heart are of particular interest. Studies have

7 shown that Mg acts as a systemic and coronary vasodilator to reduce blood pressure, and as an

anti-platelet-like element to reduce thrombus formation.

Moreover, Mg2+ plays an important role in genomic stability: it assists DNA replication, protein synthesis, DNA repair proteins, anti-oxidative mechanisms in the cell, cell cycle regulation, and apoptosis [26]. Many Americans would therefore benefit from increased magnesium consumption, and patients with heart problems may benefit as well. Intravenous magnesium was shown to reduce stent thrombosis in swine [27] and dogs [28], as well as arterial thrombus formation in rats [29]. Human studies document intravenous magnesium treatments for patients with acute myocardial infarction [30, 31] and nonacute coronary intervention with stent implantation [32]. The benefits observed in animals, however, were only present for humans in the study of the nonacute condition with stent implantation.

Experiments have shown that metals located toward the more active end of the electromotive series are much less likely to initiate a thrombus than metals located at the nobler end. Sawyer et al. [33] implanted tubes made of metals from various points on the electromotive series into canines’ descending thoracic aorta or thoracic inferior vena cava and reported that tubes made of Mg-3%Al-1%Zn alloys remained patent for periods up to 350 days. Conversely, stainless steel tubes showed complete thrombosis by 30 days from the time of insertion. The authors explained their results by noting that metals like magnesium lose positive ions in solution to take on a negative charge. This creates an electrical double layer with a positive charge in the solution, which mimics the natural state of the vascular interface and prevents blood components from transferring negative charge to the metal to initiate the coagulation cascade.

When the above factors are collectively considered, magnesium emerges as a strong candidate material for bioabsorbable implants. Clinical studies have begun to explore the use of

Mg alloy stents in animals and humans.

8 2.1.1 Cardiovascular Uses for Bioabsorbable Magnesium

Standard practice requires animal trials before biomaterials are used in humans.

Heublein et al. [17] drew interest to bioabsorbable magnesium in 2003 when they implanted stents made of the alloy Mg-2%Al-1%rare earths into coronary arteries of 11 domestic pigs.

DiMario et al. [34] implanted stents made of magnesium alloy WE43 (4% Y, 3% rare earths, Zr) into the coronary arteries of 33 minipigs, and Waksman et al. [35] implanted the WE43 stent into the coronary arteries of 11 juvenile domestic crossbred swine and 6 Gottinger minipigs. Results were favorable, so testing proceeded to clinical studies.

Biotronik’s absorbable metal stent (AMS)—which is similar to if not the same as the

WE43 stent used in animal trials—was first used in 2003-2004 to treat 20 patients ages 59-96

with critical limb ischemia in areas below the knee. Bosiers et al. [36] reported a 94.7% limb

salvage rate, no toxic effects, and 3 thrombus observations at a 12 month follow-up. The Mg

alloy stents degraded almost completely by 10 weeks. Energy dispersive x-ray spectroscopy

(EDS) identified the main constituents of the degradation products as Ca and P, which indicated

that the corrosion product contained phosphate and hydroxyl apatite as predicted by

theoretical models. The authors concluded that the results, while not statistically significant,

indicated that the Mg alloy stent may be valuable in clinical practice.

The PROGRESS-AMS (Clinical Performance and Angiographic Results of Coronary

Stenting with Absorbable Metal Stents) study [34], involving 63 patients, was initiated in 2004 in

Essen, Germany, to evaluate the Biotronik stent in human coronary arteries. Multiple vignettes

[37-39] narrated the evolution of the study, but the complete findings [40] were only published in

2007. These findings are summarized in Table 2-1. Sixty-three patients were implanted with one

or more AMS without myocardial infarction, subacute or late thrombosis, or death for at least one

year after the operation. The stents degraded completely prior to the 4 month follow-up. No

9 ischaemic episodes were reported, which suggests rapid re-endothelialization and absence of

embolism—that is, the stent corroded within the vessel wall. However, the reported TLR (re-

intervention) 1-year rates were higher for the absorbable metal stent (45%) than a bare metal stent

(28%) or drug-eluting stent (6%). The PROGRESS-AMS study showed that the AMS is both

safe and feasible, but neointimal growth and negative remodeling continue to contribute to

restenosis.

Table 2-1: Summary of findings of the PROGRESS-AMS study. Patients n = 63 Age in years, mean (SD*) 61.3 (9.5) Avg. number of stents per patient, mean (SD*) 1.13 (0.34) 4-month TLR 39.7 % 12-month cumulative TLR (n=60) 45 % 12-month cumulative stent thrombosis (n=60) 0 % 12-month cumulative death (n=60) 0 % *SD = standard deviation. Adapted from Ref. [40].

While the PROGRESS-AMS study examined the efficacy of AMSs in adults, the AMS may prove more useful for niche populations like infants or children. Conventional stents are prohibited in such patients, because conventional stents would become occlusions in growing blood vessels. Bioabsorbable stents, however, would degrade before becoming a problem. To test its efficacy, the AMS stent has been implanted in preterm babies and children born with heart defects [41-43], and the results are generally favorable. The stents restored perfusion, and vessels demonstrated growth after the stents were implanted. At no point during observation did physicians observe serum levels of magnesium above 2.5 mmol/l, which constitutes a pathological condition. Unfortunately, some stented vessels exhibited a reduced diameter after a few weeks, presumably due to stent degradation.

10 A few [44-46] authors have reviewed the literature and commented that the Mg stent degradation process still needs to be refined, but one cannot deny the promise of bioabsorbable magnesium stents, especially in niche populations like children. Overall, the clinical results suggest that bioabsorbable Mg stents should be more corrosion resistant so that they maintain their mechanical integrity for a longer period of time.

2.1.2 Orthopedic Uses for Bioabsorbable Magnesium

Magnesium alloys are also being considered for orthopedic implants. Traditional procedures favor the use of corrosion resistant metals such as stainless steel, alloys, or titanium alloys [47], but some orthopedic implants are only temporary. As a result, implants are removed with subsequent surgeries once the bones heal. These surgeries create extra expenditures for the health care industry and unnecessary morbidity for the patient.

Bioabsorbable magnesium is consequently being evaluated for use in orthopedic implants such as tissue scaffolds for bone and cartilage repair [48, 49] and bone screws for severe fractures [50].

In vivo studies report inconclusive results on the performance of orthopedic Mg alloys.

One study [51] reported brief formation of gas bubbles around magnesium rods shortly after implantation in guinea pig femurs, but others report that magnesium benefits bone mass surrounding an implant [51], bone remodeling [52], and stress shielding [53]. Permanent metallic implants possess mechanical properties that greatly exceed those of natural bone (see Table 2-2), so they shield bones from the cyclic stresses that promote bone maintenance and deposition [54].

However, magnesium’s mechanical properties are similar to those of natural bone, so its implants are expected to reduce stress shielding and promote healing in fractured and broken bones. Still, the literature [52, 55, 56] generally agrees that the corrosion resistance of orthopedic Mg implants

11 needs to be improved they are ready for clinical use. Such conclusions agree with the outcomes of the trials involving magnesium stents.

Table 2-2: Comparison of mechanical properties of natural bone with Mg and Ti alloy materials. Property Natural Bone Magnesium Ti alloy Density (g/cm3) 1.8-2.1 1.74-2.0 4.4-4.5 Elastic Modulus (GPa) 3-20 41-45 110-117 Compressive yield strength (MPa) 130-180 65-100 758-1117 1/2 Fracture toughness (MPam ) 3-6 15-40 55-115 Adapted from Ref. [53].

2.1.3 Cytotoxicity Experiments with Magnesium

The human body is incredibly complex, and its response to foreign materials, especially on the nano-scale, is often unpredictable. In vitro tests, that is, tests in a simulated physiological environment (literally, “within the glass”), are often used to evaluate the potential effects of a biomaterial on the host organism before it is actually implanted. As mentioned in preceding sections, the corrosion resistance of bioabsorbable magnesium needs to be improved for wide- spread use of magnesium biomaterials. In vitro tests may thus provide a means for assessing the cytotoxicity of corrosion reactions and alloying additions on the tissue and cells in the immediate vicinity of the reactions. Decreases in cell vitality and viability are indicators of cytotoxicity.

Several studies have already reported the results of in vitro cytotoxicity tests with magnesium. Stromal cells were cultured on Mg substrates to evaluate magnesium’s potential for orthopedic applications [55], and results showed that Mg-based substrates supported adhesion, differentiation, and growth of these cells. Bone-like matrix was also produced. The authors concluded that bone cells grown on magnesium substrates may provide a method to screen and assess the biological activity of Mg biomaterials.

12 In vitro screening may hence be used to evaluate the local toxicity of magnesium implants. As of 2006, the literature [57] had not documented any experiments reporting toxicity when high concentrations of magnesium salts were incubated with human vascular endothelial cells (EC) or smooth muscle cells (SMC). Studies have shown that low levels of Mg promote only slight dysfunction in cultured human umbilical vein ECs [58], with the cells remaining unharmed at magnesium concentrations up to 10 mM [59].

The complexity of the human body means that the results of in vitro tests may not exactly

replicate those done in vivo (literally, “in the living”). Still, in vitro tests, particularly those done

with cell cultures, are an irreplaceable part of material development. These tests will help predict

the biological response to corrosion resistant Mg alloys.

2.2 Corrosion of Magnesium

At present, in vitro cell tests and in vivo tests in animals and humans show that

conventional magnesium alloys are unable to provide the necessary corrosion resistance for the

ideal performance of bioabsorbable magnesium, particularly with regard to stents. These

implants require new alloys and new alloying techniques, but before exploring such issues, the

corrosion of magnesium must first be discussed.

Magnesium corrosion is of notable interest due to magnesium’s strong thermodynamic

tendency to oxidize. Among metals commonly used in engineering, magnesium exhibits the

lowest standard potential: Perrault [60] calculated the Mg/Mg2+ standard potential as -2.37 V vs.

NHE, but magnesium’s actual corrosion potential in dilute chloride solutions is closer to -

1.7 V vs. NHE [61]. The difference between the standard potential and corrosion potential is

attributed to the formation of an oxide film (likely Mg(OH)2) on the metal surface [62]. ASM

13 International [63-66] has amassed a considerable amount of information regarding the corrosion of magnesium, and others [61, 62, 67] have also published useful reviews.

The Pourbaix diagram provides a means for visualizing the effects of potential and pH on the thermodynamic regions of magnesium corrosion and stability. Figure 2-1, which was developed by Perrault [68], shows the potential-pH diagram for magnesium and water. This diagram more accurately accounts for the open circuit potential (OCP) of the magnesium anode when compared to Pourbaix’s original diagram [62], because Perrault used thermodynamic data

+ for magnesium hydrides (MgH or MgH2) and the monovalent magnesium ion (Mg ), which was not available to Pourbaix. This additional information led Perrault to conclude that a Mg anode cannot exist in thermodynamic equilibrium with water. However, equilibrium exists between

MgH2 and the oxidized forms of Mg when a hydrogen overpotential of 1 V is applied on the

electrode [60].

Figure 2-1: Perrault’s Pourbaix diagram for magnesium [68]. Perrault developed a more comprehensive theoretical potential-pH (Pourbaix) diagram for the magnesium-water system at 25ºC with a hydrogen overpotential of 1V.

14

Figure 2-2 is a more general diagram of the magnesium-water system, and it highlights the theoretical areas of corrosion, passivation, and immunity [66]. At physiological pH = 7.4, magnesium’s corrosion potential corresponds to the region where hydrogen is stable. Therefore,

Mg corrodes in physiological environments to produce magnesium ions (Mg+/Mg2+) and

hydrogen gas, which is shown by reaction (2-3) below. Hydrogen evolution is the dominant

reduction reaction, so dissolved oxygen does not play a significant role in Mg corrosion [62].

Corrosion is strongly deterred above pH = 11, the equilibrium pH for Mg(OH)2, which is assumed to be the major constituent of the passive film. Nevertheless, Song et al. [69] found that passive film’s formation kinetics may exceed its dissolution kinetics, so a thin layer may also be present in relatively acidic environments (pH 3-11).

Figure 2-2: Theoretical domains of magnesium’s Pourbaix diagram [66]. The potential-pH (Pourbaix) diagram for the magnesium-water system at 25ºC shows the theoretical domains of corrosion, passivation, and immunity.

15 Unfortunately, the nature of magnesium corrosion and the formation of its passive film are not thoroughly understood. Secondary reactions are likely involved, especially those

+ involving MgH2 [68] and short-lived Mg ions [70]. The general anodic and cathodic reactions

occurring on a magnesium anode can be expressed by reactions (2-1) and (2-2), respectively. The

net reaction (2-3) is the sum of reactions (2-1) and (2-2).

2eMg Mg + +→ 2eMg -2 (2-1)

- - 2 2eO2H 2 +→+ )2(OHH (2-2)

2O2HMg →+ Mg(OH) + H 22 (2-3)

2.2.1 Nature of the Passive Film

Metals almost universally rely on passive films to deter corrosion. A good passive film restricts the outward flow of cations and the inward flow of damaging anions or oxidants, and it rapidly repairs itself in cases of localized breakdown [62]. Non-crystalline films appear to exhibit these characteristics more than crystalline films, although the Mg(OH)2 film is generally

considered to be crystalline [64]. In addition, Mg(OH)2 can undergo basal cleavage to cause

cracking and curling of the film [71], and its lower density compared to magnesium metal may

cause compressive ruptures [72]. Therefore, magnesium’s passive film has been labeled quasi-

passive or partially-passive.

Nordlien et al. [73, 74] have used TEM and XPS analyses to explore the growth and

nature of the film in ambient air and distilled water. When the authors immersed magnesium in

water for 48 hours, they found a three-layer oxide film on the metal. The cellular-structured

underlying layer was 0.4-0.6 μm thick; the dense middle layer was 20-40 nm thick; and the

platelet-like upper layer was 1.8-2.2 μm thick. Their results indicate that the middle layer, a

16 dense oxide composed of MgO and Mg(OH)2, forms on Mg in ambient air, but this layer is permeable to both water and ionic species. A hydrated oxide layer, most likely Mg(OH)2, can

therefore form underneath the dense layer as water diffuses to the metal surface. This underlying

layer appears to be responsible for magnesium’s corrosion resistance—Muakami and Sato [75]

controversially reported that Mg(OH)2 is an insulator—but adhesion between the two layers seems to be weak, so the oxide may rupture easily. Furthermore, a third, platelet-like layer forms on the surface of the dense oxide when Mg is immersed in water. The authors suspect that this layer forms as Mg2+ diffuses from the underlying layer to the bulk of the solution where it forms a

Mg(OH)2 precipitate.

Other research indicates that solution chemistry has the potential to change the nature of

magnesium’s passive film. Huber [76] found that MgO is unstable in the presence of water and

- quickly replaced by Mg(OH)2. Anions such as chloride (Cl ), which are a major constituent of

many physiological fluids, induce local breakdown [77], and XRD studies [62] of Mg in chloride-

containing solutions report the existence of MgCl2•6H2O, Mg3(OH)5Cl•4H2O, and

5(Mg(OH)2)•MgCl2 in addition to Mg(OH)2 within the passive film . Hara [78] reported a stable film composed mainly of Mg(OH)2 in 0.1 M NaCl, but this film was observed to suffer local

breakdown under anodic polarization.

Magnesium’s passive film provides relatively poor protection for the underlying metal.

Therefore, this work will investigate alloying and processing approaches to create passive films

with improved corrosion resistance.

2.2.2 Impurities and Tolerance Limits

In addition to local breakdown of magnesium’s quasi-passive film, Mg corrosion may also accelerate due to the action of harmful metallurgical impurities in the magnesium metal. In

17 one of the most well known studies of magnesium, Hanawalt et al. [79] described the effects of fourteen impurity elements on the corrosion of Mg in 3% NaCl solution. Their results supported the existence of tolerance limits—values above which impurity concentrations dramatically increase the corrosion rate. Figure 2-3 shows Makar and Kruger’s [62] revision of Hanawalt and coworkers’ diagram depicting the effects of alloying additions on the corrosion rate of Mg-X alloys. The figure reveals that small amounts of Fe (0.017%), Ni (< 0.0005%), or Cu (0.1%) will dramatically increase magnesium’s rate of corrosion. Binary additions of Na, Si, lead, and Sn have a negligible effect on the corrosion rate at concentrations below a few percent, but interaction exists for combinations Si or Pb with Fe. Hanawalt et al. also discovered that additions of approximately 1% Mn or Zn increased magnesium’s tolerance limits for Fe, Ni, and

Cu, and they acted to decrease its corrosion rate when impurities exceeded their tolerance limits.

Figure 2-3: Tolerance limits for binary magnesium alloys [62]. Alloying elements can dramatically affect the corrosion behavior of binary magnesium alloys in 3% NaCl solution.

18 Subsequent experiments led the authors to conclude that harmful impurity elements seem to act as cathodic sites for hydrogen evolution [79]. As previously mentioned, magnesium’s low standard potential indicates a large thermodynamic propensity for corrosion, so micro-galvanic corrosion may occur with impurity-containing phases in the magnesium. Indeed, when the authors [79] calculated the difference between the solution potentials of Mg and the impurity material and then subtracted the hydrogen overvoltage of the impurity material, the driving forces for hydrogen evolution from Ni, Co, Fe, and Cu were among the highest reported.

Interestingly, the authors [79] obtained high-purity Mg for their experiments by evaporating commercially pure Mg to distill the impurities. Others [80, 81] have also evaporated commercially pure Mg to achieve remarkable increases in corrosion resistance. Clearly then, magnesium’s purity and hence corrosion resistance can be improved with novel processing methods such as vapor deposition.

2.2.3 The Negative Difference Effect

The Negative Difference Effect (NDE) is an interesting phenomenon exhibited by

magnesium and some of its alloys. Corrosion requires both anodic and cathodic reactions, and according to electrochemical theory, the cathodic reaction rate should decrease and the anodic reaction rate should increase as potential moves in the noble direction. Magnesium, however, does not obey the theory. Increasing potential often to increasing hydrogen evolution and larger-than-predicted rates of metal dissolution, which complicates electrochemical analyses.

Tunold [77] summarized four commonly proposed mechanisms for the NDE: (a)

breakdown of the quasi-passive film during anodic dissolution; (b) undermining of metallic particles, especially at higher potentials or anodic current densities; (c) formation of intermediate monovalent magnesium ions; and (d) formation of intermediate magnesium hydride. However, at

19 potentials and current densities that do not indicate rapid corrosion, one may still use electrochemical methods to obtain relative data for Mg and its alloys. That is, electrochemical results may not be “accurate,” but they can be used to make relative comparisons.

In a study of thin samples of aluminum, which possesses a more stable passive film but otherwise behaves similarly to magnesium, Frankel and Akiyama [82, 83] found that the ratio between the anodic dissolution current density from pits and crevices and the hydrogen evolution current density is relatively constant over a range of potentials during electrochemical testing.

Part of the cathodic reaction may occur on the working electrode—that is, some electrons may be consumed in pits as hydrogen gas evolves—so electrochemical results may underestimate the true corrosion rate. The constant ratio between current densities, however, means that electrochemical values, while less than accurate, are still useful for relative comparison. In personal communication, Frankel suggested that this ratio, if also valid for Mg (which is likely), will be higher for Mg than Al.

Since the NDE is generally associated with large rates of corrosion, it could be prevented by enhancing magnesium’s passive film [61]. Vapor deposition can accomplish this goal by improving the purity of Mg metal, by creating a more uniform distribution of impurities within the as-deposited alloy, and by incorporating alloying elements to improve the passive film.

Alloying will be discussed next in Section 2.3.

2.3 Alloying to Improve Corrosion Resistance

Magnesium alloys are an area of intense study, and the effects of alloying elements have been compiled by several sources [84-86]. However, before discussing their benefits to corrosion resistance, I should note that alloying elements can also be used to modify magnesium’s mechanical properties [85]. Aluminum is well known to increase magnesium’s strength and

20 hardness, and a content of 6% yields an ideal combination of strength and ductility [85]. It is often used in conjunction with Zn to improve strength. Both rare earth (RE) metals and Y improve tensile and creep properties at elevated temperatures, and Th is known to contribute similar benefits. Zirconium contributes a powerful grain-refining effect to Mg alloys, but it cannot be added to cast alloys containing Al or Mn, because the elements react to form precipitates. The choice of alloying elements strongly impacts magnesium’s mechanical properties; this section, however, will focus largely on the benefits that these elements confer to its resistance to corrosion.

Magnesium alloys can contain a number of elemental additions. As a result, ASTM

International (formerly known as the American Society for Testing and Materials) has created a system [87] for naming Mg alloys in order to standardize nomenclature and avoid confusion.

The system employs two letters followed by two numbers. The first letter specifies the most abundant alloying addition, and the second letter specifies the second-most abundant alloying addition. These letter codes are presented in Table 2-3. As previously mentioned, two numbers follow the two letters, and the numbers specify the respective weight percentages of the primary alloying elements. These numbers are whole numbers. The rules for rounding to obtain these numbers and the maximum impurity contents for a specific designation can be found in ASTM

B951. Letters are also assigned to the end of the alloy designation to distinguish between different alloys that meet the same specifications. The alloy designation may also be followed by an additional label, which refers to types of heat treatment.

21

Table 2-3: ASTM designation for alloying elements in magnesium. A—Aluminum M— C— Q—Silver E—Rare earths S—Silicon H—Thorium T—Tin J— V—Gadolinium K—Zirconium W—Yttrium L—Lithium Z— Adapted from Ref. [87].

Here is an example: magnesium alloy WE43 is composed of approximately 4% yttrium

and 3% rare earths, and its suffix letter B refers to its specific variation in composition within the

range allowed for the WE43 designation. For the sake of simplicity, future references to

magnesium alloys will be made using their ASTM designation.

2.3.1 Electrochemical and Atomic Considerations for Magnesium Alloys

Alloying elements have the potential to improve or exacerbate magnesium’s resistance to corrosion, so one must consider a number of factors when choosing them. Previous work [19] highlights the importance of electrochemical and atomic considerations, such as an element’s passivation characteristics, open circuit (or standard) potential, electronegativity, hydrogen exchange current density, and density, as well as secondary concerns such as atomic size, crystal structure, and oxide density, some of which will be discussed in detail here. Polmear [84] reports that alloying additions of Al, Mn, REs, Y, and Zn improve magnesium’s resistance to corrosion, so this section will focus on the benefits and shortcomings of these elements. Titanium will also be discussed.

An element’s ability to passivate in aqueous solutions can be ascertained, at least approximately, from its tendency to form a stable oxide or hydroxide as shown in a Pourbaix

22 diagram. Magnesium alloy WE43 was used in clinical trials with some success [2], so the

Pourbaix diagrams [88-90] for its alloying elements are shown in Figure 2-4. WE43, which is

composed of 3.7-4.3% Y, 2.0-2.5% Nd, 1.9% max other REs, (Gd, Dy, Er, Yb), and 0.4% min Zr,

is one of the most corrosion resistant magnesium alloys. Its corrosion rate is typically 25 mils per

year (mpy) under conditions specified by ASTM B117 salt fog test, but the rate decreases to

approximately 20 mpy when immersed in 3% NaCl [91].

Yttrium, the primary alloying element in WE43, forms a stable hydroxide within the

range of magnesium’s corrosion potential and above pH ~ 8. The REs also form stable

hydroxides under these conditions. Furthermore, the dissolution of Mg and formation of

Mg(OH)2 often shifts the surface pH of magnesium toward more alkaline values, which according to the Pourbaix diagrams should provide further stabilization for hydroxide films. Zirconium exhibits a passive region in both basic and acidic pH ranges. Aluminum, whose Pourbaix diagram is available in the literature [92], exhibits passivity for pH 4 to 9. Manganese’s Pourbaix diagram is not considered, because it improves corrosion resistance by removing harmful impurity elements such as Fe into intermetallic compounds, which may precipitate from the melt or remain in the alloy in a more harmless state [85].

Titanium is another potentially beneficial addition to magnesium alloys, but material

properties prohibit its use in cast alloys. Titanium melts at 1667ºC, which is approximately

550ºC higher than magnesium’s boiling point (1090ºC) [93]. As a result, Ti is not a viable

addition to magnesium melts, but it can be incorporated into vapor deposited alloys. Titanium’s

Pourbaix diagram is therefore presented in Figure 2-5 [94]. Note that its passive region varies

with both potential and pH—the region extends above pH ~ 5 for potentials near magnesium’s

OCP. Titanium in combination with yttrium has been shown to improve the corrosion behavior

of Mg alloys when compared to commercial alloys such as WE43 [95].

23

Figure 2-4: Pourbaix diagrams for yttrium, neodymium, and zirconium [88-90]. The Pourbaix diagrams for these element-water systems at 25º show their regions of passivity.

24

Figure 2-5: Pourbaix diagram for titanium [94]. Titanium’s potential-pH (Pourbaix) diagram at 25ºC shows that its region of passivity varies with potential and pH.

In addition to passivity, one must consider an element’s open circuit (or standard) potential, because elements with noble potentials may ennoble the open circuit potential of an Mg alloy. This, in turn, reduces its thermodynamic propensity for corrosion. Care should be taken, however, because extremely noble elements in conventional alloys will promote microgalvanic corrosion among compositional inhomogeneities. Microgalvanic corrosion may also result when intermetallic compounds form and act as cathodic sites for hydrogen reduction [96], but compositional inhomogeneities and intermetallic compounds are less of a concern for vapor-

25 deposited materials. Still, one should consider an element’s electronegativity, since intermetallic compounds are more likely to form from elements with significantly different electronegativities.

Magnesium alloys can also benefit from alloying elements that support low hydrogen exchange current densities, especially since the goal of this research is to produce alloys for use in the human body. Figure 2-6 [97] shows a representation of the periodic table and the hydrogen exchange current densities supported by various elements. Elements such as Fe, which decrease magnesium’s corrosion resistance, support a higher hydrogen exchange current density than elements such as Y, which increases magnesium’s corrosion resistance.

Figure 2-6: Hydrogen exchange current densities of various elements [97]. Hydrogen overpotential data was available for elements in bold type.

Other elemental factors such as density, atomic size, and crystal structure may also impact an alloy’s resistance to corrosion. Wolfe [19] compiled information on these and other elemental properties to compare the most promising alloying elements for Mg, which is recreated in part in Table 2-4. (Ce) is included as a representative of the RE elements. The hydrogen evolution rate (HRE) exchange current densities, io, were obtained from Figure 2-6

[97], and passivity ranges were estimated from the Pourbaix diagrams [66, 88-90, 92, 94]. One

26 can see that Al, Ti, and Y exhibit properties similar to those of Mg, so these elements appear to be the most promising additions that will improve magnesium’s corrosion resistance.

Table 2-4: Considerations for choosing alloying elements likely to impart passivity to Mg.

Eº Electronegativity ρ Crystal Atomic HER io Passivity -3 -2 (VSHE) (Pauling Scale) (g cm ) Structure Radius (Å) (A cm ) Range (pH) Mg -2.4 1.31 1.74 hexagonal 1.72 10-9 pH > 12 Al -1.7 1.61 2.70 fcc 1.82 10-9 4 < pH < 9 Ce -2.5 1.12 6.77 fcc 2.70 10-12 pH > 9 Ti -1.6 1.54 4.54 hexagonal 2.00 10-12 pH > 3 Y -2.4 1.22 4.47 hexagonal 2.27 10-12 pH > 8 -12 Zr -1.5 1.33 6.51 hexagonal 2.16 10 4 < pH < 13 Adapted from Ref. [19].

2.3.2 Physiological Considerations for Magnesium Alloys

Aluminum, yttrium, and titanium appear to be the most promising additions to Mg alloys,

but factors beyond electrochemical and atomic properties must also be considered when

designing alloys for use in the human body. Alloying elements should be nontoxic or at least

relatively nontoxic to living tissue and physiological processes, so this subsection will discuss the

toxicity of Al, Y, and Ti.

Some reports [98, 99] indicate that Al is toxic to the brain and contributes to the onset of

Alzheimer’s disease. This is clearly undesirable. Moreover, Al seems to affect parathyroid

hormone levels and interfere with osteoblast activity and hematopoietic tissue in bones [100], and

Mg-Al alloys were found to deter growth and protein expression in human smooth muscle cells

(SMCs) [101] and aortic endothelial cells (ECs) [102] when cultured in vitro. The transport and

excretion of Al are areas of active research, but at present the mechanisms of Al-toxicity are still

being explored. Therefore, aluminum is a questionable element for use in bioabsorbable

materials.

27 Yttrium may also be questionable, but for different reasons—little is known about its

potential toxic effects on the human body. Rare earths (with which Y is often grouped because of

chemical similarities) have only recently become prominent materials, so research on their

toxicity is developing. Yttrium hydrolyzes at physiological pH, and its solubility in incubated

solutions was shown to be influenced by the presence or absence of various ionic species [103].

Hirano et al. [104] intravenously injected 0.2 mg YCl3 into each of several rats and found that Y appeared in the plasma as a colloid with proteins and some minerals (including Ca), but it was

then taken up by phagocytic cells in the liver and spleen. Large amounts of Ca were deposited in

a rat’s liver and spleen when 1 mg YCl3 was intravenously injected, and the liver was found to

clear Y at a half-time of 144 days. A report on the oral consumption of Y [105] found that mice

fed 0.3 μg Y / g dry weight of food exhibited lessened growth but longer lifespan than control

mice. In vitro tests showed that a Mg-Y-RE alloy incubated with cardiovascular SMCs and ECs

inhibited SMC proliferation by 80% and SMC viability by 20%, but ECs remained viable and

showed proliferation [106]. Another study [107] indicated that fibroblasts and osteoblasts grown

on a Mg-1%Y alloy showed decreased viability. However, human SMCs incubated with 50

μg/mL YCl3 exhibited normal proliferation but upregulated inflammatory genes [108].

The contradictory results may arise from the nature of Y when it is introduced to the physiological environment. A study of metal compounds [109], including Ti, found that immunotoxicity depends on the speciation of the metal. Trivalent compounds were found to be more toxic than any other arsenic compounds, and only some Ti compounds exhibited immunotoxicity. Titanium toxicity, however, is generally studied to determine the effects of implant wear debris [110], and in vitro results often show absent or low toxicity [111] and upregulation of inflammatory genes [112].

In general, then, Y and Ti appear less toxic to a physiological environment than Al, although Y toxicity should be studied further. Still, the PROGRESS-AMS study [40] did not

28 report any harmful effects from yttrium in the stent material, and Song [113] reports that low levels of REs can be tolerated by the human body. As a result, the amount of Y in a bioabsorbable Mg alloy will likely be nontoxic to the human body. Therefore, when electrochemical, atomic, and physiological factors are considered together, Y and Ti stand out as the most promising additions to bioabsorbable magnesium alloys.

2.3.3 Electrochemical Studies of Magnesium Alloys in Simulated Body Fluid

As discussed previously, in vitro tests are a necessary component of biomaterials

research, because they provide insight into materials’ performance in a physiological

environment. Witte et al. [114] compared the corrosion of AZ91D and LAE442 using in vitro

electrochemical potentiodynamic tests in a borax-phosphate buffer (pH=7.0 at room temp.) and in vivo tests in guinea pig femurs. They used Tafel extrapolation as a means to calculate in vitro corrosion rates that were four orders of magnitude larger than in vivo corrosion rates.

Consequently, they concluded that in vitro tests do not directly simulate the in vivo environment.

This is true. However, researchers can use in vitro analyses to study magnesium’s electrochemical corrosion mechanisms and develop materials that exhibit superior corrosion resistance.

A few electrochemical studies have explored the performance of pure Mg in simulated

2+ - - 2- body fluid (SBF), where the presence of ions like Ca , HCO3 , H2PO4 , and SO4 have the potential to change magnesium’s corrosion behavior from that observed in NaCl solutions [113].

Song and Song [115] used EIS to report that the corrosion rate of 99.96% pure Mg increased with time when immersed in Hank’s SBF, while Wang et al. [116] used mass-loss experiments and found that the corrosion rate of 99.9% pure Mg decreased with time in a similar SBF. Both agree

29 that commercially pure Mg does not appear to possess the corrosion resistance required for bioabsorbable implants.

Some have proposed coatings as a means to slow magnesium corrosion [117], but the mechanical deformation required for stent deployment does not seem to favor such methods.

Consequently, alloying appears to be the most promising way to reduce the corrosion rate of magnesium. Kuwahara et al. [118] used mass loss to compare AZ31 and AZ91 alloys in Hank’s solutions with and without Ca and Mg, and they found that specimens of AZ31 corroded completely after 4 days, whereas specimens of AZ91 were more corrosion resistant. Their results are problematic for bio-applications, since AZ91 contains more Al. Nevertheless, most research on AZ alloys for bioabsorbable Mg seems to favor the AZ91 composition [119-121]. Xin et al.

[120] used EIS to analyze the corrosion of AZ91 in SBF and found that corrosion rates and hydrogen evolution were initially high, but they decreased and stabilized after two days. They also found evidence of calcium and magnesium phosphates in the corrosion product, besides magnesium oxide. Liu et al. [121] implanted Ti ions in AZ91 and found evidence of TiO and

TiO2 in the surface oxide via XPS. They discovered that Ti shifts the OCP to a more positive

value and improves the corrosion resistance at OCP.

Until recently, however, not much electrochemical work has been published regarding

Mg-Y alloys. Rettig and Virtanen [122, 123] seem to have pioneered electrochemical studies on

the corrosion processes of WE43 in SBF. EIS data showed that corrosion mechanisms differed

depending on SBF chemistry, and the mechanisms were significantly different when compared to

those observed for WE43 in 0.15 M NaCl solution [122]. They found that the dissolution rates of

WE43 were much higher in the SBFs than in NaCl. Differences in the dissolution rates were

attributed to the buffering ions in SBF, which stabilized the solution pH near sample surfaces.

Rettig and Virtanen [123] also found that the composition of the corrosion products

differs in NaCl and SBFs. Crystalline Mg(OH)2 was the primary component of WE43’s

30 corrosion layer in NaCl, but amorphous (Mg, Ca)x(PO4)y(CO3)z(OH)i formed in SBFs. Moreover, calcium was only found in the corrosion product if phosphate was also present in the SBF, but phosphate was found in the product regardless of the presence of calcium. Corrosion layers were observed to reach their final composition after a few hours and remain nearly constant thereafter.

Test sample surfaces were generally smooth with areas of localized corrosion.

These results indicated that the mechanisms of physiological Mg alloy corrosion will

likely be different than those reported for simple NaCl solution. Accordingly, electrochemical

tests will be used as part of the research presented in this thesis.

2.4 Vapor Deposited Magnesium Alloys

Physical vapor deposition (PVD) is best known for its use by microelectronics to create integrated circuits, but it can also be used to create nonequilibrium materials with structured surfaces and morphologies. As previously mentioned, magnesium boils at 1090ºC [93], so elements with higher melting points, such as Ti, cannot be easily incorporated into Mg alloys with conventional processing techniques. In addition, some elements have a low solid solubility in Mg [84], which restricts alloy chemistry. Table 2-5 lists some potential alloying elements in

order of decreasing solubility.

31

Table 2-5: Solubility data for binary magnesium alloys.

Element At. % Wt. % System Aluminum 11.8 12.7 Eutectic Yttrium 3.75 12.5 Eutectic Zinc 2.4 6.2 Eutectic Neodymium ~1 ~3 Eutectic Zirconium 1.0 3.8 Peritectic Manganese 1.0 2.2 Peritectic Calcium 0.82 1.35 Eutectic Titanium 0.1 0.2 Peritectic Adapted from Ref. [84].

Nonequilibrium processing techniques such as rapid solidification (RS) have already

been used to extend solid solubility limits and create alloys with improved corrosion resistance

[124], and several authors offer helpful reviews of Mg alloys made via rapid solidification [62,

63, 85]. RS is generally classified into spray and droplet methods, continuous casting methods,

and melt-in-situ methods, all of which are intended to extend solid solubility limits and create

materials with microcrystalline or amorphous structures [125]. RS improves corrosion resistance

by dispersing elements that may act as cathodic sites for microgalvanic corrosion [85] and

ennobling the electrode potential by extending alloying elements’ solid solubility limits [85].

PVD can also homogenize composition [126] and extend solid solubility limits [127] to improve corrosion resistance. PVD, however, also creates deposits that are more pure than the source material, and it is capable of incorporating elements that are not soluble in molten magnesium. PVD therefore has the potential to create nonequilibrium alloys with superior corrosion resistance compared to those made with RS. In addition, PVD can create alloys with structured surfaces and morphologies. Such structures, an example of which is shown in Figure

2-7, could be used to house drug-eluting polymers.

32

Figure 2-7: Structured PVD magnesium. Scale bar indicates 10 μm.

2.4.1 Morphology of Vapor Deposited Alloys

Surface defects and irregularities are known to exacerbate corrosion processes, so alloy morphology plays an important role in corrosion behavior. Ideal surfaces are dense, nonporous, and smooth. Deposition parameters, such as deposition rate, substrate temperature, substrate rotation, and bombardment can profoundly influence the surface morphology. Detailed backgrounds of PVD are available in theses submitted by previous members of this research group [19, 128].

A well-known model, which was developed by Thornton [129], describes a deposit’s morphology as a function of bombardment and substrate temperature. His model, shown in

Figure 2-8, contains four zones. Zone I is characterized by a substrate temperature (T) that is a small fraction of the arriving material’s melting temperature (TM). Shadowing plays a prominent role in film growth for ZI, so the morphology in this zone is described as porous columns with open grain boundaries, which some describe as matchstick-like. Increasing T/TM at low pressure,

however, may give adatoms (i.e. adsorbed atoms) the necessary mobility to overcome substrate

33 and nucleation roughness and form densely packed fibrous grains. Thornton called this the

Transition Zone, or Zone T, and film growth in this zone is controlled by surface diffusion. Zone

II growth is also controlled by surface diffusion, and the temperature ratio T/TM is higher, so

grains are more columnar with even less porosity. Tensile stress can develop in ZT and ZII,

because microvoids may collapse to the point where atomic attraction occurs between adjacent

columns [130]. Zone III growth is characterized by bulk diffusion and recrystallized grain

structures, which result from a combination of high T/TM and grain boundary characteristics.

Figure 2-8: Thornton’s structure zone model for vapor deposits [129]. Substrate temperature and bombardment influence deposit morphology.

Thornton’s model [129] also accounts for ion bombardment, which in addition to temperature, can increase adatom mobility and subsequently suppress porosity. Ion bombardment may also induce compressive stress in a deposit, and surface atoms may become frozen in this compressed state if T/TM is low enough [130]. This phenomenon is similar to the

34 shot-peening process used to produce compressive stresses in bulk materials, so it has been termed “ion peening” [131]. Film stress is significant, because it has been shown to negatively impact the corrosion resistance of magnesium alloy deposits [132].

The structure zone models have been shown to apply to methods of physical vapor deposition other than sputtering [133]. Substrate temperature is the most important variable for processes like EB-PVD, because bombardment is negligible except in cases where it is induced by an external source like an ion gun. The goal of this work is to increase corrosion resistance by manipulating deposition parameters to produce deposits with smooth surfaces and dense, nonporous morphologies.

2.4.2 Corrosion Studies of PVD Magnesium Alloys

Several research groups have already undertaken studies of the corrosion resistance of

PVD Mg alloys. Potentiodynamic tests of PVD Mg films in NaCl solutions showed decreased

anodic current density and extended passive regions compared to cast Mg [81, 134], and EIS

showed improved polarization resistance [134]. The Mg films were microcrystalline, and they

preferred (002) orientation [81].

Alloying elements have been observed to produce a more homogenous morphology in

Mg films [135] and improve their corrosion resistance. A group of researchers from the Defence

Research Agency, Farnborough, Hampshire, and the University of Surrey, Guildford, Surrey, has

thoroughly investigated the corrosion of Mg-Ti films made using PVD, and they have even

produced deposits up to 25 mm thick [136]. Mg-3.2%Ti deposits made using EB-PVD corroded

at 13 mdd (~10 mpy) in 0.6 M NaCl [137]. In an attempt to improve the corrosion resistance,

Mg-Ti deposits of this type were mechanically worked (via flailing) in situ, but the researchers

found that their corrosion rate was inferior to that of pure PVD Mg [138]. Blawert et al. [139]

35 have also reported inferior corrosion rates for Mg-Ti alloys—they used Tafel extrapolation and found that Mg alloys with less than 26% Ti corroded faster than commercially pure Mg in 0.5%

NaCl pH=11.

The Farnborough and Surrey group [137] found that the mode of corrosion attack on their

Mg-Ti alloys was generally uniform with some pitting, which they attributed to preexisting defects. Reports indicate that TiO2 is responsible for the improved corrosion resistance of these

Mg-Ti alloys [127]. Results for alloys with Ti content ranging from 0-44% found that alloys

containing less than 3-4% Ti or greater than 30% Ti exhibited the best corrosion resistance [140].

Additions of Ti were also found to ennoble the alloy’s OCP [141], which reduces its tendency to

corrode galvanically.

Besides Ti, PVD has also been used to create Mg-Y alloys [20, 128]. Heidersbach [128]

found that EB-PVD additions of Y up to 34 at% (64 wt%) remained in solid solution with Mg,

and potentiodynamic polarization scans in 0.1 M NaCl buffered to pH=12 showed a passive

region for alloys with larger Y content. XPS indicated that a bi-layer oxide-hydroxide film

formed when the alloys were immersed in the solution, with Mg(OH)2 and Y(OH)3 forming the

outer layer and Y2O3 forming the inner layer. Corrosion resistance was attributed to Y2O3.

Miller et al. [142] also concluded that the improved corrosion resistance of sputtered Mg-Y films

was due to the presence of oxidized Y in the passive film. Their Mg-Y films were analyzed using

potentiodynamic polarization in 0.1 M NaCl, and they found that Mg-7at%Y (Mg-22wt%Y) films

exhibited a passive region and lower current densities than pure Mg. Films containing even more

yttrium exhibited passive regions with ranges exceeding 1 V.

Ternary alloys, however, may be able to achieve even better corrosion resistance than

binary alloys. A good starting point for ternary alloys has been suggested as Mg(x)-α(y)-β(z) such that x + y + z = 100 at%; α = Y, Ca, or B (5-40 at%); and β = Cr, Ti, or Si (5-40 at%) [67].

Wolfe [19, 95] deposited Mg-Y-Ti alloys using both sputtering and EB-PVD. The alloys

36 exhibited corrosion resistance superior to commercial alloys like WE43, and film stress was shown to negatively affect their corrosion resistance. An optimal thermal heat treatment was therefore developed to relieve stress and improve corrosion resistance [132].

In general, one can see that PVD has the potential to create novel Mg alloys with superior corrosion resistance when compared to commercial alloys, and even to those made using rapid solidification. Numerous studies have explored Y and Ti, and these elements appear to be promising additions for the improvement of magnesium’s corrosion resistance.

2.5 Summary

Bioabsorbable magnesium implants are being developed for orthopedic and coronary medical procedures, but commercial magnesium alloys do not possess the corrosion resistance required for these applications. As a result, superior alloys must be developed, and in vitro tests

can be used to assess their corrosion behavior in simulated physiological environments.

Electrochemical and atomic properties indicate that aluminum, yttrium, and titanium will improve

magnesium’s corrosion resistance, but aluminum has been linked to Alzheimer’s disease in

humans, so yttrium and titanium emerge as the most promising alloying elements for

bioabsorbable magnesium alloys. Physical vapor deposition can be used to incorporate these

elements into nonequilibrium alloys that have been shown to be more resistant to corrosion than

commercial magnesium alloys.

37 Chapter 3

Experimental

This study employed multiple techniques to fabricate and characterize the magnesium alloys. Two deposition systems were used. A small-scale dual gun system was used to deposit pure Mg, binary, and ternary alloys, and an industrial-scale system was used to deposit ternary alloys. These alloys were characterized using ICP-AES, EDS, profilometry, optical microscopy, scanning electron microscopy, XRD, electrochemical corrosion characterization techniques, and cell culture assays.

3.1 Electron Beam Physical Vapor Deposition

Both thin (< 20 μm) and thick (> 100 μm) magnesium alloy deposits were fabricated using electron beam physical vapor deposition (EB-PVD).

3.1.1 Dual Gun Deposition

Thin Mg-Y-Ti alloys were made with a dual gun electron beam evaporation system, which was designed at Penn State. This system is shown in Figure 3-1. The system contains two

100 cc single pocket Telemark electron guns which operate at 10 kV and a combined maximum current of 1 A. Magnesium was placed in one crucible, and a mixture of yttrium and titanium slugs was placed in the other crucible. Quartz crystals monitor the deposition rate and thickness of material evaporated by each gun, and two Telemark Model 860 Deposition Controllers regulate the deposition rates. The bell jar-like lid is attached to a crane that lifts the lid for sample loading and chamber maintenance, and a roughing pump and cyropump can evacuate the chamber to base pressures as low as 7x10-7 Torr. A rotating two-axis substrate holder can be positioned

38 anywhere within the chamber to produce alloys with specific compositions and morphologies.

Rotation increases compositional uniformity of the alloys. Deposits of varying composition can be obtained by placing samples at regular intervals between the two evaporation sources.

Figure 3-1: Picture and schematic of the dual gun EB-PVD system designed at Penn State.

A special stage, which is cooled by liquid nitrogen, can also be installed in the chamber

to lower the substrate temperature between -50ºC and -80ºC, but this stage does not rotate. As a

result, when it is used, the sample sizes are kept small so that alloy compositions remain

relatively uniform. For this work, all alloys were deposited on oxidized silicon substrates placed

approximately 12 in. from the magnesium source (approximately 18 in. from the Y-Ti source).

A 2.5x5.0x27.0 inch magnesium ingot was obtained from Alfa Aesar and cut into

approximately 1x1x1 inch cubes for use as evaporation material. The 99.98% pure ingot’s

composition is given in Table 3-1. Care must be taken when cutting Mg, because its powder is

highly flammable. Yttrium of 99.9% purity and titanium of 99.9% purity were also obtained

39 from Alfa Aesar, and these materials were mixed and evaporated using the second electron gun.

Over 100 samples of various compositions were deposited in order to obtain alloys with superior corrosion resistance. Deposition information for noteworthy alloys is given in Table 3-2.

Table 3-1: Composition of 99.98% pure Mg ingot used to produce EB-PVD alloys (%). Zn 0.004 Al 0.0013 Cu 0.0004 Mn 0.0024 Ca 0.001 Cd < 0.0001 Fe 0.0021 Sn < 0.001 Mg > 99.98 Si 0.0020 Ni 0.0004

Table 3-2: Deposition variables for Mg alloys deposited with the dual-gun EB-PVD system.

Approx. Fabrication Approx. Mg Time Substrate Sample # Composition Alloy Rate Date Rate (Å/s) (min) Stage (Å/s) 9-21-06 16 Mg 100 - 56 - 12-1-06 24 Mg 15 - 68 - 1-2-07 33 Mg-Y 11 0.8 63 - 2-8-07 35 Mg-Y 27 0.6 116 - 3-28-07 37 Mg-Y 42 0.8 46 - 4-17-07 55 Mg-Y-Ti 21 1.0 42 - 4-19-07 56-76 (array) Mg-Y-Ti 21 1.0 48 - 5-24-07 77 Mg-Y-Ti n/a n/a n/a - 5-28-07 78 Mg-Y-Ti n/a n/a n/a - 5-30-07 80 Mg-Y-Ti 10 0.5 67 - 5-31-07 81 Mg-Y-Ti 22 0.9 67 - 6-1-07 82 Mg-Y-Ti 24 1.2 76 - 6-4-07 83 Mg-Y-Ti 19 0.9 41 - 7-6-07 84 Mg-Y 31 3.5 80 - 7-10-07 86 Mg-Ti 17 1.1 20 - 7-12-07 87 Mg-Y-Ti 31 1.5 75 - 7-16-07 88 Mg-Y-Ti 32 3.1 57 - 7-19-07 89 Mg-Y-Ti 29 3.0 50 - 7-21-07 90 Mg 11 - 48 Cooled 7-23-07 91 Mg-Y-Ti 24 2.3 52 Cooled

40 3.1.2 Industry-Scale Deposition

An industrial prototype Sciaky EB-PVD system [143] was used to deposit Mg-Y-Ti

alloys with thicknesses greater than 100 μm. Figure 3-2 shows exterior and interior views of the

chamber. The Sciaky system is located at the Penn State Materials Research Institute, and

deposits were made with the help of the Applied Research Lab.

Figure 3-2: Industrial prototype Sciaky EB-PVD system. (a) Exterior view of the system. (b) Interior view of the chamber.

The Sciaky system possesses six EB-guns and a three continuous feeds for metal ingots,

which are labeled 1-6 and A-C in Figure 3-2(b), respectively. Three ingots—Mg, Y, and Ti— were originally used to produce the thick Mg-Y-Ti alloys, but the resulting deposits did not have the desired composition because chamber geometry restricted the locations of the ingots. A hole was therefore bored through the Ti ingot, and the Y ingot was shaved so that it fit snugly into this hole. The combination-ingot was then fed through a single feed. The Mg ingot was 1.940 inches in diameter, and it was machined from the material obtained from Alfa Aesar. The Ti and Y ingots were both originally machined to be 1.05 inches in diameter. The Ti ingot was obtained

41 from www.onlinemetals.com, and the Y ingot was obtained from Hefa Rare Earth Canada Co.

Ltd. Their compositions are given in Table 3-3.

Table 3-3: Composition of Ti and Y used to produce thick alloys. Titanium Ingot (%) Yttrium Ingot (%) Fe 0.3 max Trace RE metals < 0.051 O 0.25 max Fe 0.043 C 0.1 max Ca 0.042 N 0.03 max Si 0.023 H 0.015 max Y 99 Ti 99.3 min

The final thick alloy deposition, done in early January 2008, took approximately three hours, and the final temperature of the chamber was approximately 450ºC. Glass and oxidized silicon substrates were placed approximately 16 in. from the sources. The substrates were mounted to a cooled iron substrate holder, but the glass and silicon apparently acted as insulators so that the arriving magnesium re-evaporated from these substrates. Alloys on the glass and

oxidized silicon were thus relatively thin with high contents of Y and Ti. Therefore, this study used the thick alloy specimens that deposited directly on the cooled iron substrate holder.

Figure 3-3 shows the layout of the specimens when attached to substrate holder and removed from the substrate holder. The Mg evaporation source was located below the junction of pieces 7 and 8, and the Y-Ti source was located below the junction of pieces 6, 7, and 11. Pieces

8, 10, and 11 were deposited on oxidized silicon, whereas piece 6 was deposited on glass. Piece 7 is a vapor-coated slab of WE43. Pieces 5 and 13 were deposited on foil. All other specimens were deposited on the cooled iron substrate holder. The quarter is included for size comparison.

42

Mg Y-Ti

Figure 3-3: Layout of the Mg-Y-Ti specimens made using the Sciaky EB-PVD system. The pieces are shown both attached to the substrate holder (top) and removed from the substrate holder (bottom). The location of the sources below the deposits is also shown.

43 3.2 Chemical Characterization

Energy dispersive x-ray spectroscopy (EDS) and inductively coupled plasma-atomic emission spectroscopy (ICP-AES) were used to assess chemical compositions of the as-deposited magnesium alloys and identify elements in the corrosion products.

3.2.1 Energy Dispersive X-Ray Spectroscopy

Energy dispersive x-ray spectroscopy (EDS) was used to identify elements in the magnesium alloys and their corrosion products, and it was also used to obtain estimates of the alloys’ chemical compositions. EDS was performed using a Philips XL30 ESEM, and EDAX

Genesis software was used to analyze the data. Data were obtained from alloy surfaces over a period 40-100 seconds at 2000x magnification with an accelerating voltage of 10-15 keV.

3.2.2 Inductively Coupled Plasma-Atomic Emission Spectroscopy

Inductively coupled plasma-atomic emission spectroscopy (ICP-AES) was used to determine the chemical composition and impurity content of the alloys. A Leeman Labs

PS3000UV inductively coupled plasma spectrophotometer was used for the analysis, and it was run with the help of Penn State’s Materials Characterization Laboratory (MCL) in the Steidle

Building. The spectrophotometer’s sensitivity ranged from 0.5 to 0.0001 μg/L (ppm) depending on the element, and results were given in ppm by weight. Quantitative results were obtained by calibrating the system with standards before each run.

Samples were prepared for ICP-AES analysis by cleaving the silicon substrates into

samples with an area of approximately 0.5 cm2. These samples were dissolved in screw-cap glass vials, which were obtained from VWR and Kimble Glass, containing 4-8 mL of 2 wt% HNO3.

44 The solutions became contaminated with Fe when tweezers were used to remove the silicon substrates from the acid, so an alternate method for removing the silicon had to be devised. After the Mg alloy was completely dissolved, the silicon substrate was carefully removed from each vial by turning the vial upside down so that the silicon settled on the vial’s cap. The vial was then slowly turned right-side up, the cap was removed, and the silicon was disposed of.

Control vials containing 2 wt% HNO3 were run for comparison, and Al, Ca, K, Na, and

Si were determined to be contaminants from the glass vials or local water supply. The 2 wt%

HNO3 was made using deionized water, but the water supply at University Park is notoriously hard, so small levels of contamination may be present. In addition, the acid may have leeched some elements from the glass. The results obtained for Al, Ca, K, Na, and Si concentrations were hence disregarded. A small amount of Fe was also found in the control samples, so this value was subtracted from the values obtained from the alloy analyses.

Many of the alloys were originally tested for only Mg, Y, and Ti, but another batch was prepared to test for impurities in the deposits. Impurity data were indeed obtained for the alloys, but the vials in this second batch did not contain enough solution to test for Y. Thankfully, the ratio of the concentration of Mg between respective samples from batch 1 and batch 2 was the same as the ratio for the concentration of Ti between respective samples from batch 1 and batch

2, so this ratio was used to extrapolate Y concentrations from the batch 1 data for the batch 2 data.

The respective weight percentages for each alloy were calculated by summing the concentrations of Mg, Y, and Ti and then dividing the concentration of each element by that sum.

Atomic percentages were calculated using equation (3-1), where X, Y, and Z represent elements such as Mg, Y, and Ti, and where a is the element’s atomic weight. Sample calculations are

shown in Table 3-4.

45

wt%(X) a at%(X) = X ×100 (3-1) wt%(X) wt%(Y) wt%(Z) ++ a X aY aZ

Table 3-4: Sample calculation to convert ICP results into alloy composition. Element μg/L (ppm) Control Al 0.17 0.26 Ca 0.48 0.34 Cu < 0.01 < 0.01 Fe 0.01 0.02 K 0.25 0.32 Mg 86.00 0.14 Mn < 0.01 < 0.01 Na 2.84 4.50 Ni < 0.01 < 0.01 Si 0.68 0.82 Ti 1.61 < 0.01 Y 8.00 - Zn < 0.01 < 0.01 Sum (Mg, Y, Ti): 95.61 Mg = 86.00/95.61 = 89.95 % wt% Ti = 1.61/95.61 = 1.68 % Y = 8.00/95.61 = 8.37 % Mg = 96.63 % at% Ti see eq. 3-1 = 0.92 % Y = 2.45 % *Contributions from Al, Ca, Fe, K, Na, Si, and Zn were deemed negligible when compared with the control sample.

3.3 Structural Characterization

Profilometry, optical microscopy, scanning electron microscopy, and x-ray diffraction were used to characterize the structure of the alloys. Structure and microstructure are known to influence corrosion resistance.

46 3.3.1 Profilometry

A Tencor P-10 Surface Profiler, which is located in the Earth and Engineering Science

Building and shown in Figure 3-4, was used to obtain the thickness of most alloys. A small piece of glass was placed over part of the oxidized silicon substrate during each deposition, which created a step from the silicon surface to the top of the deposit. The surface profiler was used to obtain three thickness values from this step, and these values were averaged to obtain an overall thickness. The thicknesses of the thick deposits were estimated using a SEM, because they were too thick to examine with the surface profiler.

Figure 3-4: Tencor P-10 Surface Profiler. The system uses a stylus to trace height profiles, which can be used to calculate the thickness of an alloy deposit.

3.3.2 Optical Microscopy

A Boreal Zoom Sereomicroscope was used to obtain low-magnification optical

micrographs. The microscope was attached to a computer running Windows XP with a USB

47 cable, and Motic Images 2000 software was used to capture digital images. When the microscope

was calibrated using a sample of predetermined length, as shown in Figure 3-5, images could be obtained and used to calculate sample areas for electrochemical tests. Sample areas could be traced with the mouse by changing the “Length” function to the “Irregular Area” function.

Figure 3-5: Motic Images 2000 computer interface. The optical microscope was calibrated against the diameter of a circle; then the “Length” function of the software was changed to “Irregular Area” so that samples could be photographed and their areas could be obtained for electrochemical tests.

3.3.3 Scanning Electron Microscopy

Two scanning electron microscopes (SEMs) were used to obtain micrometer-scale images of the alloys and cell cultures. A XL30 Environmental Scanning Electron Microscope

(ESEM), which is located in the Earth and Engineering Science Building, was used to obtain

48 images of alloys’ surface characteristics and cross-sectional morphology. Samples were prepared by cleaving the silicon substrates into small pieces. Some deposits did not cleave with the silicon substrate, so they were cut or mechanically worked to obtain a representative cross section.

Deposits made with the Sciaky EB-PVD system were mechanically worked until they cracked, and the fracture surface was used for imaging. All deposits were imaged by using colloidal graphite to create an electrical connection between the alloy and microscope stub. Several deposits were also imaged after being stored for over 1 year in an evacuated desiccator, which also contained desiccant.

A JEOL JSM 5400 SEM in the South Frear Laboratory was primarily used to obtain images of the cell cultures. Samples were prepared for imaging by using a Bal-Tec SCD 050

Sputter Coater to coat them with approximately 10 nm of gold.

3.3.4 X-Ray Diffraction

X-ray diffraction (XRD) was used to compare the alloys’ crystal structure with that of bulk magnesium and WE43. XRD was performed using a Scintag (now Thermo Scientific) X2

Theta-Theta Powder Diffractometer, which is located at the Materials Research Laboratory. The system uses a monochromatic Cu-Kα radiation source. It is shown in Figure 3-6.

Figure 3-6: Sintag X2 Theta-Theta Powder Diffractometer. Exterior (left) and interior (right) views are shown.

49 Magnesium alloy deposits were analyzed on the silicon oxide wafers with a zero- background holder. Oxidized silicon spacers were used to position a deposit in the x-ray plane.

Alloys were analyzed using a grazing scan of 0.02 step size at a continuous rate of 1.00º per minute. The angle of incidence was set at 2º, and scans were run from 2º to 64º, although data in the 2º to 30º region often contained significant noise. Some scans were therefore shortened to begin at 30º in order to conserve time. Bulk materials were also analyzed using grazing scans with the aforementioned parameters. XRD spectra were analyzed using Materials Data, Inc.

(MDI) Jade software.

3.4 Corrosion Characterization

Mass loss measurements are generally considered the “gold-standard” for evaluating corrosion, but such measurements take a significant amount of time, and they only assess the general mode of corrosion and the average corrosion rate over that time. Electrochemical measurements, however, assess instantaneous corrosion rates, and they can offer insight into corrosion characteristics such as passive film formation and active-passive behavior.

Electrochemical methods can also be used to evaluate the corrosion behavior of samples of small mass, such as the thin Mg alloys presented in this thesis. The thin alloy specimens are too small and delicate to test with mass loss methods, and mass loss measurements of the thick deposits are complicated by back-side contamination from the iron substrate holder. The EB-PVD alloys in this work were therefore characterized with electrochemical techniques, and WE43 was characterized using both mass loss and electrochemical techniques

Copper tape and wax were used to prepare the Mg alloy deposits for electrochemical testing. Conductive copper tape obtained from Ted Pella, Inc. was used to make an electrical connection to the surface of the deposit, and that connection was confirmed with a digital

50 multimeter (DMM). Apiezon® Wax W was used to mask the tape, silicon substrate, and corners of the sample so that a small amount covered the sample’s surface, which is shown in Figure 3-7.

The wax was mixed with trichloroethylene (TCE) until it reached a semi-liquid consistency, and it was applied to the samples using a wooden dowel with ~2 mm diameter. The wax hardened as the TCE evaporated, at which point the samples could be tested. Crevice corrosion is a particular problem for masking agents, so the samples were tested within 20 hours of preparation while the wax was still somewhat tacky. They were also inspected after testing for evidence of crevice attack, and the data were discarded if evidence was found. Motic Images 2000 software was used to calculate sample areas.

Figure 3-7: Example of a sample masked with Apiezon® Wax W. Scale bar indicates 5 mm.

WE43 was cut into small pieces for testing. Prior to masking with the wax, sample surfaces were polished to 1200 grit with LECO SiC abrasive disks and deionized water. The surface was quickly degreased with acetone after polishing.

All corrosion experiments were performed in open-air containers of Hanks Balanced Salt

Solution (HBSS) [144] at physiological temperature 37±2ºC. Temperature was monitored using either a thermometer or a thermocouple in conjunction with a hot plate. HBSS, which was

51 purchased from Cambrex or made in-house, was chosen because it simulates the ionic

composition of human plasma, as shown in Table 3-5. HBSS also possesses a pH of 7.08 to 7.40, which is similar to physiological pH (~7.4). Corrosion experiments were performed in a 250 mL beaker containing 100 mL of HBSS, except for long-term experiments (longer than a few hours), which were performed in 250 mL of HBSS. The pH of the electrolyte was tested with pH paper after the experiments were complete.

Table 3-5: Composition of HBSS and human blood plasma. Component HBSS (mg/L) Component HBSS (mmol/L) Plasma (mmol/L) + CaCl2•2H2O 186 Na 141.4 142.0 KCl 400 Cl- 144.8 130.0 + KH2PO4 60 K 5.8 5.0 2+ MgSO4•7H2O 200 Ca 1.3 2.5 NaCl 8000 Mg2+ 0.8 1.5 2- NaHCO3 350 HPO4 0.8 1.0 - NaH2PO4•7H2O 90 HCO3 4.1 27.0 2- Glucose 100 SO4 0.8 0.5 Adapted from Refs. [144, 145]

Electrochemical experiments employed a saturated calomel reference electrode (SCE) and a graphite rod counter electrode, and the magnesium sample served as the working electrode.

The magnesium sample faced the graphite counter electrode to facilitate a uniform current distribution. A picture and a diagram of the electrochemical cell is shown in Figure 3-8. The cell was open to air because O2 does not affect magnesium’s corrosion rate [63], and although carbon dioxide can affect the corrosion rate by acidifying the solution [66], the CO2 content of air is less

than 0.1%, so its influence was ignored. All electrochemical experiments were run on PC4/750

or PCI4/750 Gamry potentiostats. Gamry Framework software version 4.35 or 5.50 was used to

interface between the potentiostats and the desktop computers. Electrochemical experiments

52 were also performed on some samples after they had been stored in a desiccant-containing

evacuated desiccator for longer than 1 year.

Figure 3-8: The corrosion cell used for electrochemical tests. The glass cell contains HBSS electrolyte, a graphite counter electrode (CE), a thermometer or thermocouple, a SCE reference electrode (RE), and a Mg alloy working electrode (WE) connected to copper tape and masked with wax.

3.4.1 Mass Loss Measurements

Mass loss measurements were used to characterize alloy WE43, which was obtained from

Magnesium Elektron. Its composition is given in Table 3-6.

53

Table 3-6: Composition of Magnesium Elektron WE43. WE43 (wt%)* Y 4.2 Cu < 0.02 HRE + Nd 2.1 Zn 0.01 Zr 0.4 Si < 0.01 Li 0.06 Ni < 0.002 Zn 0.01 Fe 0.002 Mn < 0.02 Total Others < 0.30 *Mg comprises the balance.

Samples of WE43 were polished on all sides to 1200 grit using LECO SiC abrasive disks and deionized water. They were degreased with methanol or acetone immediately after polishing and weighed to obtain their masses. Their dimensions were also measured so that sample areas could be calculated. Dimensions were typically on the order of 2.3x2.3x1.0 cm. The samples were then placed on marbles in covered beakers of HBSS at 37±2ºC, where they were allowed to corrode for 1 week. The marbles served to maximize the contact between the samples and the electrolyte. After the experiments were complete, the samples were removed from the beakers and allowed to dry. Corrosion product was observed to accumulate on the samples and interfere with mass calculations (see Figure 3-9), so the samples were cleaned in boiling chromic acid

solution, which was 0.150 M CrO3 and 0.010 M Ag2CrO4, according to ASTM G1 [146]. The

samples were immersed in the cleaning solution for 1 minute; then they were removed, rinsed

with deionized water, and allowed to air dry. The cleaned samples of WE43 are shown in Figure

3-10.

54

Figure 3-9: Corrosion product on a magnesium sample. Corrosion product can interfere with mass loss calculations, so the product must be removed. Mass loss samples were placed on marbles to maximize the area exposed to the HBSS electrolyte.

Figure 3-10: Mass loss samples of WE43 cleaned according to ASTM G1. Corrosion product can be removed by immersing magnesium samples in a boiling solution of 0.150 M CrO3 and 0.010 M Ag2CrO4. Scale bar indicates 1 cm.

At this point, the samples were weighed again, and their average corrosion rate was calculated using equation 3-2, where i is the average current density (μA/cm2), Δm is the change in sample mass, n is the number of equivalents (2 for Mg), F is Faraday’s constant (96500 coulombs/equivalent), t is time (s), a is the molar mass of the material (g/mol), and A is the surface area of the sample (cm2).

55

Δm× × Fn i = (3-2) × × Aat

A sample calculation is given below.

Δm ×× Fn 0.0039g × 2e × 96500C/e i = = = 2.151 μA/cm 2 ×× Aat 604800s × 31.33g/mol × 47.18 cm 2

3.4.2 Open Circuit Potential

Electrochemical techniques were used to characterize both thin and thick alloys. The

OCP was recorded until it stabilized. Magnesium exhibits a relatively unstable open circuit potential, so the OCP was considered stable if it changed less than a few millivolts over the course of a minute. Open circuit potentials were generally observed to stabilize within 5-10 minutes of immersion, although some samples took longer to come to steady state.

3.4.3 Linear Polarization

The linear polarization technique was used to measure polarization resistance. The potential was scanned from -10 mV vs. the open circuit potential (mVoc) to +10 mVoc at a rate of

0.2 mV/s, and data points were gathered every 0.5 s. The plots generally exhibited a linear region

about zero current in accordance with theory, so the slope of the curve within this region was

recorded as the polarization resistance (see Figure 3-11).

56

2 Rp = 7192 Ω•cm

Figure 3-11: Sample Rp measurement from a linear polarization curve. The measurement was made with Gamry Echem Analyst software.

Once the polarization resistance was obtained, the instantaneous corrosion rate [147] was

2 calculated using equation 3-3, where icorr is the instantaneous corrosion rate (μA/cm ), Rp is the

2 polarization resistance (Ω•cm ), and βa and βc are the anodic and cathodic Tafel slopes,

respectively (volts per decade of current).

× ββ ca icorr = (3-3) 32 R. (+×× ββ cap )

The Tafel slopes were generally difficult to obtain from potentiodynamic polarization

curves (either by visual analysis or Gamry Echem Analyst version 5.30 software), so 0.1 V per

decade of current was assumed for each slope in order to calculate the corrosion rate. These are

common assumptions, and estimates indicate that their error is at maximum a factor of two [148].

A sample icorr calculation is shown below.

57

× ββ 0.1 V/dec × 0.1 V/dec i = ca = = 3.023 μA/cm 2 corr 2 32 R. ()+×× ββ cap 2.3 7192 Ω cm ×⋅× (0.1 V/dec + 0.1 V/dec)

Equivalent weight and alloy density are two important parameters that can be used to calculate penetration rates from current density corrosion rates, although care must be taken when reporting penetration rates for materials that do not corrode uniformly. An alloy’s equivalent weight [149] can be calculated using equation 3-4, where EW is the equivalent weight of an alloy

th composed of N elements, wt%(i) is the mass fraction of the i element, ai is the atomic weight of

th th the i element, and ni is the valence of the i element.

N × awt%(i) EW = i (3-4) ∑ n i=1 i

An alloy’s density can be approximated [19] using equation 3-5, where ρ is the density of

th an alloy composed of N elements, Vi is the volume fraction of the i element, and ρi is the density

of the ith element.

N ρ ∑V ×= ρii (3-5) i=1

An element’s volume fraction can be calculated using equation 3-6, which presupposes a

ternary alloy composed of elements X, Y, and Z.

Xwt )%( ρ V = X (3-6) X ZwtYwtXwt )%()%()%( ++ ρ X Y ρρ Z

Table 3-7 shows sample calculations for the alloy composition that was presented in

Table 3-4.

Table 3-7: Sample calculation for equivalent weight, density, and volume fraction. Alloy ρ a wt%·a Alloy ρ n wt% EW V V·ρ Element (g/cm3) (g/mol) n (g/cm3) Mg 1.74 24.305 2 0.8995 10.93 0.9584 1.67 Ti 4.51 47.867 4 0.0168 0.20 13.61 0.0069 0.03 1.86 Y 4.47 88.906 3 0.0837 2.48 0.0347 0.16

Once an alloy’s equivalent weight and density are calculated, its corrosion penetration

2 rate PR (see Appendix B) can also be calculated using equation 3-7 (icorr is in units of μA/cm ),

where K is a constant to convert units. To calculate the penetration rate in mils per year (mpy),

use K=0.129. To calculate the penetration rate in microns per day (μm/day), use K=0.00895.

Care, however, should be taken when converting current density to penetration rate, because penetration rate implies uniform corrosion.

×× EWiK PR = corr (3-7) ρ

3.4.4 Potentiodynamic Polarization

Potentiodynamic polarization curves were obtained by scanning the potential from at least -10 mVoc to several hundred millivolts above open circuit at a rate of 1 mV/s. Scans were

aborted when the alloy broke down or when current exceeded 3 mA/cm2, because IR effects were

observed to become significant. Open circuit potential, breakdown potential (the potential in the

low anodic current region where a rapid increase in current is observed due to localized surface

corrosion), and passive current density (the current density corresponding to the breakdown

potential) can be obtained from potentiodynamic scans.

59 3.4.5 Electrochemical Impedance Spectroscopy

Electrochemical impedance spectroscopy (EIS) is a useful technique, because it offers insight into both the mechanism of corrosion attack and the processes taking place on a specimen’s surface. For example, EIS can distinguish between corrosion due to a pit, pore, etc. or due to diffusion through a surface oxide. It can also provide information about the capacitance and dielectric constant of a passive film.

EIS was performed on samples by exposing them to 10 mVoc rms AC pulses beginning at a frequency of 100,000 Hz and ending at a frequency of 10 mHz. Ten data points were recorded per decade. The literature does not contain much information regarding equivalent circuits for vapor deposited magnesium, so the approximate polarization resistance, Rp, was obtained from

the curve as shown in Figure 3-12. The real component of the impedance at 100,000 Hz (the

solution resistance) was subtracted from the value of the real impedance where the curve crossed

or approached the real axis of its Nyquist plot.

Figure 3-12: Sample Rp measurement from a Nyquist plot. The measurement was made with Gamry Echem Analyst software. The left arrow indicates the point corresponding to 100,000 Hz, and the right arrow indicates the point closest to the intersection of the real axis.

60 3.5 Magnesium Cytotoxicity Characterization

In vitro cell culture tests can be used to screen potential bioabsorbable Mg alloys for their effects on the cells and tissue in particular physiological environments, such as bones or blood vessels. A549 cells and growth media additives were therefore obtained from the Department of

Pathology at Penn State’s Hershey Medical Center (HMC) so that cell culture experiments could be performed with the thin and thick EB-PVD alloys. A549 [150] cells are a transformed epithelial cell line from a human lung, and they were chosen for initial cytotoxicity screening because they are a relatively robust epithelial cell line.

3.5.1 Cell Cultures

A549 cells were cultured in medium whose primary constituent was Biowhittaker®

EMEM with EBSS and 25 mM Hepes buffer. The medium needed to be buffered since the cells were incubated in a Barnstead Lab-Line Model 120 incubator, which does not support a carbon dioxide atmosphere. Growing medium was created from a mixture of 90 mL EMEM, 10 mL fetal bovine serum, 1 mL L-glutamine, 1 mL vancomycin, 0.2 mL amphotericin, and 0.04 mL gentamicin.

Several hours after the cells were received from HMC, the old growing medium was decanted from the flask, and 4 mL of Gibco TrypLETM Express trypsin replacement was added

and rinsed around in the flask to remove excess fetal bovine serum. All fluid transfers were done

with sterile, disposable 10 mL polystyrene serological pipets from VWR. The trypsin

replacement was decanted after a few seconds of rinsing, and an additional milliliter was added to

the flask, which was then capped and placed on a 37-50ºC hot plate for approximately 1 minute.

The trypsin replacement and heat served to release the cells from the flask. After 1 minute, the

61 flask was held up to the light to observe that cells were indeed released from the flask, and 9 mL of growing medium was added to the flask to neutralize the trypsin replacement and create a suspension of cells for use in experiments or making new flasks. Less growing medium could be added to make concentrated suspensions, or more growing medium could be added to make dilute suspensions.

Cell culture experiments were performed in 6-well trays with a removable cover. Each sample was placed in its own well with approximately 6 mL of the cell suspension. Alloy samples generally exhibited an area of 0.5 cm2, and WE43 was polished to 1200 grit with LECO

SiC abrasive discs and then coated in epoxy on all but one side to simulate the surface conditions of the EB-PVD alloys. Two Corning glass slide control substrates were also used to assess the cells’ growth in regular medium and in medium containing dissolved WE43. Cell viability can be affected by the corrosion process or by elemental toxicity, so the glass substrates were included to differentiate between these variables. Samples were incubated at 37ºC in the covered 6-well trays for approximately 14 hours.

The glass control samples were approximately 3 cm2, and they were degreased with acetone and rinsed with deionized water prior to use. The one glass sample was incubated with 6 mL of the “pure” cell suspension, but the other glass sample was incubated under different conditions. A small amount of WE43 (6 mg) was dissolved in 3 mL of growing medium over a period of 14 hours to simulate the corrosion that might occur during a cell culture experiment.

The glass slide was then placed in the well containing this modified medium, and 3 mL of cell suspension was added to the well for a total of 6 mL of solution. The cells were incubated on the glass slides for the standard length of 14 hours. Longer incubation periods were prohibited by the alloy deposits, since the thinnest deposits were observed to corrode completely in less than a day.

62 3.5.2 Cell Fixation

Once the cell culture experiments were complete, the cells needed to be prepared for

SEM imaging. This was accomplished through two fixations, dehydration, drying, and gold coating, all of which were done in the Huck Electron Microscopy facility in the South Frear

Laboratory.

Karnovsky fixative served as the primary fixative, and it acted to cross-link proteins.

Growth media was removed and several mL of Karnovsky fixative was added to each well until samples were covered. Karnovsky fixative is a mixture of 25 mL 0.2 M cacodylate buffer

(pH=7.4), 2 g sucrose, 8 mL 16% ampoule conc. paraformaldehyde, 3 mL 35% ampoule conc. glutaraldehyde, and deionized water to total 50 mL. The samples soaked in the primary fixative for 1 hour. Following primary fixation, each sample was rinsed three times with 0.1 M phosphate buffer (pH=7.4), and each rinse lasted five minutes. For the secondary fixation, which stabilizes the cells’ membranes, several milliliters of 1% osmium tetroxide in 0.1 M phosphate buffer

(pH=7.4) was added to each well, and the samples were stored in a light-proof container for 1 hour. Following secondary fixation, each sample was again rinsed 3 times with 0.1 M phosphate buffer (pH=7.4) for five minutes per rinse.

At this point the samples were dehydrated with water-ethanol mixtures. Each sample was soaked in 25% ethanol for 5 minutes. The 25% solution was then removed, and each sample was soaked in 50% ethanol for 5 minutes. This procedure continued with 70%, 85%, 95%, and 100% ethanol. Three dehydration rinses were done with the 100% ethanol, after which the samples were quickly transferred to a Bal-Tec CPD 030 Critical Point Dryer for drying. The samples were flushed with liquid CO2 to remove excess ethanol, and the liquid CO2 was then brought to

its critical point to complete the drying process. At this point the samples were ready for gold

coating and SEM imaging.

63 Cell morphology was used to assess the cytotoxic effects of magnesium corrosion on the

A549 cells [151]. Normal cells were identified by their pancake-like morphology and smooth cellular membrane. Apoptotic and necrotic cells were identified by their lysed membranes. Cells were considered “indeterminate” [151] if their cellular membrane appeared intact but their morphology was more ball-like than pancake-like. Such cells may be entering the initial stages of apoptosis.

64

Chapter 4

Results

Over 90 different magnesium alloys were deposited using the dual gun EB-PVD system, and 3 runs were completed on the Sciaky system. Hundreds of experiments were performed with these samples. The reader would likely become overwhelmed if all of the results were discussed in detail, so representative samples have been selected to simplify the discussion. Alloys made with the dual gun system are referred to using their designation number and composition (in weight percent), and alloys made on the Sciaky system are referred to using their designation number preceded by the letter T (for “thick”) and composition (also in weight percent). In addition, alloys tested within 3 months after deposition are referred to as “fresh,” whereas alloys

tested after several months of storage in a desiccant-containing evacuated desiccator are referred

to as “aged.” All results are for as-deposited alloys that have not been optimized via heat-

treatment. The general thick alloy deposition process also needs to be optimized, but preliminary

results are included because of their promising nature.

4.1 Chemical Analysis of EB-PVD Mg Alloys

The compositions of selected as-deposited alloys are provided in Table 4-1. Deposition

rates for Mg, Y, and Ti are also included in the table. Deposits consisted of pure Mg, Mg-Y, Mg-

Ti, and Mg-Y-Ti chemistries. Higher Y-Ti deposition rates resulted in alloys with increased Y and Ti content. It was observed that yttrium evaporated preferentially from the Y-Ti crucible in the dual gun system, and Mg-Y-Ti alloys made with this system contained approximately 5-6 times as much Y as Ti (by weight). This was expected, since yttrium’s vapor pressure (5.31 Pa at

65 1526°C [152]) is higher than titanium’s vapor pressure (0.49 Pa at 1660ºC [153]). Table 4-2 lists the alloys’ composition in both weight percent (wt%) and atomic percent (at%).

Table 4-1: ICP-AES chemical analysis of selected EB-PVD magnesium alloys. Mg Deposition Y-Ti Deposition Alloy Type Designation Composition (wt%) Rate (Å/s) Rate (Å/s) 24 Mg100 15 - Mg 90* Mg100 11 - 35 Mg95 Y5 27 0.6 Mg-Y 33 Mg91 Y9 11 0.8 Mg-Ti 86 Mg98 Ti2 17 1.1 57 Mg92 Y7 Ti1 21 1.0 87 Mg90 Y8 Ti2 31 1.5 Mg-Y-Ti 55 Mg89 Y9 Ti2 21 1.0 (dual gun) 91* Mg86 Y11 Ti3 24 2.3 89 Mg80 Y16 Ti4 29 3.0 T1 Mg98.7 Y1.1 Ti0.2 Mg-Y-Ti T15 Mg97.2 Y2.5 Ti0.3 0.5 mm/min 0.15 mm/min (Sciaky) T3 Mg96.6 Y3.0 Ti0.4 (ingot feed rate) (ingot feed rate) T16 Mg95.8 Y3.7 Ti0.5 *Deposited on a cooled substrate.

Table 4-2: ICP-AES chemical analysis of selected EB-PVD magnesium alloys. Alloy Type Designation Composition (wt%) Composition (at%) 24 Mg100 Mg100 Mg 90* Mg100 Mg100 35 Mg95 Y5 Mg99 Y1 Mg-Y 33 Mg91 Y9 Mg97 Y3 Mg-Ti 86 Mg98 Ti2 Mg99 Ti1 57 Mg92 Y7 Ti1 Mg97 Y2 Ti1 87 Mg90 Y8 Ti2 Mg97 Y2 Ti1 Mg-Y-Ti 55 Mg89 Y9 Ti2 Mg96 Y3 Ti1 (dual gun) 91* Mg86 Y11 Ti3 Mg95 Y3 Ti2 89 Mg80 Y16 Ti4 Mg93 Y5 Ti2 T1 Mg98.7 Y1.1 Ti0.2 Mg99.6 Y0.3 Ti0.1 Mg-Y-Ti T15 Mg97.2 Y2.5 Ti0.3 Mg99.1 Y0.7 Ti0.2 (Sciaky) T3 Mg96.6 Y3.0 Ti0.4 Mg99.0 Y0.8 Ti0.2 T16 Mg95 . 8 Y3.7 Ti0.5 Mg98.7 Y1.0 Ti0.3 *Deposited on a cooled substrate.

66 4.2 Structural Characterization of WE43 and EB-PVD Mg Alloys

Microstructure and surface morphology are known to play a significant role in both corrosion and physiological reactions. The physical characteristics of WE43 and the EB-PVD

Mg alloys are discussed in this section using profilometry data, SEM images, and XRD spectra.

4.2.1 Alloy Thicknesses

EB-PVD alloys have been shown to contain a certain amount of inherent stress [132], and this stress is often directly correlated with the thickness of the deposit. Thickness must also be considered because thicker deposits can endure corrosion for longer periods of time.

Unfortunately, the dual-gun system can only produce single-layer alloys that are several microns thick. Table 4-3 compiles thickness values for selected EB-PVD magnesium alloys.

Table 4-3: Thicknesses of selected EB-PVD magnesium alloys. Alloy Type Designation Thickness (μm) 24 ~3 Mg 90* 2.66 33 3.00 Mg-Y 35 ~16 Mg-Ti 86 2.37 55 3.70 57 5.02 Mg-Y-Ti 87 10.28 (dual gun) 89 6.62 91* 3.79 T1 ~346 Mg-Y-Ti T3 ~300 (Sciaky) T15 ~300 T16 ~300 *Deposited on a cooled substrate. “~” denotes values obtained from SEM images.

67 4.2.2 Microstructure and Surface Morphology

To observe the surface morphology of bulk WE43, the metal was etched by swabbing its polished surface with a solution of 5 mL concentrated HNO3 in 95 mL methanol [154]. The

resulting structure, which is shown in Figure 4-1, contained grains that are tens to hundreds of microns in size. Figure 4-1 was used to calculate WE43’s ASTM grain size number [155], which was determined to be between 2.0 and 2.5. In contrast, the EB-PVD alloys exhibited structure on the 1-10 μm scale. Pure EB-PVD magnesium formed a columnar, porous deposit, and higher deposition rates seemed to produce deposits with larger surface morphology. Two deposits with different microstructure and surface morphologies are shown in Figure 4-2.

Figure 4-1: Etched WE43 grains. WE43 was etched with a solution of 5 mL concentrated HNO3 mixed with 95 mL methanol. Scale bar represents 500 μm.

68

Side View Top View

Figure 4-2: EB-PVD pure Mg microstructures and surface morphologies. (A) Columnar, porous Mg (alloy 90) deposited on a cooled substrate at ~11 Å/s. Similar but magnified microstructure was also seen in deposits formed on uncooled substrates at low rates of deposition (~10 Å/s, alloy 24). (B) Columnar, porous Mg (alloy 16) deposited at room temperature with a deposition rate ~100 Å/s. Scale bars represent 1 μm.

As Figure 4-3 shows, the XRD spectra of commercial Mg and EB-PVD Mg were similar.

The (002) peak was the most prominent peak for commercial and EB-PVD Mg, but the (101) peak was the most prominent peak for WE43. Interestingly, the spectrum of EB-PVD Mg showed an unidentified peak near 63.5º 2-theta. This peak did not appear on other scans of EB-

PVD Mg.

69

2000 (101)

(002) 1500

1000 (100)

Counts WE43 (102) (110) (103)

Bulk 500 Mg

EB-PVD Mg 0

30 40 50 60

2-theta Figure 4-3: Grazing angle XRD scans of bulk and EB-PVD Mg. WE43, commercial Mg, and fresh EB-PVD Mg exhibited similar spectra. The peak on the far right of the EB-PVD Mg spectrum was unidentified, but it did not show up on other scans of EB-PVD Mg.

Small additions of yttrium changed the morphology of the as-deposited EB-PVD Mg alloy, as shown in Figure 4-4. Figure 4-4(A) shows an alloy containing 9 wt% Y, whereas 4-4(B)

contains 5 wt% Y. These alloys still exhibited columnar microstructure, but there was noticeably

less porosity when compared to the pure Mg deposits.

70

Side View Top View

Figure 4-4: EB-PVD Mg-Y alloy microstructures and surface morphologies. Cross-sectional views (left) show columnar microstructures, but surface views (right) show different surface morphology. (A) Mg91 Y9 – sample 33. (B) Mg95 Y5 – sample 35. Scale bars indicate 2 μm.

Figure 4-5 compares the XRD spectrum of Mg91 Y9 (sample 33) with a polished sample of commercial magnesium. The addition of yttrium seemed to favor the (002) Mg orientation while depressing the other orientations. The reader should note that the small peak located near

53.5º 2-theta on the sample 33 spectrum is not from the magnesium alloy. It is the (311) peak from the silicon substrate, and it was often observed in the spectra from samples deposited on silicon. This peak is known to appear near 53.5º 2-theta in grazing angle geometries, as shown in

Figure 4-6.

71

33

Figure 4-5: Grazing angle XRD scans of fresh EB-PVD Mg-Y and commercial Mg. Addition of Y seemed to generate deposits with preferred (002) orientation. The small peak near 53.5º 2-theta on the sample 33 spectrum is from the silicon substrate.

Mg91 Y9 (33) 400 86 Mg98 Ti2 (86) Mg89 Y9 Ti2 (55)

300

200 Counts 55 100 33

0

51.5 52.0 52.5 53.0 53.5 54.0 54.5 55.0 55.5 2-theta Figure 4-6: Silicon (311) peaks. The Si (311) peak was commonly observed during grazing angle scans of alloys on silicon substrates. Its position is in the general vicinity of 53.5º 2-theta.

72 Relatively small additions of titanium were also observed to change the microstructure of the Mg deposits, as shown in Figure 4-7, which presents cross-sectional and surface images of

Mg98 Ti2 (sample 86). The Mg-Ti microstructure was still columnar but much less porous than vapor-deposited Mg. In addition, the surface was more uniform and less platelet-like than the pure Mg and Mg-Y deposits. The Mg98 Ti2 XRD spectrum in Figure 4-8 shows preferred (002)

Mg orientation, which is similar to the pattern observed in Mg-Y spectra, but the peaks corresponding to other orientations were more pronounced than those seen in spectra from Mg-Y.

The peaks may be more pronounced as a result of the low Ti content in sample 86.

Side View Top View

Figure 4-7: EB-PVD Mg-Ti alloy microstructure and surface morphology. Cross-section (left) shows columnar structure, and surface view (right) shows some porosity. Pictures are of Mg98 Ti2 (86). Scale bar indicates 1 μm.

73

86

Figure 4-8: Grazing angle XRD scans of fresh EB-PVD Mg-Ti and commercial Mg. Small additions of Ti (2 wt%) did not profoundly change an alloy’s spectrum.

EB-PVD Mg-Ti was imaged again after being stored in an evacuated desiccator for more than 1 year. Figure 4-9 shows that columnar microstructure persisted (left image), and the

surface morphology remained relatively unchanged (right image). Small spots of localized

oxidation, one of which is visible on the left side of the surface image, were the only visible

changes to the surface morphology.

74

Side View Top View

Figure 4-9: Aged EB-PVD Mg-Ti alloy. Columnar microstructure persisted (left image). A localized spot of oxidation is visible on the left side of the surface image (right). Otherwise, the alloy was relatively unchanged after aging. Scale bars indicate 5 μm.

Figures 4-10 and 4-11 show the surface morphology and microstructure of Mg-Y-Ti alloys shortly after deposition. One can see that Mg-Y-Ti alloys’ surfaces were relatively uniform and smooth, and their microstructures were quite dense compared to pure Mg deposits.

Small, unidentified black splotches appeared on alloy surfaces. Y and Ti seemed to exhibit a homogenizing effect on an alloy’s microstructure and surface morphology. Such properties were characteristic of many of the Mg-Y-Ti alloys.

75

Figure 4-10: EB-PVD Mg-Y-Ti alloy surface morphology. The surface of Mg89 Y9 Ti2 (sample 55) was relatively uniform and smooth. This surface was characteristic of many Mg-Y-Ti alloys. Scale bars indicate 5 μm.

Figure 4-11: EB-PVD Mg-Y-Ti alloy microstructure. Dense columnar micro structure was observed for Mg89 Y9 Ti2 – sample 55 (left) – and Mg86 Y11 Ti3 – sample 91 (right). Horizontal patterns in the alloys were a result of interruptions in the deposition process. Beads on the sample surface were likely organic contamination. Scale bars indicate 1 μm.

However, some Mg-Y-Ti alloys exhibited a columnar microstructure and porous surface morphology that was similar to the Mg-Y and Mg-Ti deposits. Figure 4-12 shows one such

example. Most of these alloys were deposited on stationary (not rotating) substrate holders.

76

Expanded View Magnified View

Figure 4-12: EB-PVD Mg-Y-Ti alloy with columnar microstructure and porous surface. Mg92 Y7 Ti1 (sample 57) exhibited more columnar structure (left) and associated porous surface morphology (right). Scale bars indicate 0.5 μm.

Many of the Mg-Y-Ti alloys demonstrated ductility and malleability, even when they were relatively thin. These samples were often difficult to cleave for SEM imaging and electrochemical testing, because the deposits would delaminate from the silicon substrate rather than cleave with the wafer. Figure 4-13 shows elongation of one alloy (left), which can be

deduced from the surface that is pulled down across the alloy’s cross-section. The figure also

shows dimple marks on the ductile fracture surface of another alloy (right).

Figure 4-13: Fracture surfaces of EB-PVD Mg-Y-Ti alloys. Elongated surfaces (left) and ductile fracture profiles (right) revealed the ductile nature of Mg-Y-Ti alloys such as Mg89 Y9 Ti2 (55) and Mg80 Y16 Ti4 (89). Scale bars indicate 1 μm.

77 Figure 4-14 shows the XRD spectrum of Mg89 Y9 Ti2 (alloy 55). One can see that even

though this sample contained the same amount of Y as sample 33 and the same amount of Ti as

sample 86, its XRD spectrum was significantly different from either Mg-Y or Mg-Ti. The peaks

were depressed in comparison with commercial Mg, and only the (002) and (103) orientations were observed. This XRD spectrum was characteristic of most Mg-Y-Ti alloys, although some generated a larger (103) peak and showed evidence of the (102) peak.

55

Figure 4-14: Grazing angle XRD scans of fresh EB-PVD Mg-Y-Ti and commercial Mg. Additions of Y and Ti appear to change the relative peak intensities of an alloy’s spectrum. The small peak near 53º on the sample 55 scan is from the silicon substrate.

A few Mg-Y-Ti samples were also imaged after storing them in an evacuated desiccator

for over 1 year. Localized areas of oxidation were visible, but some alloys, like the one shown in

4-15(B), were more oxidized than others. The surfaces of these other alloys remained relatively

78 smooth and uniform, as shown in Figure 4-15(A). Mg-Y-Ti alloys that were stored in the

desiccator continued to exhibit columnar microstructure and tensile properties, as shown in

Figure 4-16.

Attempts were made to obtain XPS information from the alloys’ surfaces, but complications arose due to sample charging. EDS was therefore used as a rough measure of sample oxidation. Some aged Mg-Y-Ti alloys showed a small increase (~ 4 at%) in oxygen content, while others showed no increase at all.

Expanded View Magnified View

Figure 4-15: Aged EB-PVD Mg-Y-Ti alloy surface morphologies. (A) Mg79 Y16 Ti5 – sample 88. (B) Mg86 Y11 Ti3 – sample 91. Scale bars indicate 10 μm.

79

Figure 4-16: Aged EB-PVD Mg-Y-Ti alloy microstructures. Columnar structure (left) and tensile properties (right) are still evident. Scale bar indicates 1 μm.

A few of the thick Mg-Y-Ti EB-PVD specimens are shown in Figure 4-17. The deposits

were columnar and relatively dense, although some porosity can be seen under high

magnification. Alloy T3, which is shown in Figure 4-17(B), contained the most Y and Ti of the alloys pictured, and one can see that its microstructure is somewhat denser than the other thick alloys.

80

Expanded View Magnified View

Figure 4-17: Structures of thick Mg-Y-Ti alloys. Relatively dense columnar structure is observed in the expanded (left) and magnified (left) views. (A) Mg89.7 Y1.1 Ti0.2 – T1. (B) Mg96.6 Y3.0 Ti0.4 – T3. (C) Mg97.2 Y2.5 Ti0.3 – T15. Scale bars indicate 100 μm (left) and 5 μm (right).

81 The thick alloys, however, exhibited defects, such as the cones and surface hillocks shown in Figure 4-18, that were generally absent from the thin alloys. These defects were common in alloy T1, but overall they occurred less frequently in the other thick alloys. The high deposition rate, lack of rotation, and uneven nature of the iron substrate holder may have contributed to their formation.

Side View Top View

Figure 4-18: Defects in thick Mg-Y-Ti alloy T1. Cone defects propagated through the film (left) and manifested themselves as hillocks on the surface (right). Scale bars indicate 100 μm.

The thick alloys generally exhibited different surface morphology, as shown in Figure

4-19. Figure 4-19(A) was more characteristic of alloys with lower Y-Ti content, and Figure

4-19(B) was more characteristic of alloys with higher Y-Ti content. In this way, the thick alloys’ appearance was similar to that of the thin alloys, since alloys with higher Y-Ti content exhibited morphologies that were more uniform and less porous than alloys with lower Y-Ti content.

82

Magnified View Expanded View

Figure 4-19: Surface morphologies of thick Mg-Y-Ti alloys. (A) Images of the surface of T15 reveal smaller, porous morphology. (B) Images of T3 reveal larger, less porous morphology. Scale bars indicate 5 μm.

The thick alloys were also similar to the thin alloys in that they exhibited preferred (002) orientation when examined with x-ray diffraction. Figure 4-20 compares sample T15 with polished commercial Mg, and strong preference was seen for the (002) orientation in comparison to the other peaks. The (103) orientation, however, was also relatively pronounced. The rightmost peak on the T15 scan was unidentified, although a similar peak was also seen in a scan of pure EB-PVD Mg (see Figure 4-3).

83

T15

Figure 4-20: Grazing angle XRD scans of aged EB-PVD T15 and commercial Mg. Mg97.2 Y2.5 Ti0.3 (T15) shows Mg (002) preferred orientation. The rightmost peak on the T15 spectrum is unidentified.

The thick alloys were also imaged after storing them in an evacuated desiccator with

desiccant for more than 1 year. Figure 4-21 reveals relatively little change for alloys T1 and T3 during that time, although surface charging indicated a small amount of oxidation. Charging often occurs on insulating materials such as oxides. EDS did not show a significant increase in the oxygen content of the alloys, although EDS is not generally considered a surface-sensitive

technique.

84

Figure 4-21: Aged thick Mg-Y-Ti alloys. Alloy T1 (left) and T3 (right) show charging during the imaging process, which may indicate a slight increase in oxidation. Scale bars indicate 5 μm.

4.3 Corrosion of Magnesium Alloys

Data obtained from mass loss experiments, electrochemical tests, and scanning electron microscopy are presented here.

4.3.1 Corrosion Rates of WE43 from Mass Loss

Polished samples of WE43 were immersed in beakers of 37ºC HBSS for 1 week; then they were cleaned and weighed to determine their mass loss. The results are presented in Table

4-4. Samples 1 and 2 were of similar initial mass and surface area, and their average corrosion

rate was 3.17 μA/cm2 over the course of the week. Samples 3, 4, 5, and 6 were slightly more massive with a larger surface area, and their average corrosion rate was 1.83 μA/cm2 over the course of the week. As expected, the rate of corrosion for WE43 differed depending on the samples’ initial mass and surface area, but the difference in corrosion rate was less than a factor of two. Edge corrosion did not appear to be significant. The final pH of the bulk solution was

85 approximately 8.0, but small aliquots of solution taken from near the metal surface exhibited a pH

between 9.0 and 9.5.

Table 4-4: Corrosion rates for WE43 samples immersed in 37ºC HBSS for 1 week. Sample # 1 2 3 4 5 6 Initial Mass (g) 1.0741 0.9335 7.4367 10.0729 9.8081 10.0434 Mass Loss (g) 0.0015 0.0014 0.0039 0.0033 0.0038 0.0045 Time (s) 604800 604800 604800 604800 604800 604800 Surface Area (cm2) 4.79 4.53 18.47 22.18 20.31 26.29 Avg. Current 3.19 3.15 2.15 1.52 1.91 1.74 Density (μA/cm2) Avg. Corrosion 3.17 3.12 2.13 1.50 1.89 1.73 Rate (mpy)

Following the acid cleaning, samples of WE43 were imaged with a SEM to determine the mode of corrosion attack. Localized corrosion predominated (see Figure 4-22), which was

primarily in the form of pitting (see Figure 4-23). Pit density was approximately 2 per 10 μm2.

As Figure 4-24 shows, some areas showed evidence of grain undermining, which is also known

as chunking. Undermining is characterized by corrosion of the areas adjacent to the grain

boundaries, because the boundaries themselves are cathodic to the rest of the grain.

86

Figure 4-22: Localized corrosion on WE43 immersed in 37ºC HBSS for 1 week. Scale bar indicates 100 μm.

Figure 4-23: Pitting corrosion on WE43 immersed in 37ºC HBSS for 1 week. Pit density was approximately 2 per 10 μm2. Scale bars indicate 10 μm.

87

Expanded View Magnified View

Figure 4-24: Undermining of grains on WE43 immersed in 37ºC HBSS for 1 week. WE43 grains were generally tens to hundreds of microns wide. WE43’s ASTM grain size was calculated to be between 2.0 and 2.5 (see Figure 4-1). Scale bars indicate 10 μm.

4.3.2 Corrosion Rates from Linear Polarization and EIS

Both linear polarization and EIS were used to determine approximate corrosion rates for

the Mg alloys in HBSS at 37±2ºC. Linear polarization and EIS were performed on the same

sample in a sequential manner, since both tests are nondestructive. The pH of the testing solution

was not observed to increase during the measurements. Figure 4-25 shows that samples were generally tarnished after testing.

Before After

Figure 4-25: Mg-Y-Ti sample before and after linear polarization and EIS tests. Alloys generally appeared tarnished (right) when compared to their initial appearance (left).

88 Table 4-5 compares the open circuit potential (OCP) of alloys that were tested shortly

after deposition (fresh) and again after storing them in a desiccator for 1+ years (aged). The

values were obtained as soon as the corrosion reaction reached steady-state, which generally

occurred 5-15 minutes after submersion. The potentials of aged alloys were generally

comparable to the potentials of fresh alloys. All alloys exhibited a corrosion potential nobler than

that of WE43. As expected, the Mg-Ti and Mg-Y-Ti alloys exhibited the noblest potentials.

Table 4-5: Average OCPs of fresh and aged Mg alloys in 37ºC HBSS

Alloy Fresh Aged Designation Composition Type # samples OCP (V) # samples OCP (V) - WE43 - 7 -1.645 n/a 24 Mg100 2 -1.643 0 - Mg 90 Mg100 2 -1.581 3 -1.589 35 Mg95 Y5 4 -1.604 0 - Mg-Y 33 Mg91 Y9 2 -1.596 4 -1.580 Mg-Ti 86 Mg98 Ti2 6 -1.520 6 -1.518 57 Mg92 Y7 Ti1 2 -1.551 0 - 87 Mg90 Y8 Ti2 2 -1.556 4 -1.558 Mg-Y-Ti 55 Mg89 Y9 Ti2 6 -1.557 0 - (dual gun) 91 Mg86 Y11 Ti3 6 -1.549 3 -1.542 89 Mg80 Y16 Ti4 2 - 5 -1.526 T1 Mg98.7 Y1.1 Ti0.2 2 -1.553 2 -1.559 Mg-Y-Ti T15 Mg97.2 Y2.5 Ti0.3 1 -1.554 3 -1.533 (Sciaky) T3 Mg96.6 Y3.0 Ti0.4 2 -1.562 2 -1.544 T16 Mg95.8 Y3.7 Ti0.5 1 -1.521 2 -1.539

Table 4-6 and Table 4-7 compare corrosion rates obtained from linear polarization (LP) and EIS tests, respectively, for fresh and aged alloys shortly after submersion in 37ºC HBSS.

One can see that the aged corrosion rates are generally lower than the fresh rates by a factor of 2 to 4. Furthermore, the corrosion rates obtained using EIS were generally comparable to or lower than the corrosion rates obtained using LP. The standard deviation of the mean (SDoM) was computed as a measure of the accuracy of the mean based on the number of values from which it

89 was calculated. The standard deviation of the mean was calculated according to Equation 4-1,

where N is the number of sample values.

σ σ = X (4-1) X N

The tables show that fresh Mg-Y-Ti alloy 55, aged Mg-Ti alloy 86, and aged Mg-Y-Ti alloys 87, 89, 91, and T3 corroded slower than or at a similar rate to freshly polished WE43.

Corrosion rates lower than the one observed for WE43 are emphasized in bold italics. Alloys 55 and 57 were consumed in other tests, so none remained after one year for aged testing.

Table 4-6: Average linear polarization corrosion rates of fresh and aged Mg alloys in 37ºC HBSS less than one hour after submersion

Fresh Alloy Aged Alloy Composition Design. # Avg. # Avg. SDoM SDoM Mg Y Ti samples μA/cm2 samples μA/cm2 WE43 93 4 - 7 17 15 n/a 24 100 - - 2 67 5 0 - - 90 100 - - 2 103 6 3 26 11 35 95 5 - 4 806 252 0 - - 33 91 9 - 2 73 16 3 44 26 86 98 - 2 4 59 22 5 7 2 57 92 7 1 2 105 10 0 - - 87 90 8 2 1 24 - 4 10 3 55 89 9 2 5 8 1 0 - - 91 86 11 3 5 19 2 4 15 3 89 80 16 4 0 - - 5 3 0.2 T1 98.7 1.1 0.2 1 103 - 3 64 6 T15 97.2 2.5 0.3 1 151 - 4 64 9 T3 96.6 3.0 0.4 1 139 - 2 29 0.1 T16 95.8 3.7 0.5 2 378 131 2 170 7 Bold italics indicate corrosion rates lower than the corrosion rate of WE43. See Appendix B for corrosion rates that have been converted into mpy and μm/day.

90

Table 4-7: Average EIS corrosion rates of fresh and aged Mg alloys in 37ºC HBSS less than one hour after submersion

Fresh Alloy Aged Alloy Composition Design. # Avg. # Avg. SDoM SDoM Mg Y Ti samples μA/cm2 samples μA/cm2 WE43 93 4 - 5 22 11 n/a 24 100 - - 1 83 - 0 - - 90 100 - - 2 62 26 3 21 8 35 95 5 - 2 185 31 0 - - 33 91 9 - 0 - - 2 26 1 86 98 - 2 2 12 6 2 3 0.4 57 92 7 1 1 32 - 0 - - 87 90 8 2 1 3 - 3 5 1 55 89 9 2 2 10 5 0 - - 91 86 11 3 4 11 1 4 12 2 89 80 16 4 0 - - 4 3 0.2 T1 98.7 1.1 0.2 1 72 - 3 32 2 T15 97.2 2.5 0.3 1 118 - 3 36 2 T3 96.6 3.0 0.4 1 74 - 2 16 0.5 T16 95.8 3.7 0.5 2 206 51 2 99 9 Bold italics indicate corrosion rates lower than the corrosion rate of WE43. See Appendix B for corrosion rates that have been converted into mpy and μm/day.

Linear polarization and EIS data was also obtained as a function of immersion time for

WE43 and selected thin and thick alloys in HBSS at 37±2ºC over a period of several hours. The data presented here are the curves showing the lowest corrosion rates with time that were obtained from the given alloys. Note that the curve shown for WE43 demonstrates corrosion rates that are significantly lower than WE43’s average corrosion rate upon submersion. Samples of WE43 were observed to corrode over a range of values, which can be seen in the SDoM parameters in Tables 4-6 and 4-7. EB-PVD alloys that corroded at similar rates generally exhibited less variation.

EIS scans were run immediately following linear polarization scans, so linear polarization and EIS corrosion rates were calculated using data from the same sample. The data for the thin and thick alloys, except alloy 55, were obtained from aged material. Figures 4-26 and 4-27 show

91 LP and EIS corrosion rates, respectively, as a function of time. Aged thick alloys T1 and T15 corroded at a much higher rate than WE43, fresh alloy 55, aged alloy 89, and aged alloy 91, and these high corrosion rates persisted for the duration of the experiments. The thick alloys’ EIS corrosion rates (Figure 4-27) showed a clear increasing trend, although the rate of T15 was increasing faster than T1.

140 WE43B 55 - Mg89 Y9 Ti2 120 89 - Mg80 Y16 Ti4 91 - Mg86 Y11 Ti3 )

2 T1 - Mg98.7 Y1.1 Ti0.2 100 T15 - Mg97.2 Y2.5 Ti0.3

80

60

40 Corrosion Rate (µA/cm Corrosion

20

0 0246810 Time (hrs) Figure 4-26: Corrosion rates determined by linear polarization as a function of time. Thick alloys corroded at the fastest rate. Data was obtained from aged EB-PVD alloys (except 55, which was fresh).

92

140 WE43B 55 - Mg89 Y9 Ti2 120 89 - Mg80 Y16 Ti4 91 - Mg86 Y11 Ti3 )

2 T1 - Mg98.7 Y1.1 Ti0.2 100 T15 - Mg97.2 Y2.5 Ti0.3

80

60

40 Corrosion Rate (µA/cm Corrosion

20

0 0246810 Time (hrs) Figure 4-27: Corrosion rates determined by EIS as a function of time. Thick alloys exhibit an increasing trend in corrosion rate. Data was obtained from aged EB-PVD alloys (except 55, which was fresh).

Figures 4-28 and 4-29 are magnifications of Figures 4-26 and 4-27, respectively. They more clearly show the corrosion rate behavior of WE43 and the thin alloys with time. Both LP

(4-28) and EIS (4-29) data showed that the corrosion rate of WE43 is relatively stable over the first few hours of submersion. The corrosion rate of aged thin alloy 89 was generally increasing, but fresh alloy 55 and aged alloy 91 were generally decreasing. Overall, aged alloy 91 appeared comparable to WE43, but the extended corrosion rates for fresh alloy 55 were lower than those of

WE43.

93

16 WE43B 14 55 - Mg89 Y9 Ti2 89 - Mg80 Y16 Ti4

) 12 91 - Mg86 Y11 Ti3 2

10

8

6

Corrosion Rate (µA/cm 4

2

0 0246810 Time (hrs) Figure 4-28: LP corrosion rates for thin alloys as a function of time. Alloy 55 showed lower corrosion rates over time when compared to WE43. Alloys other than 55 and WE43 were aged.

12 WE43B 55 - Mg89 Y9 Ti2 10 89 - Mg80 Y16 Ti4

) 91 - Mg86 Y11 Ti3 2

8

6

4 Corrosion Rate (µA/cm

2

0 0246810 Time (hrs) Figure 4-29: EIS corrosion rates for thin alloys as a function of time. Alloy 55 showed lower corrosion rates over time when compared to WE43. Alloys other than 55 and WE43 were aged.

94 4.3.3 Potentiodynamic Polarization

Potentiodynamic scans were run on WE43, thin alloys, and thick alloys. Figure 4-30,

which shows a Mg-Y-Ti alloy before and after testing, shows that potentiodynamic scans are

destructive, so these scans were run after linear polarization and/or EIS scans. The pH of the

testing solution was occasionally observed to have a slightly elevated pH (~7.50) after the

measurement was complete.

Before After

Figure 4-30: Mg-Y-Ti sample before and after a potentiodynamic polarization test. Potentiodynamic polarization scans are destructive tests.

Before running the scans, it was necessary to evaluate the effects of solution resistance on the measurements. Even if the solution resistance is small, large currents lead to large voltage drops (V = I x R), which can complicate measurements. Gamry Framework, however, can compensate for the solution resistance while running polarization scans (IR compensation).

Figure 4-31 compares potentiodynamic polarization scans of WE43 with and without IR

compensation. One can see that IR effects became significant near 3e-3 A/cm2. Therefore, this

value was used to truncate potentiodynamic data.

95

-1.40 No IR compensation IR compensation -1.45 IR error -1.50

-1.55 Voltage (V) Voltage -1.60

-1.65

-1.70 1e-9 1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1

2 log Current Density (A/cm ) Figure 4-31: IR compensation for potentiodynamic polarization scans of WE43. The plot reveals IR effects at currents above 3e-3 A/cm2.

Figure 4-32 compares potentiodynamic polarization curves for Mg98 Ti2 (alloy 86) and

Mg91 Y9 (alloy 33), which were obtained more than one year after the alloys were first deposited. One can see that the aged Mg-Ti alloy exhibited what appeared to be passive behavior with an average passive current density of 24 μA/cm2. The aged Mg-Y alloy also exhibited a polarization curve that appeared to be passive, but in this case the current density was almost one order of magnitude higher. Observation while testing verified rapid corrosion of the Mg-Ti alloy once breakdown occurred, but this may be a result of the thickness of the alloy. Alloy 86 was relatively thin (~ 2.3 μm), so 2-D pits quickly formed and propagated without opportunity for repassivation. Sample 33 was somewhat thicker (~ 3.0 μm).

96

-1.30 86 - Mg98 Ti2 33 - Mg91 Y9 -1.35

-1.40

-1.45 Voltage (V) -1.50

-1.55

-1.60 1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2

2 log Current Density (A/cm ) Figure 4-32: Potentiodynamic scans of aged Mg-Y and Mg-Ti EB-PVD alloys. Curves were obtained more than 1 year after the alloys were originally deposited.

Aged thick alloy T3 also demonstrated what appeared to be passivity when tested after being stored in a desiccator for 1 year. This can be seen in Figure 4-33. The other aged thick alloys exhibited similar potentiodynamic polarization curves that did not indicate passivity.

97

-1.35 T1 - Mg98.7 Y1.1 Ti0.2 T15 - Mg97.2 Y2.5 Ti0.3 T3 - Mg96.6 Y3.0 Ti0.4 -1.40 T16 - Mg95.8 Y3.7 Ti0.5

-1.45

Voltage (V) -1.50

-1.55

-1.60 1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2

2 log Current Density (A/cm ) Figure 4-33: Potentiodynamic scans of aged thick alloys. Potentiodynamic curves were measured after storing samples for 1 year in a desiccator. Alloy T3 tended to be nobler than the other thick alloys. It also demonstrated an apparent passive region.

Figure 4-34 compares potentiodynamic polarization curves from alloys with the lowest corrosion rates. All curves except those from alloy 55 and WE43 were from aged alloys. The curve from alloy 55 was obtained shortly after deposition. Note that the potentiodynamic scan of

Mg98 Ti2 (alloy 86) closely overlays the scan of Mg80 Y16 Ti4 (alloy 89).

One can see from the figure that the Mg-Ti and Mg-Y-Ti alloys exhibited nobler OCPs and higher breakdown potentials when compared with WE43. In addition, alloy 55 exhibited a comparable passive current density to WE43. Table 4-8 compares average OCPs, breakdown

potentials (Eb), passive regions (OCP - Eb), and passive current densities (ipass) for WE43 and EB-

PVD alloys with the lowest corrosion rates. Almost all of the EB-PVD alloys demonstrated

nobler OCPs and breakdown potentials, wider passive regions, and comparable passive current

98 densities to WE43. Caution, however, should be taken when comparing thin alloys with bulk

materials, because they may undergo different modes of corrosion [82].

-1.30 86 - Mg98 Ti2 55 - Mg89 Y9 Ti2 -1.35 89 - Mg80 Y16 Ti4 T3 - Mg97.2 Y2.5 Ti0.3 -1.40 WE43B

-1.45

-1.50

Voltage (V) -1.55

-1.60

-1.65

-1.70 1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2

2 log Current Density (A/cm ) Figure 4-34: Potentiodynamic scans comparing selected EB-PVD alloys to WE43. Aged EB- PVD alloys (except 55, which was fresh) exhibited a nobler OCP and higher breakdown potential than WE43. Passive current densities were comparable. Curves presented here exhibit the best passivity obtained from the given alloys.

Table 4-8: Average OCP, Eb, passive region, and ipass for selected Mg alloys in 37ºC HBSS # PD Avg. Avg. E Avg. Pass Avg. i Alloy Aged b pass Scans OCP (V) (V) Reg (mV) (μA/cm2) WE43 n/a 3* -1.638 -1.576 62 20 Mg98 Ti2 (86) yes 2 -1.511 -1.423 88 24 Mg89 Y8 Ti2 (55) no 2 -1.538 -1.494 44 13 Mg80 Y16 Ti4 (89) yes 2 -1.537 -1.463 74 56 Mg97.2 Y2.5 Ti0.3 (T3) yes 2 -1.534 -1.464 71 103 *Only two scans demonstrated a passive region. The third scan was not used to calculate Eb, Pass Reg, or ipass.

4.3.4 Mode of Corrosion Attack

Non-uniform corrosion is known to be the most common mode of attack on pure magnesium, because magnesium’s surface oxide is only partially protective. Magnesium alloys may favor uniform corrosion, but they still undergo a certain amount of localized attack. Figure

4-35 shows the surface of a Mg-Y-Ti alloy after being submerged in HBSS for 40 minutes.

Pitting was observed after adsorbed salts had been cleaned from the alloy’s surface. EDS

identified the primary constituents of the adsorbed salts as Na, Cl, Mg, and O; however, trace

amounts of C, Ca, K, P, S, and Y were also identified.

Figure 4-35: Pitting observed on a Mg-Y-Ti alloy exposed to HBSS for 40 minutes. Pictures show the alloy’s surface before (left) and after (right) cleaning. EDS identified adsorbed salts on the alloy’s surface (left), and pitting was observed after cleaning (right). Scale bar indicates 10 μm.

EDS also identified C, Ca, Cl, Mg, Na, O, P, Ti, and Y on the surfaces of Mg-Y-Ti alloys

after potentiodynamic polarization tests. C, Mg, and O were often the most abundant elements in

these analyses, one of which is presented in Table 4-9. The elements were more prevalent in

100 areas of localized corrosion, like the one shown on the right side of Figure 4-36, but they were also identified on relatively unaffected surfaces, like those shown on the left side of Figure 4-36 and in Figure 4-37.

Table 4-9: EDS analysis of the surface components of alloy Mg89 Y9 Ti2 after potentiodynamic testing. Element Wt% At% C 7.9 14.8 Ca 1.6 0.9 Cl 0.7 0.5 Mg 63.7 59.3 Na 0.5 0.5 O 14.2 20.0 P 1.4 1.1 Ti 1.6 0.8 Y 8.2 2.1

Expanded View Magnified View

Figure 4-36: Aged alloy T3 after potentiodynamic testing. Localized corrosion was observed, but other areas of the alloy were relatively unaffected. Scale bars indicate 50 μm.

101

Untested Surface Tested Surface

Figure 4-37: Unaffected area of aged alloy T3 after potentiodynamic testing. The tested surface (right) appeared relatively unaffected when compared to the untested surface (left). Scale bar indicates 5 μm.

Several thin EB-PVD samples were subjected to linear polarization and EIS tests, and they were then cleaned according to ASTM G1. Figure 4-38 shows the cleaned alloy surfaces.

The porous surface morphology of the pure Mg deposits is still evident in pictures A and B.

Picture C shows an Mg-Y alloy that underwent uniform corrosion with areas of localized attack.

The alloy appeared to crack due to corrosion or cleaning. Picture D shows an Mg-Ti alloy that also underwent uniform corrosion with areas of localized attack. Its surface porosity was still evident. Pictures E-F show the surfaces of various Mg-Y-Ti alloys after testing and cleaning.

Mg-Y-Ti alloys underwent uniform corrosion with areas of localized attack that were comparable in size to the areas observed on WE43 when tested under similar conditions.

The thick Mg-Y-Ti EB-PVD alloys were cleaned and imaged after being submerged in

37ºC HBSS for 16 hours, as shown in Figure 4-39. Results were similar to those observed for the

thin alloys in that corrosion was generally uniform, but areas of localized attack were observed.

Pictures A and B show areas of localized attack, and pictures C and D show areas of general

corrosion.

102

Figure 4-38: EB-PVD Mg alloy surfaces after linear polarization and EIS testing. Alloys underwent uniform corrosion with areas of localized attack. (A-B) Mg, (C) Mg-Y, (D) Mg-Ti, (E-H) Mg-Y-Ti. Scale bar indicates 5 μm.

103

Expanded View Magnified View

Figure 4-39: Thick alloy surfaces after being submerged in 37ºC HBSS for16 hours. (A-B) Some areas were characterized by localized corrosion. (C-D) Areas of general corrosion were also observed. Scale bars indicate 5 μm.

104 4.4 Cytotoxicity of Magnesium Alloys

Cytotoxicity refers to the property of being toxic to cells. Cytotoxic compounds can induce apoptosis (an internal process initiated by a cell that leads to cell death) or necrosis (a process caused by external factors that also leads to cell death). The concentration of Mg and alloying elements in the growth media and/or the rate of corrosion are potential cytotoxic factors for cells that were cultured on Mg alloy substrates.

Glass control substrates were used to differentiate between the influence of element concentration and corrosion rate on cell morphology and viability. Cells were grown on a glass substrate in media containing dissolved WE43 to determine whether dissolved elements influenced cell morphology and viability. Cells grown on this glass substrate were then compared with cells grown on WE43. Cells grown on WE43 experienced the corrosion process in addition to elevated concentrations of dissolved elements.

4.4.1 Effects of Elemental Concentrations in Growth Media

A549 cells were grown on a glass slide in normal culture medium and a glass slide in 6 mL of medium containing 6 mg of dissolved WE43. 6 mg was the mass of WE43 that normally dissolved in the growing media during the duration of a cell culture.

Figure 4-40 shows pictures of A549 cells grown on the glass slide in the normal medium.

The left image is a magnification of a clump of cells that were originally grown in the T25 culture flask and then transferred onto the glass slide. These cells maintained their attachment to each other in spite of the trypsin rinse, so they reflect the morphology of cells grown for several days in the T25 culture flask. The right image, however, shows a healthy monolayer of A549 cells

105 grown on the glass control substrate. Normal pancake-like A549 cell morphology and healthy

viability [156] were observed on the control substrate.

Figure 4-40: A549 cells grown on the glass control specimen. Normal A549 cell morphology and viability were observed. Scale bars represent 10 μm.

Figure 4-41 shows pictures of cells grown on the glass slide in media containing dissolved WE43. Most cells appeared healthy and were clearly attached to the substrate.

Comments cannot be made about the density of the cells, because the density of cells in the original medium was unknown. The concentration of Mg and other elements in the growing medium did not appear to be detrimental, because cells appeared to exhibit normal morphology and viability.

106

Figure 4-41: A549 cells grown on the glass WE43 control specimen. Normal cell morphology and viability were observed. Scale bars represent 10 μm.

4.4.2 Effects of Corrosion on Cell Morphology and Viability

Figure 4-42 displays images of cells grown on WE43. Some cells appeared indeterminate (left)—they may be healthy or apoptotic. An indeterminate cell exhibited an intact membrane, which is consistent with signs of a healthy cell, but its morphology was ball-like, which is not indicative of a healthy cell. Other cells exhibited apparent surface blebs, or fibrous textures, that indicated they were undergoing apoptosis (right). WE43’s corrosion process therefore appeared to harm most cells and prevent them from forming a uniform monolayer on the alloy surface.

107

Figure 4-42: A549 cells grown on WE43. Some cells appeared indeterminate (left), while others appeared to be apoptotic (right). Overall, WE43’s corrosion process appeared to harm the cells. Scale bars indicate 10 μm.

Cells grown on thick alloy T15, which are shown in Figure 4-43, appeared worse than those grown on WE43. Most cells grown on the thick alloys exhibited a ball-like morphology and clear breaches in their plasma membranes. In addition, cells did not appear to be attached to the underlying substrate. These cells demonstrated a necrotic appearance, and one may assume that they were damaged by the relatively high corrosion rate of alloy T15.

Figure 4-43: A549 cells grown on fresh thick alloy T15. Cells appeared apoptotic or necrotic. T15’s corrosion process appeared to harm the cells. Scale bars represent 5 μm.

108 The A549 cells grown on the thin alloys, however, were remarkably different. Their appearance was very similar to the morphology of the cells grown on the glass control substrate.

Figure 4-44 shows a clear monolayer of cells on alloy Mg98 Ti2. The cells were obviously

viable, and they demonstrated healthy, pancake-like morphologies.

Magnified View Expanded View

Figure 4-44: A549 cells grown on aged alloy Mg98 Ti2. A clear monolayer of cells covers the alloy. Scale bars represent 10 μm.

Cells grown on Mg-Y-Ti alloys also demonstrated viability and healthy morphologies.

Cells were observed to proliferate, which is shown by the cell undergoing mitosis in the left image of Figure 4-45. A monolayer of cells is shown in the right image. The cells are clearly

attached to the alloy, as shown in Figure 4-46, and their healthy pancake-like morphology is also

evident.

109

Figure 4-45: Proliferation and viability of A549 cells grown on aged alloy Mg89 Y9 Ti2. A549 cells were observed to proliferate (left) and form a monolayer (right) on Mg-Y-Ti alloys. Scale bars indicate 5 μm.

Figure 4-46: Attachment and morphology of A549 cells grown on aged alloy Mg86 Y11 Ti3. The cells are clearly attached to the alloy, and they are observed to demonstrate a healthy, pancake-like morphology. Scale bars indicate 10 μm.

110

Chapter 5

Summary and Discussion

A general discussion of the results is presented here. As a general reminder to the reader, freshly-deposited EB-PVD alloys are referred to as “fresh” alloys, and EB-PVD alloys that had been stored in a desiccator for approximately one year are referred to as “aged” alloys.

5.1 Experimental Error

Error must be considered when discussing measurements. Among other factors, error in corrosion rate calculations may be attributed to assumed Tafel slopes, linearization of the current- potential equation, neglecting reverse reactions, neglecting the solution resistance, specimen-to- specimen variation, sample surface area, and crevice corrosion.

The Stern-Geary equation [157] (equation 3-3, also shown below) was used to calculate

an alloy’s corrosion rate (icorr) in 37ºC HBSS from its polarization resistance (Rp). However, the

calculation required values for the anodic and cathodic Tafel slopes (βa and βc, respectively),

which must be experimentally obtained from potentiodynamic polarization curves. The curves

presented in Section 4.3.3 did not contain cathodic regions, because it was observed that single-

sweep cathodic-anodic potentiodynamic curves exhibited nobler OCPs than those observed

during steady-state measurements. In addition, the curves’ anodic regions did not exhibit defined

anodic Tafel slopes across at least one decade of current [158], and cathodic regions (not shown)

demonstrated behavior that indicated concentration polarization. Therefore, both βa and βc were assumed to equal 0.1 V per decade of current. These are common assumptions, and they generally yield results with a maximum constant error of a factor of two [148].

111

× ββ ca icorr = (3-3) 32 R. (+×× ββ cap )

Other sources of error, particularly for the linear polarization method, include (i) error caused by linearization of the current-potential equation—that is, measuring Rp as the slope of the

region around Ecorr instead of the differential slope as ΔE→0, (ii) error caused by neglecting

reverse reactions, and (iii) error caused by neglecting the solution resistance. These sources are

not expected to introduce significant error to these corrosion rate calculations for the following

reasons: (i) current-potential curves exhibited linear behavior about Ecorr, so the difference between the approximated linear slope and differential slope was small, (ii) magnesium possesses a large thermodynamic propensity for corrosion, so reverse reactions are unlikely, and (iii) the solution resistance (~ 50Ω to 100Ω by EIS) was generally 1% to 5% of the measured polarization

resistance. Ref. [159] contains a more detailed discussion of error.

Song [67] has argued that linear polarization cannot be used to measure the corrosion rate

of magnesium, but the literature does not appear to contain similar claims from other researchers.

As discussed in Section 2.2.3, the corrosion rates presented here are not presented as purely

accurate measurements; rather, they are presented as relative values for comparison. As will be

discussed later, however, the electrochemical corrosion rates obtained for WE43 upon

submersion were slightly higher than those obtained from mass loss specimens over the course of

a week. Therefore, the electrochemical corrosion rates appeared to be accurate.

Specimen-to-specimen variation, crevice corrosion, and sample surface area were likely

the largest contributors to error. Defects among samples may have been a factor; although care

was taken to ensure that corrosion rates resulting from defective samples were not used to

calculate average values. Compositional gradients among samples from the same alloy may have

also been a factor, but substrate rotation limited gradient effects. Furthermore, care was taken to

112 disregard data from samples that demonstrated crevice corrosion, but EB-PVD alloys were

difficult to examine because of their small thicknesses. Sample surface area was another source

of error, especially for the EB-PVD alloys, since surfaces were assumed to be planar. As one can

see from the SEM images, EB-PVD alloys often exhibited structured, not planar, surface

morphology. Therefore, surface areas may have been underestimated, and EB-PVD alloys’

corrosion rates may actually be lower than those reported.

5.2 Factors Influencing Corrosion Behavior

A multitude of factors influence an alloy’s corrosion behavior, but this section will focus primarily on microstructure, surface morphology, and passive film formation.

5.2.1 Microstructure and Surface Morphology

The microstructure and surface morphology of PVD alloys are widely recognized as factors that influence corrosion behavior, but these factors in turn depend on the processing history and composition of a deposit. Processing history is particularly important.

Miller [142] showed that preexisting defects act as limiting factors for PVD magnesium’s corrosion resistance, and depending on their severity, these defects can overcome the beneficial participation of alloying elements in the corrosion products [138]. Figure 5-1 compares two samples of EB-PVD alloy 55 and two samples of WE43, as well as their corresponding corrosion rates. A scratch defect is clearly visible in the lower right corner of the right image of alloy 55, and splotches are clearly visible on the surface of the right image of WE43. One can see that the pictures on the right side of the figure contain defects, and these defects are correlated with larger corrosion rates as determined by linear polarization (LP). Corrosion rates were not included in

113 the averages reported in Table 4-6 and Table 4-7 when they were determined to be affected by the surface preparation of the sample.

No Visible Defects Visible Defects 55 55

WE43 WE43

Figure 5-1: Surface defects correlated with corrosion rate. Top images compare two samples of alloy Mg89 Y9 Ti2 (55), and bottom images compare two samples of WE43. Defects present on the samples (right images) are clearly correlated with larger corrosion rates.

Several possibilities exist for the creation of defects in the EB-PVD alloys. Particles or

defects in the substrate could nucleate columns that shadow their surrounding area and grow into

cone-shaped defects, like the one shown in Figure 4-18. From the surface, these cones look like

small hillocks. In addition, the thick alloys exhibited noticeable striations on their surfaces,

which were correlated to the polish lines on the surface of their iron substrate, as shown in Figure

5-2. These defects could be partially responsible for the noticeably higher corrosion rate among

114 the thick alloys. As mentioned previously, the initial results for the thick alloys were promising,

but the alloys still need to be optimized.

Figure 5-2: Striations on the rear surface of a thick EB-PVD alloy. These striations were a result of the polish lines on the iron substrate holder. Striations propagated through the deposit to become surface defects. Scale bar indicates 200 μm.

Defects could also be introduced to the alloys when they were cleaved for testing. The scratch on alloy 55, shown in the upper right image of Figure 5-1, was likely introduced to the sample during cleaving. Such defects have been shown to induce pitting [142], which was the primary mode of corrosion for the EB-PVD alloys, as shown in Figure 5-3. Baldwin et al. [141]

also noted that columnar pores in the microstructure may induce pitting, so alloys with smooth,

uniform surfaces and dense microstructures are expected to corrode at lower current densities.

These types of structure were most prominent in alloys deposited on a rotating substrate. At

present, the effects of substrate rotation have not been reported elsewhere in the literature.

115

14 μA/cm2 59 μA/cm2

Figure 5-3: Pitting on EB-PVD Mg-Y-Ti and Mg-Ti EB-PVD alloys. Both surface defects and columnar pores have been linked to pitting on PVD magnesium. Mg-Y-Ti (left) and Mg-Ti (right) alloys show pitting after linear polarization and EIS testing. Corrosion rates are averages of values obtained from linear polarization tests performed on fresh alloys. Scale bars indicate 10 μm.

Figure 5-4 shows the microstructure and surface morphology of two Mg-Y-Ti alloys that were deposited with (left) and without (right) substrate rotation. Both alloys exhibited columnar microstructure, but alloy Mg89 Y9 Ti2 (55), shown on the left, exhibited a very smooth surface, whereas alloy Mg92 Y7 Ti1 (57) exhibited a more porous and granular surface. The alloys have a similar composition and were deposited at similar deposition rates (see Table 4-1), but one can see that alloy 55, with its more uniform surface morphology, corroded at 10 μA/cm2, whereas alloy 57 corroded at 32 μA/cm2. Therefore, substrate rotation, in combination with alloying, appeared to generate alloys with denser microstructures, smoother surface morphologies, and lower corrosion rates. The mechanisms for this behavior are not immediately clear, but rotation is known to ensure the most even distribution of elements within the alloy. It may also disrupt magnesium’s preferred growth by varying the impingement angle of the incoming vapor flux.

116

Figure 5-4: Effects of substrate rotation on microstructure and surface morphology. Alloy Mg89 Y9 Ti2 (55), shown on the left, was rotated during deposition, whereas alloy Mg92 Y7 Ti1 (57), shown on the right, was not rotated. Smoother surface morphology and denser microstructure was observed when alloys were rotated, and these alloys exhibited lower rates of corrosion Scale bars indicate 1 μm.

Further observations regarding the microstructure and surface morphology of the PVD alloys may be drawn from magnesium’s physical properties [84] and Thornton’s structure zone model (SZM – see Figure 2-8) [129]. EB-PVD pure magnesium nucleates and grows somewhat

differently than other vapor deposited metals, perhaps as a result of its low melting temperature.

Indeed, magnesium was observed to evaporate by sublimation, not melting. Magnesium melts at

a lower temperature (650ºC) than other common metals (iron melts at 1535ºC), so at a given

substrate temperature the T/TM ratio for Mg is larger than the T/TM ratio for other metals.

Therefore, Mg adatoms (i.e. adsorbed atoms) should be relatively mobile, even when the

substrate temperature is somewhat low. However, it was observed that pure Mg deposits

demonstrated preferential growth, which is characteristic of materials with low adatom mobility.

Preferential growth was seen more clearly when substrates were oriented at oblique angles, as

seen in Figure 5-5, even when the angle was as small as 5º from perpendicular to the vapor flux.

This type of growth deviated from behavior predicted by the structure zone model (see Figure 2-

8) [135], but Mg alloys appeared to behave as predicted by the SZM.

117

Figure 5-5: Sculptured magnesium deposited at an oblique angle to the incoming vapor flux. Pure PVD magnesium exhibited preferred orientation, which appeared to be a result of its growth mechanism. Scale bar indicates 10 μm.

Deposition temperatures often reached 100-200ºC in the dual-gun system and 450ºC in the Sciaky industrial prototype system. In the dual-gun system, one may assume that the substrate temperature also approached 100-200ºC, which means the T/TM ratio was

approximately 0.25. Ion bombardment was not a factor for EB-PVD deposits, so the T/TM ratio

places alloys made with the dual-gun system in the Zone-T–Zone-II transition of the SZM.

The thick alloys, however, were deposited on a cooled substrate, so one would expect

different mechanisms to direct nucleation and growth as the alloy grew in thickness. Figure 5-6

shows that the thick alloys did indeed appear to grow by multiple mechanisms. Frames A, B, and

C show that laminar region(s) appeared to separate the region adjacent to the substrate from the

region closer to the vapor flux. The bottom arrow in frame C points to a dark laminar region, and

the top arrow points to a horizontal fracture line. Part of alloy T16 fractured horizontally (frames

A and B), and one can see that the fracture plane appeared relatively smooth. Furthermore, if the

surface temperature of the growing thick alloy approached the final deposition temperature of the

Sciaky system, then the T/TM ratio for magnesium adatoms would be approximately 0.70. This

118 ratio correlates with Zone-II of the SZM, and columnar grains were indeed visible on thick alloy surfaces (frame D). Some recrystallization may have even occurred (frame D), which is consistent with Zone-III behavior. Increased temperature during deposit growth has been shown to produce Mg alloys that transition from Zone-II to Zone-III structures [160].

Expanded View Magnified View A B

Side View Top View C D

Figure 5-6: Changes in thick alloy microstructure with growth. (A-B) Alloy T16 fractured along the substrate-side of the deposit. The fracture surface was smooth, perhaps due to changes in the growth mechanism along that plane. (C) Distinct laminar regions were visible along the region of alloy T15 that was closest to the cooled substrate. (D) The surface of T3 demonstrated Zone-II and possible Zone-III morphology. Scale bars indicate 20 μm.

As alluded to previously, composition was also observed to influence EB-PVD alloy microstructure and morphology. As the content of alloying elements increased, alloy surfaces

119 became smoother and microstructural surface features became less prominent. This phenomenon has been observed by others [139], and as Pursel [135] notes, only a small amount of alloying element is required to completely change the growth mechanisms from those observed for pure magnesium. Figure 5-7 shows surface morphologies for EB-PVD Mg, Mg-Ti, Mg-Y, and

Mg-Y-Ti. One can clearly see that the deposits containing Ti and/or Y exhibited a smoother, more granular surface morphology when compared to the pure Mg deposit. The average linear polarization corrosion rates of the fresh material are also included in the figure, and they show that alloys Mg98 Ti2 (86) and Mg91 Y9 (33) exhibited comparable corrosion rates to pure Mg despite differences in surface morphology. Alloy Mg89 Y9 Ti2 (55), however, exhibited both a smooth surface morphology and low corrosion rate. Others have found that PVD Mg alloys containing more than 5 wt% Ti [140] or 22 wt% Y [142] demonstrated poorer corrosion behavior, but the alloys presented in this thesis contain less than those amounts. It may be that alloy stress, which is discussed later, overcomes the beneficial contributions of alloying by disrupting the passive film [132]. Benefits may only become evident after stress is relieved.

120

Figure 5-7: Comparison of surface morphologies of EB-PVD alloys. Alloying elements were observed to smooth the surface morphology of EB-PVD Mg, but the relationship between composition and fresh corrosion rates is not immediately clear. Scale bars indicate 2 μm.

The nucleation and growth of PVD magnesium is complex, and these processes seem to become more complicated with the addition of alloying elements [135]. Figure 5-8, however, shows that alloying elements in the EB-PVD alloys did not precipitate in second phases, and EB-

PVD Mg alloys continued to show the preferential Mg (002) orientation exhibited by EB-PVD pure Mg. Several authors [81, 139, 142] have reported preferential (002) orientation for PVD Mg alloys, and others [142, 161] have also observed that second phase precipitates do not appear in freshly-deposited PVD Mg-Y alloys with less than 10 atomic percent yttrium.

In regard to the PVD Mg alloys presented in this thesis, the diffraction peaks were often shifted relative to their expected location. Peak shifts were expected due to the substitutional

121 defects introduced by the alloying elements, but Miller [142] found that peak shifts for PVD

Mg-Y alloys differed from those calculated for purely substitutional effects. He pointed out that

shifts can be the result of misalignment of the x-ray system, variations in solute composition, or

phase separation, but they are most likely the result of stress in the alloys. Moreover, peaks were

often broader than the peaks of bulk Mg, which also indicates alloy stress [139]. Wolfe [132]

correlated stress with poor corrosion behavior, but this will be discussed further in Section 5.3.

Figure 5-8: XRD scans of EB-PVD alloys showing preferential (002) orientation. EB-PVD Mg and Mg alloys show preferred (002) orientation and broader diffraction peaks when compared with bulk magnesium.

In general, most of the alloys exhibited Zone-T or Zone-II dense, columnar microstructures, which often include a certain amount of tensile stress [130]. Bombardment and substrate temperature are known to affect a deposit’s microstructure, but this work did not

122 consider bombardment, and alloys deposited on cooled substrates did not exhibit microstructures

or surface morphologies that were significantly different from alloys deposited on substrates at

room-temperature. In addition, most alloys were deposited with similar rates of deposition, so it

is difficult to comment on the influence that deposition rate has on microstructure and surface

morphology. However, deposition rate is known to influence crystal growth and film structure

[162].

In general then, one can see that processing history has a profound impact on an alloy’s

corrosion rate. Defects—namely those generated during deposition or incurred while cleaving

samples—increased an alloy’s rate of corrosion. However, substrate rotation seemed to correlate

with denser alloy microstructure and smoother surface morphology, which translated to lower

corrosion rates. Composition was also observed to influence microstructure and surface

morphology, but correlations between composition and corrosion rate were not immediately

clear. XRD scans showed that alloying elements remained in solid solution, but peak shifts and

peak broadening indicated stress in the alloys, which is consistent with the behavior of Z-T and

Z-II deposits in Thornton’s structure-zone model. Alloy stress has implications for corrosion

behavior that will be further discussed in Section 5.3.

5.2.2 Passive Film Formation

Information pertaining to the formation of passive films was difficult to obtain, because the available x-ray photoelectron (XPS) system was experiencing technical difficulties.

Nevertheless, some information can be gleaned from EDS data and the literature.

Table 5-1 compares the surface composition of Mg98 Ti2 (alloy 86), Mg89 Y9 Ti2, T3, and WE43 after potentiodynamic testing. The addition of titanium did not appear to influence the surface composition of unattacked or attacked areas, although surfaces of the Ti-containing alloys

123 contained a small amount of sulfur, which was not observed in WE43. Table 5-2 compares the composition of corrosion product from WE43, which was submerged in 37ºC HBSS for 15 days, with the composition of corrosion product from WE43, which was submerged in 37ºC m-SBF for

5 days [123]. The compositions are similar.

Table 5-1: Surface composition (at%) of alloys Mg98 Ti2, Mg89 Y9 Ti2, T3, and WE43 after potentiodynamic testing in 37ºC HBSS Unattacked Surface Attacked Surface Element Mg98 Ti2 Mg89 Y9 Ti2 T3 WE43* Mg98 Ti2 T3 WE43* C 11.7 14.9 11.1 11.1 13.9 20.8 17.1 Ca 0.4 0.9 0.6 - 0.4 0.7 0.2 Cl 0.3 0.5 0.4 - 0.6 0.6 1.4 Mg 70.9 59.5 68.3 77.1 26.1 26.0 25.7 O 15.4 20.2 14.6 9.9 57.6 46.0 52.9 P 0.5 1.1 1.0 0.5 0.6 1.8 0.4 S - - - - 0.3 1.0 - Ti 0.8 0.8 1.1 - 0.6 0.7 - Y - 2.1 2.9 1.0 - 2.1 1.5 *Composition does not total 100%

Table 5-2: Composition (at%) of corrosion product on WE43 after 15 days in 37ºC HBSS compared with Ref. [123] WE43 WE43 - Ref. [123] Element (15 days) (5 days)* C 13.5 16.0 Ca 18.8 9.6 Mg 4.0 6.9 Na 0.4 - O 50.4 53.6 P 12.4 12.4 Y 0.5 - *Composition does not total 100%.

Therefore, the nature of the passive film formed on the Mg-Ti and Mg-Y-Ti alloys in

37ºC HBSS may be similar to the one formed on WE43 in 37ºC HBSS and other simulated body

124 fluids (SBFs). Rettig and Virtanen [122] found that WE43 corroded faster in SBFs than in NaCl

solutions. In addition, they found that crystalline Mg(OH)2 comprised most of its corrosion later in NaCl, but amorphous (Mg, Ca)x(PO4)y(CO3)z(OH)i was the primary component of its corrosion layer in SBFs. The EDS data appear to support the existence of a similar amorphous corrosion layer on the EB-PVD alloys.

5.3 Corrosion Rates of Magnesium Alloys

This section compares the corrosion rates calculated from linear polarization and EIS for the EB-PVD alloys and WE43. Care is taken to differentiate between results from fresh and aged

PVD alloys. The section also contrasts the electrochemical corrosion rates for WE43 with the rates calculated from mass loss tests.

5.3.1 Electrochemical Corrosion Rates

As Table 4-6 and Table 4-7 have shown, linear polarization and EIS corrosion rates were mostly in agreement. The LP and EIS corrosion rates for fresh materials differed by approximately a factor of two, although the rates of a few alloys differed by as much as a factor of 3 to 5. LP and EIS corrosion rates for the aged alloys showed better agreement—all differed by approximately a factor of two or less. However, as Table 5-3 shows, alloys with low rates of

corrosion showed good agreement between LP and EIS data. Fresh alloy 55, aged alloy 86, aged

alloy 89, and aged alloy T3 all demonstrated lower average corrosion rates than WE43, although

the LP corrosion rate of T3 was higher than the LP corrosion rate of WE43. However, the

literature on magnesium corrosion generally prefers EIS data over LP data [67].

125 Corrosion rates of the EB-PVD alloys and WE43 were also measured as a function of time, and the curves with the lowest corrosion rates were presented in Figures 4-26 through 4-29.

Thick alloys T1 and T15 demonstrated relatively high corrosion rates that were likely a result of their unoptimized condition. Defects were more common in the thick alloys, and they lacked heat treatment other than the aged stress relief. The thin alloys and WE43 demonstrated similar corrosion rates, but Mg-Y-Ti alloy 55 consistently exhibited the lowest rates with time.

Potentiodynamic results, the best of which were shown in Figure 4-34, revealed that

many of the EB-PVD alloys exhibited nobler OCPs, comparable ipass’s, nobler Eb’s, and larger

passive regions than WE43. Table 5-3 summarizes corrosion rates and potentiodynamic data for the best EB-PVD alloys.

Table 5-3: Comparison of the best EB-PVD Mg alloys with WE43

Avg. μA/cm2 Avg. Avg. Eb Avg. Passive Avg. ipass Alloy Aged 2 LP EIS OCP (V) (V) Region (mV) (μA/cm ) WE43 n/a 17 22 -1.638 -1.576 62 20 Mg98 Ti2 (86) yes 7 3 -1.511 -1.423 88 24 Mg89 Y8 Ti2 (55) no 8 10 -1.538 -1.494 44 13 Mg80 Y16 Ti4 (89) yes 3 3 -1.537 -1.463 74 56 Mg97.2 Y2.5 Ti0.3 (T3) yes 29 16 -1.534 -1.464 71 103

With a few exceptions, one can see that the EB-PVD alloys corroded at rates 2-7 times slower than WE43. Their OCPs were approximately 100 mV nobler; their Eb’s were 80-150 mV nobler; their passive regions were 10-20 mV wider, and their passive ipass’s were comparable to

WE43. Interestingly, the aged EB-PVD alloy containing only 2 weight percent Ti demonstrated

some of the best corrosion behavior. Its OCP, Eb, and passive region were the best among those

shown in Table 5-3. In addition, its passive current density was only slightly higher than that of

126 WE43. The EB-PVD alloys therefore seemed to demonstrate superior corrosion behavior when

compared with commercial alloy WE43.

5.3.2 Mass Loss Corrosion Rate for WE43

Mass loss corrosion rates were compared with electrochemical corrosion rates for polished samples of WE43. The mass loss samples were submerged in 37ºC HBSS for one week.

Results were presented in Table 4-4.

Small samples with a larger surface area to mass ratio were observed to corrode at the

average rate of 3.17 μA/cm2. Larger samples corroded at and average rate of 1.83 μA/cm2.

These values are 5.4 and 9.6 times smaller, respectively, than the average linear polarization corrosion rate of WE43 upon submersion in 37ºC HBSS. However, one must consider that the mass loss rates are averages over an extended period of time, and larger corrosion rates are expected upon initial submersion in a simulated body fluid [122].

The mass loss corrosion rates were only slightly lower than the lowest recorded electrochemical corrosion rate for freshly submerged WE43 (4.35 μA/cm2). In addition, they

were comparable to the electrochemical corrosion rates demonstrated by WE43 after several

hours of submersion, as seen in Figures 4-28 and 4-29. Magnesium alloys that demonstrate passive behavior are less likely to exhibit the negative difference effect [61], so electrochemical methods appeared to be a viable means for characterizing the corrosion behavior of commercial

WE43 and similar alloys.

127 5.3.3 The Effect of Time

The EB-PVD Mg alloys were deposited and stored in a desiccator to prevent atmospheric oxidation, but it was observed that most alloys demonstrated a lower corrosion rate after approximately one year of aging. Several explanations exist for this decrease in corrosion rate, but the most probably explanations are growth of a surface oxide or release of deposit stress.

Experimental results, however, do not indicate that a surface oxide is responsible for the alloys’ increased corrosion resistance. XPS would ideally be used to investigate surface oxides, but as previously mentioned, the technique was not available while this research was being conducted.

Therefore, data had to be obtained from other sources.

Table 4-5, which contains OCP values for fresh and aged material, does not reveal trends

that indicate growth of a surface oxide. The fresh EB-PVD alloys’ open circuit potentials were

practically the same as the OCPs of the aged alloys. The noblest increase in OCP was 21 mV for

alloy T15; thin alloy 33 demonstrated a 16 mV shift in the noble direction, but the OCP of some

alloys, like.T16, actually became less noble.

Localized oxidation was observed on aged alloys, but this oxidation did not appear to be

detrimental to their corrosion behavior. EDS did not indicate significant increases in the oxygen

content of an alloy’s surface, so oxidation may have occurred at defect locations. EDS, however,

is not generally considered a surface-sensitive technique. Figure 5-9 compares surface images of fresh and aged thin EB-PVD alloys, and Figure 5-10 compares images of fresh and aged thick

EB-PVD alloys. Localized oxidation was evident on the surfaces of the thin alloys, but overall

the surfaces of the fresh and aged alloys did not look significantly different. Images of the aged

thick alloys showed slightly more charging than images of the fresh material, but the general

surface morphologies were very similar.

128

As-Deposited Aged

Figure 5-9: Surface morphologies of fresh and aged thin Mg-Y-Ti alloys. Images of alloy (A) Mg79 Y16 Ti5 and (B) Mg90 Y8 Ti2 – alloy 87 – were taken shortly after deposition (left) and again after aging them for 1 year in an evacuated desiccator (right). Scale bar indicates 10 μm.

As-Deposited Aged

Figure 5-10: Surface morphologies of fresh and aged thick Mg-Y-Ti alloys. Images of alloy (A) T1 and (B) T3 were taken shortly after deposition (left) and again after aging them for 1 year in a desiccator (right). Scale bars indicate 5 μm.

129 Curves obtained from EIS data also failed to indicate that a surface oxide was responsible for an alloy’s improved corrosion resistance. Figure 5-11 compares EIS curves for both fresh and aged thin and thick EB-PVD Mg. The curves from the fresh material showed similar behavior to those from the aged material, which indicated that the electrochemical processes occurring on the alloys’ surfaces were similar regardless of aging. Different behavior would be expected if a surface oxide had formed.

The indicators of tensile stress—XRD peak shifts and peak broadening, Zone-T/II

morphology—have already been discussed, so tensile was likely present in the as-deposited

alloys. This tensile stress can compromise the mechanical integrity of a passive film, which

makes an alloy more susceptible to corrosion. Therefore, stress relief seems to be the most likely

explanation for the increase in corrosion resistance that was observed in the aged alloys. Wolfe

[132] annealed Mg-Y-Ti PVD alloys for 30 min. at various temperatures and found that the

optimum anneal took place at 60ºC. The annealed alloy did not demonstrate a significantly lower

corrosion rate, but the corrosion rate of Wolfe’s alloy was initially low (< 10 μA/cm2).

Annealing, however, did increase the Mg-Y-Ti alloy’s breakdown potential, extend its passive

region, and decrease its passive current density. Furthermore, Wolfe’s XPS data clearly showed

that the improved corrosion behavior was not the result of a thickening oxide film. 60ºC is a

relatively low temperature, so it is conceivable that the alloys in this work experienced stress

relief as they aged in the evacuated desiccator. Heat treatment and/or aging are important

processes for developing optimized nonequilibrium EB-PVD magnesium alloys.

130

10000 90 - fresh 90 - aged

1000 2 cm Ω |Z| /

100

10 1e-3 1e-2 1e-1 1e+0 1e+1 1e+2 1e+3 1e+4 1e+5 1e+6 Freq / Hz

10000 T3 - fresh T3 - aged

1000 2 cm

Ω 100 |Z| /

10

1 1e-3 1e-2 1e-1 1e+0 1e+1 1e+2 1e+3 1e+4 1e+5 1e+6 Freq / Hz Figure 5-11: EIS curves comparing both fresh and aged thin and thick EB-PVD Mg. The corrosion mechanism did not appear to change with aging.

131 5.4 Factors Influencing Cytotoxicity

An alloy’s corrosion rate appeared to be the strongest predictor of cell viability. As

Figure 4-41 showed, dissolved WE43 did not appear to influence the attachment or morphology

of A549 cells on a glass control substrate. However, WE43 corroded faster than the thin EB-

PVD alloys, and cells grown on WE43 demonstrated indeterminate and apoptotic morphologies.

Cells grown on fresh thick alloy T15, shown in Figure 4-43, exhibited necrotic ball-like

morphologies and perforated membranes. The thick alloys exhibited the highest corrosion rates

of the EB-PVD Mg alloys, so an alloy’s corrosion rate appeared to be the strongest predictor of

cell viability.

Nevertheless, the thick alloys documented in this thesis were the first generation of EB-

PVD Mg alloys from the Sciaky system, but multiple runs are required to refine a deposition

process The thick alloys were therefore unoptimized, but they demonstrated promising corrosion

behavior. Better cell culture results may have been observed if cells were grown on T3, which

exhibited the lowest corrosion rate of the thick specimens.

Figure 5-12 compares A549 cells grown on the glass control substrate, WE43, aged

Mg98 Ti2 (alloy 86), and aged Mg89 Y9 Ti2. Cells grown on the control substrate appeared similar to those described elsewhere in the literature [156, 163]. Cells grown on WE43 demonstrated indeterminate or apoptotic ball-like morphologies, but cells grown on the thin EB-

PVD alloys, which corroded at rates 2-7 times lower than freshly polished WE43, exhibited a more healthy pancake-shaped. Once again, one can see that low corrosion rates were correlated with cell viability.

A monolayer of cells was also observed to grow on the thin alloys, but this monolayer appeared to be more uniform on the Mg98 Ti2 alloy than on the Mg-Y-Ti alloys. Yttrium has been associated with some cytotoxic effects [106, 107], so cell viability may be negatively

132 affected by the Y content in the Mg-Y-Ti alloys. Therefore, Mg-Ti alloys, especially those containing only a few weight percent Ti, appeared to be a more biocompatible material for bioabsorbable implants.

Figure 5-12: Comparison of A549 cells grown on Mg alloys. Cells grown on the aged EB-PVD alloys most closely resemble those grown on the control substrate. Scale bar indicates 10 μm.

133

Chapter 6

Conclusions and Future Work

6.1 Conclusions

The following conclusions can be drawn from this research:

• Many deposition and processing factors influence a deposit’s structure and morphology,

but substrate rotation and elemental composition seemed to exert the most influence on

magnesium alloys made with EB-PVD. Substrate rotation and increased alloying

element composition appeared to produce alloys with comparatively dense, columnar

microstructures and smooth, uniform surface morphologies. Thin alloy microstructures

correlated with the Zone-T—Zone-II transition of Thornton’s structure zone model. X-

ray diffraction showed that solid solution was maintained in alloys containing up to 16

wt% (5 at%) yttrium and 4 wt% (2 at%) titanium. These features translated into lower

corrosion rates, although defects introduced during deposition and cleaving were

observed to negatively impact an alloy’s corrosion resistance.

• An industrial-prototype EB-PVD system was used to produce an initial batch of

Mg-Y-Ti alloys with thicknesses greater than 300 μm. The unoptimized alloys

demonstrated inferior corrosion behavior to the alloys deposited with the dual gun

system, but the industrial deposition process still needs to be refined. Defects caused by

shadowing and substrate roughness were observed to be more prevalent in thick alloys

than in thin alloys.

• Electrochemical methods—including linear polarization, electrochemical impedance

spectroscopy, and potentiodynamic polarization—can be used to characterize the

134 corrosion behavior of magnesium alloys that demonstrate passive behavior and relatively

low corrosion rates in a simulated body fluid. Corrosion rates calculated from linear

polarization and EIS data were generally in agreement, especially for aged alloys.

Electrochemical corrosion rates for WE43, which were obtained shortly after submersion

in 37ºC HBSS, were slightly higher than those obtained from mass loss measurements,

which were calculated after one week of submersion in the same electrolyte. This

relationship was expected, since initial corrosion rates are generally higher than the

average rate over time.

• Aged alloys demonstrated improved corrosion behavior when compared with their as-

deposited (fresh) counterparts. This improvement seemed to be a result of the aging

process, which appeared to give the alloys sufficient time (several months to one year) to

undergo a stress relief anneal at room temperature. Aged alloys consistently

demonstrated lower corrosion rates and superior passive behavior when compared with

commercial WE43.

• Corrosion rate appeared to be the most important factor for predicting the viability of

A549 lung epithelial cells grown on magnesium alloys. Cells grown on WE43 were

indeterminate or apoptotic, but cells grown on aged thin EB-PVD alloys demonstrated

attachment, pancake-like morphology, proliferation, and viability.

• Thin EB-PVD alloys consistently demonstrated higher corrosion resistance, superior

passivity, and lower cytotoxicity than WE43. Aged alloy Mg98 Ti2 (86) was among the

best of the EB-PVD alloys, so nonequilibrium Mg-Ti alloys may be a promising material

for bioabsorbable implants.

135 6.2 Future Work

• The thick EB-PVD alloy deposition process should be refined. More specifically, a

different substrate holder should be engineered to prevent iron from contaminating the

rear surface of the thick alloys. An ideal substrate holder would also rotate. Different

alloy compositions should be explored, especially those containing titanium in the range

of 1-5 wt%.

• The dual gun EB-PVD deposition process should be refined to control Y and Ti content

independently. Mg-Ti-Y (more Ti, less Y) alloys should be explored more extensively.

• The Mg alloys’ mechanical properties should be characterized and compared with those

of materials that are currently used for cardiac stents.

• Apiezon Wax W was observed to crack and peel after several hours of exposure to 37ºC

HBSS, so a more durable masking agent should be used for electrochemical tests. In

addition, electrochemical data could be obtained from stirred solutions to better simulate

corrosion within the circulatory system.

• XPS data should be obtained from alloy surfaces to determine chemical state information

including components of the passive film, elemental oxidation states, and the thickness of

the passive film.

• Other forms of stress relief should be explored and compared with the room temperature

anneal so that a heat treatment process can be developed. Annealing at temperatures

slightly below or slightly above 60ºC may accelerate stress relief.

• Cells derived from vascular cell lines should be used to better assess the in vitro

cytotoxicity of magnesium alloys. Cytotoxicity tests should also be assessed in a more

controlled fashion. Controlled tests will be an important factor in assessing the toxicity

or nontoxicity of yttrium.

136

• EB-PVD magnesium alloys should begin to be implanted in animals. In vivo corrosion is

significantly different from in vitro corrosion, so animal tests are a necessary future step

for developing a bioabsorbable magnesium implant.

137

References

[1] J. Kahn, "Mending Broken Hearts," in National Geographic, February 2007, pp. 40-65. [2] P. Erne, M. Schier, and T. J. Resink, "The Road to Bioabsorbable Stents: Reaching Clinical Reality?" CardioVascular and Interventional Radiology, vol. 29, pp. 11-16, 2005. [3] W. L. Hunter, "Drug-Eluting Stents: Beyond the Hyperbole," Advanced Drug Delivery Reviews, vol. 58, pp. 347, 2006. [4] D. Böse, H. Eggebrecht, M. Haude, et al., "First Absorbable Metal Stent Implantation in Human Coronary Arteries," The American Heart Hospital Journal, vol. 4, pp. 128-130, 2006. [5] U. Sigwart, J. Puel, V. Mirkovitch, et al., "Intravascular Stents to Prevent Occlusion and Restenosis After Transluminal Angioplasty," The New England Journal of Medicine, vol. 316, pp. 701-706, 1987. [6] A. Colombo, M. Ferraro, A. Itoh, et al., "Results of Coronary Stenting for Restenosis," Journal of the American College of Cardiology, vol. 28, pp. 830-836, 1996. [7] D. R. Holmes, M. B. Leon, J. W. Moses, et al., "Analysis of 1-Year Clinical Outcomes in the SIRIUS Trial: A Randomized Trial of a Sirolimus-Eluting Stent Versus a Standard Stent in Patients at High Risk for Coronary Restenosis," Circulation, vol. 109, pp. 634- 640, 2004. [8] G. W. Stone, S. G. Ellis, D. A. Cox, et al., "One-Year Clinical Results With the Slow- Release, Polymer-Based Paclitaxel-Eluting TAXUS Stent: The TAXUS-IV Trial," Circulation, vol. 109, pp. 1942-1947, 2004. [9] R. Virmani, F. Liistro, G. Stankovic, et al., "Mechanism of Late In-Stent Restenosis After Implantation of a Paclitaxel Derivative-Eluting Polymer Stent System in Humans," Circulation, vol. 106, pp. 2649-2651, 2002. [10] M. Pfisterer, H. P. Brunner-La Rocca, P. T. Buser, et al., "Late Clinical Events After Clopidogrel Discontinuation May Limit the Benefit of Drug-Eluting Stents: An Observational Study of Drug-Eluting Versus Bare-Metal Stents," Journal of the American College of Cardiology, vol. 48, pp. 2584-2591, 2006. [11] P. H. Grewe, T. Deneke, A. Machraoui, et al., "Acute and Chronic Tissue Response to Coronary Stent Implantation: Pathologic Findings in Human Specimen," Journal of the American College of Cardiology, vol. 35, pp. 157-163, 2000. [12] E. Karvouni, S. Korovesis, and D. G. Katritsis, "Very Late Thrombosis After Implantation of Sirolimus Eluting Stent," Heart, vol. 91, pp. 45-46, 2005. [13] A. Farb, A. P. Burke, F. D. Kolodgie, et al., "Pathological Mechanisms of Fatal Late Coronary Stent Thrombosis in Humans," Circulation, vol. 108, pp. 1701-1706, 2003. [14] W. W. O'Neill and M. B. Leon, "Drug-Eluting Stents: Costs Versus Clinical Benefit," Circulation, vol. 107, pp. 3008-3011, 2003. [15] P. A. Lemos, P. W. Serruys, and J. E. Sousa, "Drug-Eluting Stents: Cost Versus Clinical Benefit," Circulation, vol. 107, pp. 3003-3007, 2003. [16] J. W. Hirshfeld and R. L. Wilensky, "Drug-Eluting Stents Are Here - Now Ahat? Implications for Clinical Practice and Health Care Costs," Cleveland Clinic Journal of Medicine, vol. 71, pp. 825-828, 2004.

138 [17] B. Heublein, R. Rohde, V. Kaese, et al., "Biocorrosion of Magnesium Alloys: A New Principle in Cardiovascular Implant Technology?" Heart, vol. 89, pp. 651-656, 2003. [18] D. R. Holmes, "Opportunities for Improvement - The Disappearing Stent," Catheterization and Cardiovascular Interventions, vol. 68, pp. 618-619, 2006. [19] R. Wolfe, "Imparting Passivity to Vapor Deposited Magnesium Alloys," a Doctor of Philosophy thesis in Engineering Science and Mechanics, Pennsylvania State University, University Park, PA. May 2005. [20] P. L. Miller, "Corrosion Characteristics of Nonequilibrium Magnesium-Yttrium Alloys," a Master of Science thesis in Engineering Science, Pennsylvania State University, University Park, PA. May 1994. [21] L. Dubé and J.-C. Granry, "The Therapeutic Use of Magnesium in Anesthesiology, Intensive Care and Emergency Medicine: A Review," Canadian Journal of Anesthesia, vol. 50, pp. 732-746, 2003. [22] J. Vormann, "Magnesium: Nutrition and Metabolism," Molecular Aspects of Medicine, vol. 24, pp. 27-37, 2003. [23] "Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D, and Fluoride," Standing Committee on the Scientific Evaluation of Dietary Reference Intakes, Food and Nutrition Board, Institute of Medicine 1997, pp. 190, 224-225, 228, Available online http://books.nap.edu/catalog.php?record_id=5776. [24] L. Galland, "Magnesium, Stress and Neuropsychiatric Disorders," Magnesium and Trace Elements, vol. 10, pp. 287-301, 1991-1992. [25] B. M. Altura and B. T. Altura, "Cardiovascular Risk Factors and Magnesium: Relationships to Atherosclerosis, Ischemic Heart Disease and Hypertension," Magnesium and Trace Elements, vol. 10, pp. 182-192, 1991-92. [26] A. Hartwig, "Role of Magnesium in Genomic Stability," Mutation Research, vol. 475, pp. 113-121, 2001. [27] V. Rukshin, B. Azarbal, P. K. Shah, et al., "Intravenous Magnesium in Experimental Stent Thrombosis in Swine," Arteriosclerosis, Thrombosis, and Vascular Biology, vol. 21, pp. 1544-1549, 2001. [28] V. Rukshin, P. K. Shah, B. Cercek, et al., "Comparative Antithrombotic Effects of Magnesium Sulfate and the Platelet Glycoprotein IIb/IIIa Inhibitors Tirofiban and Eptifibatide in a Canine Model of Stent Thrombosis," Circulation, vol. 105, pp. 1970- 1975, 2002. [29] H. B. Ravn, S. D. Kristensen, V. E. Hjortdal, et al., "Early Administration of Intravenous Magnesium Inhibits Arterial Thrombus Formation," Arteriosclerosis, Thrombosis, and Vascular Biology, vol. 17, pp. 3620-3625, 1997. [30] J. Li, Q. Zhang, M. Zhang, et al., "Intravenous Magnesium for Acute Myocardial Infarction (Review)," Cochrane Database of Systematic Reviews, vol. 2, 2007. [31] E. M. Antman and MAGIC Trial Investigators, "Early Administration of Intravenous Magnesium to High-Risk Patients with Acute Myocardial Infarction in the Magnesium in Coronaries (MAGIC) Trial: A Randomised Controlled Trial," Lancet, vol. 360, pp. 1189- 1196, 2002. [32] V. Rukshin, R. Santos, M. Gheorghiu, et al., "A Prospective, Nonrandomized, Open- Labeled Pilot Study Investigating the Use of Magnesium in Patients Undergoing Nonacute Percutaneous Coronary Intervention with Stent Implantation," Journal of Cardiovascular Pharmacology and Therapeutics, vol. 8, pp. 193-200, 2003. [33] P. N. Sawyer, W. H. Brattain, and P. J. Boddy, "Electrochemical Criteria in the Choice of Materials Used in Vascular Prostheses," in Biophysical Mechanisms in Vascular

139 Homeostasis and Intravascular Thrombosis, P. N. Sawyer, Ed. New York: Appleton- Century-Crofts, 1965, pp. 337-348. [34] C. DiMario, H. Griffiths, O. Goktekin, et al., "Drug-Eluting Bioabsorbable Magnesium Stent," Journal of Interventional Cardiology, vol. 17, pp. 391-395, 2004. [35] R. Waksman, R. Pakala, P. K. Kuchulakanti, et al., "Safety and Efficacy of Bioabsorbable Magnesium Alloy Stents in Porcine Coronary Arteries," Catheterization and Cardiovascular Interventions, vol. 68, pp. 607-617, 2006. [36] M. Bosiers, K. Deloose, J. Verbist, et al., "First Clinical Application of Absorbable Metal Stents in the Treatment of Critical Limb Ischemia: 12-Month Results," Vascular Disease Management, vol. 2, pp. 86-91, 2005. [37] H. Eggebrecht, J. Rodermann, P. Hunold, et al., "Novel Magnetic Resonance-Compatible Coronary Stent: The Absorbable Magnesium-Alloy Stent," Circulation, vol. 112, pp. e303-e304, 2005. [38] D. Böse, H. Eggebrecht, and R. Erbel, "Absorbable Metal Stent in Human Coronary Arteries: Imaging with Intravascular Ultrasound," Heart, vol. 92, pp. 892-, 2006. [39] P. Barlis, J. Tanigawa, and C. Di Mario, "Coronary Bioabsorbable Magnesium Stent: 15- Month Intravascular Ultrasound and Optical Coherence Tomography Findings," European Heart Journal, vol. 28, pp. 2319, 2007. [40] R. Erbel, C. Di Mario, J. Bartunek, et al., "Temporary Scaffolding of Coronary Ateries with Bioabsorbable Magnesium Stents: A Prospective, Non-Randomised Multicentre Trial," Lancet, vol. 369, pp. 1869-1875, 2007. [41] P. Zartner, R. Cesnjevar, H. Singer, et al., "First Successful Implantation of a Biodegradable Metal Stent Into the Left Pulmonary Artery of a Preterm Baby," Catheterization and Cardiovascular Interventions, vol. 66, pp. 590-594, 2005. [42] D. Schranz, P. Zartner, I. Michel-Behnke, et al., "Bioabsorbable Metal Stents for Percutaneous Treatment of Critical Recoarctation of the Aorta in a Newborn," Catheterization and Cardiovascular Interventions, vol. 67, pp. 671-673, 2006. [43] C. J. McMahon, P. Oslizlok, and K. P. Walsh, "Early Restenosis Following Biodegradable Stent Implantation in an Aortopulmonary Collateral of a Patient with Pulmonary Atresia and Hypoplastic Pulmonary Arteries," Catheterization and Cardiovascular Interventions, vol. 69, pp. 735-738, 2007. [44] R. Erbel, D. Böse, M. Haude, et al., "Absorbable Coronary Stents: New Promising Technology?" Herz, vol. 32, pp. 308-319, 2007. [45] D. F. Williams, "On the Mechanisms of Biocompatibility," Biomaterials, vol. 29, pp. 2941-2953, 2008. [46] G. Mani, M. D. Feldman, D. Patel, et al., "Coronary stents: A materials perspective," Biomaterials, vol. 28, pp. 1689, 2007. [47] J. J. Jacobs, J. L. Gilbert, and R. M. Urban, "Corrosion of Metal Orthopaedic Implants," Journal of Bone and Joint Surgery, vol. 80, pp. 268-282, 1998. [48] F. Witte, H. Ulrich, M. Rudert, et al., "Biodegradable Magnesium Scaffolds: Part I: Appropriate Inflammatory Response," Journal of Biomedical Materials Research, vol. 81A, pp. 748-756, 2007. [49] F. Witte, H. Ulrich, C. Palm, et al., "Biodegradable Magnesium Scaffolds: Part II: Peri- Implant Bone Remodeling," Journal of Biomedical Materials Research, vol. 81A, pp. 757-765, 2007. [50] B. Denkena, F. Witte, C. Podolsky, et al., "Degradable Implants Made of Magnesium Alloys," in Proceedings of the 5th Euspen International Conference. Montpellier, France, 2005.

140 [51] F. Witte, V. Kaese, H. Haferkamp, et al., "In Vivo Corrosion of Four Magnesium Alloys and the Associated Bone Response," Biomaterials, vol. 26, pp. 3557, 2005. [52] Z. Li, X. Gu, S. Lou, et al., "The Development of Binary Mg-Ca Alloys for Use as Biodegradable Materials Within Bone," Biomaterials, vol. 29, pp. 1329-1344, 2008. [53] M. P. Staiger, A. M. Pietak, J. Huadmai, et al., "Magnesium and Its Alloys as Orthopedic Biomaterials: A Review," Biomaterials, vol. 27, pp. 1728-1734, 2006. [54] R. Decking, C. Rokahr, M. Zurstegge, et al., "Maintenance of bone mineral density after implantation of a femoral neck hip prosthesis," BMC Musculoskeletal Disorders, vol. 9, pp. (in print), 2008. [55] A. Pietak, P. Mahoney, G. J. Dias, et al., "Bone-like matrix formation on magnesium and magnesium alloys," Journal of Materials Science - Materials in Medicine, vol. 19, pp. 407-415, 2008. [56] C. Krause, D. Bormann, T. Hassel, et al., "Mechanical Properties of Degradable Magnesium Implants in Dependence of the Implantation Duration," in The International Symposium on Magnesium Technology in the Global Age, M. O. Pekguleryuz and L. W. F. Mackenzie, Eds. Montreal, Quebec, Canada: The Canadian Institute of Mining, Metallurgy and Petroleum, 2006, pp. 329-343. [57] M. Peuster, P. Beerbaum, F.-W. Bach, et al., "Are Resorbable Implants About to Become a Reality?" Cardiology in the Young, vol. 16, pp. 107-116, 2006. [58] J. A. M. Maier, C. Malpuech-Brugère, W. Zimowska, et al., "Low Magnesium Promotes Endothelial Cell Dysfunction: Implications for Atherosclerosis, Inflammation and Thrombosis," Biochimica et Biophysica Acta, vol. 1689, pp. 13-21, 2004. [59] J. A. M. Maier, D. Bernardini, Y. Rayssiguier, et al., "High Concentrations of Magnesium Modulate Vascular Endothelial Cell Behaviour In Vitro," Biochimica et Biophysica Acta, vol. 1689, pp. 6-12, 2004. [60] G. G. Perrault, "Chapter VIII-4: Magnesium," in Encyclopedia of Electrochemistry of the Elements, A. J. Bard, Ed. New York: Marcel Dekker, 1978, pp. 263. [61] G. L. Song and A. Atrens, "Corrosion mechanisms of magnesium alloys," Advanced Engineering Materials, vol. 1, pp. 11-33, 1999. [62] G. L. Makar and J. Kruger, "Corrosion of magnesium," International Materials Reviews, vol. 38, pp. 138-153, 1993. [63] B. A. Shaw and R. C. Wolfe, "Corrosion of Magnesium and Magnesium-Based Alloys," in Corrosion: Materials, vol. 13B, S. D. Cramer and B. S. Covino, Eds. Materials Park, OH: ASM International, 2005, pp. 205-227. [64] A. Froats, T. K. Aune, D. Hawke, et al., "Corrosion of Magnesium and Magnesium Alloys," in Corrosion, vol. 13, Metals Handbook, J. R. Davis, J. D. Destefani, and G. M. Crankovic, Eds., Ninth ed. Metals Park, OH: ASM International, 1987, pp. 740-754. [65] D. L. Hawke, J. E. Hillis, M. Pekguleryuz, et al., "Corrosion Behavior," in Magnesium and Magnesium Alloys, M. M. Avedesian and H. Baker, Eds. Materials Park, OH: ASM International, 1999, pp. 194-210. [66] B. A. Shaw, "Corrosion Resistance of Magnesium Alloys," in Corrosion: Fundamentals, Testing, and Protection, vol. 13A, ASM Handbook, S. D. Cramer and B. S. Covino, Eds. Materials Park, OH: ASM International, 2003, pp. 692-696. [67] G. Song and A. Atrens, "Understanding Magnesium Corrosion," Advanced Engineering Materials, vol. 5, pp. 837-858, 2003. [68] G. G. Perrault, "The Potential-pH Diagram of the Magnesium-Water System," Journal of Electroanalytical Chemistry, vol. 51, pp. 107, 1974. [69] G. Song, A. Atrens, D. Stjohn, et al., "The Electrochemical Corrosion of Pure Magnesium in 1 N NaCl," Corrosion Science, vol. 39, pp. 855, 1997.

141 [70] G. R. Hoey and M. Cohen, "Corrosion of Anodically and Cathodically Polarized Magnesium in Aqueous Media," Journal of the Electrochemical Society, vol. 105, pp. 245-250, 1958. [71] H. P. Godard, W. B. Jepson, M. R. Bothwell, et al., The Corrosion of Light Metals. New York: John Wiley, 1967, pp. 260. [72] S. Krishnamurthy, M. Khobaib, E. Robertson, et al., "Corrosion Behavior of Rapidly Solidified Mg-Nd and Mg-Y Alloys," Materials Science and Engineering, vol. 99, pp. 507-511, 1988. [73] J. H. Nordlien, S. Ono, N. Masuko, et al., "Morphology and Structure of Oxide Films Formed on Magnesium by Exposure to Air and Water," Journal of the Electrochemical Society, vol. 142, pp. 3320-3322, 1995. [74] J. H. Nordlien, S. Ono, N. Masuko, et al., "A TEM Investigation of Naturally Formed Oxide Films on Pure Magnesium," Corrosion Science, vol. 39, pp. 1397-1414, 1997. [75] K. Murakami and E. Sato, "Corrosion and Impedance of Magnesium in Alkali Solutions," Journal of Japan Institute of Light Metals, vol. 27, pp. 71-76, 1977. [76] K. Huber, "Anodic Formation of Coatings on Magnesium, Zinc, and ," Journal of the Electrochemical Society, vol. 100, pp. 376-382, 1953. [77] R. Tunold, H. Holtan, M.-B. H. Berge, et al., "The corrosion of magnesium in aqueous solution containing chloride ions," Corrosion Science, vol. 17, pp. 353, 1977. [78] N. Hara, Y. Kobayashi, D. Kagaya, et al., "Formation and Breakdown of Surface Films on Magnesium and Its Alloys in Aqueous Solutions," Corrosion Science, vol. 49, pp. 166-175, 2007. [79] J. D. Hanawalt, C. E. Nelson, and J. A. Peloubet, "Corrosion studies of magnesium and its alloys," Light Metals, vol. 5, pp. 30-36, 1942. [80] A. Yamamoto, A. Watanabe, K. Sugahara, et al., "Improvement of Corrosion Resistance of Magnesium Alloys by Vapor Deposition," Scripta Materialia, vol. 44, pp. 1039, 2001. [81] M. H. Lee, I. Y. Bae, K. J. Kim, et al., "Formation Mechanism of New Corrosion Resistance Magnesium Thin Films by PVD Method," Surface and Coatings Technology, vol. 169-170, pp. 670, 2003. [82] G. S. Frankel, "The Growth of 2-D Pits in Thin Film Aluminum," Corrosion Science, vol. 30, pp. 1203-1218, 1990. [83] E. Akiyama and G. S. Frankel, "The Influence of Dichromate Ions on Aluminum Dissolution Kinetics in Artificial Crevice Electrode Cells," Journal of the Electrochemical Society, vol. 146, pp. 4095-4100, 1999. [84] I. Polmear, "Light Alloys," 4th ed. Boston: Elsevier, 2006. [85] I. J. Polmear, "Grades and Alloys," in Magnesium and Magnesium Alloys, M. M. Avedesian and H. Baker, Eds. Materials Park, OH: ASM International, 1999, pp. 12-25. [86] E. F. Emley, Principles of Magnesium Technology. New York: Pergamon Press, 1966. [87] "ASTM Standard B951: Standard Practice for Codification of Unalloyed Magnesium and Magnesium-Alloys, Cast and Wrought," ASTM International 2008. [88] J. Van Muylder, "Scandium, Yttrium," in Atlas of Electrochemical Equilibria. Houston: National Association of Corrosion Engineers, 1974, pp. 177-182. [89] N. de Zoubov and J. Van Muylder, "," in Atlas of Electrochemical Equilibria. Houston: National Association of Corrosion Engineers, 1974, pp. 183-197. [90] M. Maraghini, E. Deltombe, N. de Zoubov, et al., "Zirconium," in Atlas of Electrochemical Equilibria. Houston: National Association of Corrosion Engineers, 1974, pp. 223-229.

142 [91] "Elektron WE43B, Datasheet: 467A," Magnesium Elektron, Manchester, UK, Available online www.magnesium- elektron.com/data/downloads/467B%20for%20PC%20111202.pdf. [92] E. Deltombe, C. Vanleugenhaghe, and M. Pourbaix, "Aluminum," in Atlas of Electrochemical Equilibria. Houston: National Association of Corrosion Engineers, 1974, pp. 168-176. [93] "Smithells Light Metals Handbook," E. A. Brandes and G. B. Brook, Eds. Boston: Butterworth-Heinemann, 1998, pp. 5. [94] J. Schmets, J. Van Muylder, and M. Pourbaix, "Titanium," in Atlas of Electrochemical Equilibria. Houston: National Association of Corrosion Engineers, 1974, pp. 213-222. [95] R. C. Wolfe and B. A. Shaw, "Corrosion Evaluation of Novel Magnesium Alloys Prepared by Magnetron Sputtering," in Corrosion and Protection of Light Metal Alloys, R. G. Buchheit, R. G. Kelly, N. A. Missert, and B. A. Shaw, Eds. Pennington, NJ: The Electrochemical Society, 2004, pp. 193-202. [96] O. Lunder, T. K. R. Aune, and K. Nisancloglu, "Effect of Mn Additions on the Corrosion Behaviour of Mould-Cast Magnesium ASTM AZ91," Corrosion, vol. 45, pp. 291-295, 1987. [97] J. M. West, Electrodeposition and Corrosion Processes. Princeton, NJ: D. Van Nostrand, 1965, pp. 50. [98] R. A. Yorkel, "The Toxicology of Aluminum in the Brain: A Review," Neurotoxicology, vol. 21, pp. 813-828, 2000. [99] S. S. Krishnan, D. R. McLachlan, B. Krishnan, et al., "Aluminum Toxicity to the Brain," The Science of the Total Environment, vol. 71, pp. 59-64, 1988. [100] E. H. Jeffery, K. Abreo, E. Burgess, et al., "Systemic Aluminum Toxicity: Effects on Bone, Hematopoetic Tissue, and Kidney," Journal of Toxicology and Environmental Health, vol. 48, pp. 649-665, 1996. [101] S. K. Lu, W. H. Lee, T. Y. Tian, et al., "Expression of Smooth Muscle Cells Grown on Magnesium Alloys," in 11th Mediterranean Conference on Medical and Biomedical Engineering and Computing 2007, vol. 16, IFMBE Proceedings, T. Jarm, P. Kramar, and A. Županič, Eds. Berlin: Springer, 2007, pp. 242-245. [102] S. K. Lu, H. I. Yeh, T. Y. Tian, et al., "Degradation of Magnesium Alloys in Biological Solutions and Reduced Pheotypic Expression of Endothelial Cell Grown on These Alloys," in 3rd Kuala Lumpur International Conference on Biomedical Engineering 2006, vol. 15, IFMBE Proceedings, F. Ibrahim, N. A. Abu Osman, J. Usman, and N. A. Kadri, Eds. Berlin: Springer, 2007, pp. 98-101. [103] J. Egawa, W. F. Neuman, and R. Gaunder, "The Interaction of Yttrium with Physiological Ligands," Radiation Research, vol. 23, pp. 503-513, 1964. [104] S. Hirano, N. Kodama, K. Shibata, et al., "Metabolism and Toxicity of Intravenously Injected Yttrium Chloride in Rats," Toxicology and Applied Pharmacology, vol. 121, pp. 224-232, 1993. [105] H. A. Schroeder and M. Mitchener, "Scandium, Chromium (VI), Gallium, Yttrium, Rhodium, Palladium, Indium in Mice: Effects on Growth and Life Span," The Journal of Nutrition, vol. 101, pp. 1431-1438, 1971. [106] A. Loos, R. Rohde, A. Haverich, et al., "In Vitro and In Vivo Biocompatibility Testing of Absorbable Metal Stents," Macromolecular Symposia, vol. 253, pp. 103-108, 2007. [107] X. Gu, Y. Zheng, Y. Cheng, et al., "In Vitro Corrosion and Biocompatibility of Binary Magnesium Alloys," Biomaterials, vol. 30, pp. 484-498, 2009. [108] A. Drynda, N. Deinet, N. Braun, et al., "Rare Earth Metals Used in Biodegradable Magnesium-Based Stents Do Not Interfere with Proliferation of Smooth Muscle Cells but

143 Do Induce the Upregulation of Inflammatory Genes," Journal of Biomedical Materials Research Part A, pp. (in print), 2008. [109] M. Di Gioacchino, N. Verna, L. Di Giampaolo, et al., "Immunotoxicity and Sensitizing Capacity of Metal Compounds Depend on Speciation," International Journal of Immunopathology and Pharmacology, vol. 20, pp. 15-22, 2007. [110] M. Niinomi, "Fatigue Performance and Cyto-toxicity of Low Rigidity Titanium Alloy, Ti-29Nb-13Ta-4.6Zr," Biomaterials, vol. 24, pp. 2673-2683, 2003. [111] A. M. Maurer, K. Merritt, and S. A. Brown, "Cellular Uptake of Titanium and from Addition of Salts or Fretting Corrosion In Vitro," Journal of Biomedical Materials Research, vol. 28, pp. 241-246, 1994. [112] D. R. Haynes, S. D. Rogers, S. Hay, et al., "The Differences in Toxicity and Release of Bone-Resorbing Mediators Induced by Titanium and Cobalt-Chromium-Alloy Wear Particles," Journal of Bone and Joint Surgery, vol. 75, pp. 825-834, 1993. [113] G. Song and S. Song, "A Possible Biodegradable Magnesium Implant Material," Advanced Engineering Materials, vol. 9, pp. 298-302, 2007. [114] F. Witte, J. Fischer, J. Nellesen, et al., "In Vitro and In Vivo Corrosion Measurements of Magnesium Alloys," Biomaterials, vol. 27, pp. 1013, 2006. [115] G.-L. Song and S.-Z. Song, "Corrosion Behavior of Pure Magnesium in a Simulated Body Fluid," Acta Physico-Chimica Sinica, vol. 22, pp. 122-126, 2006. [116] Y. Wang, M. Wei, J. Gao, et al., "Corrosion Process of Pure Magnesium in Simulated Body Fluid," Materials Letters, vol. 62, pp. 2181-2184, 2008. [117] R. Zeng, W. Dietzel, F. Witte, et al., "Progress and Challenge for Magnesium Alloys as Biomaterials," Advanced Engineering Materials, vol. 10, pp. 1-12, 2008. [118] H. Kuwahara, Y. Al-Abdullat, M. Ohta, et al., "Surface Reaction of Magnesium in Hank's Solution," Materials Science Forum, vol. 350-351, pp. 349-358, 2000. [119] M. B. Kannan and R. K. S. Raman, "In Vitro Degradation and Mechanical Integrity of Calcium-Containing Magnesium Alloys in Modified-Simulated Body Fluid," Biomaterials, vol. 29, pp. 2306-2314, 2008. [120] Y. Xin, C. Liu, X. Zhang, et al., "Corrosion Behavior of Biomedical AZ91 Magnesium Alloy in Simulated Body Fluids," Journal of Materials Research, vol. 22, pp. 2004-2011, 2007. [121] C. Liu, Y. Xin, X. Tian, et al., "Corrosion Resistance of Titanium Ion Implanted AZ91 Magnesium Alloy," Journal of Vacuum Science and Technology A, vol. 25, pp. 334-339, 2007. [122] R. Rettig and S. Virtanen, "Time-Dependent Electrochemical Characterization of the Corrosion of a Magnesium Rare-Earth Alloy in Simulated Body Fluids," Journal of Biomedical Materials Research Part A, vol. 85, 2008. [123] R. Rettig and S. Virtanen, "Composition of Corrosion Layers on a Magnesium Rare- Earth Alloy in Simulated Body Fluids," Journal of Biomedical Materials Research Part A, vol. 88, pp. 359-369, 2009. [124] C. F. Chang, S. K. Das, D. Raybould, et al., "Corrosion Resistant High Strength Magnesium Alloys by RSP," Metal Powder Report, vol. 41, pp. 302-308, 1986. [125] H. Jones, "Some Developments in Techniques for Rapid Solidification," in Processing of Structural Metals by Rapid Solidification, F. H. Froes and S. J. Savage, Eds. Materials Park, OH: ASM International, 1986, pp. 77-93. [126] H. Tsubakino, A. Yamamoto, S. Fukumoto, et al., "High-Purity Magnesium Coating on Magnesium Alloys by Vapor Deposition Technique for Improving Corrosion Resistance," Materials Transactions, vol. 44, pp. 504-510, 2003.

144 [127] C. B. Baliga, P. Tsakiropoulos, S. B. Dodd, et al., "Physical Vapour Deposited Mg-Ti Alloys," in Proceedings of the Third International Magnesium Conference. Manchester, England: Institute of Materials, 1997, pp. 627-636. [128] K. L. Heidersbach, "An Evaluation of the Corrosion Performance of Magnesium-Yttrium and Yttrium-Magnesium Nonequilibrium Alloys," a Ph.D. thesis in Engineering Science and Mechanics, The Pennsylvania State University, University Park, PA. May 1998. [129] J. A. Thornton, "Influence of Substrate Temperature and Deposition Rate on Structure of Thick Sputtered Cu Coatings," Journal of Vacuum Science and Technology, vol. 12, pp. 830-835, 1975. [130] D. L. Smith, Thin-Film Deposition: Principles and Practice. New York: McGraw-Hill, 1995. [131] F. M. d'Heurle, "Metallurgical Topics in Silicon Device Interconnections: Thin Film Stresses," International Materials Reviews, vol. 34, pp. 53-68, 1989. [132] R. C. Wolfe and B. A. Shaw, "The Effect of Thermal Treatment on the Corrosion Properties of Vapor Deposited Magnesium Alloyed with Yttrium, Aluminum, Titanium, and Misch Metal," Journal of Alloys and Compounds, vol. 437, pp. 157-164, 2007. [133] J. A. Thornton, "High Rate Thick Film Growth," Annual Reviews of Materials Science, vol. 7, pp. 239-260, 1977. [134] Y. Bohne, D. M. Seeger, C. Blawert, et al., "Influence of Ion Energy on Properties of Mg Alloy Thin Films Formed by Ion Beam Sputter Deposition," Surface and Coatings Technology, vol. 200, pp. 6527-6532, 2006. [135] S. M. Pursel, J. D. Petrilli, M. W. Horn, et al., "Effect of Alloy Addition and Growth Conditions on the Formation of Mg-Based Bioabsorbable Thin Films," in Nanostructured Thin Films, Proceedings of the SPIE, vol. 7041, G. B. Smith and A. Lakhtakia, Eds., 2008, pp. 704113-1 - 704113-11. [136] S. B. Dodd and R. W. Gardiner, "Vapour Condensation Applied to the Development of Corrosion Resistant Magnesium," in Proceedings of the Third International Magnesium Conference. Manchester, England: Institute of Materials, 1997, pp. 271-284. [137] S. B. Dodd, S. Morris, R. W. Gardiner, et al., "Preliminary Corrosion Evaluation of Some Novel Bulk Electron Beam Evaporated Magnesium Alloys," Corrosion Reviews, vol. 16, pp. 159-174, 1998. [138] T. Mitchell, S. Diplas, and P. Tsakiropoulos, "Characterisation of Corrosion Products Formed on PVD In Situ Mechanically Worked Mg-Ti Alloys," Journal of Alloys and Compounds, vol. 392, pp. 127-141, 2005. [139] C. Blawert, V. Heitmann, W. Dietzel, et al., "Corrosion Properties of Supersaturated Magnesium Alloy Systems," Materials Science Forum, vol. 539-543, pp. 1679-1684, 2007. [140] K. R. Baldwin, S. B. Dodd, and R. W. Gardiner, "The Corrosion Behaviour of Some Vapour-Deposited Magnesium-Titanium Alloys," in Proceedings of the Third International Magnesium Conference. Manchester, England: Institute of Materials, 1997, pp. 727-737. [141] K. R. Baldwin, D. J. Bray, G. D. Howard, et al., "Corrosion Behaviour of Some Vapour Deposited Magnesium Alloys," Materials Science and Technology, vol. 12, pp. 937-943, 1996. [142] P. L. Miller, B. A. Shaw, R. G. Wendt, et al., "Assessing the Corrosion Resistance of Nonequilibrium Magnesium-Yttrium Alloys," Corrosion, vol. 52, pp. 923-931, 1995. [143] D. E. Wolfe, J. Singh, R. A. Miller, et al., "Tailored Microstructure of EB-PVD 8YSZ Thermal Barrier Coatings With Low Thermal Conductivity and High Thermal

145 Reflectivity for Turbine Applications," Surface and Coatings Technology, vol. 190, pp. 132-149, 2005. [144] "Hank's Balanced Salt Solution (HBSS)," Available online http://www.cambrex.com/Content/Documents/Bioscience/HBSS.pdf. [145] A. Oyane, H.-M. Kim, T. Furuya, et al., "Preparation and Assessment of Revised Simulated Body Fluids," Journal of Biomedical Materials Research Part A, vol. 65, pp. 188-195, 2002. [146] "ASTM Standard G1: Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens," ASTM International 2003. [147] J. R. Scully, "The Polarization Resistance Method for Determination of Instantaneous Corrosion Rates," in Electrochemical Techniques in Corrosion Science and Engineering, vol. 18, Corrosion Technology, P. A. Schweitzer, Ed. New York: Marcel Dekker, 2003, pp. 125-150. [148] D. A. Jones, Principles and Prevention of Corrosion. New York: Macmillan, 1992, pp. 147. [149] S. W. Dean, "Calculation of Alloy Equivalent Weight," Materials Performance, vol. 26, pp. 51-52, 1987. [150] D. J. Giard, S. A. Aaronson, G. J. Todaro, et al., "In Vitro Cultivation of Human Tumors: Establishment of Cell Lines Derived From a Series of Solid Tumors," Journal of the National Cancer Institute, vol. 51, pp. 1417-1421, 1973. [151] L. Ramage, A. C. Jones, and C. J. Whelan, "Induction of Apoptosis with Tobacco Smoke and Related Products in A549 Lung Epithelial Cells In Vitro," Journal of Inflammation, vol. 3, 2006. [152] "Periodic Table of Elements: Yttrium," http://environmentalchemistry.com/yogi/periodic/Y.html#Chemical. Accessed April 7, 2009, [153] "Periodic Table of Elements: Titanium," http://environmentalchemistry.com/yogi/periodic/Ti.html#Chemical. Accessed April 7, 2009, [154] J. R. Davis, "Metallographic Technique for Nonferrous Metals and Special-Purpose Alloys," in Metals Handbook, 2 ed. Metals Park, OH: ASM International, 1998, pp. 1399. [155] R. E. Napolitano, "Measurement of ASTM Grain Size Number," http://mse.iastate.edu/solidification/Courses/MatE443- Website/instructions/ASTM_grainsize.pdf. Accessed April 7, 2009, [156] A. J. Carterson, K. H. zu Bentrup, C. M. Ott, et al., "A549 Lung Epithelial Cells Grown as Three-Dimensional Aggregates: Alternative Tissue Culture Model for Pseudomonas aeruginosa Pathogenesis," Infection and Immunity, vol. 73, pp. 1129-1140, 2005. [157] M. Stern and A. L. Geary, "Electrochemical Polarization I. A Theoretical Analysis of the Shape of Polarization Curves," Journal of the Electrochemical Society, vol. 104, pp. 56- 63, 1957. [158] E. McCafferty, "Validation of Corrosion Rates Measured by the Tafel Extrapolation Method," Corrosion Science, vol. 47, pp. 3202-3215, 2005. [159] Z. Nagy and J. M. Wesson, "Error Analysis of the Polarization-Resistance Technique for Corrosion-Rate Measurements," Journal of the Electrochemical Society, vol. 139, pp. 1261-1266, 1992. [160] G. Garcés, M. C. Cristina, M. Torralba, et al., "Texture of Magnesium Alloy Film Growth by Physical Vapour Deposition (PVD)," Journal of Alloys and Compounds, vol. 309, pp. 229-238, 2000.

146 [161] H. B. Yao, Y. Li, and A. T. S. Wee, "Passivity Behavior of Melt-Spun Mg-Y Alloys," Electrochimica Acta, vol. 48, pp. 4197-4204, 2003. [162] A. Leon, E. J. Knystautas, J. Huot, et al., "Magnesium Films for Hydrogen Storage," Materials Science Forum, vol. 377, pp. 85-94, 2001. [163] A. Calcabrini, S. Meschini, M. Marra, et al., "Fine Environmental Particulate Engenders Alterations in Human Lung Epithelial A549 Cells," Environmental Research, vol. 95, pp. 82-91, 2004.

147

Appendix A

Biological and Medical Terminology

Angina – Angina is chest pain associated with decreased blood flow to the myocardium.

Angiography – Angiography is a technique where x-rays are used to visualize the lumen of

blood vessels or organs in the body.

Angioplasty – Angioplasty is a medical procedure where a balloon catheter is inserted into an

occluded blood vessel and inflated to open the vessel.

Apoptosis – A naturally-occurring cellular process that leads to cell death.

Atherosclerosis – Atherosclerosis is the pathological condition in which lipids and calcium

deposit beneath the vascular endothelium

Bare-metal stent (BMS) – A bare-metal stent is a permanent metallic cage that is used to support

an occluded blood vessel after it is opened with an angioplasty procedure.

Coronary artery disease – Coronary artery disease is used to refer to a variety of conditions

related to damaged or diseased coronary arteries. It is usually related to fatty deposits and

plaque buildup.

Cytotoxicity – Cytotoxicity refers to the property of being toxic to cells. Cytotoxic compounds

can induce necrosis, a process by which cells lose membrane integrity and die. They can also

cause apoptosis or a decrease in cell viability.

Drug-eluting stent (DES) – A drug-eluting stent is a bare-metal stent that is coated with a drug-

releasing polymer that is designed to inhibit restenosis.

Embolism – An embolism refers to the medical condition whereby an embolus has migrated via

the circulatory system from one part of the body to occlude a blood vessel in another part of

the body

148 Endothelium – The endothelium is the thin layer of cells that line the lumen of the heart and

blood vessels.

In-stent restenosis (ISR) – In-stent restenosis refers to the condition where stent deployment and

placement causes restenosis in the healing blood vessel.

In Vitro – In vitro literally means “within the glass.” It refers to experiments and tests performed

in simulated physiological environments.

In Vivo – In vivo literally means “within the living.” It refers to experiments and tests performed

in living organisms, that is, in actual physiological environments

Lumen – The lumen is the cavity of a hollow tube or organ.

Minimal lumen diameter (MLD) – The minimal lumen diameter is the value used to

characterize the smallest diameter in the cross-section of an artery.

Myocardial infarction – Myocardial infarction, commonly called a heart attack, refers to the

condition where decreased blood flow results in damage to a region of myocardium.

Myocardial ischemia – Myocardial ischemia refers to the condition where the heart receives an

inadequate supply of blood.

Myocardium – The myocardium, or cardiac muscle, is the muscle that composes the heart.

Necrosis – Premature cell death caused by external factors.

Neointimal hyperplasia – Neointimal hyperplasia is smooth muscle migration and/or

proliferation and extracellular matrix deposition that often occur as a result of angioplasty and

stent procedures.

Radiolucent – Radiolucent refers to a class of materials that cannot be visualized with x-rays.

Remodeling – Remodeling is the process whereby a blood vessel’s lumen increases or decreases

in area during disease and / or healing from an intervening procedure.

Restenosis – Restenosis is the formation of arterial “scar tissue” resulting from angioplasty or

stent procedures. It is often characterized by neointimal hyperplasia and vessel remodeling.

149 Stenosis – Stenosis is an abnormal narrowing of a blood vessel.

Stent – A stent is a slotted metal tube that is used in conjunction with angioplasty to open and

support a diseased blood vessel in order to restore normal blood flow.

Target Lesion Revascularization (TLR) – Target lesion revascularization is the percentage of

patients receiving a stent as a repeat procedure to open a stented vessel occluded by restenotic

tissue.

Thrombosis – Thrombosis is the medical condition where a blood clot forms inside of a blood

vessel.

Thrombus – A thrombus is a blood clot inside of a blood vessel.

150

Appendix B

Corrosion Rates Converted into Penetration Rates

Table B-1: Average linear polarization and EIS corrosion rates of freshly-deposited Mg alloys in 37ºC HBSS.

Linear Polarization EIS Alloy Design. Composition μA μm μA μm Type mpy mpy cm2 day cm2 day - WE43 - 17 1.2 17 22 1.5 22 24 Mg100 67 4.2 60 83* 5.2* 75* Mg 90 Mg100 103 6.5 93 62 3.9 56 35 Mg95 Y5 1217 52.5 754 - - - Mg-Y 33 Mg91 Y9 73 4.9 70 - - - Mg-Ti 86 Mg98 Ti2 59 3.7 53 12 0.7 10 57 Mg92 Y7 Ti1 105 6.9 99 32* 2.1* 30* 87 Mg90 Y8 Ti2 24* 1.6 23 3* 0.2* 2* Mg-Y-Ti 55 Mg89 Y9 Ti2 8 0.5 7 10 0.7 10 (dual gun) 91 Mg86 Y11 Ti3 19 1.4 20 11 0.7 10 89 Mg80 Y16 Ti4 ------T1 Mg98.7 Y1.1 Ti0.2 103 6.5 93 72* 4.5* 65* Mg-Y-Ti T15 Mg97.2 Y2.5 Ti0.3 151 9.7 139 118* 7.6* 109* (Sciaky) T3 Mg96.6 Y3.0 Ti0.4 139 8.9 128 74* 4.7* 68* T16 Mg95.8 Y3.7 Ti0.5 378 24.3 349 206 13.2 190 *Value obtained from one measurement.

151

Table B-2: Average linear polarization and EIS corrosion rates of aged Mg alloys in 37ºC HBSS.

Linear Polarization EIS Alloy Design. Composition μA μm μA μm Type mpy mpy cm2 day cm2 day - WE43 - n/a n/a 24 Mg100 ------Mg 90 Mg100 26 1.7 24 21 1.4 19 35 Mg95 Y5 ------Mg-Y 33 Mg91 Y9 44 3.0 43 26 1.7 25 Mg-Ti 86 Mg98 Ti2 7 0.4 6 3 0.2 2 57 Mg92 Y7 Ti1 ------87 Mg90 Y8 Ti2 10 0.7 9 5 0.3 5 Mg-Y-Ti 55 Mg89 Y9 Ti2 ------(dual gun) 91 Mg86 Y11 Ti3 15 1.0 15 12 0.8 11 89 Mg80 Y16 Ti4 3 0.2 2 3 0.2 3 T1 Mg98.7 Y1.1 Ti0.2 64 4.1 58 32 2.1 29 Mg-Y-Ti T15 Mg97.2 Y2.5 Ti0.3 64 4.1 59 36 2.3 33 (Sciaky) T3 Mg96.6 Y3.0 Ti0.4 29 1.8 26 16 1.1 15 T16 Mg95.8 Y3.7 Ti0.5 170 10.9 156 99 6.4 91